Bisphosphonate alternative regimens for the prevention of osteoporotic fragility fractures: BLAST-OFF, a mixed-methods study

Eligibility

To be eligible for inclusion, studies needed to report on patients’, clinicians’, academics’ and/or manager/payers’ experiences and preferences regarding BP regimes for adults (≥ 18 years) with osteoporosis. BP needed to be mentioned by name, or there needed to be sufficient information that was specific to BP (e.g. reference to the special instructions for use of oral BP) to deduce that study findings related to BP, as agreed by two clinically experienced authors independently. Papers describing experiences of osteoporosis more generally were included if there were findings relating to BP treatment in the study abstract. Studies were only included if they were qualitative in design or mixed methods with a qualitative component, relevant to a developed country setting and written in English. Studies were excluded that involved paediatric patients, patients and clinicians receiving/recommending other treatments for osteoporosis and studies in which BP were being used for other indications (e.g. malignancy or Paget’s disease).

Search methods

Systematic searches were conducted in seven bibliographic databases {MEDLINE, EMBASE, AMED, CINAHLPlus, PsycINFO, ASSIA and Web of Science [Social Science Citation Index (SSCI) and Conference Proceedings Citation Index-Social Science and Humanities (CPCI-SSH)]} from inception to 15 July 2019. The search strategy utilised database subject headings and text word searching in title, abstract or keywords, combining terms for: (1) BP; (2) experiences and preferences and (3) qualitative research, based on DeJean et al.’s search filter. Search terms were adapted as appropriate for each database platform.

In addition, grey literature was searched [DART Europe, Open Grey and National Digital Library of Theses and Dissertations (NDLTD)]; the reference lists of all included studies and relevant systematic reviews identified were checked, and key studies were citation tracked.

Study selection

Two-stage screening of articles against eligibility criteria was undertaken. Firstly, titles and abstracts were screened, then full texts. At both stages, screening was conducted independently by sets of two reviewers (co-applicants: NC, EC, ZP), and articles were excluded by agreement. Disagreements were resolved through discussion or by third-reviewer adjudication.

Data extraction

For each paper, data extraction was completed independently by two researchers (co-applicants: ZP and JW, or EC and FM). Key findings from the results sections of papers relating to BP were extracted; a ‘key finding’ was defined as any sentence or statement relating to views or experiences of BP from the results section of the paper or abstract. Wherever possible, the key finding was extracted as written by the author, with minimal edits only for clarification, description of context or for consistency across papers. For each paper, two authors extracted key findings independently and subsequently agreed on a final list of key findings for each paper. Data were also extracted on participant numbers and demographics, data collection technique, setting and country. Additionally, if available for patients, information was extracted on their BP use, including type of drug and current status (adherent, non-adherent, decliner).

Quality appraisal

The quality of each study was assessed using the Critical Appraisal Skills Programme (CASP) qualitative tool. This tool consists of 10 items split into 3 sections (qualitative suitability, data analysis and overall quality). The first two sections consist of items related to qualitative suitability and data analysis, which were evaluated as ‘yes’, ‘no’, ‘unclear’ or ‘partial’. The final question was an assessment based on the overall quality of the paper; this was informed by responses to the previous items (indicating methodological quality) and by the relevance of the study to the review objectives and was rated as ‘high’, ‘moderate’ or ‘low’. All papers were quality-appraised by three researchers independently (FM, SB, JW). Disagreements were resolved through discussion with a fourth reviewer (ZP).

Synthesis

We used a framework synthesis approach informed by the ‘best fit’ model described by Carroll et al. 55 The ‘best fit’ method offered a means to test, reinforce and build on an existing published model, conceived for a different but relevant purpose. This approach was chosen as a published theory identified from the literature that conceptualised acceptability – the theoretical framework of acceptability (TFA). The TFA is a relatively new framework which was developed to inform the understanding of acceptability of complex interventions and consists of seven constructs: affective attitudes – the emotions elicited by an intervention; intervention coherence – the extent to which an intervention makes sense; perceived effectiveness – the perceived extent to which intervention will achieve purpose; burden – the amount of effort required to participate in an intervention; self-efficacy – individual’s confidence that they can perform the behaviour(s) required to participate in the intervention; opportunity costs – the extent to which benefits, profits or values must be given up to engage in an intervention; and ethicality – the extent to which an intervention has a good fit with an individual’s values. The framework also incorporates temporal perspectives on anticipated and experienced acceptability at three time points: before (prospective), during (experienced) and after (retrospective) experience of an intervention.

The TFA has not previously been used to evaluate drug acceptability. We anticipated the seven constructs of the TFA would be relevant to engagement with drug treatment; for example, burden could relate to treatment burden associated with administering the drug or side effects. However, one aspect which did not appear to be explicitly conceptualised within the framework was patient beliefs about medicines. Studies across a range of long-term conditions, healthcare systems and cultures have consistently shown that engagement with treatment is influenced by patients’ personal evaluation of the medicine in question. 56 Particularly important is how they judge their personal need for treatment relative to their concerns about it. For this reason, we therefore included the Necessity-Concerns Framework (NCF) to further explore the TFA domain relating to intervention coherence.

The first author initially conducted inductive open coding on the extracted data, before mapping the codes to a draft framework derived from a priori themes (the domains of the TFA). Authors then met to first discuss the themes and compare findings for each study and the ‘fit’ to the draft framework. A preliminary synthesis was achieved using tabulation of studies, organising the studies into groups relating to temporal perspectives and research questions and exploring relationships between studies and between groups.

A final coding framework was agreed at a second meeting of authors. A second author (FM) recoded the original key findings, where necessary, to the new framework to ensure all findings were represented. Finally, relationships between themes and TFA and NCF domains were explored by further group discussion. We used the Grades of Recommendation, Assessment, Development, and Evaluation Confidence in the Evidence from Qualitative Reviews (GRADE-CERQual) approach to determine confidence in our synthesised findings. 57

Intervention coherence (high confidence)

Both before starting and during treatment, patients considered the perceived need or necessity for BP based on their views of osteoporosis, including its seriousness and controllability, symptoms and their perception of their own health. Perceived need was weighed up against concerns about medication, including suspicion of drugs in general and specific concerns about BP safety, by both patients and HCPs. HCPs sometimes used principles of ethicality to support perceptions of low necessity and their reluctance to prescribe. The decision process of balancing necessity against concerns was influenced by the doctor–patient relationship and wider societal influences, including friends, family and the general media. This process influenced whether HCPs reported recommending BP. For patients, the decision process could be explicit or tacit, was revisited over time and influenced both whether they initiated treatment and subsequently adhered.

Perceived effectiveness (high confidence)

Both patients and HCPs expressed doubt or uncertainty about the mechanism of effectiveness of BP and expressed a range of treatment expectations, including strengthening bone – improving bone density, preventing worsening of osteoporosis – maintaining bone density and/or total fracture prevention. Patients wanted proof or evidence of effectiveness through more structured monitoring and follow-up and were disincentivised to continue treatment in the absence of evidence of perceived effectiveness.

Self-efficacy (high confidence)

Measures to help patients integrate medication taking into their daily routines (supporting routinisation), and the provision of information and support, enhanced their feeling of having control over their health and confidence to adhere to BP. Clinicians reported barriers to supporting adherence related to perceptions of their knowledge and attitudes, with several knowledge gaps and uncertainties reported and the perception that osteoporosis was not a priority. Finally, service-level barriers which impaired clinicians’ self-efficacy in recommending and managing patients on BP included uncertainty about professional roles and responsibilities, capacity, access to IV drugs and communication and IT systems.

Affective attitudes (moderate confidence)

The emotions elicited by BP were closely related to intervention coherence. BP were associated predominantly with negative emotions of fear (of side effects) and annoyance (with special instructions); however, positive emotions of reassurance and hope were noted in two studies, linked to the anticipated protection that BP could incur.

Burden (moderate confidence)

The burden or effort of oral BP was described mostly relating to the special instructions to take oral BP or experienced side effects, although costs incurred were also a potential source of burden. Only one study included the experience of a patient on an IV BP. This patient described low treatment burden as she only had to go once a year and felt no side effects (62).

Opportunity costs (low confidence)

There were few descriptions of ‘benefits, profits or values’ being given up to take BP. However, circumstances where competing priorities challenged adherence or initiation of BP were described relating to comorbid conditions. The presence of comorbid conditions was described as resulting in less time to support discussion about BP in consultations and resulted in recommendation of, and adherence to, BP being given relative low priority.

Strengths and limitations

A strength of this review is the comprehensive search, use of underpinning theoretical framework, inclusion of clinician views in addition to patients and use of the GRADE-CERQual to give confidence in our findings, which has facilitated a clear identification of where further research is needed. Areas where we have identified moderate or low confidence are in need of further research and specifically relate to the influence of multimorbidity on sense-making, burden and self-efficacy in BP users, the extent to which IV BP may overcome issues related to treatment burden and self-efficacy and the impact of BP on affective attitudes and emotions. Furthermore, we have identified gaps in our understanding of how clinicians make decisions in practice and how views of BP may be influenced by gender. Given that many osteoporosis drugs have a different evidence base and licensing arrangements in men, this is an area in need of further study.

The main limitation of this review relates to the lack of clarity in many of the included studies in the results sections about which osteoporosis treatments or BP were being referred to, meaning that in some cases we may have overinterpreted findings relating to BP that were about other osteoporosis drugs. However, all of our review findings were identified from a comparison of data from several studies, and as BP represent the mainstay of osteoporosis treatment, we consider that overinterpretation is unlikely. As there was frequently little detail about medication participants were taking or referring to, it is also possible that we have missed relevant studies. Only two studies reported the views of managers, but unfortunately neither of these studies distinguished professional roles in the presentation of results, so a further need exists to explore perceptions of this group and perceptions of payors and academics. Finally, although the population from which each study sampled was reasonably well described, it was not always possible to appreciate if the setting was primary or secondary care; the majority of studies appeared to recruit from primary care, which may explain the lack of findings related to IV BP and limit the transferability of our findings to non-primary care settings.

Sampling

The inclusion criteria for patients participating were adults who had taken or received BP for the prevention of fragility fractures within the previous 24 months, and they needed to have the capacity to provide informed consent. The purpose of the interviews was to explore patients’ experiences of BP treatment regimens for the prevention of fragility fractures, focusing on which BP were most acceptable to patients. Originally, we planned to recruit participants via regional primary and secondary care clinicians and the regional Clinical Research Network. Due to COVID-19, recruitment methods were adapted.

With ethical approval, in January 2020 [North West-Preston Research Ethics Committee (REF: 19/NW/0714)], semistructured interviews were conducted between June 2020 and August 2020 and March 2021. A study advertisement in the Spring 2020 edition of the Royal Osteoporosis Society (ROS) newsletter invited individuals to take part in one telephone semistructured interview. Replies were used as part of purposive sampling, thus ensuring that the sample included enough participants who had experience of oral BP, IV BP and those who had experience of both types of treatment. Once major COVID-19 restrictions had been lifted, the research team were able to engage with clinicians across the region, via professional networks, to support the recruitment of patients who were receiving IV BP in the community. Such experiences were sought since community provision of IV BP is not usual practice across the UK.

A total of 78 participants with a mean age of 69.9 years were recruited through the advertisement in the ROS newsletter and through engagement with clinicians via professional networks. Forty-three patients had most recently taken oral BP (for the majority of these participants, the current or most recent oral BP that they had taken was ALN tablets). Thirty-seven participants had most recently received IV ZOL BP infusions in hospital or community settings. Interviews ranged in duration from 20 to 60 + minutes. Table 4 provides an overview of participants’ demographics.

Clinicians, specialist experts and service lead sampling

In order to understand the wider contexts of BP treatments and the service systems surrounding them, general practitioners (GPs), secondary care clinicians, specialist experts (including those involved in research), as well as those providing and leading novel treatments were recruited for qualitative interviews. These groups were purposefully sampled to include those with a good knowledge of the BP regimens in use and involved the following approaches.

First, GPs were contacted through a snowball approach, beginning with the existing professional networks of the study team. Although it was originally planned to focus GP recruitment on the practices in which patient samples were drawn from, due to the COVID-19 changes identified in patient recruitment, existing networks allowed the identification of GPs both with and without specialist/research involvement and commissioning/service leadership for osteoporosis and BP treatment. Study team members identified potential participants, and a research advertisement was also placed in the West Midlands Comprehensive Local Research Network (CLRN) newsletter for research-active GPs. GPs who were interested in taking part were invited to contact the study team and were then sent a formal invitation e-mail/letter and Participant Information Sheet.

Second, we contacted specialist clinicians, including those involved in research and service leadership. These respondents were identified through snowball sampling, beginning with the study team. Eligible participants identified by the study team were sent a Study Information pack, which included an informal invitation e-mail/letter and participant information sheet.

Third, we sampled from two specific areas where different or novel first-line BP regimens are used. This included participants from around the country (Table 5), where first-line ALN is recommended with a programme of blood test monitoring which is not usual practice elsewhere in the UK. Potential participants were sent a formal invitation e-mail/letter and participant information sheet.

In total, we recruited 23 clinicians and clinical specialists, including those active in research and service leadership. Interviews ranged in duration from 20 to 60 + minutes, and the background of participants is provided in Table 5.

Data collection

Interview schedules for participants were developed in collaboration with the study team and steering group, which included patient and public involvement and engagement (PPI) representatives and comprised questions about patients’ experiences of the osteoporosis diagnosis, perceptions about their BP treatment regimen(s) and clinician and service factors. The interview schedule was piloted with two PPI representatives and refined as appropriate. All participants provided informed consent. All interviews were conducted over the telephone.

Data analysis

Interviews were digitally recorded, transcribed verbatim and anonymised. The interview transcripts were uploaded to NVivo (version 12) (QSR International, Warrington, UK) and were subjected to intense open coding to identify early ideas and issues (referred to in NVivo as ‘nodes’). Two researchers independently coded the first five transcripts and then compared analyses, allowing interpretations of the data to be critically assessed, refined and agreed. Once first-level nodes had been agreed, the remaining transcripts were coded according to these by two researchers. Newer subnodes were added over time to enable specific and relevant issues to be categorised effectively. These subnodes were developed and agreed by two researchers. Once all the transcripts had been coded, the process of iterative categorisation86 was used to provide a clear and rigorous written trail reflecting the development of themes from initial nodes. This involved identifying key NVivo nodes as particularly pertinent to the research question and exporting the content to Word. Table 6 describes these nodes and what aspects of the data they captured.

Once exported into Word, the data that had been coded to each node were examined and systematically reread. Summary and interpretive notes were added, reflecting on the content of the theme in relation to other themes, research questions and prior literature. The results below first cover the degree to which different forms of treatment were acceptable to patients, with data interpretation considered in relation to our definition of ‘acceptability’ based on Sekhon et al.’s (2017) framework,87 which proposes that it is a multifaceted construct underpinned by seven key domains. These domains are collectively known as the TFA and their descriptive definitions are conveyed in Table 7. The codes were then formally mapped to the TFA as a framework, providing appropriate constructs to capture key dimensions of treatment acceptability and engagement.

Acceptability of treatment regimes

Patients’ acceptability and engagement behaviours were captured through the TFA domains. Participant identifiers at the end of each quote indicate their role as patient, clinician, specialist, research expert or service lead. For patient participants, it also indicates the BP treatment they reported experience of. O = oral BP treatment only; IV = intravenous BP treatment only; Dif = experience of different types of BP treatment.

Intervention coherence and perceived effectiveness

Participants' views on ‘intervention coherence’ and ‘perceived effectiveness’ were often closely related, with both requiring participants to make sense of how, and to what extent, the medication could be said to be ‘working’. For the BP medicine, this commonly involved developing both a conception of the future risk posed by reducing bone density and the potential reduction of such risk. Many patients described the treatment as a way of avoiding a negative consequence of ongoing bone deterioration. This is sometimes derived from the personal experience of seeing a family member suffer:

The state of my father really, because mentally he was one hundred percent, and physically he was an absolute wreck. His spine had bent completely.

(BO42_Dif)

The desire to avoid fractures was identified as a key motivation for engaging in treatment, particularly for patients who had already experienced one:

Oh, just because I’d had, I’d fallen and broken my hip and I didn’t want to fall and break something else again.

(BO76_IV)

This motivation had also been observed by secondary care clinicians:

In more, let’s say 70- or 80-year-olds who have had a hip fracture, then they tend not to question it at all, they just go for whichever you tell the best we will take it, that sort of thing.

(B002c)

Moreover, one specialist nurse suggested that certain types of fractures prompted patients to take osteoporosis and the prospect of treatment seriously:

I mean a lot of patients are really scared of like the Dowager’s Hump. The spinal fractures tend to, if you tell somebody they’ve got a fracture in their spine they’re more likely to buy into treatment and understanding because they desperately don’t want to curve over. So that’s a huge thing, but when patients have just like fractured a wrist or a humerus or a hip, yeah, then you do struggle to get them to buy into actually it is quite serious.

(B012c)

As BP do not necessarily address any felt symptoms, the coherence of the medication involved envisioning it as providing a level of protection from either such acute or chronic health problems.

And when they said I would be having it I felt good because it’s protecting me, that’s what I thought you know.

(BO10 IV)

In some instances, patients felt they had sufficient support to understand this protection and were reassured by a good relationship with HCPs.

I felt that everything had been explained and I just thought [treatment] was a way of preventing it getting worse really.

(BO68_Dif)

This was particularly the case where participants felt they had perceived tangible evidence of the treatment working, which drove them to continue taking the medication. This was often prompted by receiving positive results from Dual Energy X-ray Absorptiometry (DEXA) scans denoting improved T-scores. (A T-score is an indication of how close the person’s bone density is to the average peak bone density.)

My hip then improved over three years to -1.1, so OK, it was a slight improvement but as the consultant said, ‘it is an improvement, and it hasn’t got worse’. And I think well, something is working somewhere along the line … That encouraged me then just to keep going with it.

(BO 9_O)

Others disclosed feeling motivated by the fact that they had not sustained fractures since beginning treatment:

I haven’t had any more vertebral fractures and just to take the positives from that, but yeah, and I’m able to walk and do all sorts of things.

(B065p_Dif)

In these cases, patients developed hope that treatment would maintain or improve their bone density. Therefore, working to attain these goals also encouraged patients to engage in treatment:

And if it adds another five years or longer, ten years of me having a more stable life, I want that, I’ll go for it, thank you.

(B055p_O)

However, uncertainties about the effectiveness of BP treatment were also very common, and several questioned the degree to which their efforts to adhere were working. Underpinning this, participants described a wide variability in the level of information, support and/or feedback. This included perceived limitations or contradictions in the way in which the medication was explained.

I was waking up in the middle of the night, thinking I’m not doing it, I’m not going to take it. The side effects aren’t explained to you, well not to me when they were discussing this was the medication that you’re going to get.

(B027p_O)

And my first initial reaction was ‘you’re taking something for osteoporosis that causes a fracture!’ And it didn’t make sense. It was almost too much to take on-board at the time. And from that to say, ‘will you just pop in and collect your prescription and start taking it’- ‘mm-mm, no way’!

(B013p_O)

Further, patients often perceived wide gaps in their ongoing care. For example, a patient spoke about requesting a scan, as opposed to waiting for this to be initiated by her GP, in order to ascertain whether her bone health had improved, which was important for her:

…That’s why I asked for the scan, because I thought, I’m doing all this, I’m doing all this running, I’m doing these weights, I’m doing this diet, I’m taking all the tablets, I’m doing everything that I can do, I needed to know if it was actually making a difference ... I can’t say what was making the difference but all together it was making a difference and that made me feel better sticking with everything.

(B071p_O)

Similarly, others on oral BP described disappointment at the poor follow-up and lack of opportunities to discuss effectiveness, which left them uncertain as to whether the treatment was still appropriate:

I do think that somebody should have carried out a proper, as I said before, a proper review after five years ... and said you know ‘I think you need another DEXA scan to see if that’s the right drug.’

(B031p_O)

The only time I was made aware of the reading from the DEXA scan was when it was the first one in 2010. And so, the subsequent ones, I’ve never had a patient’s copy, or I don’t really know whether there’s been an improvement.

(B019p_O)

Moreover, even when follow-up was planned, the timescale for accessing another DEXA scan after oral BP had been prescribed was sometimes perceived to be too long:

I found that the DEXA scan that I was eventually referred to was quite informative. But I feel the gap between, well 5 years before you can have another one. I feel as if I’m a little bit in the dark about what’s going on and I’ve got an enquiring mind, so I like to know more.

(B038p_O)

The variation in treatment plans and follow-up in primary care was reflected in GPs’ responses. While some had a specialist interest in osteoporosis and had set up systems for periodic checks on patients (e.g. 3 months following treatment initiation), others commented on the lack of scope to check patient adherence or review medication. One GP also mentioned the lack of incentives in the form of quality outcome payments.

Well if we had infinite resources then I would have a sort of annual review in general practice but that’s never going to happen because GPs are just too busy at the moment.

(B006c)

I think we’re as I’ve already discussed, not good at picking up when compliance drops off. Because actually we haven’t got a formal strategy for auditing or reviewing, non-requesting of medication and that’s actually a quite tricky to identify, but perhaps not impossible.

(B016c)

Patients receiving IV treatment generally had fewer concerns over effectiveness and coherence of the medication. This could be seen as due to the fact that this group had more regular scheduled appointments, as well as the way that the administration of the treatment required the presence of a HCP. This enabled more opportunities for patients to interact with clinicians to discuss the reasons for taking the medication and also how to deal with any side effects:

So, she took me through absolutely everything to make sure I didn’t suffer once I’d had the Zoledronate [intravenous treatment]. And I think it was so much easier for her to tell me face to face. I mean she could have done it on the phone, but as opposed to just sending the instructions, which they did too, they did send instructions in the mail as well.

(B050p_Dif)

The care particularly centred on the second infusion was so reassuring … I felt so well looked after. And having it explained to me I think several times in several different ways because people realised that I just wasn’t taking stuff in.

(B066p_IV).

This was further enhanced when IV treatment was provided at home, with the nursing team taking on larger responsibility for organising the appointment, and then being able to focus exclusively on one patient for the duration of the treatment. This included opportunity to discuss effectiveness and the findings or previous scans:

Well, I think they give reassurance, and they give you their time, because I was very impressed with the chap, who came last week, because when I asked about my DEXA scan results he had to go out to the car to get his laptop and he, you know he wasn’t rushing me. He got his laptop, and he sorted out exactly what the data was from the DEXA scan and told me all the data.

(B081p_IV)

Further, patients on IV BP also tended to have well-established time points for follow-up:

Oh, I’m just sent a letter with an appointment date and time. And I can sort of change it, if it’s not convenient … it’s always been alright.

(B080p_IV)

Secondary care consultants confirmed that appointments and follow-up plans were well established for patients on IV BP, underpinned by clear rationales, including scheduling a point at which such treatment could be reviewed:

The nurses follow them up regularly to do the infusion. After three infusions, we will see them, look at the DEXA to see whether they need to remain a bit longer on the bisphosphonates.

(B002c)

This was reflected in the views of the community team delivering IV treatment in community settings who valued the opportunity to provide direct and individual care.

So from, just comparing to when I worked in the hospital and obviously being in their home address, there is only one patient, so your attention is a hundred percent on that person or the people that are in the room with you.

(B022c)

Having clarity around when DEXA scans were going to be scheduled was important for enabling patients to see a potential end point for treatment. One patient pointed out that this is what made IV BP treatment particularly favourable:

And it’s time limited as well, knowing that it’s just for three years and then I will have a bone scan and hopefully it’s better, it’s quite nice as well because you see light at the end of the tunnel with yet one more health condition.

(B066p_IV)

Moreover, a few patients seemed to have higher expectations of IV treatment itself, even before scan results were attained. They associated the format of this type of treatment as facilitating a direct, and seemingly stronger, intervention for the actual problem of osteoporosis. This was reflected in patients’ comparisons with previous oral BP treatment:

I had no sort of medical expertise at all, but to have an infusion would seem to me to get to the heart of the matter more than a pill!

(B051p_Dif)

While this might not have been an explicit message from health professionals, the IV treatment was often seen as a ‘step up’ in the intensity of care for patients who had previously been on oral BP. In this context, a greater efficacy of IV treatment was seen to make sense. Some clinicians providing IV treatments did indeed suggest these would be more effective than oral BP, as captured by a specialist nurse below:

So if we give IV we know there’s no absorption problems, it’s getting straight in, so it will be effective basically, if it’s not going to be effective IV then it’s never going to be effective sort of thing.

(B003c)

In most instances, however, clinicians described IV treatments as recommended due to side effects, non-adherence and/or intolerance to ‘first-line’ oral treatments.

Opportunity costs and burden

When the BP treatments required patients to make sacrifices (i.e. ‘opportunity costs’) and/or the regimens were regarded as burdensome, this contributed to overall negative views, which balanced against whether the treatment was seen as effective and coherent. Although a small portion of patients who were on or who had been on oral BP found the regimen relatively straightforward when weighed against the benefits, a larger portion identified wide-ranging costs and burdens. Most commonly, patients described struggling with the general treatment routine, which was regarded as complex, and the disruption this caused to their morning habits.

I find it a total burden to be honest … I’m an early riser, and I do like to wake up early and have a cup of tea, I know this is all trivial, but just have to swallow a huge amount of water with this tablet, and then sit around for half an hour, trying to occupy myself – it just, it’s just alien.

(B037p_O)

This could be seen to be exacerbated when the patients had other commitments, priorities or health issues:

To take the pill and to sort out the timing of it and everything, I just couldn’t be dealing with that, because I had quite a lot of stresses in other areas of my life, emotional stresses and stuff like that.

(B051p_Dif)

As a point of contrast, others highlighted that their personal circumstances facilitated dealing with the burdens.

It’s a bit of a pain I have to admit, if you want to get on and you’re going out to take it. But it wasn’t – that had nothing to do with the reason why I stopped taking it. I was able to fit that into my daily life quite easily. I can imagine had I been working it might have been a little harder.

(B032p_O)

The common problems for patients on oral BP were the restrictions around not being able to eat straight away and also needing to remain upright. The latter was particularly challenging for patients who had previously suffered from particular fractures or other conditions which affected their posture:

I mean, in fact you can sort of sit bolt upright but I’m not very good at that, so I always seem to hunch a bit and since I had the fractures that’s really not possible.

(B014p_Dif)

Other patients highlighted the practical difficulties of the oral BP treatment regimen which prevented them from taking them properly, such as not being able to swallow the tablets:

I really couldn’t get them down my throat, because I knew that I would start gagging and a good chance of vomiting and I can’t live like this, you know.

(B054p_Dif)

The requirement to follow strict instructions caused high levels of burden for some patients, to the extent that they felt that their mental well-being was compromised:

I try to keep active for an hour – because I’m frightened, you’re frightened taking that particular Alendronic Acid. I don’t like taking it, I find it a real inconvenience.

(B027p_O)

General practitioners in the study also recognised the challenges their patients faced in terms of medication burden.

You’re not meant to take particular drugs at the same time, isn’t it, either, I don’t think, that then messes up timing for other medications, so yes, if it was a pill that you just take at the same time as your anti-hypertensives and your aspirin and your lansoprazole then I think compliance would be much better.

(B010c)

Some patients on oral BP discussed developing strategies to cope with the more challenging aspects of the regimen, thus reducing costs and burdens. This included planning ahead, thinking positively, engaging with the ROS and setting reminders:

I’ve got an alarm on my phone to remind me that it’s Wednesday and I’ve got to take my Alendronic Acid.

(B071p_O)

In some cases, this also included adapting instructions to make the oral BP regimen more bearable:

So, I was really suffering and in the end I just thought I’m going to make my own rules up here. So, I was doing an hour, I was doing a pint [of water], sometimes more than a pint, I was not moving, because if I moved I suffered all day.

(B050p_Dif)

In comparison with the costs and burdens of oral treatment, patients generally identified practical challenges and inconveniences relating to accessing IV BP treatment, as opposed to the treatment per se. This included travelling to hospital appointments, issues around parking and navigating one’s way around the hospital:

I think I had to go to something like atrium three which didn’t happen to be marked and it was all a bit, it was a bit confusing and a bit open, but I found it. I blundered my way towards it. It wasn’t ideal.

(B023p_Dif)

Such issues were removed for patients receiving IV treatment at home. Having home-based IV treatment also enabled some patients to cope better with the regimen since they would have faced difficulties with travelling due to the physical restrictions caused by osteoporosis and/or comorbidities:

I would prefer it at home, because it is a bit of a struggle for me to get to places, you know with the multiple sclerosis.

(B075p_Dif)

Moreover, having IV BP treatment at home was valued since it enabled patients to receive such treatment even more comfortably:

That’s great, because you see, I went to have scans in the hospitals, and I knew what a scan was like and a drip and all that kind of thing. And I was just, well I was amazed to be able to sit in my favourite chair.

(B082p_Dif)

In addition, the fact that the IV treatment regimen required a nurse to administer, it took much of the work out for the patients and did not seem to cause much inconvenience for patients having it at home:

They just come into the dining room, the chair is pulled out, they set their gear up on the table and we’re away, no problem … I mean how could it be better? You just sit down, she puts a needle in, pops it in and then Bob’s your Uncle! No problem!

(B077p_Dif)

Thus, IV treatment seemed to offer patients a regimen that was lower burden, with fewer opportunity costs. Some patients who had direct experience of both oral and IV BP even suggested that side effects from the latter treatment regimen were more favourable:

Basically, it’s a morning and a day of feeling woozy and that’s it for in my case eighteen months, you know, compared with wrestling yet another lot of pills to take every week.

(B014p_Dif)

This was similarly reflected in the experience of those delivering IV treatments for patients at home.

They’re in their own home at the end of the day. If they can’t be comfortable there where can they be comfortable? So we have even done a few infusions in the garden sometimes because the weather is nice.

(B022c)

A small number of patients did comment on difficulties with the administration of IV treatment, for example, having the needle inserted into the vein, and clinicians reported that a number of patients were concerned about needles. There were also some patients who felt happy with the oral BP regimen, and a small number drew on points of comparison with the IV treatment to justify their preference:

I just [worry] about side effects and if you put intravenously, you know, how you counter them. And I thought I’m not going down this road at the moment.

(B013p_O)

Generally, though, there were more positive accounts relating to IV treatment regimens, with patients enthusing about the overall ‘experience’. They recounted the staff approaches that contributed to little burden and opportunity costs on the part of the patients:

It’s wonderful, they even come round and offer you a cup of tea and a sandwich or whatever. Oh, they’re just really friendly, really welcoming and getting you sat down, ask you which arm you want in, making sure you’re comfortable, you’ve got something to put your feet on, you’ve got something to read. No, I couldn’t fault them at all.

(B063p_Dif)

Regardless of treatment type, a regimen which led to no side effects or manageable side effects enabled patients to cope with their treatment regimens, supporting acceptability and engagement. But for patients where side effects were particularly severe, this was enough to cause them to stop engaging in the treatment:

And I just thought the side effects from the Alendronate just weren’t worth it, they were impacting my life.

(B032p_O)

For some patients who did not experience severe side effects, this did not preclude the worry that these could occur. Such concern served to be a burden for some, while living with the risk of such side effects was one of the costs required for engaging in the treatment:

You know, it does worry me that if I have to have a tooth out there could be complications … it’s adding to my decision to stop when I’ve been on it for five years.

(B005p_O)

Mostly, those on IV treatments were less concerned with side effects. While one patient on oral BP did worry more about the potential side effects from the IV treatment, the comparison between the two could also be seen in light of higher levels of support for patients receiving IV treatment, as noted above. Professional support and care may have helped to both reduce side effects as well as reduce the concern of these occurring.

Ethicality

In the context of BP treatment acceptability and engagement, the TFA domain of ‘ethicality’ was linked to the extent to which patients regarded the treatment regimens as aligning with their individual values and whether they perceived the treatment to be fair and suitable for their needs. This often related to the way in which they had come to be prescribed the treatment and the manner in which the treatment had been offered to them.

For some patients who were prescribed ALN oral BP, they felt that their personal wishes had been dismissed, which led to negative attitudes towards the treatment:

I wrote to my doctor and said I’d read all the information, so when she started prescribing, ‘pretty please prescribe the soluble tablet’, which to my way of thinking was the least harmful or least – the one I could tolerate most, of the options. And that got ignored, and I got sent the prescription for the Alendronic tablet, and I refused to take it. I took it once, and I just found it so difficult I felt I’m not taking it – I’d rather not have it.

(B037p_O)

Even for patients who had come to accept any inconveniences of the oral BP treatment regimen and were engaging in the treatment, many still expressed dissatisfaction with the fact that they had not been presented with other treatment options:

I shouldn’t complain really because so many people have so many things, and their lives are really dictated by their meds. But what annoys me is the reluctance to actually discuss any options.

(B044p_O)

One patient described the ramifications of her GP discounting the advice of the osteoporosis nurse to offer the patient a soluble form of ALN to minimise the risk of GI side effects:

So basically, I’ve always had a dodgy tummy. And so, if they give me medication, they give me the stomach liner thing as well. And so [the osteoporosis nurse] said to me ‘Well you might be advised to take the bisphosphonate in a soluble solution form’. So, when I told the GP this, he said ‘No, you can’t have that, you’ve got to have the Alendronate tablet’. And I assume it’s because it was cheaper. So, I started off on the Alendronate. Well, it got to about four months in, or three to four months in and I was, it had made my tummy bad.

(B018p_O)

On the other hand, other patients did not find it a problem that they had not been presented with alternative options, and in fact, this encouraged them to engage in the treatment because of the implications of osteoporosis:

I was, you know I never enjoyed the fact that I had to have my glass of water and potter around the house but well, there was sort of no option, so I was perfectly happy. I knew what, I know what osteoporosis can do.

(B023p_Dif)

Moreover, some patients felt that oral BP were right for them, particularly when they could identify improvements in health which they attributed to the treatment. As such, they did not wish to pursue other options:

I mean when I look back to what it was like when I first had the spinal fractures, you know it was unbelievably painful, and then now as I say, I can mow the grass, pick up weeds, do all these sort of things – clean windows. And so I mean I’m very happy on it really. I can’t see any reason to change it.

(B011p_O)

For some patients on IV BP treatment, they had previously been on oral BP and could make direct comparisons between both regimens. IV BP treatment was generally viewed as easier to engage in with optimal frequency due to being just once a year. This suggested that the IV treatment regimen was a better fit with their lifestyle and therefore more suitable:

I know it’s only little things really but the convenience of having something once a year compared to 52 times a year is amazing.

(B074p_Dif)

While this could be seen as pointing to the generally lower burden of IV treatment, it could also be identified that such comparisons played into overall perceptions of equity; namely, it was more common for those on IV treatment to see themselves as being treated fairly.

Self-efficacy and affective attitude

In the context of accepting and engaging in BP treatments, the balancing of the aforementioned TFA domains could be seen to shape patients’ overall feelings of self-efficacy and affective attitudes towards their regimens. As per Table 1, self-efficacy reflected patients’ confidence in their abilities to undertake the necessary behaviours to engage in the treatment. Affective attitude is related to patients’ feelings towards their treatment.

Positive affective attitudes of patients, including gratitude and hope for the future were apparent when they understood the reasons for the treatment, felt they had made sense of its potential effects and felt that they were well supported by clinicians and the wider service. This often appeared to be facilitated by regular contact with professionals and access to diagnostic tests to track the progress of bone health.

The consultant that diagnosed it, because he looked back at this x-ray and saw it and he just put everything into action and was very clear … so professional … And yes, so that was very positive for me. And the actual treatments, because I’ve always felt quite secure and quite happy with them.

(B060p_IV)

No, but I do appreciate, that you know the trouble, I do appreciate having the treatment that I’ve had. Because, if it helps my bone density then that’s really good.

(B081p_IV)

Many of the patients on IV treatment had previously had difficulties in adhering to regimen for oral BP and were therefore pleased to have ‘arrived at’ a more acceptable form of treatment. Having experience of both types of regimens led to some patients enthusing about IV treatment. Being able to reflect on negative experiences of oral BP led to them conveying positive affective attitudes and feelings about IV treatment.

No, when considering the tablets, I was rather pleased that I was getting this one … I thought I would come off the better actually.

(B076p_IV)

And then of course I hit the jackpot with the infusion and that’s marvellous.

(B014p_Dif)

[Intravenous treatment is] just wonderful … I’ve never been up nor down or anything, and compared to what I was suffering with, fabulous. For me it’s been great.

(B030p_Dif)

Others on IV treatment who had not previously been on oral treatments made speculative comparisons, weighing up the negatives against the positives. The IV treatment was often regarded as the better option, leading to feelings of satisfaction:

I’ve talked to people afterwards who just have the tablet, and they’re surprised that I had the infusion, but I was quite happy to have the infusion.

(B081p_IV)

Equally, some patients on oral BP felt positively towards the tablets and that they were in control of managing them themselves:

I managed the five years without missing them, I think. It was convenient having the tablets because once or twice we were on holiday, to just take them with you and do the same thing.

(B064p_Dif)

Moreover, this patient also highlighted that the COVID-19 pandemic had made the prospect of going to hospital to receive IV treatment more burdensome, thus leading to a more negative affective attitude:

I’m not looking forward to going to the hospital with all these COVID problems.

(B064p_Dif)

Further, despite curiosity about alternative treatments, when the experience of oral BP had been unproblematic, patients described being happy to remain on their current regimen:

Not that I would want to change over, but it was only just something that I read. So no, I wouldn’t want that anyway. I’m quite happy with my tablets.

(B011p_O)

It might be nice to know that they were aware that there are alternatives, but I don’t take a massive amount of notice of what they are because, at the moment, I’m quite happy on the Risedronate.

(B024p_O)

This was also the case for patients who had had a treatment break and needed to resume the oral BP regimen:

I was very happy to go back on Alendronic Acid having had no trouble previously.

(B039p_O)

Regardless of treatment type, in circumstances where opportunity costs and burdens were low, patients expressed high levels of self-efficacy, sharing their confidence in being able to execute the necessary actions to partake in the treatment regimen:

…so I think all in all [the tablet] definitely agrees with me, and it’s so easy to take. You just get into a routine you know, like Friday morning, I know exactly what I’m doing and what times I’m doing it at you know.

(B011p_O)

…they were a good, shaped tablet, you know, they were oblongs – not big round ones. I’ve had all sorts of trouble with big round tablets. But they’re a good shape. I had no trouble in swallowing it with a full glass of water.

(B064p_Dif)

Sometimes, patients discussed the steps that were taken to reduce the burden of undergoing treatment and how such measures made them feel more positive and able to cope with the regimens, thus increasing their self-efficacy. By nature of the oral BP regimen, such strategies had to be instigated by the patients themselves. For example, one patient discussed keeping herself occupied during the period when she needed to remain upright after taking the tablet:

Interviewer: Did you manage to fit it into your weekly routine?

Oh yes, because I got loads of jobs done while I was upright and swallowing loads of water.

(B053p_Dif)

In contrast, for patients receiving IV treatment, it was often the healthcare staff who could facilitate a smooth and comfortable experience for them:

I know what [the staff are] there for and yeah, I just, put your arm out and keep it in a safe place. So, they just, put you in a comfortable position. And then they just do it, so … I don’t really feel anything to be honest.

(B078p_IV)

Also, when the treatment aligned with one’s personal beliefs, this also led to positive affective attitudes and higher levels of self-efficacy. Patients openly expressed their satisfaction when they believed that they were on the most appropriate treatment for osteoporosis:

I need to be on a treatment. I think [intravenous is] the best one for me at the moment but I’m fully aware I am at high fracture risk anyway.

(B023p_Dif)

Some patients described positivity at being on what they perceived to be the right treatment, particularly if their experience of previous treatments had been negative:

I was really sort of rather pleased that I’d finally achieved, you know, some other treatment and I didn’t have to keep taking the wretched [oral] bisphosphonates.

(B014p_Dif)

Therefore, perceived fairness of treatment allocation, including the timing of this allocation, also appeared to be related to the patients’ affective attitudes.

Strengths and limitations

A key strength of this study is the fact that a large sample of participants were recruited to provide in-depth insight into different experiences of BP treatment regimens. The main limitation is that findings were largely drawn from a sample of participants who had membership with the ROS, with a smaller number recruited through NHS services. This may have caused the sample to be biased, for example, it may have largely comprised individuals who had the financial means to fund membership and who were possibly taking a more proactive approach to their health by investing in resources. This may restrict applicability to other patient groups, such as those who are financially disadvantaged and those who are less proactive in their care and treatment. Moreover, our sample of patients receiving IV BP comprised those receiving ZOL only and not 3-monthly IV IBN treatment. However, the paper has been able to effectively demonstrate to some extent the relevance of TFA domains in explaining attitudes and behaviours around acceptability and engagement in BP treatment regimens. Moreover, utilising the TFA domains to explain acceptability and engagement of IV BP treatments is particularly novel. It will be useful to explore whether such findings apply to other patient groups.

The study has identified several questions for further research, including whether it would be feasible and appropriate to offer specific patients first-line IV ZOL treatment for osteoporosis and how acceptability and engagement can be optimised. In addition, it has also highlighted the possibility of treating long-term conditions in alternative ways, such as in the community, which may be favourable for an ageing population, where hospital travel may be challenging due to comorbidities and the COVID-19 pandemic. These uncertainties will feed into a research priority-setting exercise alongside other questions which have arisen from the wider research study programme.

Eligibility criteria

The eligibility criteria of both reviews regarding the population and interventions of interest were the same and have been described elsewhere. 92 Only studies in which the interventions of interest (ALN, IBN-IV, IBN-oral, RIS and ZOL) were assessed within their licensed doses for treating osteoporosis. Studies that reported data for both licensed and unlicensed dose study groups were considered eligible only if data for the licensed groups were separately reported. Studies reporting comparisons among the interventions of interest were considered eligible for inclusion. In the updated review, we targeted the following outcomes: fragility fractures, BMD at the femoral neck, mortality, adverse effects and health-related quality of life (HRQoL), with only randomised controlled trials (RCTs) eligible for inclusion. In the adherence review, the outcomes of interest were persistence and compliance, quantified either as continuous (e.g. absolute numbers or rates) or discrete measures (e.g. absolute number of participants being persistent/compliant based on pre-specified thresholds). In this review, RCTs, non-randomised parallel comparative studies and observational (both prospective and retrospective) studies were considered eligible for inclusion. In the RCTs, persistence was indirectly inferred by assessing the total number of participants who dropped out at 12 and 24 months; in the observational studies, persistence was inferred by assessing the total number of participants who discontinued their treatment based on treatment refill gaps, using data from claim databases or medical records. In the observational studies, compliance was indirectly measured by assessing ‘treatment continuity’ and using percentages/absolute numbers of medication possession ratios (MPR) and number of users with MPRs over a pre-specified threshold. Reports published as abstracts or conference presentations were excluded where insufficient details were reported. RCTs which were judged otherwise eligible but did not report outcome data per treatment arm or reported zero dropouts for both arms were excluded. Studies which reported the outcomes of interest for BP groups collapsed or studies reporting comparisons based solely on the frequency of administration (e.g. daily vs. weekly) were also excluded.

Search strategy and information sources

A set of comprehensive search strategies were undertaken to systematically identify eligible studies in both systematic reviews. The search strategies comprised the following main elements: searching of electronic databases (including unpublished data and trial registries), extensive keyword hand-searching and scrutiny of bibliographies of retrieved papers.

The following databases were searched in both reviews:

• MEDLINE^® In-Process and Other Non-Indexed Citations and MEDLINE (Ovid), including PubMed

• EMBASE (Ovid)

• Cochrane Database of Systematic Reviews (Wiley Interscience)

• Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley Interscience)

• Cumulative Index to Nursing and Allied Health Literature (CINAHL, EBSCO)

• Database of Abstract of Reviews of Effects (Wiley Online Library)

• HTA Database (CRD Database)

• NHS Economic Evaluation Database (CRD Database)

• OpenGrey

• Science Citation Index (ISI Web of Knowledge)

• CPCI – Science (Web of Science)

• ClinicalTrials.gov.

Searches of the updated review covered the period from September 2014 to 1 March 2021. Searches of the adherence review covered the period from January 2000 to 25–26 March 2021. In both reviews, all potentially relevant citations were downloaded to Endnote X8 Reference Manager bibliographic software (version 8.0; Clarivate Analytics, Philadelphia, PA, USA).

Study selection, data collection process and data items

Selected studies were imported into Rayyan online software. 107 Two independent reviewers screened studies for relevance based on titles/abstracts and later full texts (AB and TL) with disagreements resolved through discussion or by consulting a third reviewer (JLB/OS). Two independent reviewers (AB, TL) conducted full-text screening with a high level of agreement observed in both reviews (κ = 0.91 in the update review and κ = 0.84 in the adherence review). A standardised and pilot-tested data extraction form was used to extract relevant data in both reviews. One reviewer (AB) extracted data, with a second reviewer (TL) independently checking at least 50% and 80% of the extracted records in the updated and adherence review, respectively. Data extracted consisted of the following categories: (1) descriptive statistics (e.g. number recruited and randomised, participants’ characteristics); (2) baseline data on outcomes of interest (e.g. comorbidities, fractures at baseline, alcohol use, number of falls); (3) moderators of action [e.g. glucocorticoids (GC) use, patients with osteoporosis, history of fractures/fractures at baseline]; (4) intervention characteristics (e.g. drug type, administration mode, concomitant treatments) and (5) statistics and relevant data on the outcomes of interest expressed either as continuous or binary outcomes [clinical outcomes such as fractures, BMD changes, HRQoL, adverse events mortality, number of participants who dropped out from RCTs, number of users who discontinued with BP treatment, number of users with varying compliance levels based on pre-specified thresholds, mean/range MPR, mean number of infusions/tablet counts, proportion of days covered (PDC) percentage, mean duration of BP treatment]. Authors were contacted when there was lack of data on outcomes of interest and/or further information was needed in order to attest eligibility of relevant studies.

Geometry of networks

Both treatment-placebo (PLB) and treatment-active comparisons were examined, and network plots were created for all outcomes. In both reviews, nodes indicate the different treatments included in the analysis, and thickness of edges connecting the nodes indicates the number of studies informing each comparison (thicker lines indicate more populated comparisons). In the adherence review, node size indicates the number of studies included in each node, and thickness of lines indicates the overall sample size informing each comparison (thicker edges indicate more populated pairwise comparisons).

Risk of bias within individual studies

The methodological quality of the included RCTs was independently assessed at the study level by two reviewers (AB, JLB) using the Cochrane Collaboration risk of bias tool 1.0 (94). The Cochrane Collaboration risk of bias tool 1.0 addresses the following specific domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment and incomplete outcome data and selective outcome reporting. Studies were rated with a low risk of bias in the randomisation sequence if they provided an explicit statement on how they performed the randomisation. Open-label trials were rated as high risk in the ‘blinding’ category, whereas higher than 20% attrition at 12 months’ follow-up resulted in high-risk rating in the ‘incomplete outcome data’ category. The methodological quality of the included observational studies was independently assessed at the study level by one reviewer (AB), with a second reviewer (JLB) independently checking 50% of the included studies. Any disagreements were resolved through discussion. The assessment of methodological quality in observational studies was undertaken using the Cochrane Collaboration Risk Of Bias In Nonrandomised Studies of Interventions tool (ROBINS-I). 108 Risk-of-bias plots were created by using the ‘robvis’ tool. 104

Summary measures and methods of analysis

Assessment of inconsistency

Consistency of evidence for the NMAs of RCTs was assessed using the node-splitting method117–119 in R Studio (R version 4.0.3). Differences between direct and indirect evidence in all network loops were calculated, with p values < 0.05 indicating the presence of significant inconsistency. In the case of fracture data, inconsistency was assessed for vertebral fractures only. For non-vertebral fractures, no indirect evidence was available. For hip fractures, an assessment of inconsistency was not performed because the direct evidence between ALN and RIS was provided by one small, and unbalanced in terms of sample size, study with zero events in one arm. For wrist fractures, an assessment of inconsistency was not performed because the direct evidence between ALN and RIS was provided by the same small study, and the only direct evidence between ALN and oral IBN-oral was provided by the only three-arm study included in the NMA. For BMD data, the assessment of inconsistency was performed after excluding an outlier study, which was the only study informing the direct relationship between ZOL and ALN, and the three-arm study, which was the only study providing direct evidence for the relationship between RIS and IBN-oral. For the overall adverse event outcome, an assessment of inconsistency was not formally performed because the fit of the model with the data was poor. For myalgia, headache and pyrexia, assessment of inconsistency was not performed because there was no indirect evidence. For influenza-like symptoms, an assessment of inconsistency was not performed because there was only one small study with zero events in the control arm informing the direct relationship between IBN-oral and PLB and three small studies with zero events in control arms informing the direct relationship between ZOL and PLB. Due to the multiple arms reported per study in retrospective observational studies included in the adherence review, a formal assessment of inconsistency in persistence NMA was not performed.

Credibility of the findings

Credibility of findings on persistence was assessed in RCTs only by following the confidence in network meta-analysis (CINeMA) approach,120 where the credibility of findings is accounted for by the assessment of: (1) within-study bias, (2) reporting bias, (3) indirectness, (4) imprecision, (5) heterogeneity and (6) incoherence. In this NMA, HR/OR ranging from 0.8 to 1.25 was used to indicate clinical significance in binary outcomes, while a MD of 2.71 (1/2 SD of baseline control arms) was used to indicate clinical significance in the continuous outcomes (i.e. BMD femoral neck). CINeMA’s freely available web application121 was used to assess credibility of the findings.

Updated review

A PRISMA flow diagram shows the selection of papers for inclusion and exclusion in the updated systematic review (Figure 4). A total of 6623 articles were retrieved, of which 1889 were duplicates. Overall, 4535 studies were excluded following title and abstract screening, and 170 were excluded following full-text screening. Data from 25 newly identified trials obtained from 29 published reports were added to the data obtained from 43 trials identified in the previous review (92) resulting in a total of 68 trials out of 47,007 participants.

FIGURE 4.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of the selected studies in the updated review. Reproduced from Bastounis A, et al. JBMR Plus 6:e10620. https://doi.org/10.1002/jbm4.10620.

Adherence review

A PRISMA flow diagram (Figure 5) shows the selection of papers for inclusion and exclusion. A total of 10,030 articles were retrieved, of which 1729 were duplicates. Overall, 7976 studies were excluded following title and abstract screening, and 220 were excluded following the full-text screen. Data were extracted from 59 RCTs drawn from 69 published reports and 43 observational studies drawn from 45 published reports, resulting in a total population of 2,656,659 participants. Table 9 lists all the studies (see Appendix 2 list of references).

FIGURE 5.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of the selected studies in the adherence review. Reproduced from Bastounis A, et al. Osteoporos Int 2022;33(6):1223–3. https://doi.org/10.1007/s00198-022-06350-w. Epub 21 February 2022. PMID: 35188591; PMCID: PMC9106630.

Overall, for persistence, 16,577 participants were included in the NMAs of RCTs, and 985,484 BP users were included in the NMAs of retrospective observational studies.

Synthesis of results on persistence: measured using dropouts in randomised controlled trials

A NMA was used to compare the effects of ALN, RIS, ZOL and IBN-oral relative to PLB on the total number of dropouts at 12 months. Overall, data were available from 30 two-arm and 1 three-arm RCTs. The network provided nine direct treatment comparisons. Each of the direct comparisons between ALN versus ZOL and RIS versus IBN-oral was informed by one small study. Eight contrasts were checked for inconsistency between direct and indirect evidence. None of the comparisons showed significant evidence of inconsistency, as assessed using Bayesian p values (p > 0.05). The model fitted the data relatively well (difference < 3), with a Dres of 64.8 being close to the number of data points included in the analysis, which was 63 (DIC = 369). The between-study SD was estimated to be 0.16 (95% CrI 0.009 to 0.41), implying mild heterogeneity in treatment effects between RCTs. Users of ZOL and RIS were less likely to drop out compared to PLB users, with none of these effects being statistically significant. The lowest likelihood of dropping out was detected in ZOL users OR = 0.73 (95% CrI 0.51 to 1.05; probability: 0.88; SUCRA: 0.95).

A NMA was used to compare the effects of ALN, RIS, ZOL and IBN-oral relative to PLB on the total number of dropouts at 24 months. Overall, data were available from 21 two-arm and 1 three-arm RCTs. The network provided eight direct treatment comparisons. Each of the direct comparisons between ALN versus IBN, ALN versus RIS and RIS versus IBN-oral were informed by one small study. Three contrasts were checked for inconsistency between direct and indirect evidence. None of the comparisons showed significant evidence of inconsistency, as assessed using Bayesian p values (p > 0.05). The model fitted the data well, with a Dres of 45.51 being close to the number of data points included in the analysis, which was 45 (DIC = 276.7). The between-study SD was estimated to be 0.34 (95% CrI 0.09 to 0.64), implying mild to moderate heterogeneity in treatment effects between RCTs but with reasonable uncertainty. Users of ALN, RIS and IBN-oral were less likely to drop out compared to PLB users, with none of these effects being statistically significant. The lowest likelihood of dropping out was detected in IBN-oral users OR = 0.72 (95% CrI 0.31 to 1.66; probability: 0.54; SUCRA: 0.72).

Synthesis of results on persistence: measured using discontinuation of treatment data from observational studies

A NMA was used to compare the effects of RIS, ZOL, IBN-oral and IBN-IV relative to ALN on the absolute number of people who discontinued their BP treatments. Overall, data were available from 24 retrospective observational studies. The model fitted the data well, with a Dres of 73.27 being close to the number of data points included in the analysis, which was 73 (DIC = 762.5). The between-study SD was estimated to be 0.24 (95% CrI 0.19 to 0.31), implying mild heterogeneity in treatment effects between observational studies with reasonable uncertainty. Users of ZOL and IBN-IV were less likely to discontinue compared to ALN, with the effects of the former being statistically significant (Table 10). The lowest likelihood for discontinuation was detected in ZOL users HR = 0.73 (95% CrI 0.61 to 0.88; probability: 0.88; SUCRA: 0.97). Heterogeneity of effects was explored by undertaking a post hoc meta-regression on the absolute number of discontinuers using refill-gap as a moderator variable. Although slightly decreased in magnitude, the direction of effects remained the same. The model fit remained almost the same, with a Dres of 72.63 (DIC: 755.9). The between-study SD was estimated to be 0.29 (95% CrI 0.20 to 0.42), implying mild heterogeneity in treatment effects between RCTs with reasonable uncertainty. Higher medication effects on discontinuation were detected in participants with longer refill gap thresholds, although the results were not statistically significant β = −0.23 [95% confidence interval (CI) −0.72 to 0.21].

Updated review

Overall, 44 trials provided data for femoral neck BMD, whereas 27 and 19 trials provided data for vertebral and non-vertebral fractures, respectively. Only 14 and 10 trials provided data for hip and wrist fractures, respectively. ZOL was found to be the most effective treatment in preventing the occurrence of vertebral fractures and increasing femoral neck BMD. ZOL was also found to be comparably effective to RIS and ALN in preventing non-vertebral fractures and hip fractures, respectively. ZOL’s effects in preventing hip and vertebral fractures and increasing femoral neck BMD were found to be clinically significant. In addition, treatment effects in preventing vertebral fractures were found to be stronger in people with osteoporosis compared to PLB. Uptake of ZOL was also found to be accompanied by more frequently reported adverse events; however, these events are likely to be short-lived. Based on these updated estimates, ZOL could be considered as the first-line treatment for people who experience or are at increased risk of fragility fractures.

These findings arguably have important implications for clinical decision-making in terms of the preferred therapeutic approach for people with varying fracture risk. It has recently been suggested that anabolic treatments should be preferred as the first-line treatment for people who are at high risk for developing osteoporotic fractures. 151 Although recent evidence has shown that anabolic treatment is more effective than BP in reducing fracture risk in females who are at high risk of developing fractures,152,153 its effectiveness has only been tested against oral BP. There is therefore an urgent need for future comparative studies to test the effectiveness of anabolic treatments versus ZOL in reducing the fracture risk in high-risk populations. This becomes more apparent when the imminent fracture risk and the need to expedite clinical decision-making154,155 are taken into account. Based on our findings, ZOL seems to be a promising treatment that could decrease the imminent fracture risk for high-risk populations within 24 months after administration. Future studies should investigate whether ZOL or anabolic treatments are more effective in reducing imminent fracture risk in high-risk populations.

Adherence review

For persistence, results from the NMA from RCTs showed that ZOL users may be less likely to drop out from trials at 12 months, although these effects were marginally non-significant. Results from the NMA using data from the observational studies showed that ZOL and IBN-IV users were less likely to discontinue their treatment over time, with ZOL users being statistically significantly more persistent compared to oral BP users. Data drawn from the vote-counting synthesis were in line with the results of NMAs, where ZOL and IBN-IV users were more likely to persist with their treatment, with ZOL users being more persistent compared to their IBN-IV counterparts. Users of ALN and RIS showed comparable persistence rates; however, when we restricted our analysis to weekly administration, ALN users were found to be more likely to persist to treatment over time. Due to the paucity of data and the heterogeneity in reporting compliance data, we were unable to perform NMAs, but synthesis based on vote counting found that compliance to ZOL was greater within 24 months after the initiation of their treatment. Users of IBN-IV were found to be more compliant compared to IBN-oral users. Users of ALN were found to be more compliant than RIS users, while mixed evidence was observed in the comparison between ALN and IBN-oral users.

These findings have important implications for clinical practice and future research. In general, persistence to BP treatment was found to decrease after 12 months, stressing the need to address adherence barriers according to BP treatment and people with different clinical profiles. Nevertheless, ZOL users were found to be less likely to discontinue their treatment over time, and they showed higher compliance rates. These findings are partly in line with the results from the dropout NMA at 12 months. Without ignoring the interplay of individual and contextual factors, which affect participation and adherence in clinical trials,156,157 we can assume that most ZOL users are likely to receive at least two infusions before discontinuing their treatment. The use of ZOL has been generally recommended for at least 3 years5 and, although reduced adherence rates have been observed in ZOL users after the first year,158 simpler drug regimens can improve adherence rates. 159 Results of vote-counting synthesis on oral BPs were partly in line with the NMA results. ALN and RIS users showed comparable persistence rates; however, ALN users were found to be more compliant than their RIS counterparts. When we restricted our synthesis to weekly administration of both BP, weekly ALN users were found to be more persistent to treatment compared to RIS weekly users. Given that ALN is the most widely prescribed medication, clinical decision-making should consider, alongside its clinical effectiveness, ways in which ALN users could be assisted to receive medication properly and remain on treatment long-term.

Brief overview the of model structure

The model used a discrete event simulation (DES) framework to simulate lifetime costs and QALYs for a cohort of patients treated using each of the alternative BP treatment regimens (ALN, RIS, IBN-oral, IBN-or, ZOL). A schematic of the model structure is shown in Figure 6. The key clinical events modelled are fractures at the hip, vertebrae, wrist or proximal humerus, all-cause mortality and fracture-related mortality, with the latter only possible following hip or vertebral fractures. Fractures are associated with an acute cost in the year of fracture and an ongoing cost in subsequent years. Costs in the model are estimated from a NHS and Personal Social Services (PSS) perspective, including costs incurred in primary and secondary care and social care provided in the home. In addition, hip fractures are also associated with an increased risk of new admission to a residential care home, with an associated cost for a proportion of patients whose residential care is not self-funded. Fractures are associated with a reduction in HRQoL, with separate decrements applied in the first and subsequent years after fracture, with a further decrement applied to patients admitted to a nursing home following fracture. The prevention of fractures therefore results in QALY gains through the avoidance of these HRQoL decrements, in addition to the QALY gains achieved by preventing fracture-related mortality. Future costs and benefits are discounted at 3.5% per annum.

FIGURE 6.

Clinical events that can occur during a patient’s lifetime in the DES.

Patient cohort

The model is a patient-level simulation that takes into account the heterogeneous patient characteristics present within the cohort simulated. The cohort simulated all patients eligible for fracture risk assessment within CG146. It therefore includes both men and women, patients with and without a prior fracture, those with steroid-induced osteoporosis, those with secondary osteoporosis and those with other risk factors for fragility fractures. In addition, a proportion of the cohort is assumed to reside in a care home at the start of simulation. The characteristics of each individual within the population are simulated. Life expectancy, body mass index (BMI), steroid use, prior fracture and residential status (care home or community dwelling) are sampled dependent on age and sex, whereas the remaining Qfracture risk factors are sampled based on their prevalence within the Qfracture cohort. The Qfracture algorithm estimates the risk of fracture for each simulated individual. As cost-effectiveness varies with baseline risk of fracture, results were presented for fracture risk categories defined according to deciles of absolute risk. This allowed identification of the optimal strategy for patients according to their fracture risk. As the primary aim of the analysis was to compare alternative BP to ALN, the presentation of the expected value of parameter information (EVPI) analysis focused on those groups where ALN was recommended. The NICE quality standard (QS149) provides a table of treatment thresholds that vary by age, with the intervention threshold in the lowest age group being a 10-year absolute fracture risk of 5.9% (NICE 2017). Based on this, we have presented EVPI results for the 8th, 9th and 10th risk categories, as the average risk in the 8th risk category is 5.5% (range of 4.6–6.6%).

Treatment offset

For all BP, it was assumed in the model that the treatment effect persists for some time after treatment is stopped, as the effect of the treatment on BMD does not reverse immediately at the point that treatment is discontinued. The time taken for the treatment effect to reduce to zero is described as the offset period in the model. In keeping with the assumptions made in previous cost-effectiveness analyses, we assumed that the offset period for ALN was equivalent to the treatment duration. This was based on evidence that it took 5 years for hip BMD to return to pre-treatment levels when treatment with ALN was discontinued after 5 years in the FLEX study. 166,167

For RIS, two studies reported findings that gains in hip BMD were lost in the year following treatment discontinuation. 160,168 For IBN-oral data were limited, but one study which used a daily dose of IBN found that hip BMD returned to pre-treatment levels 1 year after completing 1 year of treatment. 169 Based on this, we have assumed that RIS and IBN-oral have a 1-year offset period in the base-case analysis but have explored an offset period equal to treatment duration in a sensitivity analysis. IV ZOL is thought to have a longer offset period than ALN, as femoral neck BMD was found not to have returned to baseline after 3 years of IV ZOL, followed by 3 years of PLB in the HORIZON-PFT extension study. 168 Based on this, we have applied the assumption used in previous models, which is that 3 years of treatment with IV ZOL is expected to provide up to 10 years of effect. We have therefore assumed in the base-case analysis that the offset period for IV ZOL is 2.3 times the treatment duration [i.e. (10 − 3)/3 = 2.3]. We have explored setting the offset period equal to treatment duration for IV ZOL in a sensitivity analysis. For IBN-IV, we have made the same assumption as previous models,170 which have assumed that the offset time for IBN-IV is equivalent to that for IBN-oral as there is a lack of studies reporting treatment efficacy following discontinuation for IBN-IV.

Adverse effects

The review of RCTs described in Chapter 4 identified that arthralgia, headache, myalgia, pyrexia and influenza-like symptoms were all statistically significantly more common for patients having IV ZOL than for patients having PLB. No other BP treatment was found to have a statistically significant higher risk of these adverse effects when compared to PLB, although no data were available for IBN-IV. The cluster of symptoms associated with IV ZOL appears to occur within the same time period (within 3 days of infusion) and is sometimes referred to as an ‘acute-phase reaction’. 174 They are also more common at the time of the first infusion. We have therefore assumed that they occur together as a cluster of symptoms and have used the data reported for influenza-like symptoms to estimate the risk of patients experiencing an acute-phase reaction, which could include arthralgia, headache, myalgia, pyrexia or other influenza-like symptoms. The NMA for influenza-like symptoms was used to estimate the absolute difference in these symptoms between patients treated with IV ZOL and those treated with PLB. This gave an increased risk of influenza-like symptoms of 6.8% (95% CI 3.3% to 12.7%) attributable to treatment with IV ZOL. An identical risk was applied to IBN-IV as there were no comparative data and it was assumed that the acute-phase reaction would be similar for other IV BP. Symptoms were assumed to last 3 days and occur only on the first dose (HRQoL data applied during this period are discussed below). For patients experiencing flu-like symptoms following administration of IV BP, a QALY loss of 0.005 was applied based on the approach used in TA464 in which a utility multiplier of 0.35 was applied for a period of 3 days. This was based on a study reporting a utility value of 0.34 in patients with flu-like symptoms relative to a baseline value of 0.97 prior to flu-like symptoms. 175

The model developed to inform TA464 included a risk of 3% for GI adverse effects for oral BP based on a review of observational studies by Lloyd et al. 45 In that analysis, the risk of experiencing a GI adverse effect was estimated from a study reporting adverse effects for ALN, and it was assumed that an equal event rate would apply to all three oral BPs. However, in this analysis, we are interested in capturing any uncertainty regarding the difference in side effects between the different oral BPs. Therefore, in this analysis, we have applied the RRs versus ALB obtained from the NMA, to the absolute risk of 3%, to estimate the risk of GI adverse effects for RIS and IBN-oral. This gave risks of 3% (95% CI 2% to 5%) and 4% (95% CI 2% to 8%), respectively, for RIS and IBN-oral. We did not include any GI adverse effects for IV BP, as although they are reported, they were considered to be part of the ‘acute-phase reaction’, which has already been incorporated within the risk of influenza-like symptoms. For patients experiencing GI adverse effects, we have applied a QALY loss of 0.0075 and a cost of £39 for a GP appointment176 at the start of treatment.

Approach to sensitivity analysis

A probabilistic sensitivity analysis (PSA) was conducted to estimate the mean costs and QALYs gained when taking into account the uncertainty in the parameter values used in the model. The sampling of parameters was consistent with the methods previously reported by Davis et al. 45 The exceptions were the parameters capturing efficacy, treatment persistence and adverse events, which have all been updated in this analysis. For the estimates of treatment efficacy, we used the convergence diagnosis and output analysis (CODA) samples from the NMA described in Chapter 4. We assumed that the mean treatment persistence for ALN was normally distributed using the 95% CI calculated from the Kaplan–Meier data from Guyot et al. 163 We used beta distributions to sample the proportion discontinuing IV ZOL at 1 and 2 years165 and generated sampled estimates of mean treatment persistence for IV ZOL. For the remaining oral and IV treatments, we applied the CODA samples from the HRs for treatment persistence relative to the baseline treatment persistence curves for ALN and IV ZOL, respectively. For the adverse effects of GI symptoms on oral BP, we applied the CODA samples for the relative risk of GI symptoms in patients having RIS and IBN-oral compared to ALN. This was applied to a fixed risk of 3% for ALN. For the adverse effect of flu-like symptoms in patients having IV ZOL, we used the CODA samples from the NMA described in Chapter 4 to estimate the absolute risk of flu-like symptoms for IV ZOL and applied this identically to both IV BP.

For the base-case cost-effectiveness estimates, we ran the model for 2 million patients with a single set of parameter samples per patient. As cost-effectiveness is dependent on absolute fracture risk, results were presented by risk category, with each category representing a decile of absolute fracture risk and therefore informed by approximately 200,000 patients.

Structural sensitivity analyses were conducted to explore whether the results were sensitive to different model assumptions. To reduce model computation time, the structural sensitivity analyses were conducted using mid-point parameter inputs rather than using the full PSA version of the model.

Base-case results

The adverse clinical outcomes avoided for each BP treatment compared to no treatment are summarised in Table 14. These are based on average outputs from the probabilistic model (run for a cohort of 2 million patients, resampling parameter inputs for each patient). It can be seen that the fractures avoided by IV ZOL treatment are higher than for the other BP treatments. This is largely driven by the longer duration of treatment persistence and the longer offset period. Of the remaining treatments, ALN prevents the most hip and vertebral fractures, which translates into fewer fatal fractures and fewer new admissions to nursing/residential care. Although IBN-IV prevents slightly more fractures overall than ALN, this is driven by it preventing more wrist fractures, which have a smaller impact on costs and utilities than hip and vertebral fractures.

The base-case cost-effectiveness results from the probabilistic model are summarised in Table 15, which shows the incremental cost-effectiveness ratio (ICER) for each treatment relative to no treatment within each risk category. It can be seen that ALN has an ICER under £20,000 per QALY relative to no treatment from the sixth risk category, RIS from the eighth risk category and IBN-oral from the ninth risk category (10-year average risks of 2.7%, 5.5% and 8.4%, respectively). Table 16 shows the ICERs for each of the alternative BP relative to ALN. It can be seen that all of the alternative BPs, except IV ZOL, are dominated by ALN in that they have a higher cost and lower QALY gain. The ICERs for IV ZOL compared to ALN are greater than £30,000 per QALY across all risk categories. While ICERs are useful when determining which treatment is cost-effective compared to no treatment, it can be more helpful to use incremental net monetary benefit (INMB) to determine the most cost-effective intervention. Treatments which are cost-effective compared to no treatment have an INMB relative to no treatment that is greater than zero, and the optimal treatment has the highest INMB. Figure 7 shows the variation in INMB (when valuing a QALY at £20,000) with absolute risk. Results for risk levels above 40% were not plotted, as the estimates are unstable at high levels as they are based on very small numbers of patients.

FIGURE 7.

Incremental net monetary benefit for BP compared to usual care plotted against absolute risk.

From Figure 7 it can be seen that each of the BP, with the exception of IBN-IV, achieves an ICER under £20,000, relative to no treatment, once a certain level of absolute risk is reached because their INMB becomes positive. The risk levels required for the BP to achieve an ICER under £20,000 compared to no treatment are 1.9%, 4.0%, 6.7% and 26.9% for ALN, RIS, IBN-oral and IV ZOL. However, even at higher levels of absolute risk, ALN is the most cost-effective intervention, as it has the highest INMB. Figure 7 also shows the CIs around the estimates of INMB. These widen at higher levels of absolute risk, but there is no crossing of the CIs until the risk is above 30%. The CIs for INMB for ALN and RIS do overlap at higher levels of absolute risk, and they overlap substantially when the risk reaches 40%. This suggests there may be some uncertainty regarding the most cost-effective treatment, but only at very high levels of absolute risk.

Scenario analysis results

A scenario analysis was conducted to explore the impact of assuming that all treatments have an offset time equal to treatment duration, which is consistent with what was assumed for ALN. This scenario analysis was run using a cohort of 2 million patients but with parameters fixed at their mid-point values. Figure 8 shows the base-case results when running the model in this manner, and Figure 9 shows the scenario analysis assuming an offset time equal to treatment duration for all drugs. It can be seen by comparing Figures 8 and 9 that this had the impact of bringing the INMBs estimates for RIS and IBN-oral closer to those for ALN. Therefore, both RIS and IBN-oral have an ICER under £20,000 compared to no treatment in the 7th risk category in this scenario. It also reduced the INMBs for IV ZOL and increased the INMBs for IBN-IV. However, the overall conclusion that the INMB was highest for ALN across all 10 risk categories was unchanged.

FIGURE 8.

Incremental net monetary benefit compared to usual care for 10 risk categories when using mid-point parameter inputs – base case.

FIGURE 9.

Incremental net monetary benefit compared to usual care for 10 risk categories when using mid-point parameter inputs – assuming offset equal to treatment duration.

A scenario analysis was conducted exploring the impact of assuming that clinicians use two 4 mg vials to make up the 5 mg dose of IV ZOL. This reduced the risk level at which IV ZOL achieved an ICER under £20,000 versus usual care from 26.9% to 19.9%. However, it did not alter the conclusion that ALN was more cost-effective than IV ZOL at all levels of absolute risk. A scenario analysis was also conducted to explore increasing the annual cost of prescribing oral BP to reflect prescribing time and dispensing fees for monthly prescriptions in addition to drug costs. A cost analysis exploring the differences in costs between monthly and quarterly prescribing of repeat prescriptions from a UK NHS perspective assumed that GPs take just under a minute (49 seconds) to process a repeat prescription. 184 Applying the unit cost per hour of GP activity (£156 per hour) and the cost of pharmacy dispensing fees (£1.29 per item) would result in an additional cost of £40.96 per annum for oral BP or £78.64 across the average duration of treatment persistence. Under these assumptions, IV ZOL would still not be the optimal treatment, even at higher levels of absolute risk. However, the risk level at which ALN would achieve an ICER under £20,000 compared to no treatment would increase to 5.0%. This would have no impact on current recommendations as the lowest intervention threshold, according to the NICE Quality Standard, is a 10-year absolute fracture risk of 5.9%.

Expected value of parameter information results

Figure 10 shows the partial EVPI indexed to overall EVPI for risk categories 8–10 for various groups of parameters. It can be seen that the HRs for hip fracture account for a high proportion of the overall EVPI, and this is fairly consistent across risk categories 8–10 (denoted D8–D10 on the plot). It can also be seen that the HRs for any type of fracture for IBN-oral are also quite important and that both of these are being driven by a high EVPI for the HR of hip fracture in IBN-oral as a single parameter. The decision uncertainty associated with adverse events and treatment persistence is minimal in comparison. This is because there is a great degree of uncertainty around the HR for hip fracture for IBN-oral. Consequently, the probability that IBN-oral is the most cost-effective BP is 22%, 21% and 10% in risk categories 8, 9 and 10, respectively (when valuing a QALY at £20,000), whereas the probability that RIS is the most cost-effective BP is less than 5% in each of these risk categories, and the probability that either of the IV BP is the most cost-effective BP is < 0.1%.

FIGURE 10.

Expected value of perfect information for groups of parameters in risk categories 8–10 indexed to total EVPI.

Strengths and limitations

Limitations of this analysis include the minimal comparative data available for IBN-IV, which meant that it was necessary to assume that IBN-IV had efficacy similar to IBN-oral and adverse events similar to ZOL. Therefore, the results for IBN-IV should be treated with caution. In the NMA conducted to inform the NICE appraisal of BP, studies that reported outcomes for the daily 2.5 mg IBN-oral were included in the analysis, despite the fact that this was not a licensed dose of IBN-oral. However, in the previous analysis, efficacy estimates from this unlicensed dose were applied in the economic model for the licensed doses of IBN-oral (i.e. 150 mg monthly IBN-oral and the quarterly 3 mg IBN-IV), where data specifically from these licensed doses were lacking in the NMA. The approach taken in our review and economic evaluation was to limit the data to trials reporting outcomes for licensed doses of IBN, which better reflects the evidence gap for IBN-oral and IBN-IV. This is preferable given that one of the aims of this project was to identify where further research to reduce uncertainties in the current evidence base would be useful. Finally, our analysis was conducted from a UK NHS and PSS perspective, and the conclusions are not expected to apply to other countries or healthcare settings.

Step 1: gathering uncertainties

Uncertainties were gathered from (1) Chapters 2–5 and (2) existing published research recommendations. Over a series of four group meetings, the group study team reviewed and discussed the findings from (1) Chapters 2–5 and generated a list of potential arising uncertainties. A final meeting involved a patient advisory group (PAG) to further inform the process. Separately, a systematic search of relevant electronic databases and websites of professional organisations was conducted to identify (2) research recommendations highlighted within recent clinical guidelines. Databases searched included Epistemonikos, NICE, SIGN, Guidelines international network, Guidelines.co.uk and TRIP database. Inclusion criteria were (1) international guidelines from non-low- or middle-income country (LMIC), (2) about osteoporosis (including glucocorticoid osteoporosis), (3) published since 2016 and (4) developed on behalf of a professional organisation. Exclusion criteria were: (1) guidelines from LMIC, (2) about osteoporosis only in the context of another specific health condition, (3) published before 2016 and (4) written by individuals not representing a broader organisation. Attempts were made to translate guidelines that were not in the English language. Relevant sections on recommendations for research were extracted, and a list of research recommendations was produced. Subsequently, research recommendations were considered as in- or out-of-scope initially by two members of the study team (ZP, NC) and then approved by the whole team, with in-scope recommendations defined as relating to the use of BP. This generated a list of research recommendations.

Step 2: processing and refining uncertainties

Using stakeholder input, we refined the list of uncertainties from (1) and research recommendations (2) into research questions. One 3-hour stakeholder meeting was convened with patients and carers, clinicians (medical and non-medical) and academics to include representatives from primary and secondary care. Potential participants were invited from the ROS Effectiveness Working Group of the Bone Research Academy, Nottingham osteoporosis patient support group and clinical networks of the study team. We recorded the professional role and sex of attendees, but we did not collect data on age or ethnicity. The list of uncertainties and list of research recommendations [outputs from (1) and (2)] were circulated to attendees before the meeting. In the meeting, within small groups, the list of uncertainties were discussed and refined, with some uncertainties combined as appropriate. Attendees and study team members had the opportunity to suggest additional uncertainties during this process. The uncertainties were also categorised into groups. Each uncertainty was then refined into a research question with particular attention to defining the population and setting, intervention, comparison and outcomes of interest. 197 These were then combined with (2) forming a final list of research questions (3).

In order to validate that the research questions (3) were true research questions and not already answered, a search was subsequently conducted of the Cochrane Database of Systematic Reviews, PubMed and references of NICE guidelines, SIGN clinical guidelines, NOGG guidelines and ROS guidance for any relevant systematic reviews. If no systematic review was found to exist, the research question proceeded to Step 3.

Step 3: prioritisation

A full-day online workshop was convened in February 2022, aiming for between 12 and 30 participants to include a mix of patients, carers and primary and secondary care clinicians. Potential participants were invited as per the Step 2 workshop; in addition, the workshop was advertised on Twitter and via the Keele research User Group to particularly target lay, non-medical and primary care representatives. People were allocated on a first-come, first-served basis with the aim of achieving a balance of attendees across professional and lay groups. Study team members attended and acted as facilitators but did not vote or discuss ranking. Information on participant interests and disclosures was collected and reviewed to ensure balance across the group. Participants were sent the research questions in advance and asked to rank their top twenty questions before the workshop. Participants were permitted to send in pre-ranking if interested but unable to attend the workshop. In the workshop, an adapted nominal group technique was used. As per updated JLA guidance for online workshops, a four-step approach was used (removing a fifth plenary step, which has been difficult to operationalise online). 198 The workshop started with a plenary session to introduce the task and explain the background. Thereafter, four small groups compared and discussed their initial pre-workshop rankings. After a break, the same groups then produced their own combined ranking of at least the top 20 questions. The ranking of the four small groups was then combined and shared with the group in a plenary session. Finally, a second round of group prioritisation took place, to revise the shared ranking, in new small groups. These small group rankings were combined, reviewed and agreed as the final prioritised list.

Patient and public involvement

Members of the Nottingham ROS (NotROS) Support Group were involved in a series of meetings to discuss the design of the BLAST-OFF research programme and confirmed that understanding acceptability of BP from a range of perspectives was important. A PAG helped the study team identify the research uncertainties emerging from BLAST-OFF and public contributors were involved in both stakeholder groups (Steps 2 and 3).

Step 1: gathering uncertainties

The study team and PAG identified 22 uncertainties. Eleven uncertainties were informed by Chapter 2, 9 by Chapter 3, 11 by Chapter 4 and 7 by Chapter 5. The PAG talked about the importance of outcomes other than fracture, for example, meeting people’s information needs. They discussed and particularly informed uncertainties relating to how patients could be supported to make decisions, how treatment could be made easier and how effectiveness could be monitored.

Sixty-nine potential clinical guidelines were identified, of which 17 included relevant research recommendations (Figure 12).

FIGURE 12.

Search for clinical guidelines results.

Sixteen research recommendations were informed from the clinical guidelines; six of these overlapped with uncertainties from our study. In addition, the clinical guideline research recommendations highlighted populations in need of specific study, including men, people without BMD-defined osteoporosis, frail older adults, those with cognitive impairment and those with glucocorticoid-induced osteoporosis.

Step 2: processing and refining uncertainties

Eleven people attended Workshop 1. Characteristics of those attending are listed in Table 17.

The group was asked to consider the specific populations highlighted in the research recommendations when rewording and refining all the research uncertainties; younger adults emerged as a further group from discussion where further research was needed. Following the workshop, the uncertainties and research recommendations were finalised into 33 distinct research questions.

Step 3: prioritisation

Thirty-three questions went forward for prioritisation, organised into five categories relating to patient factors and patient support; clinical support and policy; safety; effectiveness and delivery. Twenty people attended Workshop 2, with a further individual (a GP) submitting individual rankings for consideration in the first small group work without attending. Characteristics of attendees were similar to those shown in Table 17.

The final top 10 priorities are shown in Box 1. Research questions 11–20 were also ranked, with the remainder unranked (attached in Appendices 6 and 7, Boxes 2–3).