Vitamin K to prevent fractures in older women: a systematic review and economic evaluation

Notes

Permissions

Chapter 1 The aim of the review

The aim of the review was to address the question, ‘What is the clinical and cost-effectiveness of vitamin K in preventing fractures in postmenopausal women at high risk of fracture?’ The cost-effectiveness of vitamin K must be discussed with reference to the costs and clinical effectiveness of other licensed interventions in order to provide advice on the likely position of vitamin K within a treatment algorithm. Vitamin K has been explicitly compared with two bisphosphonates (alendronate and risedronate) and strontium ranelate.

The authors of the review have been involved with several evaluations of the clinical and cost-effectiveness of interventions for preventing fractures. 1–5 In the more recent work,4,5 data on the risk of osteoporotic fracture were provided under an academic-in-confidence agreement; permission to use these data were not granted for this report and therefore a different methodology has been adopted for calculating fracture risk.

Chapter 2 Background

The internationally agreed definition of osteoporosis is a systemic skeletal disease characterised by low bone mass and microarchitectural deterioration of bone tissue, with a consequent increase in bone fragility and susceptibility to fracture. 6

The clinical significance of osteoporosis lies in the fractures that arise; without a fracture a woman suffering from osteoporosis will not suffer morbidity. The most common fractures include vertebral compression fractures and fractures of the distal radius and the proximal femur (hip fracture). In addition, when the skeleton is osteoporotic, fractures occur more commonly at many other sites including the pelvis, proximal humerus, distal femur and ribs. The incidence of fracture is strongly related to age, with a steady increase in incidence as a woman ages. 7

Fractures of the spine often go undetected; it is estimated that only one-third of fractures seen in trials in which morphometric criteria are used to establish the presence of a fracture come to clinical attention. 8 There is a good deal of uncertainty surrounding the impact of undetected ‘morphometric’ fractures on the quality of life of the sufferer, and any cost impacts that such fractures have.

Osteoporotic fractures occurring at the spine and the distal radius are associated with significant morbidity, but the most serious consequences arise in patients with hip fracture, which is associated with an increase in mortality in the year following the hip fracture. 9

It has been estimated that the cost of treating osteoporotic fractures in female postmenopausal patients in the UK in 2000 was approximately £1.5–1.8 billion per annum. 10,11 It has been estimated that these costs will increase to £2.1 billion by 2010. 11 The key components of the costs associated with osteoporotic fractures are hip fractures and subsequent nursing home care that is required for a proportion of these patients.

This report is focused on postmenopausal women because of the deterioration of bone quality following the menopause.

Description of osteoporosis, osteopenia and severe (established) osteoporosis

The definition of osteoporosis based on bone mineral density (BMD) has been developed because BMD can be measured with precision and accuracy, allowing definitive diagnoses of osteoporosis. However, it is acknowledged that other factors such as abnormalities within the skeleton and risk of falls are also important in determining the risks of fracture. Nevertheless, BMD alone forms the basis for the diagnosis of osteoporosis.

The units used in this report for assessing the BMD of a woman will be T-scores and Z-scores. A T-score is defined as the number of standard deviations (SD) from the average BMD of healthy young women. A Z-score is defined as the number of SDs from the average BMD of women of the same age as the patient.

Two thresholds of BMD have been proposed for Caucasian women based on the T-score. 12,13 The first, osteoporosis, denotes a value for BMD that is 2.5 SD or more below the young adult mean value (T-score –2.5 SD or less). The second, osteopenia, denotes a T-score that lies between –1 and –2.5 SD below the young adult mean value. These values refer to the use of dual X-ray absorptiometry at the lumber spine, hip (total hip or femoral neck) and the forearm. Other measurement methods, such as quantitative ultrasound or quantitative computerised tomography, or other sites, such as the calcaneus, do not produce comparable results.

The class of osteoporosis is further divided into patients with severe (or established) osteoporosis, which is defined as a T-score of –2.5 SD or less plus at least one documented fracture. In this report severe osteoporosis will be used to define patients who have a T-score of –2.5 SD or less with a previous fracture. The term osteoporosis will be used to define patients with a T-score of –2.5 SD or less without a previous fracture.

Since the introduction of working definitions of osteoporosis, much attention has focused on their application to epidemiology, clinical trials and patient care. Several problems have emerged, however, largely because of the development of new measurement techniques applied to many different sites. It is now clear that the same T-score derived from different sites and techniques yields different information on fracture risk, even when adjustments are made for age. Thus, the T-score cannot be used interchangeably with different techniques and at different sites.

The site that we have chosen to use is measurement at the femoral neck, as this is the reference site for diagnosis because of the limitations of early dual X-ray absorptiometry machines. 14 Accordingly, the statistical relationships that have been established between increased fracture risk at the hip and Z-score (the T-score of the women minus average T-score for that age and sex) have been undertaken at this site. 15,16

Epidemiological data

The prevalence of osteoporosis by age

Raw data were taken from a UK population-based study by Holt et al. 17 and used to calculate the relationship between T-score and age. The prevalence of osteoporosis within the UK has also been estimated from these data. This data set contained observations on 5713 women aged between 50 and 85 years and used the National Health and Nutritional Evaluation Study III (NHANES III) reference data for women aged 20–29 years.

The percentage of women with a T-score of –2.5 SD or less, as measured at the femoral neck, was recorded. These data are shown in Figure 1; there is a marked increase in the percentage with a T-score of –2.5 SD or less with age. The database taken from the Holt et al. study17 had relatively few women aged between 80 and 84 years (n = 40). The confidence interval around the prevalence at this age is wide (Figure 1). Assuming, however, that the midpoint values are correct, and multiplying these prevalence rates by the respective population of England and Wales,18 it is estimated that there are 0.95 million women suffering with osteoporosis. Assuming that the lower 95% confidence intervals are correct would result in a predicted 0.69 million women with osteoporosis; conversely, assuming that the upper 95% confidence intervals are correct would result in a predicted 1.22 million women with osteoporosis.

FIGURE 1.

The estimated prevalence of female osteoporosis by age band.

The average T-score at the femoral neck at each age band was calculated from the UK population data in the Holt et al. study. 17 A linear relationship was assumed and the T-score was assumed to be 2.0251–(0.0512 × age in years). The assumed average T-score at the midpoint of each age band is given in Table 1. It is seen that, above 85 years of age, the average T-score for women almost reaches the threshold for osteoporosis.

TABLE 1 - Average T-scores for women by age band

Compared with the NHANES III reference data for women aged 20–29 years.

Age (years)	Average UK T-score17a	Z-score at threshold of osteoporosis (T-score of –2.5 SD)a
50–54	–0.66	–1.84
55–59	–0.92	–1.58
60–64	–1.17	–1.33
65–69	–1.43	–1.07
70–74	–1.69	–0.81
75–79	–1.94	–0.56
80–84	–2.20	–0.3
85–89	–2.45	–0.05

Fractures considered to be osteoporotic

Historically, four fracture sites were considered in estimating the cost-effectiveness of interventions to reduce fractures. These were the hip, spine, wrist and proximal humerus. Recent modelling work4 has increased the number of sites included because of evidence that further sites are considered to be related to osteoporosis. 19 The additional fracture types included are fractures of the pelvis, humeral shaft, tibia, fibula, scapula, ribs and sternum and other femoral fractures.

Chapter 3 Clinical effectiveness

Results

Quantity and quality of research available

Number of studies of clinical efficacy identified

The electronic literature searches identified 1078 potentially relevant articles. Of these, 14 articles related to five trials that compared vitamin K with a relevant comparator in postmenopausal women with osteoporosis or osteopenia (Figure 2).

FIGURE 2.

Clinical effectiveness: summary of study selection and exclusion – electronic literature searches.

Number and type of studies included

A total of five individual RCTs met the review inclusion criteria. The various publications relating to these studies are listed in Appendix 3.

Number and type of studies excluded, with reasons

As may be seen, a substantial number of the references identified by the electronic searches related to studies that did not meet the inclusion criteria and were thus excluded as part of the sifting process. Details are therefore given only of those references that were excluded at the full paper stage. These references are listed in Appendix 4 together with the reasons for their exclusion.

Quantity and quality of research available

Phylloquinone (vitamin K₁)

Only one study of phylloquinone met the review’s inclusion criteria. This was the ECKO study,71 a double-blind placebo-controlled trial that used a daily dose of phylloquinone of 5 mg, nearly 60 times the US recommended daily dietary phylloquinone intake of 90 μg/day for adult women. 24 The study population of postmenopausal women with osteopenia but without osteoporosis was predominantly (88%) European Canadian (i.e. Caucasian). The daily dietary calcium and vitamin D intakes of all participants were assessed and supplements were given as required to bring these to 1500 mg and 800 IU respectively. 71

The ECKO trial was originally planned as a 2-year study. However, because of interest in the long-term safety and antifracture efficacy of vitamin K, participants enrolled before 15 March 2004 were subsequently invited to continue in a 2-year extension. In total, 325 participants who had completed the first 24 months were invited to participate in this extension, and 261 (80%) agreed. However, because the extension study terminated before 66% (172/261) of those who had been enrolled had completed it, only 17% (73/440) of the original study participants completed the full 4 years. 71

For further details of study design and reporting quality see Appendix 4.

Menatetrenone (vitamin K₂)

Four trials of menatetrenone met the review’s inclusion criteria: these were the studies by Iwamoto et al. 72 and Shiraki et al. ,73,74 the Osteoporosis Fracture (OF) study75 and the Yamaguchi Osteoporosis Prevention Study (YOPS). 76 All were open-label trials that had as their population postmenopausal women with osteoporosis. All were carried out in Japan and, as none commented on the ethnicity of the study populations, probably all participants, but certainly the majority, were likely to be of Japanese ethnic origin. The trial by Shiraki et al. was originally planned as a 2-year study,73 but subsequently an extension study74 was carried out for 3 years, thus apparently bringing the total study period to 5 years.

Three trials (OF study,75 Shiraki et al. study73,74 and YOPS76) essentially compared menatetrenone with no treatment. Two of these trials stated that calcium was given to both the intervention and control groups. The dose of calcium used in the OF study was not specified. 75 The Shiraki et al. trial originally gave participants a daily dose of 150 mg of elemental calcium,73 although the later extension study74 used a dose of 200 mg/day. The trial by Iwamoto et al. 72 strictly encouraged all participants (apparently including those randomised to calcium) to consume 800 mg of calcium and 400 IU of vitamin D a day in their meals. The YOPS trial76 made no mention of the use of calcium in any of the treatment groups.

In addition, two trials, the Iwamoto trial72 and the YOPS trial,76 compared menatetrenone alone with etidronate alone, and the Iwamoto trial compared menatetrenone alone with calcium alone. The YOPS trial also compared menatetrenone with hormone replacement therapy (HRT), calcitonin and alfacalcidol; these comparisons are not reported here.

All of the trials used a daily dose of 45 mg of menatetrenone (although this was not clear from the published data relating to the OF study, clarification was obtained from Eisai UK77). Although there is no recommended dietary intake for menatetrenone, 45 mg appears to be a high dose.

Menatetrenone is available in the form of Glakay capsules, which contain no more than 175 mg of hydrogenated oil,45 substantially less than the 35 g of fat said to be required for the optimum absorption of menatetrenone. 40 However, none of the trials specified that participants were advised to consume the study medication with a meal containing fat.

Three of the four trials were published as journal articles. However, the extension study of Shiraki et al. 74 was only published in abstract form, whereas the results of the OF study75 were only reported in a press release.

For further details of study design and reporting quality see Appendix 4.

Assessment of effectiveness

Phylloquinone (vitamin K₁)

Phylloquinone: antifracture efficacy

The ECKO trial71 reported only clinical fractures. These included all fragility and non-fragility fractures except those of the fingers and toes. Phylloquinone was associated with a significant reduction in the risk of such clinical fractures (Table 4). As the trial was carried out in women with osteopenia, not osteoporosis, relatively few fractures were reported.

TABLE 4 - Phylloquinone in postmenopausal osteopenia: all clinical fractures

Study	Dose	Numbers in each group suffering clinical fractures
ECKO trial 200871	5 mg/day	All clinical fractures: Phylloquinone: 9/217 (4.2%) Placebo: 20/223 (9.0%) RR: 0.46 (95% CI 0.22 to 0.99) Clinical fragility fractures: Phylloquinone: 4/217 (1.8%) Placebo: 11/223 (4.5%) RR: 0.41 (95% CI 0.13 to 1.29)

The study was not powered to identify a significant reduction in the risk of clinical fragility fractures (defined as low trauma fractures such as those sustained by falling from standing height).

Phylloquinone: adverse effects

In the ECKO trial,71 no serious adverse events were attributed to phylloquinone therapy. Indeed, it was suggested that phylloquinone might have anticancer effects as only three women in the phylloquinone group developed cancer compared with 12 in the placebo group (RR 0.26, 95% CI 0.07 to 0.90). The incidence of nausea and vomiting was similar in the intervention and placebo groups (5.1% versus 4.5%, p = 0.77).

The MEDLINE search identified no large, long-term studies of the safety of phylloquinone therapy in the treatment of osteoporosis, and no postmarketing surveillance data could be identified. The Expert Group on Vitamins and Minerals35 noted no toxicity related to oral vitamin K₁, although a review by Vermeer et al. 26 referred to unpublished preliminary studies which suggested that vitamin K₁ supplementation in doses higher than 1 mg/day might contribute to periodontal disease.

Phylloquinone: health-related quality of life

In the ECKO trial, health-related quality of life was not significantly different between the intervention and placebo groups.

Phylloquinone: continuance and compliance

In the ECKO trial, 91.2% of participants randomised to phylloquinone completed the first 2 years of the study, compared with 90.6% in the placebo group, suggesting relatively high continuance with study medication in both groups. However, 25% of eligible women in the phylloquinone group who were invited to participate in the study extension refused to continue, compared with 15% in the placebo group. 71 The reasons for their refusal were not reported.

The ECKO trial assessed compliance by counting leftover pills at follow-up visits; participants who took 80% or more of the study medication were considered compliant. 71 Compliance at different time periods is presented in Table 5 and, as may be seen, is higher in the placebo group. The difference is particularly marked at 4 years, even though participants had been given the opportunity to refuse to participate in the study extension, and refusal was substantially higher in the phylloquinone group than in the placebo group.

TABLE 5 - Compliance with phylloquinone (data from the ECKO trial71)

Time point	Phylloquinone	Placebo
2 years	82.5%	83.9%
3 years	84.9%	89.9%
4 years	82.4%	93.3%

Menatetrenone (vitamin K₂)

Menatetrenone: antifracture efficacy

Vertebral fracture

All four trials reported vertebral fracture data (Table 6). However, the extension study of Shiraki et al. 74 reported the number of fractures rather than the number of women suffering those fractures and thus the RR of fracture for that period could not be calculated.

TABLE 6 - Menatetrenone in postmenopausal osteoporosis or osteopenia: vertebral fracture data

Study	Dose	Fracture definition	Number in each group suffering vertebral fracture
Iwamoto 200172	45 mg/day	A decrease of at least 20% in any vertical height ratio; or central/anterior or central/posterior height less than 0.8; or anterior/posterior height less than 0.75	Menatetrenone: 2/23 (8.7%) Etidronate: 2/25 (8.0%) Calcium: 6/24 (25.0%) RR menatetrenone vs etidronate: 1.09 (95% CI 0.17 to 7.10) RR menatetrenone vs calcium: 0.35 (95% CI 0.08 to 1.55)
OF study 200575	45 mg/day77	Not specified	Menatetrenone + calcium: 227/1619 (14.0%) Calcium: 223/1638 (13.6%) RR: 1.03 (95% CI 0.87 to 1.22)
Shiraki 2000,73 200274	45 mg/day	2000 data: semiquantitative; ≥ 20% decline in any of the three vertebral heights compared with baseline 2002 data: clinical fractures only	2000: Menatetrenone + calcium: 13/91 (14.2%) Calcium: 30/99 (30.3%) RR: 0.47 (95% CI 0.26, 0.85) 2002: Number of fractures: Menatetrenone + calcium: 33 Calcium: 54
YOPS 200476	45 mg/day	A decrease of at least 20% in one of the ratios of vertebral height in intact vertebrae, or a decrease of at least 4 mm in vertebrae fractured at baseline; also semiquantitative assessment	Menatetrenone: 9/66 (13.6%) Etidronate: 8/66 (12.1%) No treatment: 17/66 (25.8%) RR menatetrenone vs etidronate: 1.13 (95% CI 0.46 to 2.74) RR menatetrenone vs no treatment: 0.53 (95% CI 0.25 to 1.10)

The two trials that compared menatetrenone with cyclical etidronate found no difference between the two interventions in relation to vertebral fracture risk (Figure 3). However, it should be noted that both studies would have been substantially underpowered to demonstrate a difference in efficacy between two active interventions.

FIGURE 3.

Menatetrenone vs etidronate: vertebral fracture.

Of the trials that compared menatetrenone, with or without calcium, with no additional active treatment, only one, that by Shiraki et al. ,73 reached statistical significance. In this trial, menatetrenone plus calcium was associated with a reduction in the risk of radiographic fractures relative to calcium alone. In two other trials, Iwamoto et al. 72 and YOPS,76 the point estimates also favoured menatetrenone but neither trial was large enough to achieve statistical significance. However, interim analyses from the substantially larger OF study75 found no difference in the risk of vertebral fracture between women randomised to menatetrenone plus calcium and those randomised to calcium alone (Table 6).

Meta-analysis of data from all four trials suggests that menatetrenone is not associated with a significant reduction in the risk of vertebral fracture compared with no active treatment (Figure 4). However, this meta-analysis demonstrates substantial statistical heterogeneity, as indicated by the chi-squared test for heterogeneity, with its very low p-value of 0.01, and the moderately high I² statistic of 71.4%, which measures the percentage of variation that is due to heterogeneity rather than chance. 78 Meta-analysis of data from the three earlier trials indicates that this heterogeneity is due to the inclusion of the OF study (Figure 5); without data from this trial menatetrenone appears to be associated with a statistically significant reduction in the risk of radiographic vertebral fracture. This highlights the extent to which the results of the OF study differ from those of the other studies, prompting exploration of the possible reasons for this difference and consideration of which results are more likely to represent the true antifracture efficacy of menatetrenone. Such an exploration can only be conjectural because of deficiencies in the published data relating to all four trials.

FIGURE 4.

Menatetrenone vs calcium or no additional active treatment: vertebral fracture risk – meta-analysis using fixed-effects model.

FIGURE 5.

Menatetrenone: vs calcium or no additional active treatment, excluding the OF study:75 vertebral fracture risk.

The OF study appears to differ from the other three trials in terms of its population (Table 7). All four trials took as an inclusion criterion the presence of osteoporosis diagnosed using the Japanese diagnostic criteria (see Appendix 5, Table 31). Despite this, the OF study appears to have recruited participants at substantially lower risk of vertebral fracture than did the smaller studies; the fracture rate in the control group of the OF study, at under 14%, is less than half the mean fracture rate (28%) seen in the control groups of the smaller trials (Table 6). Because the OF study has not published data relating to the baseline characteristics of its population, it is not possible to determine whether it recruited a lower proportion of patients with pre-existing fractures than did the smaller trials. However, it seems unlikely that the difference in the OF study results can be attributed to hypothetical differences in the study populations; data from the FIT trial fracture79 and non-fracture80 arms indicate that an intervention with antifracture efficacy will result in a very similar RR of vertebral fracture in populations at higher and lower risk of fracture.

TABLE 7 - Menatetrenone: population

Study	Percentage of study population with prevalent vertebral fractures at study entry		Percentage of group receiving no active treatment suffering incident vertebral fractures	Definition of incident vertebral fracture
	Menatetrenone	No active treatment
Iwamoto 200172	30.4	29.2	25.0	Decrease of ≥ 20% in any vertical height ratio
Shiraki 200073	38.7	35.8	30.3	20% decline in any of the three vertebral heights
YOPS 200476	30.3	30.3	25.8	Decrease of ≥ 20% in any vertical height ratio
OF study 200575	Not specified	Not specified	13.6	Not specified

Alternatively, the lower fracture rate seen in the control group of the OF study may be due not to a difference in the study populations but to the use of a different definition of vertebral fracture. Whereas the other trials stated that they reported radiographic fractures, the OF study did not provide a fracture definition; it is therefore possible that only clinical fractures have been reported. However, this seems unlikely as the OF study describes its primary end point as ‘new incidence of vertebral fracture’, in contrast to its secondary end point, ‘new incidence of clinical fracture’ of any sort, including vertebral fracture. 75 Moreover, although the reporting of only clinical vertebral fractures would have resulted in lower absolute fracture rates, evidence from the FIT trial50 suggests that it should not have had a noticeable effect on the RR of fracture.

Because the results of the smaller trials are systematically different from those of the larger OF study, it is possible that the difference in antifracture efficacy may relate to differences in methodological quality, most probably relating to the methods of randomisation and allocation concealment. All four trials were poorly reported, making it impossible to exclude the possibility of low methodological quality, but it is perhaps more likely that the three smaller trials were of lower quality than the larger one.

Finally, the heterogeneity between the trials may derive from publication bias. Small trials are particularly vulnerable to the play of chance, and it is impossible to exclude the possibility that, in addition to the published studies that favour menatetrenone, there may be a number of small unpublished trials whose results are either neutral or unfavourable to menatetrenone. This conjecture is strengthened by the fact that the considerably larger OF study, which yielded non-significant results, has not been published in journal form. The number of studies is too small to allow publication bias to be assessed using a funnel plot.

Thus, there seems no reason why the OF study should be excluded from the meta-analysis on the basis of clinical heterogeneity and, indeed, it could be argued that its results should be prioritised above those of the smaller trials as being less vulnerable to the play of chance. However, if data from all four trials are combined in a meta-analysis, conservatively using the random-effects model, which gives considerably less weight to the larger OF study than does the fixed-effects model, menatetrenone is still not associated with a statistically significant reduction in vertebral fracture risk (RR 0.63, 95% CI 0.36 to 1.11; Figure 6).

FIGURE 6.

Menatetrenone vs calcium or no additional active treatment: vertebral fracture risk – meta-analysis using random-effects model.

Vertebral fracture: post-hoc subgroup analyses

Post hoc subgroup analysis undertaken by the OF study analysts75 found that menatetrenone was associated with a statistically significant reduction in vertebral fracture in women with five or more fractures at baseline (Figure 7). This result should be treated with caution because post hoc subgroup analyses do not represent true randomised comparisons. Moreover, in this particular instance, the subgroup was extremely small, representing only 4% of those originally enrolled in the study.

FIGURE 7.

Menatetrenone vs no additional active treatment in women with at least five vertebral fractures at study entry: vertebral fracture.

The OF study analysts75 also claimed that menatetrenone was associated with less height loss than no treatment in women aged 75 years and older, women more than 30 years past the menopause and women who had at least five vertebral fractures at study entry. These claims should again be treated with caution because they derive from post hoc subgroup analyses. Moreover, the underlying data were not presented, making it impossible to assess either the proportion of participants included in the first of the two subgroups or the magnitude of the treatment effect.

Non-vertebral fracture

Three trials reported non-vertebral fracture data (Table 8 and Figure 8); however, none was powered to identify a statistically significant difference in the incidence of non-vertebral fracture. In 2002, Shiraki reported only the number of fractures,74 not the number of participants suffering a fracture, and therefore it was not possible to calculate a RR. The OF study took as its secondary outcome measure the incidence of all new clinical fractures of the upper forelimb, femur, radius and vertebrae81 but stated that these data would not be analysed until after completion of the 12-month observational follow-up period. 75 Although almost 3 years have elapsed since Eisai announced the intermediate results of the OF study, the clinical fracture data do not appear to have been released, raising the suspicion that menatetrenone did not reduce the risk of fracture.

TABLE 8 - Menatetrenone in postmenopausal osteoporosis or osteopenia: all non-vertebral fractures

There were said to have been three fractures among women in this group and it is therefore possible that one woman may have suffered more than one fracture.

Study	Dose	Number in each group suffering non-vertebral fracture
Iwamoto 200172	45 mg/day	Menatetrenone: 0/23 Etidronate: 0/25 Calcium: 0/24 RR menatetrenone vs etidronate: not estimable RR menatetrenone vs calcium: not estimable
Shiraki 2000,73 200274	45 mg/day	2000: Menatetrenone + calcium: 1/91 (1.1%) Calcium: 5/99 (5.1%) RR: 0.22 (95% CI 0.03 to 1.83) 2002: Number of long bone fractures: Menatetrenone: 4 Control: 10 RR not calculable as number of patients with fracture not reported
YOPS 200476	45 mg/day	Menatetrenone: 0/66 Etidronate: 1/66 (1.5%) No treatment: 3a/66 (4.5%) RR menatetrenone vs etidronate: 0.33 (95% CI 0.01 to 8.04) RR menatetrenone vs no treatment: 0.14 (95% CI 0.01 to 2.71)

FIGURE 8.

Menatetrenone vs calcium or no additional active treatment: non-vertebral fracture risk.

Hip, wrist and other non-vertebral fractures

Three trials reported hip fracture data; however, none was big enough to yield a statistically significant result (Table 9), nor did their combined results achieve significance (Figure 9).

TABLE 9 - Menatetrenone in postmenopausal osteoporosis or osteopenia: hip fracture data

Study	Dose	Number of women in each group suffering hip fracture
Iwamoto 200172	45 mg/day	Menatetrenone: 0/23 Etidronate: 0/25 Calcium: 0/24 RR menatetrenone vs etidronate: not estimable RR menatetrenone vs calcium: not estimable
Shiraki 200073	45 mg/day	Menatetrenone + calcium: 0/91 Calcium: 2/99 RR: 0.22 (95% CI 0.01 to 4.47)
YOPS 200476	45 mg/day	Menatetrenone: 0/66 Etidronate: 0/66 No treatment: 1/66 RR menatetrenone vs etidronate: not estimable RR menatetrenone vs no treatment: 0.33 (95% CI 0.01 to 8.04)

FIGURE 9.

Menatetrenone vs calcium or no additional active treatment: hip fracture risk.

Three trials reported wrist or forearm fractures (Table 10). None reported fractures of the humerus. Again, none of the trials was big enough to yield a statistically significant result, nor did their combined results achieve significance (Figure 10).

TABLE 10 - Menatetrenone in postmenopausal osteoporosis or osteopenia: wrist or forearm fracture data

It is possible that both of these fractures occurred in the same woman.

Study	Dose	Number of women in each group suffering wrist or forearm fracture
Iwamoto 200172	45 mg/day	Menatetrenone: 0/23 Etidronate: 0/25 Calcium: 0/24 RR menatetrenone vs etidronate: not calculable RR menatetrenone vs calcium: not calculable
Shiraki 200073	45 mg/day	Menatetrenone + calcium: 1/91 Calcium: 2/99 RR: 0.54 (95% CI 0.05 to 5.90)
YOPS 200476	45 mg/day	Menatetrenone: 0/66 Etidronate: 1/66 No treatment: 2a/66 RR menatetrenone vs etidronate: 0.33 (95% CI 0.01 to 8.21) RR menatetrenone vs no treatment: 0.20 (95% CI 0.01 to 4.09)

FIGURE 10.

Menatetrenone vs calcium or no additional active treatment: wrist or forearm fracture risk.

Menatetrenone: adverse effects

In the included trials of menatetrenone, reporting of adverse events was generally poor. The OF study75 noted a significantly higher incidence of skin and skin appendage lesions in patients receiving menatetrenone (0.5 per 100 patient-years compared with 0.1 in the control group, p < 0.001).

Although the MEDLINE adverse events search identified no large, long-term studies of the safety of menatetrenone therapy in the treatment of postmenopausal osteoporosis or osteopenia, it identified two relatively small studies82,83 of the effect of menatetrenone therapy on haemostatic activity in patients with osteoporosis or osteopenia. These suggested that menatetrenone, at a dose of 45 mg/day, did not induce a thrombotic tendency in such patients.

The Expert Group on Vitamins and Minerals35 noted no toxicity related to oral vitamin K₂, and the review by Vermeer et al. 26 stated that menatetrenone had been used in Japan on a large scale and as yet no adverse side effects had been reported. However, the product leaflet for Glakay capsules45 stated that adverse reactions were reported in 81 of 1885 patients (4.3%) taking the product (Table 11).

TABLE 11 - Glakay capsules (menatetrenone): adverse reactions45

ALT, alanine aminotransferase; AST, aspartate aminotransferase; BUN, blood urea nitrogen; GOT, glutamic oxaloacetic transaminase; GPT, glutamic pyruvic transaminase.

	≥ 0.1% to 5%	< 0.1%	Incidence unknown
Gastrointestinal	Stomach discomfort, abdominal pain, nausea, diarrhoea, dyspepsia	Thirst, anorexia, glossitis, constipation	Vomiting, stomatitis
Hypersensitivity	Rash, pruritus, redness
Psychoneurological	Headache	Light-headedness, numbness	Dizziness
Cardiovascular		Increase in blood pressure	Palpitations
Hepatic	Elevation of AST (GOT), ALT (GPT) and γ-GPT, etc
Urinary	Elevation of BUN, etc		Urinary frequency
Other	Oedema, eye abnormalities		Malaise, arthralgia

Menatetrenone: health-related quality of life

Only one study of menatetrenone reported health-related quality of life. The OF study75 claimed that, in the first 12 months of the study, walking, and the degree and duration of back pain at rest, improved more in women who received menatetrenone and calcium than in those who received calcium alone. However, as the data underlying this claim were not presented, the improvements cannot be quantified. Moreover, it is not clear whether such data were only collected for the first 12 months of the study or whether they were collected for the full 36 months but only reported for the first 12 months; the latter would suggest that no difference between the groups in relation to these factors was seen following the first 12 months.

Menatetrenone: continuance and compliance

None of the trials of menatetrenone provided information on compliance with study medication. However, two trials provided information on the proportion of participants who completed follow-up for the planned length of the study (Table 12).

TABLE 12 - Proportion of participants completing the study protocol

Study	Duration of study	Proportion of participants completing study protocol
Shiraki 200073	2 years (data not available for extension study)	Menatetrenone + calcium: 92% Calcium: 87%
YOPS 200476	2 years	Menatetrenone: 95% Etidronate: 94% No treatment: 91%

Chapter 4 Economic analysis

The assessment group undertook a systematic literature review to identify any economic evaluations of vitamin K. No relevant papers were found and thus, to our knowledge, this report is the first assessment of the cost-effectiveness of vitamin K. We assess the cost-effectiveness of vitamin K compared with no treatment, two bisphosphonates (alendronate and risedronate) and strontium ranelate, at combinations of age and T-score, assuming that a BMD scan has already been undertaken and thus the T-score is known without cost. Previous modelling work4 has estimated the likely combinations of age and clinical risk factors at which BMD scanning would be cost-effective; this modelling work was updated following a fall in the price of alendronate, the intervention that appeared most cost-effective, from £301 to £54 per annum,5 and further updating of this work is beyond the remit of this project. It is assumed that all women in the model have an adequate baseline intake of calcium and vitamin D as RCT data on the effectiveness of interventions have been compared against such a population. It is noted that some of the parameters used in previous modelling work5 may have become outdated. These have been updated when appropriate in producing the results for this report. Any changes in parameters between the previously published work using an alendronate price of £54 per annum and this report are detailed within the text and are additionally summarised later in this chapter (see Summarising changes within the parameters used in this report and in preceding work).

Methods for economic analyses

The assessment group has constructed a peer-reviewed model to estimate the cost-effectiveness of osteoporosis interventions. 88 The most recent work undertaken4,5 used academic-in-confidence data that were used in the development of the FRAX tool. 89 Permission to re-use the academic-in-confidence data for this project was not granted. As such, an alternative model for predicting the risks of fracture was developed based on age, BMD and previous fracture history. The results produced by the model were compared, and recalibrated when appropriate, with the data provided in the recent work5 to try and ensure that the results were relatively consistent.

The key inputs to the model are the efficacy data for each intervention in terms of reducing the incidence of hip, vertebral, wrist and proximal humerus fractures. As detailed in Chapter 2, other fracture types are subsumed into these groups, but for reasons of brevity we will refer only to the four main fracture sites.

The model calculates the number of fractures that occur and provides as output data the costs associated with osteoporotic fractures and the quality-adjusted life-years (QALYs) accrued by a cohort of 100 osteoporotic women, with each fracture being detrimental to health and incurring a cost. When the costs of the intervention are included, the incremental cost compared with no treatment can be calculated and divided by the gain in QALYs to calculate a cost per QALY ratio.

The cost per QALY of vitamin K treatment has been evaluated against a no treatment option to provide information on whether vitamin K can be given cost-effectively in the absence of other osteoporosis interventions. Incremental analyses against alendronate, risedronate and strontium ranelate have also been conducted to provide information on the likely placement of vitamin K within a treatment algorithm.

As the results were being calculated it was apparent that vitamin K₁ was potentially a cost-effective treatment for the prevention of osteoporotic fractures. The expected net benefit of sampling (ENBS) of conducting an RCT that compared alendronate and vitamin K₁ was estimated. Such analysis will provide data on whether such a trial would be an efficient use of resources and, if so, on the number of patients that would need to be recruited to maximise the ENBS.

This section is divided into the following subsections:

• the structure of the model, which will discuss the formulation of the appraisal model and the modelling assumptions made

• the costs associated with each event contained within the model

• the utility multipliers associated with each event contained within the model

• comparison of the results produced using the new model with those produced by previous work

• changes in the parameters used for this evaluation compared with those used in previous evaluations

• methodology for calculating the ENBS of an RCT comparing vitamin K with alendronate.

The structure of the cost-effectiveness model

The individual patient model

The model used to calculate cost-effectiveness ratios is an updated version of the previously reported Sheffield Health Economic Model for Osteoporosis (SHEMO). 4,88,90 This model deviated from approaches previously used, which have been based on cohort analyses using the standard techniques of decision analysis and state transition models. 91,92

The basic design of SHEMO is similar in many ways to the conventional state transition models used in the area of osteoporosis, in which women pass through states using a set of time-dependent transition probabilities and each state has its associated costs, mortality rates and health state utility values. However, it differs in a crucial respect to the conventional cohort design as individual women pass through the model one at a time. The model simulates for each patient whether or not an event occurs in the forthcoming year that would impact on the costs and utility associated with the woman.

The full patient history is recorded and factors such as previous fractures and current residential status can therefore be used to determine the likelihood of events in the next time period. Following the simulated event, the quality of life of the patient and costs incurred in that time period are calculated. These values have taken into account any residual costs or quality of life impacts from previous fractures. The model simulates at 1-year intervals until either the patient dies or a user-defined time horizon is reached. This process is repeated until the chosen number of women has been simulated. The rationale for using the individual patient approach is that it provides more accuracy and flexibility than a cohort approach, which is bounded by a limited number of transition states. 4,90 A diagram of the model structure is provided in Figure 11.

FIGURE 11.

The structure of the individual patient model. 90

The exact values of p2: p14 will be determined by patient age, patient history regarding the presence of previous fracture at each site, and the residential status of the patient. These probabilities are calculated for each individual at the beginning of each year. The cycle is repeated for all non-absorbing states until the time horizon is reached.

The basic probabilities for moving from transition state to transition state have been taken from epidemiological data, as described later in this chapter. Once a fracture is sustained within the model the risk is increased in accordance with the data reported by Klotzbuecher et al. ,93 as described later in this chapter.

As a patient moves into a transition state there is an initial one-off cost incurred and an ongoing cost that is assumed to last until the end of the simulation. By using such a methodology, states with high ongoing costs can be distinguished from those in which the costs incurred are all in the initial year. In circumstances in which a patient has already been in the state being entered, it has been assumed that only the one-off costs will be incurred, with the ongoing costs from that state remaining constant. For example, if the consequences of a vertebral fracture comprised an initial cost of £600 and a recurrent cost of £300 per year, a further vertebral fracture in the same individual would cost a further £600 but the recurrent costs would not increase from £300 per year. This may underestimate the costs involved, but few data were found on the additional ongoing costs of second events. Following the introduction of additional fracture sites, the methodology of not duplicating the long-term fracture costs may be slightly unfavourable to the intervention. As a tibia fracture is now grouped with a proximal humerus fracture, if both fractures had been sustained then only one long-term cost would be included.

When a patient moves into a transition state this affects their quality of life. It has been assumed that there will be a QALY multiplier effect within the first year and a separate QALY multiplier that will apply for the remaining years of the simulation. By using this methodology, states from which the patient will recover but not to the level prior to the event can be modelled. It is assumed that, when a patient suffers a transition state for a second or more time, only the first year reduction in quality of life will be taken into consideration, using similar logic to that employed for costs. It is noted that in some cases this will underestimate the loss in QALYs, for example second hip or wrist fractures on a different side to the first, or a second vertebral fracture. However, because of insufficient data the approach of assuming no extra residual QALY loss from a second incident was taken. As in the explanation given when discussing costs, the inclusion of more than one fracture in some states may be slightly unfavourable to the intervention.

It has been assumed that, for a year in which death occurs, the QALYs gained are half those for the previous year, costs incurred are equal to half of the ongoing annual costs, and only half of the drug acquisition cost is paid.

Having established a baseline ‘no treatment’ cost for the cohort the incremental effects from pharmaceutical treatments have been calculated. The efficacy of each treatment is modelled by the use of RRs in entering a transition state. It is expected that a cohort using a treatment with a RR of 0.5 for hip fracture would, in the next time period, have half the number of hip fractures as the same cohort receiving no treatment (RR 1) assuming an equal death rate. For each intervention the RRs were sampled from the relevant meta-analysis of efficacy undertaken.

The effect of treatment on fracture probability was assumed to be instantaneous and to persist unchanged throughout the treatment period. A 5-year treatment period was assumed, which corresponds to the duration of exposure in RCTs, particularly those undertaken in the past 10 years. In addition to the treatment RR, the model incorporates fall times, which have been defined as the time from when the treatment is stopped to the time that the RR returns to 1 compared with no treatment. It is assumed that the RR returns to 1 in a linear manner during a fall time period of 5 years.

The time horizon of the individual patient model was constrained to a 10-year period because of the likely treatment effects being confined within this period, the uncertainty around future medical costs and technologies that may become available, and the gap in the evidence base concerning the effect of fractures on quality of life after a period of 10 years.

Diseases for which possible links with osteoporosis treatments may exist, such as Alzheimer’s disease, venous thrombolic events and cancer, were excluded from this cost-effectiveness analysis.

The construction of a meta-model

Gaussian process modelling was used to transform the results of the individual patient model into a meta-model that allowed instantaneous calculation of the incremental costs and QALYs associated with an input parameter configuration. 88 The advantage of the Gaussian process technique is that given the same starting assumptions the results for a new drug with defined RRs can be instantly calculated with the benefits associated with an individual patient methodology retained.

Subsequent adjustments to the results produced by the meta-model

Limiting the time horizon of the model to 10 years will underestimate the effect of mortality, particularly in younger women, which has been shown to markedly impact on the cost-effectiveness ratio. 1–5 The benefits of reduced mortality beyond the initial 10-year time horizon have been estimated. The methodology for estimating the QALYs gained through prevented mortality is explained in Appendix 6.

The discount rates used for the individual patient model2,4 were 6% per annum for costs and 1.5% per annum for utility in accordance with the prevailing NICE rates at the commissioning of the initial project. In the interim period the discount rates have been set to 3.5% per annum for both costs and utility94 and therefore the results need to be adjusted to take the discount rate change into consideration. The summated discounted value of 10 single units over 10 years is 7.36, 8.32 and 9.22 when using discount rates of 6%, 3.5% and 1.5% respectively. These values were used to adjust the model outputs, with QALYs for each scenario multiplied by 0.90 (8.32/9.22) and costs associated with fracture multiplied by 1.13 (8.32/7.36). Intervention costs (drug costs, GP visits and BMD scans) that were assumed to be incurred only in the first 5 years were multiplied by 1.07 to reflect the shorter time horizon of such costs, using the discounted values of five single units over 5 years. Although the methodology used for altering the discount rates is likely to introduce some inaccuracies, as the costs and QALYs will not be equally spread across the 10-year period (as the intervention is used only in the initial 5-year period), it is a pragmatic solution and is not expected to markedly change the conclusions.

Each treatment option has also been assigned GP costs in addition to drug acquisition costs. Following (NICE Osteoporosis) Guideline Development Group (GDG) advice, and considering that elderly women have their complete medication (for all diseases reviewed) annually, it was assumed that, following initiation, osteoporosis treatment would result in no additional costs for women aged 75 years or over, and would result in one-third of women below 75 years of age requiring an additional GP appointment per annum. It was also assumed that no follow-up BMD scans would be required.

Compliance has been modelled at 50% in line with published estimates. 95 Women who are non-compliant are assumed to incur the costs of 3 months of treatment whilst receiving no benefit.

The individual patient model assumed a 5-year treatment period and a residual benefit that declined linearly to zero over a 5-year period. For vitamin K, which has a short duration in the body, a 5-year residual benefit was not deemed appropriate and it was assumed that any protective benefit would cease upon discontinuation of treatment. Analyses of the results from the individual patient model were undertaken to determine the effect of a decrease in residual benefit from 5 years to 0 years. Multiplicative factors were estimated that would be applied to the incremental costs (excluding intervention costs) and incremental QALYs associated with 5 years of residual benefit. For women with a T-score of –2.5 SD, these factors ranged from 0.73 for costs and 0.67 for QALYs at 50 years of age to corresponding values of 0.78 and 0.75 at 75 years of age.

The incidence of hip, vertebral, wrist and proximal humerus fractures by age

Data on the incidence of hip, vertebral, wrist and proximal humerus fractures were taken from a large-scale Scottish study. 7 Exponential distributions have been applied to these distributions to smooth the data. The distributions are shown in Figures 12–15; all R-squared values are greater than 0.90, showing that these are good fits. These distributions are also plausible as the rates do not decrease as a woman ages. The summarised risks are provided in Table 13 and Figure 16.

FIGURE 12.

The goodness of fit when replacing the raw data for hip fracture7 with a statistical distribution.

FIGURE 13.

The goodness of fit when replacing the raw data for vertebral fracture7 with a statistical distribution.

FIGURE 14.

The goodness of fit when replacing the raw data for wrist fracture7 with a statistical distribution.

FIGURE 15.

The goodness of fit when replacing the raw data for proximal humerus fracture7 with a statistical distribution.

TABLE 13 - The female population risk of hip, vertebral, proximal humerus and wrist fractures estimated from the smoothed Singer et al. data7

Age (years)	Probability of hip fracture	Probability of vertebral fracture	Probability of proximal humerus fracture	Probability of wrist fracture
50–54	0.03%	0.09%	0.31%	0.07%
55–59	0.06%	0.13%	0.36%	0.09%
60–64	0.11%	0.19%	0.43%	0.12%
65–69	0.20%	0.28%	0.51%	0.15%
70–74	0.38%	0.40%	0.61%	0.20%
75–79	0.73%	0.59%	0.72%	0.26%
80–85	1.38%	0.85%	0.86%	0.35%
85–89	2.62%	1.23%	1.02%	0.46%

FIGURE 16.

The annual female population risk of hip, vertebral, proximal humerus and wrist fractures estimated from the smoothed Singer et al. data. 7

The incidence of fractures other than hip, vertebral, wrist and proximal humerus fractures

As detailed in Stevenson et al. ,4 other fracture sites considered were incorporated into the four main fracture types in order to use the previously calculated meta-model. 88 The earlier modelling work4 used the raw data without adjustment to estimate the increase in the expected numbers of fractures at each site. For this report it was decided to smooth these data by fitting statistical distributions, which were used to estimate the increased fracture risk. The goodness of each fit between a statistical relationship and the raw data is shown in Figures 17–19. Whereas the increase against age was presumed to be linear for hip and proximal humerus fracture, the best fit for increasing the number of wrist fractures was seen to be parabolic. The estimated increase in fracture risk at each age is given in Table 14. This would adjust the average risk of fracture for the female population in the UK at each age to those given in Table 15 and Figure 20.

FIGURE 17.

The goodness of fit when replacing the increased incidence of hip fracture type as detailed in Stevenson et al. 4 with a statistical distribution.

FIGURE 18.

The goodness of fit when replacing the increased incidence of wrist fracture type as detailed in Stevenson et al. 4 with a statistical distribution.

FIGURE 19.

The goodness of fit when replacing the increased incidence of proximal humerus fracture type as detailed in Stevenson et al. 4 with a statistical distribution.

TABLE 14 - The increase in incidence of hip, wrist and proximal humerus fractures to incorporate fractures at other sites, as used in the model

Note that the incidence of vertebral fractures was not increased.

Age (years)	Increase in hip fracture incidence to incorporate pelvis and other femoral fractures	Increase in proximal humerus fracture incidence to incorporate tibia and fibula fractures	Increase in wrist fracture incidence to incorporate rib, sternum, clavicle and scapula fractures
50–54	26%	87%	63%
55–59	25%	75%	33%
60–64	23%	63%	17%
65–69	22%	51%	14%
70–74	20%	39%	24%
75–79	19%	27%	48%
80+	17%	15%	86%

TABLE 15 - The female population risk of hip, vertebral, proximal humerus and wrist fractures estimated from the smoothed Singer et al. data7 and incorporating other osteoporotic fractures

Age (years)	Probability of hip fracture	Probability of vertebral fracture	Probability of proximal humerus fracture	Probability of wrist fracture
50–54	0.04%	0.09%	0.50%	0.12%
55–59	0.07%	0.13%	0.48%	0.15%
60–64	0.13%	0.19%	0.50%	0.19%
65–69	0.25%	0.28%	0.58%	0.23%
70–74	0.46%	0.40%	0.75%	0.28%
75–79	0.87%	0.59%	1.07%	0.34%
80–85	1.62%	0.85%	1.59%	0.40%

FIGURE 20.

The average annual female population risk of hip, vertebral, proximal humerus and wrist fractures estimated from the smoothed Singer et al. data7 and incorporating other osteoporotic fractures. ph, proximal humerus; vert, vertebral.

Fracture risk at the threshold for osteoporosis

Table 20 and Figure 21 give the estimated fracture risk at each site by age for women at the threshold of osteoporosis (T-score –2.5 SD). No data on the fracture risks for patients with severe osteoporosis have been given, as the risks would be dependent upon the site of the previous fracture, as detailed in Table 6. As the population age increases, the risk at the threshold for osteoporosis may become lower than that of the average population (compare Tables 19 and 20). This is due both to the large proportion of women with severe osteoporosis and to the small differential between the average population BMD and the T-score of –2.5.

TABLE 20 - The estimated annual fracture risk by age for a woman with a T-score of –2.5 and no previous fracture

Age band (years)	Fracture site (including fractures grouped with the main site)
	Hip	Vertebral	Wrist	Proximal humerus
50–54	0.26%	0.25%	0.88%	0.27%
55–59	0.29%	0.29%	0.75%	0.28%
60–64	0.33%	0.33%	0.68%	0.29%
65–69	0.40%	0.37%	0.67%	0.29%
70–74	0.48%	0.41%	0.73%	0.28%
75–79	0.71%	0.49%	0.90%	0.28%
80–84	1.04%	0.58%	1.17%	0.28%
85–89	1.67%	0.70%	1.55%	0.29%

FIGURE 21.

The estimated annual fracture risk by age for a woman with a T-score of –2.5 and no previous fracture. ph, proximal humerus; vert, vertebral.

It is seen that at a T-score of –2.5 SD the risk of hip fracture greatly increases from 70 years of age. The rate of proximal humerus fractures remains fairly stable regardless of age, whereas that of vertebral fractures exhibits a relatively steady increase as the woman ages. The risk of a ‘wrist’ fracture initially decreases (as the influence of tibia and fibula fractures wanes) but this effect is overridden by the large natural increase in wrist fractures that are estimated to occur in the elderly.

Comparison and calibration of the model against previously published work

Recent modelling work4,5 undertaken using SHEMO for NICE and the National Coordinating Centre for Health Technology Assessment used data regarding the risks of fracture that were provided under an academic-in-confidence agreement. These data were not permitted to be used for this report and, as previously described, the model has reverted to estimating the risks of fracture based on age, gender and T-score alone, with increases in risks associated with clinical risk factors (including previous fracture).

Analyses have been carried out to compare the results of the new model with those of the model that used academic-in-confidence data. The most recent work5 incorporating the current price of alendronate could not be used for the comparison as different efficacies were used for some of the clinical risk factors. Instead the results are compared with previous work in which the price of alendronate was £173 per annum. 119 With the exception of the underlying risks of fracture, which have now been calculated using age, gender, T-score and previous fracture status, and a slight adjustment in the costs, because of different proportions of additional fractures (see Table 4), the input parameters were identical. These results are shown in Figures 22 and 23. It is seen that the change in cost per QALY in relation to age is less pronounced in the new model compared with the model using academic-in-confidence data, with similar results being obtained for women aged 70 years and older with a T-score of –2.5 SD to –3.0 SD (Figure 22). This is favourable to the interventions as the cost per QALY in the younger patients has been underestimated.

FIGURE 22.

Comparing the results under the new model and those when using the academic-in-confidence data allowed for previous modelling work assuming that women have no additional clinical risk factors and a T-score of –2.5 SD to –3.0 SD. AiC, academic-in-confidence.

FIGURE 23.

Comparing the results under the new model and those when using the academic-in-confidence data allowed for previous modelling work assuming that women have no additional clinical risk factors except for a previous fracture and a T-score of –2.5 SD. AiC, academic-in-confidence.

Additional analyses were undertaken to determine an appropriate level of increased fracture risk when a previous fracture has been sustained.

The new model was rerun using four different ratios for the initial fracture risk due to a previous fracture. These were increases of 25%, 50%, 75% and 100%. The results produced by these factors were compared with the results for women with a previous fracture and a T-score of –2.5 SD to –3.0 SD. 119 These data are depicted in Figure 23. It is seen that increases of 100% and 75% underestimate the cost per QALY at all ages, whereas an increase of 25% overestimates the cost per QALY at all ages. An increase of 50% appears the most appropriate as it produces similar results in those aged 70 years and over. This factor still underestimates the cost per QALY in younger ages, which will be favourable to the interventions.

The expected net benefit of sampling of a proposed RCT comparing alendronate and vitamin K₁

The rationale for conducting an RCT

Both alendronate and vitamin K₁ are relatively cheap compared with risedronate and strontium ranelate, with the former group priced below £61 per annum and the latter group priced at more than £250 per annum (Table 26). Both alendronate and vitamin K₁ have relatively good midpoint estimates for efficacy at preventing fracture (Table 27), although the evidence base for alendronate is much stronger because of the large number of RCTs conducted with bisphosphonates and the wide confidence intervals for vitamin K. Given these mid-point efficacy estimates, it is thus not surprising that, as detailed later, the most cost-effective intervention using standard cost per QALY thresholds94 appears to be either vitamin K₁ or alendronate.

To address the uncertainty in the decision regarding which treatment to prescribe, an RCT directly comparing alendronate and vitamin K₁ would be beneficial, although there has been no previous work undertaken on whether such a trial would be cost-effective. We employ expected value of sample information (EVSI) techniques. The method is fully described elsewhere120–122 and has recently been employed to determine the cost-effectiveness of an RCT to look at the long-term efficacy of bisphosphonates;117 it will be summarised in this report.

A major unknown in the knowledge base is the uncertainty around the efficacy of vitamin K₁ in preventing fractures. An RCT of vitamin K₁ against calcium alone would provide these data but may be unethical as patients allocated to the calcium arm would be denied alendronate, which, as shown later, is estimated to be cost-effective in postmenopausal women. Thus, ethically an RCT should compare vitamin K₁ directly against alendronate. The information that would be provided by the RCT would be the number of patients who fracture at each site in the two arms of the study. This would allow a RR of fracture at each site to be computed for alendronate compared with vitamin K₁. Note that as there is not a ‘no treatment’ arm the RCT would not provide any direct additional data on the efficacy of either alendronate or vitamin K₁ against no treatment.

The analyses undertaken assume that the proposed RCT would have a duration of 5 years and that an equal number of patients would be randomised to the alendronate and vitamin K₁ arms. It is recommended that within the proposed trial it is ensured that women are replete of calcium as is standard in trials of osteoporosis interventions.

Simulating the prior expectation of the comparative efficacy of alendronate and vitamin K₁

To undertake EVSI a prior expectation of the RR of fracture for alendronate compared with vitamin K₁ is required. This was obtained at each fracture site by sampling 1000 RRs for alendronate compared with no treatment and 1000 RRs for vitamin K₁ compared with no treatment (Table 27). The RRs for each intervention were combined to form 1000 estimates of the RR of alendronate compared with vitamin K₁. Note that this assumes independence between the efficacy of alendronate and the efficacy of vitamin K₁.

A statistical distribution was fitted to the generated RRs of alendronate compared with vitamin K1. Figure 24 shows the underlying simulated data on the RR of alendronate compared with vitamin K₁ at the hip and the statistical distribution fitted. Figures 25 and 26 give corresponding data for vertebral and the combined wrist and proximal humerus fractures respectively. A summary of the assumed distribution is provided in Table 29.

FIGURE 24.

The fitted distribution for the relative risk of alendronate compared with vitamin K₁ for hip fracture.

FIGURE 25.

The fitted distribution for the relative risk of alendronate compared with vitamin K₁ for vertebral fracture.

FIGURE 26.

The fitted distribution for the relative risk of alendronate compared with vitamin K₁ for wrist and proximal humerus fracture.

TABLE 29 - The summarised prior distribution of the relative risk of alendronate compared with vitamin K1

Site	Distribution	Mean (SD) of the log of the value
Hip	Log-normal	0.4463 (0.3821)
Vertebral	Log-normal	0.2287 (0.3941)
Wrist and proximal humerus	Log-normal	0.5558 (0.3924)

Calculating the expected value of sample information

The optimal decision given current information is that which yields the greatest expected net benefit,123 defined as max t Eθ NB(t, θ) where t represents the treatments being compared (t = 1,2, . . . T), θ a multivariate probability distribution based on current evidence and NB(t, θ) the net benefit of treatment t associated with θ. Undertaking an RCT will provide additional data on a subset of the unknown parameters θI (in this example the RR of alendronate compared with vitamin K₁ in preventing fracture at each site) but no direct evidence on the complimentary parameters denoted θI c, for example the disutility of a hip fracture. Calculation of EVSI involves simulating the collection of data samples and for each simulated sample data set examining the potential effect on the decision between treatment options. The data collected, which will be affected by both the underlying uncertainty in the parameters and the lack of precision associated with finite sample sizes, are denoted Xθ_I and a posterior distribution of θ_I,θ_I ^c | Xθ_I is calculated, which synthesises the prior distribution with the information produced by the RCT. Further ‘inner level’ Monte Carlo sampling then quantifies the effect of the simulated data obtained from the RCT on the decision, that is, whether the RCT would produce evidence to change the decision from that which is estimated to be optimal given current information. The optimal decision following data collection is the treatment option toptimal|XθI, which maximises the expected net benefit [max _t E(θ_I,θ_I ^c | Xθ_I)NB(t, θ_I,θ_I ^c)]. From this, the expected net benefit, given the updated data, of the treatment that was optimal using current information [E (θ_I,θ_I ^c | Xθ_I)NB(t_baseline, θ_I,θ_I ^c), where t_baseline is that which maximises Eθ NB(t, θ)] is subtracted to provide the additional expected net benefit obtained by making a decision after obtaining the sample data set Xθ_I. It is noted that when updating the data does not change the optimal decision the EVSI for that value of Xθ_I is zero. The expected EVSI for the RCT is calculated as the average EVSI across all sampled data sets Xθ_I. That is:

EVSI = E X θ I ( [ max t E ( θ I , θ I c | X θ I ) NB( t , θ I , θ I c ) ] − [ E ( θ I , θ I c | X θ I ) NB( t baseline , θ I , θ I c ) ] )

For this case study, the process of simulating the number of fractures expected in a proposed RCT of specified size, updating the prior distribution and calculating which is the better treatment duration from an inner probabilistic sensitivity analysis (PSA) process was undertaken for each of the 1000 parameter configurations previously used in calculating the better duration of treatment given current information. This produced a range of 1000 different simulated results for the specified trial design. To estimate the likely number of fractures in each arm the likely fracture risks of women recruited to the RCT needed to be estimated. It was assumed that the risks of those recruited would be equivalent to the risks associated with women aged 70–74 years with a T-score of –3 SD but no previous fracture, which are 0.93%, 0.84%, 0.80% and 0.31% per annum for vertebral, hip, wrist and proximal humerus fractures respectively.

Estimating the number of fractures in the RCT

For simplicity, we consider only trials with equal numbers of women, denoted n, in the two arms. The number of fractures simulated to occur in each arm of an RCT was estimated using a normal distribution with a mean equal to n multiplied by the probability of having a fracture, and variance equal to n multiplied by the product of the probability of fracture and the probability of not fracturing. The simulated trial results provide the number of people who fracture in the alendronate arm (x₁) and the number of people who fracture in the vitamin K₁ arm (x₂).

Updating the prior distribution with the data from the RCT to form a posterior distribution

The prior distributions of the RR of alendronate compared with vitamin K₁ were seen to be log-normal, thus the natural log of the RR of alendronate compared with vitamin K₁ is distributed normally with a mean m₁ and a variance of v_1. We approximate the distribution of the log RR from the hypothetical RCT by a normal distribution with mean z, where z is the log RR of alendronate compared with vitamin K₁, and variance s, where s = (n – x₁)/(n.x₁) + (n – x₂)/(n.x₂). Note that this approximation has used an estimate of the true sampling variance based on the data x₁ and x₂, and so uncertainty in the sampling variance has been ignored. The posterior distribution of the log of the RR of alendronate compared with vitamin K₁ is then approximately normal, with mean m₂ and variance v₂ given by v₂ = 1/(1/v₁ + 1/s) and m₂ = v₂ × (m₁/v₁ + z/s). Note that, as the number of women in the RCT increases, the posterior distribution becomes less sensitive to the choice of prior distribution and moves closer to that observed from the RCT.

An illustrative example of forming a posterior distribution

An illustrative example is provided using the prior distribution of the RR of alendronate compared with vitamin K₁ for vertebral fractures (Table 27). The natural log of the prior distribution of the RR of alendronate compared with vitamin K₁ is distributed normally with a mean of 0.2287 and a variance of 0.1553. It is assumed that we sample a ‘true’ log RR of alendronate compared with vitamin K₁ of 0.3; however, due to lack of precision from a finite sample size, this value of the log RR of alendronate compared with vitamin K₁ may not be observed. For illustrative purposes we assume that a simulated RCT of 1000 women produced 25 vertebral fractures in the alendronate cohort and 20 vertebral fractures in the vitamin K₁ cohort {thus exhibiting a RR of alendronate compared with vitamin K₁ of 0.2231 [log (25/20)] rather than the ‘true’ value of 0.3}. The distribution of z is thus estimated to be normally distributed with a mean of 0.2231 and a variance of 0.0880 [(1000–20)/(1000 × 20) + (1000–25)/(1000 × 25)]. The posterior distribution of the natural log of the RR of alendronate compared with vitamin K₁ would then be estimated to be normal with a variance of 0.0561 [1/(1/0.1553 + 1/0.0880)] and a mean of 0.2251 [0.0561 × (0.2287/0.1553 + 0.2231/0.0880)]. Monte Carlo sampling would be based on this distribution to provide estimates of the log RR of alendronate compared with vitamin K₁. These values would be used in a standard PSA to determine which of the two treatments was optimal.

Transforming the posterior distribution into RRs compared with no treatment for alendronate and vitamin K₁

Our mathematical model has been constructed whereby the efficacy of an intervention compared with no treatment is used to calculate the incremental costs and QALYs associated with treatment. The posterior distribution is for the log RR of alendronate compared with vitamin K₁. This must be converted into a RR for each intervention compared with no treatment. To do this it is assumed that the RRs for alendronate compared with no treatment are correct, because of the large number of patients (almost 5000 per arm) recruited to the bisphosphonate RCTs. In the PSA, the RR of alendronate is sampled and the RR of vitamin K₁ compared with no treatment is directly derived from a sampled RR of alendronate compared with vitamin K₁ and a RR of alendronate compared with no treatment. For example, if alendronate was sampled to have a RR of 0.8 compared with no treatment and a RR of 2 compared with vitamin K₁ then the RR of vitamin K₁ compared with no treatment would be estimated to be 0.4 (0.8/2).

Calculating the optimal decision following the RCT data

Given the simulated data from the RCT and the resulting posterior distribution of the natural log of the RR of alendronate compared with vitamin K₁, the optimal decision was re-evaluated using PSA and the expected net benefit of the optimal decision obtained. This expected net benefit was compared with the expected net benefit of the optimal decision based on prior information alone. The average of the increase in mean net benefit across all 1000 simulated trial results gives the estimated EVSI. To calculate if the RCT is a cost-effective use of resources two further sets of data are needed: how much the RCT will cost and how many patients will benefit from the additional information.

The costs of recruiting to the RCT

The cost of running the proposed RCT has been assumed to be £1000 per recruit (Clinical Trials Research Unit, ScHARR, University of Sheffield, 2007, personal communication) with additional intervention acquisition costs of £111.27 (£51 + £60.27) per annum. Although in reality there will be fixed costs and some form of economies of scale to be exploited, this value appears to be a reasonable approximation to the costs of successfully funded bids in the UK.

The expected number of women who will benefit from the increased knowledge

Estimating the number of women who may benefit from the extra research is more complicated. Table 40 in Appendix 7 reports that an estimated 20.9% of women aged 70–74 years are osteoporotic; as the number of women in this age range is estimated to be 1,130,516124 this would equate to approximately 236,000 women who could benefit from the better information regarding the relative efficacy of alendronate and vitamin K₁. Assuming that one of these interventions is likely to remain the mainstay of treatment for 10 years and with an assumed 50,000 women becoming eligible for treatment per year, this would equate to an estimated 736,000 women who would benefit, which has been rounded up to 1,000,000 to account for patients over 75 years of age or under 70 years of age who may also be eligible. It is believed that compliance with osteoporosis interventions is in the region of 50%95 and, thus, the estimated number of women who could benefit from an RCT assessing the relative efficacy of alendronate and vitamin K₁ is approximately 500,000. This value is adjusted to 200,000 and 1,000,000 in sensitivity analyses.

The scenarios undertaken

Two scenarios were run, the first using a cost per QALY threshold of £20,000 and the second using a cost per QALY threshold of £30,000. In both scenarios RCTs were evaluated assuming that 1000, 2000, 5000 or 10,000 women were recruited per arm. Sensitivity analyses were undertaken on the number of women who would benefit from the better data and the underlying fracture risks of those recruited within the trial.

Chapter 5 Results

Results for women with and without a previous fracture

The results from the cost-effectiveness analyses are provided in detail in Appendix 11, with summary values provided in Figures 27–30. Note that, for ease of interpretation, when an intervention dominates no treatment the value in the figure has been shown as zero.

FIGURE 27.

The cost-effectiveness of osteoporosis interventions compared with no treatment in women with a T-score of –2.5 SD and without a previous fracture.

FIGURE 28.

The cost-effectiveness of osteoporosis interventions compared with no treatment in women with a T-score of –3.0 SD and without a previous fracture.

FIGURE 29.

The cost-effectiveness of osteoporosis interventions compared with no treatment in women with a T-score of –2.5 SD and with a previous fracture.

FIGURE 30.

The cost-effectiveness of osteoporosis interventions compared with no treatment in women with a T-score of –3.0 SD and with a previous fracture.

Results of the EVSI analyses

Two scenarios were examined, which differed in the cost per QALY threshold used (£20,000 or £30,000). In all cases it was assumed that the efficacy data from the ECKO study71 were applicable at all fracture sites. The results are presented graphically in Figures 31 and 32. In both figures the expected EVSI is given, which is seen to decline as the marginal addition of patients provides less information (i.e. moving from 1000 to 2000 patients per arm is expected to provide more benefit than moving from 9000 to 10,000 patients as at 9000 patients the uncertainty in the relative efficacies will have been substantially reduced). The costs of the trial are linear. The ENBS, which is the EVSI minus the costs of the trial, is curved, with the optimal number of patients per arm shown by the turning point. It is seen that in both scenarios an RCT of 2000 women per arm appears to be optimal.

FIGURE 31.

The expected net benefit of sampling (ENBS) assuming a £20,000 cost per QALY threshold. EVSI, expected value of sample information.

FIGURE 32.

The expected net benefit of sampling (ENBS) assuming a £30,000 cost per QALY threshold. EVSI, expected value of sample information.

Chapter 6 Discussion

Vitamin K₁ has the potential to be a cost-effective intervention for preventing osteoporotic fractures, as it is likely to have a relatively inexpensive acquisition cost and a low RR of fracture prevention. However, this conclusion is heavily dependent on the efficacy of the intervention at the hip and the spine. At present there has been only one RCT of vitamin K₁,116 which has reported efficacy in the reduction of fractures as one group rather than at different sites. Analyses assuming that this efficacy is applicable at all sites indicate that vitamin K₁ would be the most cost-effective intervention. However, supplementary analyses that took an extreme position that vitamin K₁ had no effect on hip or vertebral fractures indicated that alendronate, the cheapest bisphosphonate, would be more cost-effective in this scenario. Data are urgently needed on the efficacy of vitamin K₁ at individual sites. It is noted that vitamin K₁ in the preparation used in the RCT is not currently available in the UK, although preparations at double this dose do exist. We have used the price of the higher-strength formulation but indicate that the price of 5 mg of vitamin K₁ is likely to be different to that used in this evaluation.

Expected value of sampling information has been conducted, which shows that an RCT comparing alendronate with vitamin K₁ and recruiting 2000 patients per arm would represent a cost-effective use of resources and would allow a more informed decision to be made over which drug, alendronate or vitamin K₁, is the more cost-effective. Although this analysis necessarily made assumptions regarding the likely efficacy of both drugs compared with no treatment the choice of 2000 women per arm consistently produced high ENBS. It has been assumed that clinicians and osteoporotic women are sufficiently in equipoise between the benefits of alendronate and vitamin K₁ to allow an RCT to be conducted, although it is recognised that some clinicians may have strong prevalent opinions on the relative merits of the two interventions. The authors note that the uncertainty in the efficacy of vitamin K₁, particularly in Western women, will not be resolved unless an RCT is undertaken, and that it would be ethically more appropriate to provide alendronate rather than placebo in the comparator arm.

The model constructed for this analysis differed from that used in recent NICE appraisals4,5 (Table 28). However, comparison of the results produced by the two modelling approaches shows that the cost per QALY ratios are unlikely to be substantially changed by the alternative methodology; however, it is noted that the cost-effectiveness ratios produced in this report may be favourable to the intervention for younger women (50–60 years of age).

Our analysis has not tried to position interventions within a care pathway as this has been the focus of a recent NICE appraisal,125 which has drawn on our earlier work5 and considered the costs of identifying women who should receive BMD scans. However, alendronate (and vitamin K₁ when efficacy is assumed to be equal at all sites) has a cost per QALY of below £20,000 for all women aged 50 years and over who are osteoporotic, which implies that these treatments would normally be considered cost-effective. 106

No formal evaluation of vitamin K₂ has been undertaken for a number of reasons. This intervention is currently not permitted as a food supplement in the EU because there is no evidence for its independent role in health26 and the price of the intervention is unknown. Additionally, the fracture efficacy data have wide confidence intervals, all of which spanned unity, and the only large (n > 1500 patients per arm) RCT reported a RR of 1.01 for vitamin K₂ compared with calcium or no active intervention.

No evaluation of treating with a combination of alendronate alongside vitamin K₁ has been undertaken. If vitamin K₁ is shown to be efficacious in future RCTs then the cost-effectiveness of combination treatment is a subject for future research.

Chapter 7 Conclusions

It would not be prudent to recommend the position of vitamin K₁ within a treatment algorithm without further information. EVSI analyses have been undertaken and it is recommended that an RCT comparing alendronate and vitamin K₁ and recruiting 2000 women per arm would represent a cost-effective use of resources. The cost implications to the NHS of undertaking such a trial have been estimated to be in the region of £4 million.

Vitamin K₂ has not been evaluated as it is not licensed as a food supplement in the EU and there is no statistically significant evidence that this intervention is effective at reducing fractures.

Acknowledgements

The clinical experts on the team who commented on the draft report were Professor Tahir Masud, Professor in Musculoskeletal Gerontology, University of Derby, and Dr Peter Selby, Senior Lecturer in Medicine, Department of Medicine, Manchester Royal Infirmary. Andrea Shippam, Project Administrator, ScHARR, organised the retrieval of papers and helped in preparing and formatting the report. The authors wish to thank all of the above.

References

1. Kanis JA, Brazier JE, Stevenson M, Calvert NW, Jones ML. Treatment of established osteoporosis: a systematic review and cost–utility analysis. Health Technol Assess 2002;6.
2. Stevenson M, Lloyd Jones M, De Nigris E, Brewer N, Davis S, Oakley J. A systematic review and economic analysis of alendronate, etidronate, risedronate, raloxifene and teriparatide for the prevention and treatment of postmenopausal osteoporosis. Health Technol Assess 2005;9.
3. Kanis JA, Brazier J, Stevenson M, McCloskey EV, Lloyd Jones M, Davis S. Glucocorticosteroid-induced osteoporosis: a systematic review and cost–utility analysis. Health Technol Assess 2007;11.
4. Stevenson M, Davis S, Lloyd Jones M, Beverley C. The clinical effectiveness and cost-effectiveness of strontium ranelate for the prevention of osteoporotic fragility fractures in postmenopausal women. Health Technol Assess 2007;11.
5. Stevenson M. Analyses of cost-effective BMD scanning and treatment strategies for generic alendronate, and the cost-effectiveness of risedronate and strontium ranelate in those people who would be treated with generic alendronate. London: NICE; 2008.
6. Consensus development conference: diagnosis, prophylaxis, and treatment of osteoporosis. Am J Med 2003;94:646-50.
7. Singer BR, McLauchlan GJ, Robinson CM, Christie J. Epidemiology of fractures in 15,000 adults: the influence of age and gender. J Bone Joint Surg 1998;80:243-8.
8. Kado DM, Browner WS, Palermo L, Nevitt MC, Genant HK, Cummings SR. Vertebral fractures and mortality in older women: a prospective study. Study of Osteoporotic Fractures Research Group. Arch Intern Med 1999;159:1215-20.
9. Keene GS, Parker MJ, Pryor GA. Mortality and morbidity after hip fractures. BMJ 1993;307:1248-50.
10. Torgerson DJ, Dolan P. The cost of treating osteoporotic fractures in the United Kingdom female population – the author replies. Osteoporos Int 2000;11:551-2.
11. Burge RT, Worley D, Johansen A, Bhattacharyya S, Bose U. The cost of osteoporotic fractures in the UK: projections for 2000–2020. J Med Econ 2001;4:51-2.
12. World Health Organization . Assessment of Fracture Risk and Its Application to Screening for Postmenopausal Osteoporosis 1994.
13. Kanis JA, Melton LJ, Christiansen C, Johnston CC, Khaltaev N. The diagnosis of osteoporosis. J Bone Miner Res 1994;9:1137-41.
14. Kanis JA, Gluer CC. An update on the diagnosis and assessment of osteoporosis with densitometry. Osteoporos Int 2000;11:192-20.
15. Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 1996;312:1254-9.
16. Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Delams P, et al. Predictive value of bone mineral density for hip and other fractures. J Bone Miner Res 2005;20:1185-94.
17. Holt G, Khaw KT, Reid DM, Compston JE, Bhalla A, Woolf AD, et al. Prevalence of osteoporotic bone mineral density at the hip in Britain differs substantially from the US over 50 years of age: implications for clinical densitometry. Br J Radiol 2002;75:736-42.
18. National Statistics, Population Estimates 2008. www.statistics.gov.uk/.
19. Kanis JA, Oden A, Johnell O, Jonsson B, De Laet C, Dawson A. The burden of osteoporotic fractures: a method for setting intervention thresholds. Osteoporos Int 2001;12.
20. Tsugawa N, Shiraki M, Suhara Y, Kamao M, Tanaka K, Okano T. Vitamin K status of healthy Japanese women: age-related vitamin K requirement for gamma-carboxylation of osteocalcin. Am J Clin Nutr 2006;83:380-6.
21. Booth S, Suttie JW. Dietary intake and adequacy of vitamin K1. J Nutr 1998;128:785-8.
22. Adams J, Pepping J. Vitamin K in the treatment and prevention of osteoporosis and arterial calcification. Am J Health Syst Pharm 2005;62:1574-81.
23. Kaneki M, Hodges SJ, Hosoi T, Fujiwara S, Lyons A, Crean SJ, et al. Japanese fermented soybean food as the major determinant of the large geographic difference in circulating levels of vitamin K2: possible implications for hip-fracture risk. Nutrition 2001;17:315-21.
24. Duggan P, Cashman KD, Flynn A, Bolton-Smith C, Kiely M. Phylloquinone (vitamin K-1) intakes and food sources in 18-year-old to 64-year-old Irish adults. Br J Nutr 2004;92:151-8.
25. Schurgers LJ, Teunissen KJF, Hamulyak K, Knapen MHJ, Vik H, Vermeer C. Vitamin K-containing dietary supplements: comparison of synthetic vitamin K1 and natto-derived menaquinone-7. Blood 2007;109:3279-83.
26. Vermeer C, Shearer MJ, Zittermann A, Bolton-Smith C, Szulc P, Hodges S, et al. Beyond deficiency: potential benefits of increased intakes of vitamin K for bone and vascular health. Eur J Nutr 2004;43:325-35.
27. British Medical Association and Royal Pharmaceutical Society of Great Britain . British National Formulary 2007. www.bnf.org/bnf/.
28. Kaneki M, Hosoi T, Ouchi Y, Orimo H. Pleiotropic actions of vitamin K: protector of bone health and beyond?. Nutrition 2006;22:845-52.
29. Binkley NC, Suttie JW. Vitamin K nutrition and osteoporosis. J Nutr 1995;125:1812-21.
30. Feskanich D, Weber P, Willett WC, Rockett H, Booth SL, Colditz GA. Vitamin K intake and hip fractures in women: a prospective study. Am J Clin Nutr 1999;69:74-9.
31. Booth SL, Tucker KL, Chen H, Hannan MT, Gagnon DR, Cupples LA, et al. Dietary vitamin K intakes are associated with hip fracture but not with bone mineral density in elderly men and women. Am J Clin Nutr 2000;71:1201-8.
32. Rejnmark L, Vestergaard P, Charles P, Hermann AP, Brot C, Eiken P, et al. No effect of vitamin K1 intake on bone mineral density and fracture risk in perimenopausal women. Osteoporos Int 2006;17:1122-32.
33. Compston JE. Boning up on vitamin K – commentary. Gut 2001;48.
34. Booth SL. Vitamin K status in the elderly. Curr Opin Clin Nutr Metab Care 2007;10:20-3.
35. Expert Group on Vitamins and Minerals . Risk Assessment. Vitamin K 2003. www.food.gov.uk/multimedia/.
36. Schaafsma A, Muskiet FA, Storm H, Hofstede GJ, Pakan I, van der Veer E. Vitamin D(3) and vitamin K(1) supplementation of Dutch postmenopausal women with normal and low bone mineral densities: effects on serum 25-hydroxyvitamin D and carboxylated osteocalcin. Eur J Clin Nutr 2000;54:626-31.
37. Fairfield KM, Fletcher RH. Vitamins for chronic disease prevention in adults: scientific review [erratum appears in JAMA 2002; 288: 1720]. JAMA 2002;287:3116-26.
38. Thane CW, Paul AA, Bates CJ, Bolton-Smith C, Prentice A, Shearer MJ. Intake and sources of phylloquinone (vitamin K1): variation with socio-demographic and lifestyle factors in a national sample of British elderly people. Br J Nutr 2002;87:605-13.
39. Gijsbers BLMG, Jie K-SG, Vermeer C. Effect of food composition on vitamin K absorption in human volunteers. Br J Nutr 1996;76:223-9.
40. Uematsu T, Nagashima S, Niwa M, Kohno K-I, Sassa T, Ishii M, et al. Effect of dietary fat content on oral bioavailability of menatetrenone in humans. J Pharm Sci 1996;85:1012-16.
41. Orimo H, Hosoda Y, Fujiwara S, Mizuno S, Hashimoto T, Tamaki T, et al. Hip fracture incidence in Japan. J Bone Miner Metab 1991;9:15-9.
42. Merck &amp; Co. Inc . Tablets Mephyton® (phytonadione) Vitamin K1 2002. www.fda.gov/medwatch/.
43. British Medical Association and Royal Pharmaceutical Society of Great Britain . British National Formulary 2008. www.bnf.org/bnf/.
44. Roche Products . Konakion® Tablets 10 mg 2005.
45. Eisai Co . Ltd. Vitamin K2 Preparation for Treatment of Osteoporosis – Glakay® Capsules 15 mg (menatetrenone Preparation) 2005. www2.eisai.co.jp/.
46. Knapen MH, Schurgers LJ, Vermeer C, Knapen MHJ, Schurgers LJ, Vermeer C. Vitamin K2 supplementation improves hip bone geometry and bone strength indices in postmenopausal women. Osteoporos Int 2007;18:963-72.
47. Solgar . Vitamin K2 (MK-7) 2008. www.the-health-store.co.uk/.
48. Cambridge Laboratories . Menadiol Diphosphate Tablets 10mg 2002. http://emc.medicines.org.uk/.
49. National Institute for Health and Clinical Excellence . Final Appraisal Determination. Alendronate, Etidronate, Risedronate, Raloxifene, Strontium Ranelate and Teriparatide for the Secondary Prevention of Osteoporotic Fragility Fractures in Postmenopausal Women 2008. www.nice.org.uk/nicemedia/pdf/FADOsteoSecondary180708.pdf.
50. Black DM, Thompson DE, Bauer DC, Ensrud K, Musliner T, Hochberg MC, et al. Fracture risk reduction with alendronate in women with osteoporosis: the Fracture Intervention Trial. J Clin Endocrinol Metab 2000;85:4118-24.
51. Jiang G, Eastell R, Barrington NA, Ferrar L. Comparison of methods for the visual identification of prevalent vertebral fracture in osteoporosis. Osteoporos Int 2004;15:887-96.
52. National Osteoporosis Foundation Working Group on Vertebral Fractures . Assessing vertebral fractures. J Bone Miner Res 1995;10:518-23.
53. Genant HK, Wu CY, van Kuijk C, Nevitt MC. Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 1993;8:1137-48.
54. Genant HK, Jergas M. Assessment of prevalent and incident fractures in osteoporosis research. Osteoporos Int 2003;14:S43-S55.
55. Grigoryan M, Guermazi A, Roemer FW, Delmas PD, Genant HK. Recognizing and reporting osteoporotic vertebral fractures. Eur Spine J 2003;12:S104-S112.
56. Ferrar L, Jiang G, Adams J, Eastell R. Identification of vertebral fractures: an update. Osteoporos Int 2005;16:717-28.
57. Cummings SR, Stone KL, Lui LL, Hillier TA, Bauer DC, Genant HK, et al. Are traumatic fractures osteoporotic? 2002. www.abstractsonline.com/.
58. Brewer T, Colditz GA. Postmarketing surveillance and adverse drug reactions: current perspectives and future needs. JAMA 1999;281:824-9.
59. Cranney A, Tugwell P, Wells G, Guyatt G. Systematic reviews of randomized trials in osteoporosis: introduction and methodology. Endocr Rev 2002;23:497-50.
60. Cramer JA. Effect of partial compliance on cardiovascular medication effectiveness. Heart 2002;88:203-6.
61. Insull W, Troendle A, Silvers A, Dunne CW. Substantial non-compliance to dose and time prescriptions for medications treating hypercholesterolaemia. Atherosclerosis 1995;115.
62. Farmer KC. Methods for measuring and monitoring medication regimen adherence in clinical trials and clinical practice. Clin Ther 1999;21:1074-90.
63. Stephenson BJ, Rowe BH, Haynes RB, Macharia WM, Leon G. Is this patient taking the treatment as prescribed?. JAMA 1993;269:2779-81.
64. Voss S, Quail D, Dawson A, Backstrom T, Aguas F, Erenus M, et al. A randomised, double-blind trial comparing raloxifene HCl and continuous combined hormone replacement therapy in postmenopausal women: effects on compliance and quality of life. Br J Obstet Gynaecol 2002;109:874-85.
65. Howell N, Trotter R, Mottram DR, Rowe PH. Compliance with statins in primary care. Pharm J 2004;272:23-6.
66. NHS Centre for Reviews and Dissemination . Undertaking Systematic Reviews of Research on Effectiveness 2001.
67. Gillespie WJ, Avenell A, Henry DA, O’Connell DL, Robertson J. Vitamin D and vitamin D analogues for preventing fractures associated with involutional and post-menopausal osteoporosis. Cochrane Database Syst Rev 2001;1.
68. Berlin JA. Does blinding of readers affect the results of meta-analyses?. Lancet 1997;350:185-6.
69. Clark HD, Wells GA, Huet C, McAlister FA, Salmi LR, Fergusson D, et al. Assessing the quality of randomized trials: reliability of the Jadad scale. Control Clin Trials 1999;20:448-52.
70. Cochrane Review Manager 4.2.7. Oxford: The Cochrane Collaboration; 2004.
71. Cheung AM, Tile L, Lee Y, Tomlinson G, Hawker G, Scher J, et al. Vitamin K supplementation in postmenopausal women with osteopenia (ECKO trial): a randomized controlled trial. PLoS Med 2008;5.
72. Iwamoto J, Takeda T, Ichimura S. Effect of menatetrenone on bone mineral density and incidence of vertebral fractures in postmenopausal women with osteoporosis: a comparison with the effect of etidronate. J Orthop Sci 2001;6:487-92.
73. Shiraki M, Shiraki Y, Aoki C, Miura M. Vitamin K2 (menatetrenone) effectively prevents fractures and sustains lumbar bone mineral density in osteoporosis. J Bone Miner Res 2000;15:515-21.
74. Shiraki M. Vitamin K2 effects on the risk of fractures and on lumbar bone mineral density in osteoporosis – a randomized prospective open-label 3-year study. Osteoporos Int 2002;13.
75. Eisai Co. Ltd . Eisai Announces the Intermediate Analysis of Anti-Osteoporosis Treatment Post-Marketing Research to Investigate the Benefits of Menatetrenone As Part of the Ministry of Health, Labour and Welfare’s Pharmacoepidemiological Drug Review Program 2005. www.eisai.co.jp/enews/.
76. Ishida Y, Kawai S. Comparative efficacy of hormone replacement therapy, etidronate, calcitonin, alfacalcidol, and vitamin K in postmenopausal women with osteoporosis: the Yamaguchi Osteoporosis Prevention Study. Am J Med 2004;117:549-55.
77. C Underhay, Medical Information Pharmacist, Eisai Ltd 2008.
78. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-60.
79. Black DM, Cummings SR, Karpf DB, Cauley JA, Thompson DE, Nevitt MC, et al. Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Lancet 1996;348:1535-41.
80. Cummings SR, Black DM, Thompson DE, Applegate WB, Barrett-Connor E, Musliner TA, et al. Effect of alendronate on risk of fracture in women with low bone density but without vertebral fractures: results from the Fracture Intervention Trial. JAMA 1998;280:2077-82.
81. Anon . Randomized, Open, Parallel, Active Controlled Study on Fracture Prevention in Antiosteoporosis Treatment (OF Study) 2005. http://clinicaltrials.gov/.
82. Asakura H, Myou S, Ontachi Y, Mizutani T, Kato M, Saito M, et al. Vitamin K administration to elderly patients with osteoporosis induces no hemostatic activation, even in those with suspected vitamin K deficiency. Osteoporos Int 2001;12:996-1000.
83. Ushiroyama T, Ikeda A, Ueki M. Effect of continuous combined therapy with vitamin K(2) and vitamin D(3) on bone mineral density and coagulofibrinolysis function in postmenopausal women. Maturitas 2002;41:211-21.
84. Ding H, Koinuma N, Stevenson M, Ito M, Monma Y. The cost-effectiveness of risedronate treatment in Japanese women with osteoporosis. J Bone Miner Metab 2008;26:34-41.
85. Storm T, Thamsborg G, Steiniche T, Genant HK, Sorensen OH. Effect of intermittent cyclical etidronate therapy on bone mass and fracture rate in women with postmenopausal osteoporosis. N Engl J Med 1990;322:1265-71.
86. Watts NB, Harris ST, Genant HK, Wasnich RD, Miller PD, Jackson RD, et al. Intermittent cyclical etidronate treatment of postmenopausal osteoporosis. N Engl J Med 1990;323:73-9.
87. Black D. Epidemiology of Osteoporosis – Update from the ASBMR 2002. www.medscape.com/.
88. Stevenson MD, Oakley J, Chilcott JB. Gaussian process modeling in conjunction with individual patient simulation modeling: a case study describing the calculation of cost-effectiveness ratios for the treatment of established osteoporosis. Med Decis Making 2004;24:89-100.
89. FRAX . FRAX 2008 n.d. www.sheffield.ac.uk/FRAX.
90. Stevenson MD, Brazier JE, Calvert NW, Lloyd Jones M, Oakley J, Kanis JA. Description of an individual patient methodology for calculating the cost-effectiveness of treatments for osteoporosis in women. J Oper Res Soc 2005;56:214-21.
91. Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making 1993;13:322-38.
92. Sox HC, Blatt MA, Higgins MC, Marton KI. Medical decision making. Oxford: Butterworth-Heinemann; 1998.
93. Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA, Berger M. Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis. J Bone Miner Res 2000;15:721-39.
94. Guide to the methods of technology appraisals. National Institute for Clinical Excellence 2008.
95. Lloyd Jones M, Wilkinson A. Adverse events and persistence with therapy in patients taking oral alendronate, etidronate or risedronate: systematic reviews. London: NICE; 2006.
96. Oden A, Dawson A, Dere W, Johnell O, Jonsson B, Kanis J. Lifetime risk of hip fractures is underestimated. Osteoporos Int 1998;8:599-603.
97. Meyer HE, Tverdal A, Falch JA, Pederden JI. Factors associated with mortality after hip fracture. Osteoporos Int 2000;11:228-32.
98. Poor G, Atkinson EJ, O’Fallon WM, Melton LJ. Determinants of reduced survival following hip fractures in men. Clin Orthop Relat Res 1995;319:260-5.
99. Todd CJ, Freeman C, Camilleri-Ferrante C. Anglian audit of hip fracture 2. Cambridge: Health Services Research Group, University of Cambridge; 1999.
100. Parker MJ, Anand JK. What is the true mortality of hip fractures?. Public Health 1991;105:443-6.
101. Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA. Mortality after all major types of osteoporotic fracture in men and women: an observational study. Lancet 1999;353:878-82.
102. Jalava T, Sama S, Pylkkanen L, Mawer B, Kanis JA, Selby P, et al. Association between vertebral fracture and increased mortality in osteoporotic patients. J Bone Miner Res 2003;18:1254-60.
103. Cauley JA, Thompson DE, Ensrud KC, Scott JC, Black D. Risk of mortality following clinical fractures. Osteoporos Int 2000;11:556-61.
104. Kanis JA, Oden A, Johnell O, De Laet C, Jonsson B. Excess mortality after hospitalisation for vertebral fracture. Osteoporos Int 2004;15:108-12.
105. Johnell O, Kanis JA, Oden A, Sembo I, Redlund-Johnell I, Petterson C, et al. Mortality after osteoporotic fractures. Osteoporos Int 2004;15:38-42.
106. Government Actuary Department . Expectation of Life: United Kingdom Females 1999.
107. Browner WS, Seeley DG, Vogt TM, Cummings SR. Non-trauma mortality in elderly women with low bone mineral density. Study of Osteoporotic Fractures Research Group. Lancet 1991;338:355-8.
108. Johansson C, Black D, Johnell O, Oden A, Mellstrom D. Bone mineral density is a predictor of survival. Calcif Tissue Int 1998;63:190-6.
109. Van Der Klift M, Pols HA, Geleijnse JM, Van Der Kuip DA, Hofman A, De Laet CE. Bone mineral density and mortality in elderly men and women: the Rotterdam Study. Bone 2002;30:643-8.
110. Jonsson B, Christiansen C, Johnell O, Hedbrandt J, Karlsson G. Cost-effectiveness of fracture prevention in established osteoporosis. Scand J Rheumatol 1996;25:30-8.
111. Kanis JA, Adams J, Borgstrom F, Coope, C, Jonsson B, Preedy D, et al. The cost-effectiveness of alendronate in the management of osteoporosis. Bone 2008;42:4-15.
112. Johnell O, Kanis J, Jonsson BA, Oden A, Johansson H, De Laet C. The burden of hospitalised fractures in Sweden. Osteoporos Int 2005;16:222-8.
113. Kind P, Dolan P, Gudex C, Williams A. Variations in population health status: results from a United Kingdom national questionnaire survey. BMJ 1998;316:736-41.
114. Stevenson M, Davis S. Analyses of the cost-effectiveness of pooled alendronate and risedronate, compared with strontium ranelate, raloxifene, etidronate and teriparatide. London: NICE; 2006.
115. Curtis L, Netten A. Unit costs of health and social care. Canterbury: PSSRU, University of Kent; 2007.
116. Cheung AM, Tile L, Lee Y, Tomlinson G, Hawker G, Scher J, et al. Vitamin K Supplementation in Postmenopausal Women With Osteopenia: The ECKO Trial n.d.
117. Stevenson MD, Oakley JE, Lloyd Jones M, Brennan A, Compston JE, McCloskey EV, et al. The cost-effectiveness of an RCT to establish whether 5 or 10 years of bisphosphonate treatment is the better duration for women with a prior fracture. Med Decis Making 2009.
118. Lloyd Jones M, Wilkinson A. Adverse effects and persistence with therapy in patients taking oral alendronate, etidronate or risedronate: a systematic review. Sheffield: ScHARR, University of Sheffield; 2006.
119. Stevenson M. Analyses of Cost-Effective BMD Scanning and Treatment Strategies for Generic Alendronate, Risedronate, Strontium Ranelate, Raloxifene and Teriparatide Following Corrections to the Methodology Associated With Lower Efficacy in Some Risk Factors. NICE 2006.
120. Ades AE, Lu G, Claxton K. Expected values of sample information calculation in medical decision making. Med Decis Making 2004;24:207-27.
121. Brennan A, Kharroubi SA, Chilcott JB, O’Hagan A. A two level Monte Carlo approach to calculation expected value of sample information: how to value a research design. Sheffield: University of Sheffield; 2002.
122. Brennan A, Kharroubi SA. Efficient computation of partial expected value of sample information using Bayesian approximation. J Health Econ 2007;26:122-48.
123. Stinnett A, Mullahy J. Net health benefits: a new framework for the analysis of uncertainty in cost-effectiveness analyses. Med Decis Making 1998;18:S68-S80.
124. Government Actuary Department . Interim Life Tables. England and Wales 2008. www.gad.gov.uk/.
125. National Institute for Health and Clinical Excellence . Osteoporosis – Secondary Prevention Including Strontium Ranelate 2008.
126. Alper BS. Evidence-based medicine. Vitamin K appears to reduce post-menopausal fracture rates. Clin Advisor 2007;10.
127. Clouatre D, Shastri S. Natto K2, the unique vitamin K: new benefits for bone and cardiovascular health. Total Health 2004;26:48-9.
128. Ishida Y, Kawai S. Comparative efficacy of hormone replacement therapy, bisphosphonates, calcitonin, vitamin D and vitamin K in postmenopausal women with osteoporosis. J Bone Miner Res 2003;18.
129. Iwamoto I, Kosha S, Noguchi S, Murakami M, Fujino T, Douchi T, et al. A longitudinal study of the effect of vitamin K2 on bone mineral density in postmenopausal women a comparative study with vitamin D3 and estrogen-progestin therapy. Maturitas 1999;31:161-4.
130. Iwamoto J, Takeda T, Ichimura S. Effect of combined administration of vitamin D3 and vitamin K2 on bone mineral density of the lumbar spine in postmenopausal women with osteoporosis. J Orthop Sci 2000;5:546-51.
131. Kenney JJ. Vitamin K improves bone health. Commun Food Health 2006;Aug.
132. Meunier PJ. Calcium, vitamin D and vitamin K in the prevention of fractures due to osteoporosis. Osteoporos Int 1999;9:S48-S52.
133. Anon . Effect and Safety of Menatetrenone on Treatment of Postmenopausal Osteoporosis 2007. http://clinicaltrials.gov/.
134. Purwosunu Y, Muharram, Rachman IA, Reksoprodjo S, Sekizawa A. Vitamin K2 treatment for postmenopausal osteoporosis in Indonesia. J Obstet Gynaecol Res 2006;32:230-4.
135. Radecki TE. Calcium and vitamin D in preventing fractures: vitamin K supplementation has powerful effect. BMJ 2005;331.
136. Rejnmark L, Vestergaard P, Charles P, Hermann AP, Brot C, Eiken P, et al. No effect of vitamin K intake on bone mineral density and fracture risk in perimenopausal women. J Bone Miner Res 2005;20.
137. Rejnmark L, Vestergaard P, Charles P, Hermann P, Brot C, Eiken P, et al. No effect of vitamin K1 intake on bone mineral density and fracture risk in perimenopausal women. Calcif Tissue Int 2006;78.
138. Shiraki M. A head-to-head comparative study for evaluation of effectiveness among drugs for osteoporosis. Jpn J Geriatr 2006;43:45-7.
139. Ishida Y, Kawai S. A two-year randomized controlled trial of hormone replacement therapy, etidronate, calcitonin, vitamin d, or vitamin k, in women with postmenopausal osteoporosis. J Bone Miner Res 2002;17.
140. Ishida Y, Soh H, Ogawa S, Kawahara S, Murata H. A one-year randomized controlled trial of hormone replacement therapy, bisphosphonate, calcitonin, vitamin D, and vitamin K in women with postmenopausal osteoporosis. J Bone Miner Res 2000;15.
141. Orimo H, Sugioka Y, Fukunaga M, Muto Y, Hotokebuchi T, Gorai I, et al. Diagnostic criteria of primary osteoporosis. J Bone Miner Metab 1998;16:139-50.
142. Kanis JA, Johnell O, Oden A, Sembo I, Redlund-Johnell I, Dawson A, et al. Long-term risk of osteoporotic fracture in Malmö. Osteoporos Int 2000;11:669-74.
143. Meunier PJ, Slosman DO, Delmas PD, Sebert JL, Brandi ML, Albanese C, et al. Strontium ranelate: dose-dependent effects in established postmenopausal vertebral osteoporosis – a 2-year randomized placebo controlled trial. J Clin Endocrinol Metab 2002;87:2060-6.
144. Meunier PJ, Roux C, Seeman E, Sergio O, Badurski JE, Spector TD, et al. The effects of strontium ranelate on the risk of vertebral fracture in women with postmenopausal osteoporosis. N Engl J Med 2004;350:459-68.
145. Servier Laboratories . Submission to the National Institute for Health and Clinical Excellence for the appraisal. Strontium Ranelate for the Prevention of Postmenopausal Osteoporotic Fractures 2005.
146. Reginster JY, Seeman E, De Vernejoul MC, Adami S, Compston J, Phenekos C, et al. Strontium ranelate reduces the risk of nonvertebral fractures in post-menopausal women with osteoporosis: TROPOS study. J Clin Endocrinol Metab 2005;90:2816-22.
147. Reginster JY, Felsenberg D, Boonen S, Diez-Perez A, Rizzoli R, Brandi ML, et al. Effects of long-term strontium ranelate treatment on the risk of nonvertebral and vertebral fractures in postmenopausal osteoporosis: results of a five-year, randomized, placebo-controlled trial. Arthritis Rheum 2008;58:1687-95.
148. Marquis P, Roux C, de la Loge C, Diaz-Curiel M, Cormier C, Isaia G, et al. Strontium ranelate prevents quality of life impairment in post-menopausal women with established vertebral osteoporosis. Osteoporos Int 2008;19:503-10.
149. Roux C. Strontium ranelate: short- and long-term benefits for post-menopausal women with osteoporosis. Rheumatology 2008;47:iv20-iv22.
150. O’Donnell S, Cranney A, Wells GA, Adachi JD, Reginster JY, O’Donnell S, et al. Strontium ranelate for preventing and treating postmenopausal osteoporosis [update of Cochrane Database Syst Rev 2006; 3:CD005326]. Cochrane Database Syst Rev 2006;4.
151. EMEA Scientific Discussion 2004. www.emea.eu.int/humandocs/Humans/EPAR/protelos/protelos.htm.
152. Meunier PJ, Reginster JY. Design and methodology of the phase 3 trials for the clinical development of strontium ranelate in the treatment of women with postmenopausal osteoporosis. Osteoporos Int 2003;14:S66-S76.
153. Hosking D, Adami S, Felsenberg D, Andia JC, Valimaki M, Benhamou L, et al. Comparison of change in bone resorption and bone mineral density with once-weekly alendronate and daily risedronate: a randomised, placebo-controlled study. Curr Med Res Opin 2003;19:383-94.
154. Adami S, Passeri M, Ortolani S, Broggini M, Carratelli L, Caruso I, et al. Effects of oral alendronate and intranasal salmon calcitonin on bone mass and biochemical markers of bone turnover in postmenopausal women with osteoporosis. Bone 1995;17:383-90.
155. Bone HG, Downs RW, Tucci JR, Harris ST, Weinstein RS, Licata AA, et al. Dose-response relationships for alendronate treatment in osteoporotic elderly women. J Clin Endocrinol Metab 1997;82:265-74.
156. Bone HG, Greenspan SL, McKeever C, Bell N, Davidson M, Downs RW, et al. Alendronate and estrogen effects in postmenopausal women with low bone mineral density. Alendronate/Estrogen Study Group. J Clin Endocrinol Metab 2000;85:720-6.
157. Carfora E, Sergio F, Bellini P, Sergio C, Falco D, Zarcone R. Effect of treatment of postmenopausal osteoporosis with continuous daily oral alendronate and the incidence of fractures. Gazz Med Ital 1998;157:105-9.
158. Chesnut CH, McClung MR, Ensrud KE, Bell NH, Genant HK, Harris ST, et al. Alendronate treatment of the postmenopausal osteoporotic woman: effect of multiple dosages on bone mass and bone remodeling. Am J Med 1995;99:144-52.
159. Durson N, Durson E, Yalcin S. Comparison of alendronate, calcitonin and calcium treatments in postmenopausal osteoporosis. Int J Clin Pract 2001;55:505-9.
160. Black DM, Schwartz AV, Ensrud KE, Cauley JA, Levis S, Quandt SA, et al. Effects of continuing or stopping alendronate after 5 years of treatment: the Fracture Intervention Trial Long-term Extension (FLEX): a randomized trial. JAMA 2006;296:2927-38.
161. Greenspan SL, Schneider DL, McClung MR, Miller PD, Schnitzer TJ, Bonin R, et al. Alendronate improves bone mineral density in elderly women with osteoporosis residing in long-term care facilities. A randomized, double-blind, placebo-controlled trial. Ann Intern Med 2002;136:742-6.
162. Kaadan N. The preventive effect of alendronate on bone looses (sic) and vertebral fractures in postmenopausal osteoporosis in Aleppo City. Bone 2002;30.
163. Liberman UA, Weiss SR, Broll J, Minne HW, Quan H, Bell NH, et al. Effect of oral alendronate on bone mineral density and the incidence of fractures in postmenopausal osteoporosis. The Alendronate Phase III Osteoporosis Treatment Study Group. N Engl J Med 1995;333:1437-43.
164. Lindsay R, Cosman F, Lobo RA, Walsh BW, Harris ST, Reagan JE, et al. Addition of alendronate to ongoing hormone replacement therapy in the treatment of osteoporosis: a randomized, controlled clinical trial. J Clin Endocrinol Metab 1999;84:3076-81.
165. Or APC, Chan WS, Lau EMC, Leung PC. Effect of alendronate treatment for the Chinese postmenopausal women with osteoporotic vertebral fractures. Bone 2001;28.
166. Pols HA, Felsenberg D, Hanley DA, Stepan J, Munoz-Torres TJ, . Multinational, placebo-controlled, randomized trial of the effects of alendronate on bone density and fracture risk in postmenopausal women with low bone mass: results of the FOSIT study. Foxamax International Trial Study Group. Osteoporos Int 1999;9:461-8.
167. Rossini M, Gatti D, Zamberlan N, Braga V, Dorizzi R, Adami S. Long-term effects of a treatment course with oral alendronate of postmenopausal osteoporosis. J Bone Miner Res 1994;9:1833-7.
168. Clemmesen B, Ravn P, Zegels B, Taquet AN, Christiansen C, Reginster JY. A 2-year phase II study with 1-year of follow-up of risedronate (NE-58095) in postmenopausal osteoporosis. Osteoporos Int 1997;7:488-95.
169. Fogelman I, Ribot C, Smith R, Ethgen D, Sod E, Reginster JY. Risedronate reverses bone loss in postmenopausal women with low bone mass: results from a multinational, double-blind, placebo-controlled trial. BMD-MN Study Group. J Clin Endocrinol Metab 2000;85:1895-900.
170. Harris ST, Watts NB, Genant HK, McKeever CD, Hangartner T, Kelle M, et al. Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis: a randomized controlled trial. Vertebral Efficacy With Risedronate Therapy (VERT) Study Group. JAMA 1999;282:1344-52.
171. McClung M, Benson W, Bolognese M, Bonnick S, Ettinger M, Harris S, et al. Risedronate increases bone mineral density at the hip, spine and radius in postmenopausal women with low bone mass. Osteoporos Int 1998;8.
172. McClung MR, Geusens P, Miller PD, Zippel H, Benson WG, Roux C, et al. Effect of risedronate on the risk of hip fracture in elderly women. Hip Intervention Program Study Group. N Engl J Med 2001;344:333-40.
173. Reginster J, Minne HW, Sorensen OH, Hooper M, Roux C, Brandi ML, et al. Randomized trial of the effects of risedronate on vertebral fractures in women with established postmenopausal osteoporosis. Vertebral Efficacy with Risedronate Therapy (VERT) Study Group. Osteoporos Int 2000;11:83-91.
174. Tonino RP, Meunier PJ, Emkey R, Rodriguez-Portales JA, Menkes CJ, Wasnich RD, et al. Skeletal benefits of alendronate: 7-year treatment of postmenopausal osteoporotic women. Phase III Osteoporosis Treatment Study Group. J Clin Endocrinol Metab 2000;85:3109-15.
175. Bone HG, Hosking D, Devogelaer JP, Tucci JR, Emkey R, Tonino RP, et al. Ten years’ experience with alendronate for osteoporosis in postmenopausal women. N Engl J Med 2004;350:1189-99.
176. Emkey R, Reid I, Mulloy A, Correa-Rotter R, Favus M, Bone H, et al. Ten-year efficacy and safety of alendronate in the treatment of osteoporosis in postmenopausal women. J Bone Miner Res 2002;17.
177. Ste-Marie LG, Sod E, Johnson T, Chines A. Five years of treatment with risedronate and its effects on bone safety in women with postmenopausal osteoporosis. Calcif Tissue Int 2004;75:469-76.
178. Sorensen OH, Crawford GM, Mulder H, Hosking DJ, Gennari C, Mellstrom D, et al. Long-term efficacy of risedronate: a 5-year placebo-controlled clinical experience. Bone 2003;32:120-6.
179. Mellstrom DD, Sorensen OH, Goemaere S, Roux C, Johnson TD, Chines AA, et al. Seven years of treatment with risedronate in women with postmenopausal osteoporosis. Calcif Tissue Int 2004;75:462-8.
180. Ensrud K, Barrett-Connor E, Schwartz A, Santora A, Bauer D, Suryawanshi S, et al. Randomized trial of effect of alendronate continuation versus discontinuation in women with low BMD: results from the Fracture Intervention Trial Long-term Extension. J Bone Miner Res 2004;19:1259-69.
181. Steinbuch M, D’Agostino RB, Mandel JS, Gabrielson E, McClung MR, Stemhagen A, et al. Assessment of mortality in patients enrolled in a risedronate clinical trial program: a retrospective cohort study. Regul Toxicol Pharmacol 2002;35:320-6.
182. Tucci JR, Tonino RP, Emkey RD, Peverly CA, Kher U, Santora AC. Effect of three years of oral alendronate treatment in postmenopausal women with osteoporosis. Am J Med 1996;101:488-501.
183. Woo S-B, Hellstein JW, Kalmer JR. Systematic review: bisphosphonates and osteonecrosis of the jaws. Ann Intern Med 2006;144:753-61.
184. Nevitt MC, Thompson DE, Black DM, Rubin SR, Ensrud K, Yates AJ, et al. Effect of alendronate on limited-activity days and bed-disability days caused by back pain in postmenopausal women with existing vertebral fractures. Arch Intern Med 2000;160:77-85.
185. Buist DS, LaCroix AZ, Black DM, Harris F, Blank J, Ensrud K, et al. Inclusion of older women in randomized clinical trials: factors associated with taking study medication in the fracture intervention trial. J Am Geriatr Soc 2000;48:1126-31.
186. Borgstrom F, Zethraeus N, Johnell O, Lidgren L, Ponzer S, Svensson O, et al. Costs and quality of life associated with osteoporosis-related fractures in Sweden. Osteoporos Int 2006;17:637-50.
187. Lawrence TM, White CT, Wenn R, Moran CG. The current hospital costs of treating hip fractures. Injury 2005;36:88-91.
188. Stevenson MD, Davis SE, Kani JA. The hospitalisation costs and out-patient costs of fragility fractures. Women’s Health Med 2006;3:149-51.
189. Kanis JA, Pitt F. Epidemiology of osteoporosis. Bone 1992;1:S7-S15.
190. Puffer S, Torgerson DJ, Sykes D, Brown P, Cooper C. Health care costs of women with symptomatic vertebral fractures. Bone 2004;35:383-6.
191. 2008. www.scb.se/templates/.

Appendix 1 MEDLINE clinical effectiveness search strategy

1. exp osteoporosis/

2. Osteoporo$.tw.

3. Bone diseases, metabolic/

4. 1 or 2 or 3

5. (Bone adj6 densit$).tw.

6. Bone density/

7. (Bone or bones).mp. [mp=title, original title, abstract, name of substance word, subject heading word]

8. exp Densitometry/

9. Tomography, x-ray computed/

10. Densit$.tw.

11. 9 and 10

12. 8 or 11

13. 7 and 12

14. 4 or 5 or 6 or 13

15. exp Vitamin K/

16. vitamin k1.tw.

17. vitamin k 1.tw.

18. menaquinone$.tw.

19. phylloquinone$.tw.

20. phytomenadione$.tw.

21. phytonadione$.tw.

22. aquamephyton$.tw.

23. konakion$.tw.

24. phyllohydroquinone$.tw.

25. vitamin k2.tw.

26. vitamin k 2.tw.

27. menaquinone$.tw.

28. vitamin k quinone$.tw.

29. vitamin k3.tw.

30. vitamin k 3.tw.

31. vitamin k sodium bisulfite.tw.

32. menadione$.tw.

33. 2-methyl-1, 4-napthalenedione.tw.

34. 2-methyl-1, 4-napthoquinone$.tw.

35. menadione bisulfite$.tw.

36. menadione sodium bisulfite$.tw.

37. vicasol.tw.

38. vikasol.tw.

39. phytonadione.tw.

40. or/15–39

41. 14 and 40

Appendix 2 Randomised controlled trial data extraction form

Appendix 3 Publications relating to the trials that met the inclusion criteria for the review

The major publication for each study is indicated with an asterisk.

Appendix 4 References excluded from the review of clinical effectiveness after a full reading

Appendix 5 Evidence tables

Appendix 6 Calculation of the additional QALYs lost through a death from a hip fracture, vetebral fracture or proximal humerus fracture

The initial individual patient model runs used a time horizon of 10 years. This, however, would mean than any mortality prevented within this period would not be given full weight, which would bias against beneficial treatments, and adjustments were needed to correct for this error.

To adjust for this factor, an estimation of the QALYs that could be gained by the prevention of mortality at each age was made. Calculations were only needed from the end of the 10-year modelling horizon as any QALY impacts within this period would be explicitly calculated within the model. The methodology for this was as follows.

The life expectancy for a patient at the threshold of osteoporosis was calculated from standard life tables. It was assumed that any increase in mortality rate due to low bone mass would continue until death or an age of 110 years.

As the final utility value of each patient within the individual patient model was not estimated by the Gaussian model, it was assumed, slightly favouring the interventions, that individuals would have a utility equal to that of the general population as reported by Kind et al. 113 QALYs were discounted at 1.5% per annum, starting from the time of intervention, so that the results were consistent with those produced by the individual patient model.

Using these assumptions it was estimated that an average patient alive at the end of the model would accrue QALYs as given in Table 34.

Having established the gains associated with preventing mortality, the expected number of potentially preventable deaths through hip fracture was calculated. The methodology for this was based on the standard rate of hip fracture at each age and the expected mortality associated with hip fraacture at that age.

For example, the expected hip fracture rate at age 60 years for healthy women at the threshold of osteoporosis is estimated to be 0.1%. For women with severe osteoporosis it is assumed that this risk can be doubled in accordance with data reported by Klotzbuecher et al. 93 This would equate to an estimate of the hip fracture rate of 0.2% per annum, or 1.0% over a 5-year treatment period, assuming no additional mortality, which is one hip fracture for a cohort of 100 women.

The mortality rate following hip fracture is estimated to be 6% at age 60 years (Table 21), resulting in an estimated maximum of 0.06 hip fractures that were preventable over the intervention period. The number that were preventable is assumed to be equal to the sampled RR for each treatment; thus, if a RR of hip fracture of 0.5 was estimated then it was assumed that 0.03 deaths associated with hip fractures would be saved. When the RR was greater than 1, the model assumed that an additional number of deaths would occur and subtracted the expected QALYs from that estimated for the intervention.

The expected numbers of additional QALYs for women with severe osteoporosis suffering death from hip fracture are given in Table 35.

An alternative methodology had to be employed for deaths assumed to be associated with vertebral fractures because unlike mortalities associated with hip fracture these were not explicitly calculated within the 10-year time horizon.

It was assumed that all deaths from vertebral fracture would happen in year 3, the midpoint of the treatment period. A 66% increase in the mortality rate in the year of a vertebral fracture was assumed, as reported by Center et al. 101 and assuming that all of these deaths were attributable to the vertebral fracture. By calculating the expected number of vertebral fractures per year and the expected associated mortality, assuming 5 years of no treatment, the maximum number of QALYs that could be prevented were estimated. These are shown in Table 36.

It was assumed that the number of mortalities that could be prevented is proportional to the RR of the treatment. Hence, a treatment with a RR of 0.5 for vertebral fracture would be assumed to prevent 50% of the mortalities from vertebral fracture.

A similar methodology has been used for mortality associated with fractures of the proximal humerus. The maximum number of QALYs lost because of proximal humerus fracture and assumed to be preventable are shown in Table 37.

Appendix 7 Calculating the risk of fracture for women with a Z-score of 0 and no previous fracture

An estimate of the fracture risk for women with average BMD and no previous fracture has been calculated assuming that there are three sets of patient type at each age: women with a T-score of –2.5 SD or less with a previous fracture (group A), women with a T-score of –2.5 SD or less without a previous fracture (group B) and women with an average BMD and without previous fracture (group C).

The T-score values of the average female population and of those that are osteoporotic are provided in Table 38.

It is assumed that a previous fracture will increase the risk of subsequent fractures (at all sites) twofold compared with group C. The increased risk due to a woman being osteoporotic will be calculated from the data in Table 38 and Tables 17 and 18 of the main report. An example of how to calculate the RRs for hip fracture in groups A and B is provided for women aged 70–74 years.

At 70–74 years of age the average T-score is –1.69 SD and the decrease in Z-score to those who are osteoporotic is 1.31 (Table 38). This will equate to an increased risk of hip fracture in group B of 2.78^1.31 = 3.82 (see Table 18).

The RR of women in group A is assumed to be double that of group B, i.e. 2 × 2.78^1.31 = 7.63.

The calculated RRs of groups A and B, by age and fracture site, are given in Table 39.

To estimate the risk of fracture within group C the percentages of patients suffering osteoporosis and severe osteoporosis must be estimated. Data on the percentage of women who are osteoporotic (including severe) at each age have been calculated from the average BMD, assuming that BMD is normally distributed and has a SD of 1. Estimates of the percentage of women with severe osteoporosis at each age have been calculated using data from Kanis et al. ,142 which indicate that the percentages of all fractures that are first fractures are approximately 90% below the age of 70 years and 80% above the age of 70 years. Assuming these figures to be applicable in the UK the percentage of the population with severe osteoporosis can be estimated from the incidence of fracture since age 50 years142 and expected mortality rates,35 assuming that all fractures at the hip, spine, wrist or proximal humerus were caused by osteoporosis. The estimated proportions of the female population with osteoporosis and severe osteoporosis are given in Table 40 (calculations not shown).

At 70 years of age it is expected that 15.6% of women will have severe osteoporosis and 5.3% will have osteoporosis; 79.1% of women will not be osteoporotic (Table 40).

The average incidence of hip fracture (ignoring fractures at the pelvis and other femoral fractures) at age 70–74 years is estimated to be 0.38% per annum (Table 2).

This will comprise:

which equates to:

As 218% × group C risk equals the average incidence of 0.38% per annum, the risk in group C must equal 0.38%/2.18 = 0.18%. The risks in group A and group B will be 0.18% × 7.63 and 0.18% × 3.82, respectively, which correspond to 1.34% and 0.67%, respectively, per annum.

The risk for a woman aged 70–74 years can now be estimated at any T-score. Thus, at the threshold of osteoporosis, the Z-score decrease is 0.81 (see Table 1) and the risk of a hip fracture will be 0.18% × 2.78^0.81 = 0.40% per annum.

When fractures at the pelvis and other femoral fractures are considered the fracture risk is increased by 20% (see Table 14), which would increase the risk of fracture to 0.48% per annum.

This methodology was repeated for all fracture sites and all ages. Sensitivity analyses previously conducted4 showed that the average risk for a healthy woman did not change markedly with small changes in the percentages of patients with severe osteoporosis.

Appendix 8 Systematic searching for evidence relating to Vitamin K and adverse effects in osteoporotic patients

An electronic search of MEDLINE was conducted in January 2008. This sought to identify studies of any type that dealt with the adverse effects associated with the administration of any type of supplementary vitamin K to adults with primary osteoporosis or osteopenia. The following search strategy was used:

1. (ae or po or to or co or de).fs.

2. adverse event$.tw.

3. adverse effect$.tw.

4. side effect$.tw.

5. safe$.tw.

6. 1 or 2 or 3 or 4 or 5

7. exp Vitamin K/

8. vitamin k1.tw.

9. vitamin k 1.tw.

10. menaquinone$.tw.

11. phylloquinone$.tw.

12. phytomenadione$.tw.

13. phytonadione$.tw.

14. aquamephyton$.tw.

15. konakion$.tw.

16. phyllohydroquinone$.tw.

17. vitamin k2.tw.

18. vitamin k 2.tw.

19. vitamin k quinone$.tw.

20. vitamin k3.tw.

21. vitamin k 3.tw.

22. vitamin k sodium bisulfite.tw.

23. menadione$.tw.

24. 2-methyl-1, 4-napthalenedione.tw.

25. 2-methyl-1, 4-napthoquinone$.tw.

26. menadione bisulfite$.tw.

27. menadione sodium bisulfite$.tw.

28. vicasol.tw.

29. vikasol.tw.

30. phytonadione.tw.

31. or/7–30

32. 6 and 31

33. exp osteoporosis/

34. Osteoporo$.tw.

35. Bone diseases, metabolic/

36. 33 or 34 or 35

37. (Bone adj6 densit$).tw.

38. Bone density/

39. (Bone or bones).mp. [mp=title, original title, abstract, name of substance word, subject heading word]

40. exp Densitometry/

41. Tomography, x-ray computed/

42. Densit$.tw.

43. 41 and 42

44. 40 or 43

45. 39 and 44

46. 36 or 37 or 38 or 45

47. 32 and 46

FIGURE 33.

Adverse effects: summary of study selection and exclusion – electronic literature searches.

Appendix 9 Strontium ranelate for the prevention of osteoporotic fracture in postmenopausal women

A systematic review of the clinical effectiveness of strontium ranelate for the prevention of osteoporotic fracture in postmenopausal women was carried out on behalf of NICE in 2005. 4 Three studies that compared strontium ranelate with placebo met the review’s inclusion criteria. These were:

• the STRATOS study,143 a randomised, multicentre, double-blind, 2-year phase II dose-ranging study whose aim was to identify the smallest dose of strontium ranelate that was effective in treating postmenopausal vertebral osteoporosis, using BMD of the lumbar spine adjusted for bone strontium content as the primary end point

• the SOTI study,144 a randomised, multicentre, double-blind, phase III study designed to evaluate the efficacy of strontium ranelate against vertebral fracture in postmenopausal women with osteoporosis and a history of vertebral fracture (although only 86.9% of the study population actually had prevalent vertebral fractures145)

• the TROPOS study,146 a randomised, multicentre, double-blind, phase III study designed to assess the efficacy of strontium ranelate in reducing the incidence of osteoporosis-related non-vertebral fractures in postmenopausal women with osteoporosis with or without fracture.

At that time, only 3-year data were available for the SOTI and TROPOS studies.

In August 2008, the search strategy used for the previous review was rerun in MEDLINE. No new trials were identified, but four publications were identified that presented new data from two of the three studies previously identified:

• 5-year results from the TROPOS study147

• quality of life data from the SOTI study148

• 4-year vertebral fracture data from the SOTI study (presented in a review article149)

• fracture data from the TROPOS study (included in a Cochrane review150).

Marquis et al. ,148 in their article on quality of life data from the SOTI study, variously describe SOTI as a 5-year and a 3-year study, and Roux149 included in his review 4-year data from the SOTI study. However, Roux did not reference the source of those data and we have been unable to identify any publication relating specifically to the SOTI study that presents data relating to efficacy at 4 or 5 years.

The additional data from the publications listed above have been incorporated into our earlier assessments of clinical effectiveness and the results relating to the licensed 2-g daily dose of strontium ranelate are summarised in the following sections.

Strontium ranelate: fracture data

Vertebral fracture

All three studies reported only those fractures that occurred in previously intact vertebrae. In the TROPOS study,146 vertebral radiographs were not mandatory; although they were taken in as many patients as possible, baseline and follow-up radiographs were available for only 71% of the study population.

Meta-analysis of the 3-year fracture data from the SOTI and TROPOS studies found a RR of radiographic fracture over that period of 0.63 (95% CI 0.56 to 0.71) (Figure 34). It was not possible to include in the meta-analysis the results of the STRATOS study or the 5-year data from the TROPOS study because of the way that the data were published. However, the 5-year RR calculated by the authors for the TROPOS study (0.76, 95% CI 0.65 to 0.88) is not inconsistent with the 3-year results, although the point estimate is rather less favourable to strontium ranelate.

FIGURE 34.

Strontium ranelate: incident radiographic vertebral fracture.

The pooled data from the SOTI and TROPOS studies suggest that it would be necessary to treat 13 women for 3 years to avoid a radiographic vertebral fracture (95% CI 10 to17). However, because the number needed to treat is related to the absolute rather than the relative risk of fracture, the number needed to treat for 3 years is noticeably lower in the SOTI study than in the TROPOS study (9 versus 16), even though the relative risk of fracture is very similar in both studies. This is because the absolute risk of a fracture is higher in the SOTI study, which set out to recruit only participants with osteoporosis with previous fracture; the 3-year radiographic fracture rate in the placebo group in the SOTI study is 30.7%, whereas in the TROPOS study, which recruited participants with osteoporosis with or without previous fracture, it is 17.6% (Table 41).

TABLE 41 - Strontium ranelate: vertebral fracture data

SR, strontium ranelate.

PJ Meunier, Hôpital Edouard Herriot, Lyon, 11 March 2005, personal communication.

Study	Fracture definition	Number in each group suffering vertebral fracture	Number needed to treat for a given period to avoid an event (95% CI)
STRATOS143	A decrease of at least 20% in one of the ratios of vertebral height	Radiographic fracture, 2 years: SR: 42.0%; placebo: 54.7% RR (calculated by study investigators): 0.77 (95% CI 0.54 to 1.09)	Not calculable
SOTI144	Semiquantitative (method of Genant et al.53)	Radiographic fracture, 3 years: SR: 139/719 (19.3%); placebo: 222/723a (30.7%); RR 0.63 (95% CI 0.52 to 0.76), p < 0.0001 Clinical fracture, 3 years: SR: 75/719 (10.4%); placebo: 117/723a (16.2%); RR 0.64 (95% CI 0.49 to 0.85), p < 0.001 Unspecified fracture, 4 years (author’s calculation):149 RR 0.67 (95% CI 0.53 to 0.81), p < 0.001	Radiographic fracture, 3 years: 9 (6 to 14) Clinical fracture, 3 years: 18 (11 to 44) Unspecified fracture, 4 years: not calculable
TROPOS146	Semiquantitative (method of Genant et al.53)	3 years:150 SR: 202/1817 (11.1%); placebo: 321/1823 (17.6%); RR 0.63 (95% CI 0.54 to 0.74) 5 years:147 SR: 307/? (20.8%); placebo: 384/? (24.9%); RR 0.76 (95% CI 0.65 to 0.88), p < 0.001	3 years: 16 (11 to 24) 5 years: not calculable

Non-vertebral fracture

All three studies reported non-vertebral fractures, although only the SOTI and TROPOS studies presented the data in such a way as to enable them to be included in a meta-analysis (Table 42). Meta-analysis of the 3-year data from these studies indicated that strontium ranelate was associated with a RR of any non-vertebral fracture of 0.86 (95% CI 0.75 to 0.98, p = 0.03) (Figure 35), with the number needed to treat for 3 years to avoid an event being 58 (95% CI 31 to 471).

TABLE 42 - Strontium ranelate: all non-vertebral fractures

95% CI not calculated because the 95% CI for the absolute risk reduction extends from a negative number (indicating that treatment may be harmful) to a positive number (indicating that treatment may be beneficial).

Study	Number in each group suffering non-vertebral fracture	Number needed to treat for 3 years to avoid an event (95% CI)
STRATOS143	SR 2 g: 9.2%; placebo: 7.7% As the number of women in each group was not stated, it was not possible to calculate the RR, nor was this reported by the study investigators	Not calculable
SOTI144	All non-vertebral fractures: SR: 112/826; placebo: 122/814; RR 0.90 (95% CI 0.71 to 1.15), p = 0.41	71a
TROPOS146	3 years:146 SR: 233/2479; placebo: 276/2453; RR 0.84 (95% CI 0.71–0.99) 5 years:147 SR: 312/2479; placebo: 359/2456; RR 0.86 (95% CI 0.75 to 0.99), p = 0.04	3 years: 54 (28–647) 5 years: 50 (25–839)

FIGURE 35.

Strontium ranelate: all incident non-vertebral fractures.

Hip, wrist and other non-vertebral fractures

None of the studies was powered to identify a statistically significant difference in the incidence of fracture at any specific peripheral fracture site, and none reported a significant reduction in hip or wrist fracture in relation to its full intention to treat population (Tables 43 and 44).

TABLE 43 - Strontium ranelate in postmenopausal osteoporosis or osteopenia: hip fracture data

95% CI not calculated because the 95% CI for the absolute risk reduction extends from a negative number (indicating that treatment may be harmful) to a positive number (indicating that treatment may be beneficial).

Study	Number of women in each group suffering hip fracture	Number needed to treat for 3 years to avoid an event (95% CI)
STRATOS143	Not reported	–
SOTI144	Not reported	–
TROPOS147	3-year data (actual numbers not reported):150 RR 0.85 (95% CI 0.61 to 1.19) 5-year data:147 SR: 88/2479; placebo: 98/2456; RR 0.89 (95% CI 0.67 to 1.18)	3 years: not calculable 5 years: 228a

TABLE 44 - Strontium ranelate in postmenopausal osteoporosis or osteopenia: wrist fracture data

95% CI not calculated because the 95% CI for the absolute risk reduction extends from a negative number (indicating that treatment may be harmful) to a positive number (indicating that treatment may be beneficial).

Study	Number of women in each group suffering wrist fracture	Number needed to treat for 3 years to avoid an event (95% CI)
STRATOS143	Not reported	–
SOTI144	Not reported	–
TROPOS147	5-year data:147 SR: 86/2479; placebo: 87/2456; RR 0.98 (95% CI 0.73 to 1.31), p = 0.89	1367a

TABLE 45 - Strontium ranelate in postmenopausal osteoporosis or osteopenia: humerus fracture data

Study	Number of women in each group suffering humerus fracture	Number needed to treat for 3 years to avoid an event (95% CI)
STRATOS143	Not reported	–
SOTI144	Not reported	–
TROPOS147	5-year data:147 SR: 26/2479; placebo: 43/2456; RR 0.60 (95% CI 0.37–0.97), p = 0.04	143 (74–2157)

Adverse effects

In the STRATOS, SOTI and TROPOS studies, a 2-g daily dose of strontium ranelate was not associated with a statistically significant increase in all-cause mortality (RR 0.99, 95% CI 0.64 to 1.53). 150 However, there was an increased death rate due to cardiac disorders in patients receiving active treatment during the first year of therapy, but not thereafter, and deaths that could be related to thrombosis/embolism (including pulmonary embolism, cerebrovascular accident and intestinal infarction) were also more common in patients receiving active treatment. 151 The 2-g dose of strontium ranelate was not associated with a statistically significant increase in serious adverse events (RR 1.01, 95% CI 0.92 to 1.09), although it was associated with an increased risk of diarrhoea (RR 1.38, 95% CI 1.02 to 1.87). 150 Patients in the SOTI and TROPOS studies who discontinued study therapy because of adverse events did so mainly because of nausea, but diarrhoea was also associated with a statistically significant increase in the likelihood of discontinuing therapy. 151

Postmarketing surveillance has identified isolated cases of hypersensitivity syndrome. This syndrome is very rare (16 cases in 570,000 patient-years of treatment) and typically occurs within 2–6 weeks of initiating therapy, generally resolving upon discontinuation. However, it is potentially fatal. 149

Quality of life

The SOTI and TROPOS studies both recorded health-related quality of life every 6 months using the Short-Form 36 (SF-36) health questionnaire; the SOTI study also used the QUALIOST questionnaire. 152 No quality of life results have been published for the TROPOS study. In the SOTI study strontium ranelate therapy was associated with a slight improvement and placebo with a slight deterioration in quality of life as assessed by the QUALIOST specific scale; the difference between the groups, although small, was statistically significant. No significant differences were seen between the strontium ranelate and placebo groups on the SF-36 before imputation of missing data, and after imputation of missing data only the general health perception score was significantly better in the strontium ranelate group than in the placebo group. Strontium ranelate was also associated with a 31% reduction relative to placebo in the number of patients reporting back pain (p = 0.023). 148

Continuance and compliance

All three studies presented data relating to compliance (Table 46).

TABLE 46 - Compliance with study treatment

Study	Definition of compliance	How measured	Compliance
STRATOS143	Not given	Unused tablets returned at study visits Drug concentrations	Mean global compliance: 93 ± 13%; said to be no relevant differences between groups
SOTI144	Not given	Not reported	Number compliant in each group: strontium ranelate: 83%; placebo: 85%
TROPOS147	Percentage of sachets of study medication given to the patient that were actually taken	Unused sachets of study medication returned at each 6-monthly visit	Global compliance: 81.6%

All three studies provided information on the proportion of participants who completed follow-up (see Table 47). Although it is clear that, in the STRATOS study, this figure represents the proportion who continued to take the study medication for the length of the study period, it is not clear what proportion of participants who completed follow-up in the SOTI and TROPOS studies were still taking the study medication at the end of the study period. However, the publication of the 5-year TROPOS data147 gives the mean overall length of exposure to study medication, slightly over 3 years (Table 47). In total, 19% of participants were said to have discontinued study medication or withdrawn from the study because of adverse events, 21% for non-medical reasons, 0.2% because of protocol deviations and 0.3% because of aggravated osteoporosis. 147

TABLE 47 - Proportion of participants completing study

Study	Proportion of participants completing study protocol
STRATOS143	Proportion of participants completing study protocol (2 years): SR 0.5 g: 77%; SR 1 g: 73%; SR 2 g: 77%; placebo: 81%
SOTI144	Proportion of participants completing follow-up at 3 years: SR 2 g: 76%; placebo: 77%
TROPOS146	Proportion of participants completing follow-up at 3 years: SR 2 g: 66%; placebo: 64% Proportion of participants completing follow-up at 5 years:147 SR 2 g: 1384/2554 (54%); placebo: 1330/2537 (52%) Mean duration of exposure to randomised treatment:147 1126 ± 668 days

Appendix 10 The second-generation bisphosphonates alendronate and risedronate for the prevention of osteoporotic fracture in postmenopausal women

A systematic review of the clinical effectiveness of the second-generation bisphosphonates alendronate and risedronate for the prevention of osteoporotic fragility fractures in postmenopausal women was undertaken in 2007 on behalf of the NCCHTA. This review updated that carried out for NICE in 2003. 2

In total, 23 randomised controlled studies were identified that met the inclusion criteria of the 2007 review. These compared alendronate or risedronate (or, in the case of Hosking et al. ,153 either) with either placebo or no treatment in postmenopausal women with osteoporosis or osteopenia, and reported fracture outcomes. The studies were as follows:

• alendronate:

  1. – Adami et al. 154

  2. – Bone et al. 155

  3. – Bone et al. 156

  4. – Carfora et al. 157

  5. – Chesnut et al. 158

  6. – Durson et al. 159

  7. – the Fracture Intervention Trial (FIT) fracture arm79

  8. – the Fracture Intervention Trial (FIT) non-fracture arm80

  9. – the FLEX trial160

  10. – Greenspan et al. 161

  11. – Hosking et al. 153

  12. – Kaadan162

  13. – Liberman et al. 163

  14. – Lindsay et al. 164

  15. – Or et al. 165

  16. – Pols et al. 166

  17. – Rossini et al. 167

• risedronate:

  1. – Clemmesen et al. 168

  2. – Fogelman et al. 169

  3. – Harris et al. 170

  4. – Hosking et al. 153

  5. – McClung et al. 171

  6. – McClung et al. 172 (the Hip Intervention Program)

  7. – Reginster et al. 173

All studies except the FLEX trial appear to have evaluated the efficacy of alendronate or risedronate in women who were essentially bisphosphonate naive. The FLEX trial evaluated the efficacy of a further 5 years of alendronate therapy in participants in the FIT study who had already received alendronate for at least 3, and for a mean of 5, years during and after that study. 160 It therefore did not seem appropriate to include the results of the FLEX trial in meta-analyses of studies that recruited bisphosphonate-naive women.

All studies of alendronate except that by Hosking et al. 153 used a daily dose. Hosking et al. compared weekly alendronate both with placebo and with daily risedronate.

A number of articles presented data relating to extensions of earlier studies. These were:

• a 4-year extension174 followed by a 3-year extension175,176 of the study by Liberman et al. ;163 only 247 of the original 994 participants (25%) took part in the final extension study

• a 2-year extension of the study by Harris et al. ,170 open only to women who had both successfully completed the original study and undergone iliac crest bone biopsies at baseline and 36 months;177 only 86 of the 2458 original participants (3%) took part in the extension

• two extensions178,179 of the study by Reginster et al. ;173 these included only 265 (33%) and 164 (20%), respectively, of the 814 women randomised in the original study to either placebo or a 5-mg dose of risedronate.

These extension studies were not included in the 2007 review because the attrition rates were such that they could no longer be considered truly randomised.

Details of the included studies are summarised in Table 48.

Nine studies154–156,159,161,164,167,168,173 stated that they recruited women with osteoporosis but used definitions that included women who, by the current World Health Organization (WHO) definition, had osteopenia rather than osteoporosis; they are therefore classified as such in Table 48. Harris et al. 170 sought to include women who had either two or more vertebral fractures, regardless of T-score, or one vertebral fracture and a T-score at the lumbar spine of –2 or below; however, at baseline, only 79% of the placebo group, 80% of the 5-mg risedronate group and 85% of the 2.5-mg group had prevalent vertebral fractures, and it is therefore possible that some of the participants had osteopenia without prevalent fracture. Kaadan162 and Or et al. 165 did not specify the definition of osteoporosis used.

The 2001 study by McClung et al. 172 was designed specifically to study the effect of risedronate on the risk of hip fracture in elderly women with osteoporosis or other risk factors for hip fracture. Two distinct groups were recruited: women aged 70–79 years with osteoporosis, and women aged 80 years or older with either at least one non-skeletal risk factor for hip fracture or osteoporosis. Each group was randomised separately to treatment, and the proportion of younger and older women with various risk factors was said to be balanced among the treatment groups. Only 16% of the older stratum was recruited on the basis of low femoral neck BMD; 58% were recruited solely on the basis of clinical risk factors such as a recent fall-related injury. In total, 39% of the younger stratum had evidence of at least one vertebral fracture at baseline. 172

Alendronate and risedronate: fracture data

Vertebral fracture

Eleven studies of alendronate and four studies of risedronate provided information on the incidence of radiographic vertebral fractures. However, only seven of the alendronate studies and three of the risedronate studies used throughout a dose currently licensed in the UK for the treatment of postmenopausal osteoporosis (i.e. 10 mg/day or 70 mg/week of alendronate or 5 mg/day of risedronate), and two of those (the FLEX study160 and Liberman et al. 163) used a range of doses that included a licensed dose but only presented pooled data relating to all doses (Table 49). In the Fracture Intervention Trial,79,80 the alendronate dose was increased from 5 mg to 10 mg after 2 years.

TABLE 49 - Alendronate in postmenopausal osteoporosis or osteopenia: comparisons with placebo or no treatment – radiographic vertebral fracture data

Study	Bisphosphonate dose	Definition of morphometric fracture	Number of women in each group suffering vertebral fracture
Alendronate
Adami 1995154	10 and 20 mg/day	Not applicable	Clinical fracture data only presented; site not specified
Bone 1997155	1, 2.5 and 5 mg/day	20%	Alendronate 1 mg: 4 Alendronate 2.5 mg: 3 Alendronate 5 mg: 4 Placebo: 6 RRs not calculable as denominators not available; difference between groups said by investigators not to be statistically significant
Bone 2000156	10 mg/day	Not applicable	Clinical fracture data only presented
Carfora 1998157	5 and 10 mg/day; 20 mg/day for 15 months/placebo for 15 months	Not given	Alendronate 5 mg: 5.88% Alendronate 10 mg: 2.94% Alendronate 20 mg: 8.82% Placebo: 11.8% RRs not calculable as the actual numbers of women suffering fracture were not stated
Chesnut 1995158	5, 10, 20 and 40 mg/day	Not given	There were no vertebral fractures in any subject
Durson 2001159	10 mg/day	20%	Alendronate: 12/38 (31.6%) Control: 14/35 (40.0%) RR 0.79 (95% CI 0.42 to 1.47)
FIT fracture arm 199679	5 mg/day, increased after 2 years to 10 mg/day	20%	Alendronate: 78/981 (8.0%) Placebo: 145/965 (15.0%) RR 0.53 (95% CI 0.41 to 0.69)
FIT non-fracture arm 199880	5 mg/day, increased after 2 years to 10 mg/day	20%	Alendronate: 43/2057 (2.1%) Placebo: 78/2077 (3.8%) RR 0.56 (95% CI 0.39 to 0.80) [The reduction in RR was significant in those women whose initial T score was –2.5 or less (RR 0.50, 95% CI 0.31 to 0.82), but not in those with initial T-scores greater than –2.5]
FLEX 2006160	5 or 10 mg/day	> 20% with a semiquantitative confirmation	Denominator not specified Alendronate: 60 (9.8%) Placebo: 46 (11.3%) RR 0.86 (95% CI 0.60 to 1.22) (authors’ calculation, adjusted for centre and risk stratum)
Greenspan 2002161	10 mg/day	Not applicable	Clinical fracture data only presented; site not specified
Hosking 2003153	70 mg/week	Not applicable	Clinical fracture data only presented; site not specified
Kaadan 2002162	10 mg/day	Not stated	Alendronate: 3.3% Placebo: 6.3% RR not calculable as the actual numbers of women suffering fracture were not stated
Liberman 1995163	5, 10 and 20 mg/day, decreased to 5 mg/day after 2 years	20%	Pooled alendronate groups: 17/526 (3.2%) Placebo: 22/355 (6.2%) RR 0.52 (95% CI 0.28 to 0.97)
Lindsay 1999164	10 mg/day	Not applicable	No symptomatic vertebral fractures were identified in either group
Or 2001165	10 mg/day	Not stated	Alendronate: 2.5% Placebo: 9.8% RR not calculable as the actual numbers of women suffering fracture were not stated
Pols 1999166	10 mg/day	Not applicable	Vertebral fractures not investigated
Rossini 1994167	20 mg/day	Not stated	No subjects suffered vertebral fracture during the study period
Risedronate
Clemmesen 1997168	2.5 mg daily or cyclically	15% or 25% (different fracture definitions used by the Danish and Belgian centres)	Gives number of vertebral fractures identified at each centre but not number of women suffering those fractures. States that there was a tendency towards a lower incidence and rate of new vertebral fractures in the group taking daily continuous risedronate, but this was not statistically significant
Fogelman 2000169	2.5 and 5 mg/day	Any vertebral height ratio below 3 SD of the mean for the study population	Risedronate 2.5 mg: 8/60 Risedronate 5 mg: 8/112 Placebo: 17/125 RR, 5 mg vs placebo, 0.53 (95% CI 0.24 to 1.17)
Harris 1999170	2.5 and 5 mg/day	15% + semiquantitative method	Risedronate 5 mg: 61/696 Placebo: 93/678 RR 0.64 (95% CI 0.47 to 0.87)
Hosking 2003153	5 mg/day	Not applicable	Clinical fracture data only presented; site not specified
McClung 1998171	2.5 and 5 mg/day	Not reported	Not reported
McClung 2001172	2.5 and 5 mg/day	Not reported	Not reported
Reginster 2000173	2.5 and 5 mg/day	15% + semiquantitative method	Risedronate 5 mg: 53/344 Placebo: 89/346 RR 0.60 (95% CI 0.44 to 0.81)

Ideally, only data relating to the licensed doses of alendronate and risedronate would have been included in the meta-analysis of vertebral fracture data. However, only one study, the small 1-year study by Durson et al. ,159 provided usable data relating to alendronate taken at a licensed dose throughout the study, and it did not produce a statistically significant result. Despite using a dose of only 5 mg for the first 2 years, both arms of the longer, and much larger, high-quality Fracture Intervention Trial79,80 demonstrated a greater treatment effect, as did the study by Liberman et al. 163 (the latter pooled data relating to alendronate given at doses of either 5 or 10 mg for 3 years or 20 mg for 2 years followed by 5 mg for 1 year). Data from these studies have therefore been included in the meta-analysis of vertebral fracture risk, together with data from the arms of the risedronate studies by Fogelman et al. ,169 Harris et al. 170 and Reginster et al. 173 that used the licensed 5-mg dose.

The meta-analysis indicates that the second-generation bisphosphonates alendronate and risedronate are associated with a risk of vertebral fracture relative to placebo of 0.58 (95% CI 0.50 to 0.67) in women with osteoporosis or osteopenia (Figure 36).

FIGURE 36.

Alendronate or risedronate: incident vertebral fracture in postmenopausal osteoporosis (with or without previous fracture) or osteopenia.

Those studies that could not be incorporated into the meta-analysis because they only presented rates and not actual numbers of women with fracture (Carfora et al. ,157 Kaadan162 and Or et al. 165) also found that a 10-mg dose of alendronate was associated with a reduction in vertebral fracture rate (Table 49).

Non-vertebral fracture

Thirteen studies of alendronate and seven studies of risedronate presented data relating to non-vertebral fracture (Table 50).

TABLE 50 - Alendronate or risedronate in postmenopausal osteoporosis or osteopenia: comparisons with placebo or no treatment –non-vertebral fracture data

Study	Bisphosphonate dose	Fracture definition	Number of women in each group suffering non-vertebral fracture
Alendronate
Adami 1995154	10 and 20 mg/day	Not given; may include clinical vertebral fractures	Alendronate 10 mg: 1/68 (1.5%) Alendronate 20 mg: 1/72 (1.4%) Placebo: 3/71 (4.2%) RR, 10 mg vs placebo, 0.35 (95% CI 0.04 to 3.26)
Bone 1997155	1, 2.5 and 5 mg/day	Not given	Alendronate 1 mg: 15/86 (17.4%) Alendronate 2.5 mg: 9/89 (10.1%) Alendronate 5 mg: 9/93 (9.7%) Placebo: 16/91 (17.6%) RR, 5 mg vs placebo, 0.55 (95% CI 0.26 to 1.18)
Bone 2000156	10 mg/day	Any clinical fracture (most were said to be non-vertebral, occurring at sites such as foot, ankle and rib; most occurred as a result of trauma)	Alendronate 10 mg: 4/92 (4.3%) Placebo: 4/50 (8.0%) RR 0.68 (95% CI 0.19 to 2.42)
Carfora1998157	5, 10 and 20 mg/day	Not given	RR, alendronate vs placebo, 0.55 (authors’ calculation; confidence intervals and numbers of women suffering fractures not supplied)
Chesnut 1995158	5, 10, 20 and 40 mg/day	Not given	13 non-vertebral fractures occurred in 12 subjects. These were evenly distributed across treatment groups and were not considered related to therapy
FIT fracture arm 199679	5 mg/day, increased after 2 years to 10 mg/day	Any clinical non-vertebral fracture that was not pathological (e.g. due to malignant disease), due to excessive trauma or involving the face or skull	Alendronate: 122/1022 (11.9%) Placebo: 148/1005 (14.7%) RR 0.81 (95% CI 0.65 to 1.01)
FIT non-fracture arm 199880			Alendronate: 261/2214 (11.8%) Placebo: 294/2218 (13.3%) RR 0.89 (95% CI 0.76 to 1.04)
FLEX 2006160	5 or 10 mg/day	Any non-vertebral fracture other than pathological, skull and excessive trauma fractures	Alendronate: 132/662 (19.9%) Placebo: 93/437 (21.3%) RR 0.93 (95% CI 0.71 to 1.21) (authors’ calculation, adjusted for centre and risk stratum)
Greenspan 2002161	10 mg/day	Not given	Alendronate: 13 (8%) Placebo: 18 (11%) RR not calculable as the number of women in each group was not stated
Hosking 2003153	70 mg/week	Not given; may include clinical vertebral fractures	Alendronate: 6/219 (2.7%) Placebo: 2/108 (1.9%)
Liberman 1995163	5, 10 and 20 mg/day, decreased to 5 mg/day after 2 years	All symptomatic fractures, not excluding those due to trauma	Alendronate: 45/597 (7.5%) Placebo: 38/397181 (9.6%) RR 0.79 (95% CI 0.52 to 1.19)
Lindsay 1999164	10 mg/day	Any clinically apparent fracture	Alendronate: 15/214 (7.0%) Control: 9/214 (4.2%) RR 1.67 (% CI 0.75 to 3.73)
Pols 1999166	10 mg/day	Any clinically confirmed fracture	Alendronate: 19/950 (2.0%) Control: 37/958 (3.9%) RR 0.52 (95% CI 0.30 to 0.89)
Risedronate
Clemmesen 1997168	2.5 mg daily or cyclically	Not given, but all fractures occurred after falls	Continuous risedronate: 4/44 (9.1%) Cyclical risedronate: 9/44 (20.5%) Placebo: 4/44 (9.1%) RR, continuous risedronate vs placebo, 1.00 (95% CI 0.27 to 3.75)
Fogelman 2000169	2.5 and 5 mg/day	Not given	Risedronate 2.5 mg: 4/184 (2.2%) Risedronate 5 mg: 7/177 (4.0%) Placebo: 13/180 (7.2%) RR, 5 mg vs placebo, 0.55 (95% CI 0.22 to 1.34)
Harris 1999170	2.5 and 5 mg/day	All radiographically confirmed fractures of the clavicle, humerus, wrist, pelvis, hip or leg, whether or not associated with trauma	Risedronate 5 mg: 33/812 (4.1%) Placebo: 52/815 (6.4%) RR 0.64 (95% CI 0.42 to 0.97)
Hosking 2003153	5 mg/day	Not given; may include clinical vertebral fractures	Risedronate: 6/222 (2.7%) Placebo: 2/108 (1.9%) RR 1.46 (95% CI 0.30 to 7.11)
McClung 1998171	2.5 and 5 mg/day	Not given	Non-vertebral fractures were said to be few in number and comparable between groups. More specific data were not available
McClung 2001172	2.5 and 5 mg/day	All radiographically confirmed fractures of the clavicle, humerus, wrist, pelvis, hip or leg	Risedronate: 583/6197 (9.4%) Placebo: 351/31342 (11.2%) RR 0.84 (95% CI 0.74 to 0.95)
Reginster 2000173	2.5 and 5 mg/day	All radiographically confirmed fractures of the clavicle, humerus, wrist, pelvis, hip or leg, whether or not associated with trauma	Risedronate 5 mg: 36/406 (8.9%) Placebo: 51/406 (12.6%) RR 0.71 (95% CI 0.47 to 1.06)

As in the case of vertebral fractures, ideally only data relating to the current licensed doses of alendronate and risedronate would have been included in the meta-analysis. However, as before, data from the FIT fracture and non-fracture arms and from the study by Liberman et al. have been included. In addition, data have been included from one large risedronate study (McClung et al. 172), which only reported pooled data relating to participants receiving 2.5-mg and 5-mg doses of risedronate. Their reason for doing so was that the study by Reginster et al. 173 had shown that both doses were effective in reducing the risk of vertebral fractures. However, Reginster et al. 173 discontinued the 2.5-mg dose after 2 years on the basis that McClung et al. 171 had shown that the 5-mg dose had a more consistent effect on BMD and a similar safety profile, and both they and Harris et al. 170 published only 1-year data relating to the 2.5-mg dose and 3-year data relating only to the 5-mg dose. Consequently, the meta-analysis includes the 3-year data from the studies by Harris et al. 170 and Reginster et al. ,173 which relate to the licensed 5-mg dose, and does not include data relating to the 2.5-mg dose either from those studies or from the study by Clemmesen et al. 168

The meta-analysis indicates that the second-generation bisphosphonates alendronate and risedronate are associated with a risk of non-vertebral fracture relative to placebo of 0.82 (95% CI 0.74 to 0.90) in women with osteoporosis or osteopenia or at high risk of hip fracture (Figure 37).

FIGURE 37.

Alendronate or risedronate: incident non-vertebral fracture in women with postmenopausal osteoporosis (with or without previous fracture) or osteopenia or at high risk of hip fracture.

As may be seen, the results are less consistent than those for vertebral fracture. There is no apparent reason why, in the study by Lindsay et al. ,164 the fracture rate should have been higher in women receiving HRT plus alendronate than in those receiving alendronate alone. In the study by McClung et al. ,172 risedronate appears less effective (RR 0.84, 95% CI 0.74 to 0.96) than is suggested by meta-analysis of data from the studies by Fogelman et al. ,169 Harris et al. 170 and Reginster et al. 173 relating to the 5-mg risedronate dose in women with osteoporosis or osteopenia, with or without previous fracture (RR 0.66, 95% CI 0.50 to 0.87). This finding may be spurious, as the confidence intervals around the two point estimates overlap. Alternatively, in the study by McClung et al. 172 the estimated efficacy of risedronate may have been reduced by the inclusion of data relating either to women receiving a 2.5-mg dose of risedronate or to women selected for risk factors other than low BMD, or both. However, although McClung et al. 172 did not publish the data for all non-vertebral fractures in such a way as to enable the inclusion in the meta-analysis of only those women selected for low BMD (or indeed only those receiving a 5-mg dose), they did publish an analysis which indicated that the RR of non-vertebral fracture in women in the younger osteoporotic stratum who received risedronate was not very different, at 0.8 (95% CI 0.7 to 1.0), from that in the study as a whole.

Subgroup data from the FIT non-fracture arm80 suggest that alendronate may have a significant effect on non-vertebral fractures in osteoporotic women (RR 0.64, 95% CI 0.58 to 0.82), but not in those who are only osteopenic (RR 1.08, 95% CI 0.87 to 1.35).

Hip fracture

Few studies reported specifically on hip fracture (Table 51).

TABLE 51 - Alendronate or risedronate in postmenopausal osteoporosis or osteopenia: comparisons with placebo or no treatment – hip fracture data

Study	Bisphosphonate dose	Number of women in each group suffering hip fracture
Alendronate
FIT fracture arm 199679	5 mg/day, increased after 2 years to 10 mg/day	Alendronate: 11/1022 (1.1%) Placebo: 22/1005 (2.2%) RR 0.49 (95% CI 0.24 to 1.01)
FIT non-fracture arm 199880	5 mg/day, increased after 2 years to 10 mg/day	Alendronate: 19/2214 (0.9%) Placebo: 24/2218 (1.1%) RR 0.79 (95% CI 0.44 to 1.44)
FLEX 2006160	5 or 10 mg/day	Alendronate: 20/662 (3.0%) Placebo: 13/437 (3.0%) RR 1.02 (95% CI 0.51 to 2.10) (authors’ calculation, adjusted for centre and risk stratum)
Greenspan 2002161	10 mg/day	Alendronate: 2 Placebo: 4 As the number of women in each group was not stated, it was not possible to calculate a RR
Liberman 1995163	5, 10 and 20 mg/day, decreased to 5 mg/day after 2 years	Alendronate: 1/597 (0.2%) Placebo: 3/397182 (0.8%) RR 0.22 (95% CI 0.02 to 2.12)
Risedronate
Harris 1999170	2.5 and 5 mg/day	Risedronate 5 mg: 12/812 (1.5%) Placebo: 15/815 (1.8%) RR 0.80 (95% CI 0.38 to 1.70)
McClung 2001172	2.5 and 5 mg/day	Risedronate: 137/6197 (2.2%) Placebo: 95/3134 (3.0%) RR 0.73 (95% CI 0.56 to 0.94) Younger osteoporotic group: Risedronate: 55/3624 (1.5%) Placebo: 46/1821 (2.5%) RR 0.60 (95% CI 0.41 to 0.89) Older high-risk group: Risedronate: 82/2573 (3.2%) Placebo: 49/1313 (3.7%) RR 0.85 (95% CI 0.60 to 1.21)
Reginster 2000173	2.5 and 5 mg/day	Risedronate 5 mg: 9/406 (2.2%) Placebo: 11/406 (2.7%) RR 0.82 (95% CI 0.34 to 1.95)

Studies were included in the meta-analysis on the same basis as for the meta-analysis of all non-vertebral fractures. Pooled data for all participants for both the 2.5-mg and 5-mg doses of risedronate from the 2001 study by McClung et al. 172 were again included as usable data were not provided separately for the two doses. However, the authors calculated that, in the younger osteoporotic stratum, the risk of hip fracture relative to placebo was 0.5 (95% CI 0.3 to 0.9) in women receiving 2.5 mg of risedronate and 0.7 (95% CI 0.4 to 1.1) in those receiving 5 mg, suggesting that the higher dose did not confer increased protection.

The meta-analysis indicates that the second-generation bisphosphonates alendronate and risedronate are associated with a risk of hip fracture relative to placebo of 0.72 (95% CI 0.58 to 0.88) in women with osteoporosis (with or without previous fracture) or osteopenia or at high risk of hip fracture (Figure 38).

FIGURE 38.

Alendronate or risedronate: incident hip fracture in women with postmenopausal osteoporosis (with or without previous fracture) or osteopenia or at high risk of hip fracture.

Adverse events

The studies included in our 2003 review2 generally did not identify an increase in adverse events in women taking bisphosphonates. Moreover, Steinbuch et al. 181 undertook a retrospective cohort study in which they used routinely collected data to compare mortality rates in women enrolled in the risedronate and placebo groups of the North American risedronate trials (Harris et al. ,170 McClung et al. 171 and the Hip Intervention Program172) over the period from initiation of study medication to 31 December 1997. They found no association between risedronate and an increase in all-cause mortality, cancer mortality or stroke mortality; although risedronate was associated with a non-significant increase in mortality from coronary artery disease, it was associated with a significant decrease in stroke mortality.

The more recent studies summarised here do not suggest that second-generation bisphosphonates are associated with an increase in adverse events other than gastrointestinal disturbances. Hosking et al. 153 found that, although there was no significant increase in serious upper gastrointestinal adverse events in women randomised to weekly alendronate or daily risedronate compared with those randomised to placebo, women randomised to either bisphosphonate were more likely than those randomised to placebo to discontinue study medication because of upper gastrointestinal adverse events. Or et al. 165 also found that gastrointestinal disturbances were more common in the alendronate than in the placebo group.

A systematic review183 has indicated that bisphosphonates are associated with an increased risk of osteonecrosis of the jaw. Although this risk appears to be linked primarily to intravenous bisphosphonates, some cases were reported in patients taking daily alendronate (10 mg) for osteoporosis. However, no cases of osteonecrosis of the jaw were identified in the long-term users of oral alendronate in the FLEX study. 160

Quality of life

Only one study, that by Durson et al. ,159 set out to measure the effect of alendronate treatment on health-related quality of life, as measured by the Nottingham Health Profile (NHP). At 12 months they found statistically significant improvements in the NHP scores for pain, social isolation, energy level and physical ability in the alendronate group, but not in the control group; pain, as measured on a visual analogue scale, decreased significantly from baseline in the alendronate group but not in the control group. Or et al. 165 noted no obvious improvement in quality of life in the alendronate group compared with the placebo group, but did not state how this was measured.

The FIT fracture arm collected data on the effects of alendronate on back pain and days of functional limitation or bed rest. 184 Women in the alendronate group had significantly fewer days in bed because of back pain than women in the placebo group (mean of 1.9 days over a 3-year period versus 5.1 days, p = 0.001), and fewer days of limited activity because of such pain (mean of 61.8 days versus 73.2, p = 0.04).

None of the studies of risedronate reported on its effect on quality of life.

Continuance and compliance

Continuance with medication decreases over time. In the studies reviewed here, the percentage of subjects receiving daily alendronate who completed the study protocol ranged from 100% at 1 year in the very small study by Rossini et al. to 81% at 4 years in the FIT non-fracture arm and 72% at 6 years in the FLEX trial; surprisingly, the figure for weekly alendronate was lower, at 78% at 3 months (Table 52). The figures for daily risedronate ranged from 82% at 1 year to 60% at 3 years (Table 52). The FIT trial found that discontinuation of study medication was greatest in the first month post randomisation: 4.8% of participants had withdrawn at 3 months, and 11.1% at 12 months. Although clinical adverse events formed the most common reason for withdrawal, causing 6.9% of women to withdraw, the proportion of women discontinuing treatment was comparable in the alendronate and placebo groups (RR 2.10, 95% CI 1.47 to 2.99). 185

TABLE 52 - Continuance in randomised controlled trials: percentage of patients receiving a licensed bisphosphonate dose still taking bisphosphonate therapy by year

NR, not reported.

6 months of blinded treatment followed by 6 months of blinded follow-up.

First 12 weeks only.

Study	Year 1	Year 2	Year 3	Year 4	Year 5	Year 6
Alendronate 10 mg/day
Bone 2000156	NR	74
FIT fracture arm 199679 (5 mg/day, increased to 10 mg/day after 2 years)	NR	NR	89
FIT non-fracture arm 199880 (5 mg/day, increased to 10 mg/day after 2 years)	NR	NR	NR	81
FLEX 2006160	NR	NR	86	NR	NR	72
Liberman 1995163	92	89	84
Lindsay 1999164	95
Pols 1999166	88
Rossini 1994167	100a
Alendronate 70 mg/week
Hosking 2003153	78b
Risedronate 5 mg/week
Fogelman 2000169	NR	78
Harris 1999170	NR	NR	60
Hosking 2003153	80b
Reginster 2000173	82	NR	62

Four studies defined compliance as taking at least 75% of the study medication since the last clinic visit. In both arms of the FIT trial,79,80 96% of participants who continued to take the study medication (alendronate or placebo) were found to be compliant with that medication at the final clinic visit, as were 96% of participants in the FLEX trial160 who were still taking the study medication at 36 months. Hosking et al. 153 also found that, although in their study continuance at 12 weeks was perhaps disappointing, 95% of those taking either bisphosphonate were compliant with that medication over that period according to patient-completed medication diaries validated by tablet counts.

Summary

The aggregated results suggest that the second-generation bisphosphonates alendronate and risedronate have a protective effect in relation to vertebral fracture in women with osteoporosis or osteopenia, with or without previous fracture, and that they also have a protective effect in relation to non-vertebral fracture generally, and hip fracture specifically, in women with osteoporosis or osteopenia, with or without previous fracture, or who are at increased risk of hip fracture.

Appendix 11 The calculation of the costs of fracture using Healthcare Resource Groups (HRGs)

Healthcare Resource Groups (HRGs) detail the costs that are expected to be incurred by a trust when treating a patient with a certain condition. These costs can be modified if the patient has an exceptionally long duration of stay, which is defined as beyond the ‘trim point’, with additional costs per day after this period. These costs have been centrally calculated, across a large number of NHS trusts, and this is the approach recommended by NICE for calculating costs.

The HRGs used in the estimation of costs are shown in the following table.

The costs estimated for a hip, clinical vertebral, wrist and proximal humerus fracture have been provided.

Based on the work previously undertaken it has been assumed that:

• the costs for a hip fracture will also incorporate pelvis and other femoral fractures

• the costs for a wrist fracture will also incorporate rib, scapula, sternum and clavicle fractures

• the costs for a proximal humerus fracture will also incorporate tibia, fibula and humeral shaft fractures.

Appendix 12 Detailed results

This appendix will be subdivided into results for osteoporotic women without a previous fracture and results for osteoporotic women with a previous fracture. These results will be reported by combinations of age and T-score.

Results for women without a previous fracture

Women aged 50–54 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 50–54 years with a T-score of –2.5 SD are given in Table 53.

TABLE 53 - Cost-effectiveness of interventions in women aged 50–54 years with a T-score of –2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	25,985	1.71	15,239	8804 to 40,684
Alendronate	23,947	1.36	17,653	12,091 to 28,137
Risedronate	125,793	1.36	92,727	73,093 to 126,186
Strontium ranelate	164,317	0.90	182,942	115,564 to 540,523
Vitamin K1b	40,263	0.47	84,842	54,355 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 39. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 39.

Cost-effectiveness acceptability curves for interventions in women aged 50–54 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 50–54 years with a T-score of –3.0 SD are given in Table 54.

TABLE 54 - Cost-effectiveness of interventions in women aged 50–54 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	14,825	2.60	5694	895 to 26,623
Alendronate	15,171	2.14	7097	2691 to 15,514
Risedronate	116,966	2.14	54,728	41,069 to 81,612
Strontium ranelate	160,368	1.35	118,627	68,244 to 648,233
Vitamin K1b	40,430	0.60	67,481	43,340 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 40. However, it is seen that if the efficacy associated with vitamin K₁ is only applicable to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 40.

Cost-effectiveness acceptability curves for interventions in women aged 50–54 years with a T-score of –3.0 SD and no previous fracture.

Women aged 55–59 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 55–59 years with a T-score of –2.5 SD are given in Table 55.

TABLE 55 - Cost-effectiveness of interventions in women aged 55–59 years with a T-score of –2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	23,594	1.90	12,407	6825 to 34,856
Alendronate	20,879	1.47	14,185	9412 to 23,095
Risedronate	122,721	1.47	83,375	65,777 to 112,756
Strontium ranelate	161,455	0.99	163,585	104,850 to 450,867
Vitamin K1b	38,674	0.50	77,149	38,256 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 41. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 41.

Cost-effectiveness acceptability curves for interventions in women aged 55–59 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 55–59 years with a T-score of –3.0 SD are given in Table 56.

TABLE 56 - Cost-effectiveness of interventions in women aged 55–59 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	11,583	2.84	4075	Dominating to 21,912
Alendronate	10,622	2.26	4700	909 to 11,670
Risedronate	112,442	2.26	49,750	37,506 to 73,282
Strontium ranelate	155,739	1.46	106,789	62,990 to 467,595
Vitamin K1b	37,512	0.63	59,266	38,256 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 42. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 42.

Cost-effectiveness acceptability curves for interventions in women aged 55–59 years with a T-score of –3.0 SD and no previous fracture.

Women aged 60–64 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 60–64 years with a T-score of –2.5 SD are given in Table 57.

TABLE 57 - Cost-effectiveness of interventions in women aged 60–64 years with a T-score of –2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	22,238	2.17	10,235	4973 to 31,641
Alendronate	20,846	1.70	12,270	7731 to 20,594
Risedronate	122,683	1.70	72,209	56,644 to 98,655
Strontium ranelate	162,628	1.14	142,055	90,743 to 386,976
Vitamin K1b	40,572	0.56	72,608	46,719 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 43. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 43.

Cost-effectiveness acceptability curves for interventions in women aged 60–64 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 60–64 years with a T-score of –3.0 SD are given in Table 58.

TABLE 58 - Cost-effectiveness of interventions in women aged 60–64 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	10,354	3.19	3244	Dominating to 20,762
Alendronate	11,522	2.55	4517	691 to 11,443
Risedronate	113,337	2.55	44,435	33,317 to 65,618
Strontium ranelate	158,284	1.66	95,172	55,947 to 383,786
Vitamin K1b	40,771	0.71	57,815	37,307 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 44. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 44.

Cost-effectiveness acceptability curves for interventions in women aged 60–64 years with a T-score of –3.0 SD and no previous fracture.

Women aged 65–69 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 65–69 years with a T-score of –2.5 SD are given in Table 59.

TABLE 59 - Cost-effectiveness of interventions in women aged 65–69 years with a T-score of –2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	16,300	2.43	6719	1689 to 28,573
Alendronate	16,306	1.95	8349	3816 to 16,248
Risedronate	118,135	1.95	60,492	46,580 to 85,872
Strontium ranelate	160,538	1.31	122,893	75,829 to 370,798
Vitamin K1b	40,351	0.52	77,817	49,921 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 45. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 45.

Cost-effectiveness acceptability curves for interventions in women aged 65–69 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 65–69 years with a T-score of –3.0 SD are given in Table 60.

TABLE 60 - Cost-effectiveness of interventions in women aged 65–69 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	1309	3.55	368	Dominating to 21,912
Alendronate	4641	2.91	1596	Dominating to 14,305
Risedronate	106,444	2.91	36,613	26,528 to 55,835
Strontium ranelate	155,174	1.89	82,245	45,579 to 380,423
Vitamin K1b	40,488	0.65	61,838	39,753 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 46. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 46.

Cost-effectiveness acceptability curves for interventions in women aged 65–69 years with a T-score of –3.0 SD and no previous fracture.

Women aged 70–74 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 70–74 years with a T-score of –2.5 SD are given in Table 61.

TABLE 61 - Cost-effectiveness of interventions in women aged 70–74 years with a T-score of –2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

Dominating denotes more health provided at a lower or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	11,426	2.78	4113	Dominating to 25,268
Alendronate	12,560	2.23	5635	1335 to 13,269
Risedronate	114,683	2.23	51,319	38,900 to 74,558
Strontium ranelate	158,787	1.48	107,188	63,930 to 354,275
Vitamin K1b	41,571	0.57	73,312	47,802 to dominating

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 47. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 47.

Cost-effectiveness acceptability curves for interventions in women aged 70–74 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 70–74 years with a T-score of –3.0 SD are given in Table 62.

TABLE 62 - Cost-effectiveness of interventions in women aged 70–74 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–6325	4.05	Dominating	Dominating to 14,915
Alendronate	–933	3.28	Dominating	Dominating to 6350
Risedronate	100,860	3.28	30,714	21,512 to 48,318
Strontium ranelate	152,594	2.12	71,955	38,418 to 367,355
Vitamin K1b	42,007	0.72	58,702	38,495 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 48. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 48.

Cost-effectiveness acceptability curves for interventions in women aged 70–74 years with a T-score of –3.0 SD and no previous fracture.

Women aged 75–79 years without previous fracture

The mean results for each intervention compared with no treatment for women aged 75–79 years with a T-score of –2.5 SD are given in Table 63.

TABLE 63 - Cost-effectiveness of interventions in women aged 75–79 years with a T-score of -2.5 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–2020	3.48	Dominating	Dominating to 17,433
Alendronate	3383	2.76	1226	Dominating to 8757
Risedronate	105,184	2.76	30,714	27,362 to 58,785
Strontium ranelate	154,586	1.78	71,955	47,443 to 438,311
Vitamin K1b	40,795	0.53	77,424	49,981 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 49. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 49.

Cost-effectiveness acceptability curves for interventions in women aged 75–79 years with a T-score of –2.5 SD and no previous fracture.

The mean results for each intervention compared with no treatment for women aged 75–79 years with a T-score of –3.0 SD are given in Table 64.

TABLE 64 - Cost-effectiveness of interventions in women aged 75–79 years with a T-score of –3.0 SD and no previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–25,815	5.06	Dominating	Dominating to 14,915
Alendronate	–14,001	4.02	Dominating	Dominating to 6350
Risedronate	87,759	4.02	21,807	13,434 to 38,567
Strontium ranelate	146,678	2.53	58,090	27,634 to 470,415
Vitamin K1b	41,316	0.89	46,674	38,495 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 50. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 50.

Cost-effectiveness acceptability curves for interventions in women aged 75–79 years with a T-score of –3.0 SD and no previous fracture.

Results for women with a previous fracture

Women aged 50–54 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 50–54 years with a T-score of –2.5 SD are given in Table 65.

TABLE 65 - Cost-effectiveness of interventions in women aged 50–54 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	18,911	2.52	7500	2761 to 26,132
Alendronate	18,062	2.09	8625	4720 to 15,615
Risedronate	119,902	2.09	57,255	44,474 to 79,065
Strontium ranelate	161,219	1.41	114,241	72,240 to 320,586
Vitamin K1b	40,327	0.68	59,678	38,270 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 51. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 51.

Cost-effectiveness acceptability curves for interventions in women aged 50–54 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 50–54 years with a T-score of –3.0 SD are given in Table 66.

TABLE 66 - Cost-effectiveness of interventions in women aged 50–54 years with a T-score of –3.0 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	2170	3.86	562	Dominating to 16,895
Alendronate	4898	3.26	1502	Dominating to 7812
Risedronate	106,706	3.26	32,721	23,719 to 50,372
Strontium ranelate	155,294	2.09	74,373	41,314 to 384,617
Vitamin K1b	40,579	0.85	47,537	30,596 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 52. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 52.

Cost-effectiveness acceptability curves for interventions in women aged 50–54 years with a T-score of –3.0 SD and with a previous fracture.

Women aged 55–59 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 55–59 years with a T-score of –2.5 SD are given in Table 67.

TABLE 67 - Cost-effectiveness of interventions in women aged 55–59 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	15,324	2.82	5441	1388 to 21,588
Alendronate	13,460	2.27	5935	2459 to 11,751
Risedronate	115,294	2.27	50,833	39,453 to 70,322
Strontium ranelate	156,925	1.55	101,563	65,044 to 267,052
Vitamin K1b	37,945	0.72	53,007	34,025 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 53. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 53.

Cost-effectiveness acceptability curves for interventions in women aged 55–59 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 55–59 years with a T-score of –3.0 SD are given in Table 68.

TABLE 68 - Cost-effectiveness of interventions in women aged 55–59 years with a T-score of –3.0 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–2692	4.22	Dominating	Dominating to 11,688
Alendronate	–1926	3.45	Dominating	Dominating to 4547
Risedronate	99,876	3.45	28,985	20,968 to 44,285
Strontium ranelate	148,351	2.25	65,972	37,095 to 282,316
Vitamin K1b	34,807	1.27	27,329	38,495 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 54. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 54.

Cost-effectiveness acceptability curves for interventions in women aged 55–59 years with a T-score of –3.0 SD and with a previous fracture.

Women aged 60–64 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 60–64 years with a T-score of –2.5 SD are given in Table 69.

TABLE 69 - Cost-effectiveness of interventions in women aged 60–64 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	13,290	3.20	4149	Dominating to 20,229
Alendronate	13,411	2.60	5161	1833 to 11,565
Risedronate	115,237	2.60	44,347	34,351 to 62,489
Strontium ranelate	158,684	1.77	89,445	57,137 to 246,718
Vitamin K1b	40,791	0.78	52,159	33,663 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 55. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 55.

Cost-effectiveness acceptability curves for interventions in women aged 60–64 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 60–64 years with a T-score of –3.0 SD are given in Table 70.

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 56. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 56.

Cost-effectiveness acceptability curves for interventions in women aged 60–64 years with a T-score of –3.0 SD and with a previous fracture.

Women aged 65–69 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 65–69 years with a T-score of –2.5 SD are given in Table 71.

TABLE 71 - Cost-effectiveness of interventions in women aged 65–69 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in Cost per QALY (£)
Vitamin K1	4383	3.60	1217	Dominating to 16,987
Alendronate	6600	2.99	2208	Dominating to 8117
Risedronate	108,415	2.99	36,263	27,068 to 53,153
Strontium ranelate	155,549	2.02	76,845	45,675 to 230,525
Vitamin K1b	40,458	0.74	54,549	35,050 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 57. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 57.

Cost-effectiveness acceptability curves for interventions in women aged 65–69 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 65–69 years with a T-score of –3.0 SD are given in Table 72.

TABLE 72 - Cost-effectiveness of interventions in women aged 65–69 years with a T-score of –3.0 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–18,103	5.29	Dominating	Dominating to 9277
Alendronate	–10,9877	4.42	Dominating	Dominating to 2851
Risedronate	90,879	4.42	20,576	18,617 to 33,783
Strontium ranelate	147,504	2.89	51,015	33,130 to 238,938
Vitamin K1b	40,665	0.94	43,398	27,971 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 58. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 58.

Cost-effectiveness acceptability curves for interventions in women aged 65–69 years with a T-score of –3.0 SD and with a previous fracture.

Women aged 70–74 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 70–74 years with a T-score of –2.5 SD are given in Table 73.

TABLE 73 - Cost-effectiveness of interventions in women aged 70–74 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–2927	4.11	Dominating	Dominating to 14,924
Alendronate	981	3.39	289	Dominating to 6122
Risedronate	102,787	3.39	30,307	21,865 to 45,710
Strontium ranelate	152,923	2.28	67,144	38,251 to 224,900
Vitamin K1b	42,291	0.79	53,522	35,219 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 59. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 59.

Cost-effectiveness acceptability curves for interventions in women aged 70–74 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 70–74 years with a T-score of –3.0 SD are given in Table 74.

TABLE 74 - Cost-effectiveness of interventions in women aged 70–74 years with a T-score of –3.0 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–29,554	6.00	Dominating	Dominating to 8391
Alendronate	–19,258	4.97	Dominating	Dominating to 1426
Risedronate	82,503	4.97	16,612	9843 to 29,107
Strontium ranelate	143,633	3.23	44,457	21,477 to 227,060
Vitamin K1b	42,944	1.00	43,039	28,552 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 60. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 60.

Cost-effectiveness acceptability curves for interventions in women aged 70–74 years with / T-score of –3.0 SD and with a previous fracture.

Women aged 75–79 years with a previous fracture

The mean results for each intervention compared with no treatment for women aged 75–79 years with a T-score of –2.5 SD are given in Table 75.

TABLE 75 - Cost-effectiveness of interventions in women aged 75–79 years with a T-score of –2.5 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–23,097	5.18	Dominating	Dominating to 9598
Alendronate	–12,784	4.20	Dominating	Dominating to 2716
Risedronate	88,988	4.20	21,187	13,497 to 35,546
Strontium ranelate	143,633	3.23	53,727	27,028 to 265,902
Vitamin K1b	41,125	0.75	54,524	35,361 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 61. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 61.

Cost-effectiveness acceptability curves for interventions in women aged 75–79 years with a T-score of –2.5 SD and with a previous fracture.

The mean results for each intervention compared with no treatment for women aged 75–79 years with a T-score of –3.0 SD are given in Table 76.

TABLE 76 - Cost-effectiveness of interventions in women aged 75–79 years with a T-score of –3.0 SD and with a previous fracture

Per 100 women.

Assuming no effect on hip or vertebral fractures.

Dominating denotes more health provided at a lower or equal acquisition cost.

Dominated means that less health is provided at a higher or equal acquisition cost.

	Incremental cost compared with no treatment (£)a	Incremental QALYs compared with no treatmenta	Cost per QALY compared with no treatment (£)	95% CI for range in cost per QALY (£)
Vitamin K1	–58,789	7.54	Dominating	Dominating to 6220
Alendronate	–38,860	6.09	Dominating	Dominating to dominating
Risedronate	62,851	6.09	10,314	4232 to 22,238
Strontium ranelate	134,759	3.85	35,006	13,506 to 295,993
Vitamin K1b	41,503	0.95	43,554	28,397 to dominated

It is seen that both alendronate and vitamin K₁ have relatively low cost per QALY ratios. Incremental analyses suggest that vitamin K₁ is the most cost-effective intervention. The dominance of alendronate and vitamin K₁ is shown in Figure 62. However, it is seen that if the efficacy associated with vitamin K₁ is applicable only to wrist and proximal humerus fractures and not to hip or vertebral fractures then the QALYs gained are dramatically reduced.

FIGURE 62.

Cost-effectiveness acceptability curves for interventions in women aged 75–79 years with a T-score of –3.0 SD and with a previous fracture.