Rapid fetal fibronectin testing to predict preterm birth in women with symptoms of premature labour: a systematic review and cost analysis

Aim

The aim of this project was to assess the impact of including fFN testing in the assessment of women with symptoms of preterm labour on NHS resource use and to propose possible changes in maternal management.

Objectives

1. To assess the clinical effectiveness and accuracy of the fFN test (commercial rapid test kit) in predicting PTB in symptomatic women.

2. To assess, from an NHS perspective, the cost-effectiveness of the use of fibronectin (rapid fFN testing) in the assessment of women with symptoms of threatened preterm labour, in comparison with no testing (current usual care).

The scope of this assessment did not include an evaluation of the effectiveness of treatment interventions to prevent PTB.

Population

Studies including pregnant women with singleton or twin gestations who have signs and symptoms of preterm labour (e.g. uterine contractions, dull backache, pelvic pressure, change in volume or consistency of vaginal discharge, and menstrual-like or intestinal cramping) before 37 weeks' gestation.

Setting

Secondary care.

Intervention

Studies assessing swab testing for fFN using a commercial rapid test kit before 37 weeks' gestation plus usual care, for the management of women with symptoms of preterm labour. Studies using rapid fFN test in participants after 37 weeks' gestation or studies assessing fFN for detecting any other risks than PTB were excluded from this review.

Comparator (clinical effectiveness studies only)

Usual care, without fibronectin testing, for managing PTB.

Reference standard (for test accuracy studies only)

Spontaneous PTBs which occur before 37 weeks' gestation, before 34 weeks' gestation, or within 7–10 days of testing.

Outcomes

• Incidence of spontaneous PTB before 37 weeks' gestation, before 34 weeks' gestation, or within 24 hours, 48 hours, or 7–10 days of testing (time required for corticosteroids to exert beneficial effects and the potential for in utero transfer and tocolytic administration) – primary outcome measure.

• Changes in maternal management.

  1. admission to hospital

  2. use of corticosteroids

  3. changes in frequency of monitoring

  4. changes from usual care.

• Outcomes in the newborn, morbidity, mortality.

• Outcomes of maternal health.

• Diagnostic accuracy of the test.

• Cost-effectiveness.

Study design

• Randomised trials in which participants are assigned to the intervention group or comparator group, and which report patient-relevant outcomes (changes to maternal management, maternal health outcomes, newborn morbidity and mortality) and/or incidence of PTB (before 37 weeks).

• Diagnostic cohort studies, published since the completion of the searches for the previous systematic review by Honest et al. ,10 in order to provide an updated estimate of test accuracy.

Included test accuracy studies were required to report sufficient data to construct 2 × 2 contingency tables [i.e. numbers of true-positive (TP), false-negative (FN), false-positive (FP) and true-negative (TN) test results].

The following study/publication types were excluded:

• studies with < 10 participants

• pre-clinical and animal studies

• reviews, editorials, and opinion pieces

• case reports and diagnostic case–control studies.

Clinical effectiveness

The clinical effectiveness searching was undertaken in two stages. In the first stage, RCTs and systematic reviews filters were applied to identify effectiveness studies. In the second stage, these filters were removed to allow identification of accuracy studies.

Effectiveness studies

These searches were an update of Honest et al. 6 and were limited by date from 2000 to September 2011. The following databases were searched for relevant studies:

• MEDLINE (OvidSP): 2000 to September week 1 2011

• MEDLINE In-Process & Other Non-Indexed Citations and Daily Update (OvidSP): 2000 to 15 September 2011

• EMBASE (OvidSP): 2000 to week 36 2011

• Cochrane Database of Systematic Reviews (CDSR) (Wiley): 2000 to Issue 9 2011

• Cochrane Central Register of Controlled Trials (CENTRAL) (Wiley): 2000 to Issue 3 2011

• Database of Abstracts of Reviews of Effects (DARE) (Wiley): 2000 to Issue 3 2011

• HTA Database (Wiley): 2000 to Issue 3 2011

• Science Citation Index (SCI) (Web of Knowledge): 2000 to 19 September 2011

• Cumulative Index to Nursing and Allied Health Literature (CINAHL) (EBSCOhost): 2000 to 9 September 2011

• Maternity and Infant Care (OvidSP): 2000 to August 2011

• National Institutes of Health (NIH) ClinicalTrials.gov (URL: www.clinicaltrials.gov/): 2000 to 19 September 2011

• Current Controlled Trials (URL: www.controlled-trials.com/): up to 19 September 2011

• WHO International Clinical Trials Registry Platform (ICTRP) (URL: www.who.int/ictrp/en/): up to 19 September 2011

• EU Clinical Trials Register (EUCTR) (URL: www.clinicaltrialsregister.eu/): up to 19 September 2011.

Accuracy studies

These searches were an update of Honest et al. 10,35 and were limited by date from 2005 to November 2011. Search strategies differed from those used by Honest et al. in that they did not include methodological terms for test accuracy studies.

The following databases were searched for relevant studies:

• MEDLINE (OvidSP): 2005 to November week 3 2011

• MEDLINE In-Process & Other Non-Indexed Citations (OvidSP) 2005 to 28 November 2011

• MEDLINE Daily Update (OvidSP): 2005 to 16 November 2011

• EMBASE (OvidSP): 2005 to week 47 2011

• Maternity and Infant Care (OvidSP): 2005 to November 2011

• CDSR (Wiley): 2005 to Issue 11 2011

• CENTRAL (Wiley): 2005 to Issue 4 2011

• DARE (Wiley): 2005 to Issue 4 2011

• HTA Database (Wiley): 2005 to Issue 4 2011

• CINAHL (EBSCOhost): 2005 to 29 November 2011

• SCI (Web of Knowledge): 2005 to 29 November 2011

• NIH ClinicalTrials.gov (URL: www.clinicaltrials.gov/): 2005 to 29 November 2011.

Identified references were downloaded in EndNote X5 software (Thomson Reuters, CA, USA) for further assessment and handling.

The bibliographies of retrieved articles and relevant systematic reviews were checked for additional studies.

Full search strategies are reported in Appendix 1.

Clinical effectiveness

Of the five included studies, four studies were published in full,47–50 whereas the remaining study was only published as a conference abstract. 51 Two studies (Lowe et al. 48 and Grobman et al. 50) were published as both full reports and conference abstracts;52,53 for these studies data extraction was based on the full reports. All the included studies were RCTs published 2002 or later. Four studies determined the impact of fFN testing on the maternal management. 47–51 Grobman et al. 50 also determined health care costs. Three studies were conducted in USA,48–50 one study was conducted in Portugal51 and Dutta and Norman47 was conducted in Scotland. Two studies were funded by Adeza Biomedical Corporation (Sunnyvale, CA, USA), which is the manufacturer of the fFN testing assay. 49,50 Participant recruitment was over a period of 1–2 years in all cases. An overview of the study design, objectives and outcomes reported by all studies is provided in Table 1. Further details of the inclusion/exclusion criteria, characteristics of study participants and details of the index test (fFN) are reported in the data extraction tables presented in Appendix 2.

All studies followed standard methods for rapid fFN testing. Two studies reported the use of Adeza Tli™ to perform fFN testing. 49,50 These studies randomised patients, after rapid fFN testing, to the intervention group, in which case the physicians had knowledge of fFN test results, or the control group, in which case the physicians were unaware of the test results. 49,50 The test results were communicated to the treating physician by the resident physician50 or by laboratory personnel. 49 In the remaining studies, the participants were randomised to rapid fFN testing performed for managing preterm labour or to a control group in which rapid fFN testing was not performed.

The inclusion and exclusion criteria of the studies differed to some extent. All studies included women with signs and symptoms of preterm labour. All studies included women with an estimated gestation age within the range of 24–36 weeks, except for the study of Lowe et al. ,48 which recruited women with gestational ages between 23 and 24 weeks. The study by Lowe et al. 48 was the only one to allow the inclusion of multiparous women with cervical dilation of 3–4 cm. Three studies excluded women with cervical dilation ≥ 3 cm. 47,49,50 The remaining study was published as an abstract only and did not report the exclusion criteria. 51 Two studies included women with singleton or twin gestations. 48,49 However, Grobman et al. 50 and Dutta and Norman47 reported that they included only women with singleton gestations.

The five studies included a total of 541 women with symptoms of preterm labour; sample size ranged from 66 to 108 participants. All studies, with the exception of that by Osório et al. ,51 reported some form of sample size calculation for the primary outcome(s). Lowe et al. 48 calculated a sample size of 50 participants in each arm to detect a significant reduction in the length of stay of at least 1.3 days with 80% power. Dutta and Norman47 powered their study to detect a 40% reduction in the number of hospital admissions; the estimated sample size required for 85% power to detect a significant difference was 304 participants. Grobman et al. 50 was powered to detect 20% reduction in total health-care cost in the fFN group. Plaut et al. 49 estimated a required sample size of 500 women to detect a significant difference in transport to tertiary care centre; however, it is unclear how this estimate was calculated and the study was terminated prematurely because of low enrolments. All studies appeared to have made power calculations based on the whole study population, rather than on the index test (fFN) negative population; the latter option would be more appropriate as, if conservative management is the norm, only test-negative patients have the potential for changed management and outcomes. All the included studies were, therefore, likely to be underpowered. In all included trials, treatment decisions were at the discretion of clinicians, not based on fFN results alone. The trials may therefore provide important information about the consequences when fFN is used in clinical context. All studies reported that there was no significant difference in baseline demographic and clinical characteristics between patients in the index test and control groups.

None of the included studies was judged to be ‘high risk’ of bias overall (Figure 2). An overall rating of ‘high risk’ was defined as ‘high risk’ for any of three key domains: randomisation sequence, allocation concealment and blinded outcome assessment. Poor reporting resulted in a high number of ‘unclear’ risk of bias ratings across studies. Low et al. 's48 was the only study to be judged at ‘low risk’ of bias for two of the key domains. One study was a conference abstract so was judged to be at unclear risk of bias for all domains as the information reported was not sufficient to make any definitive judgement. 51 The complete risk of bias assessment along with relevant quotes from the included papers and review authors' judgements is provided in Appendix 3. The quality assessment of the included studies for all the domains of Cochrane risk of bias tool is summarised below. One study was judged to be at ‘high risk’ of bias for the ‘incomplete outcome data domain because drop outs and protocol violations were excluded from the analyses,47 and one study was judged to be at ‘high risk’ of bias for ‘selective outcome reporting’, focusing on the outcome measure where a significant effect was observed and, additionally, because it was stopped prematurely due to low enrolment. 49

Overall, for all the domains across all the included studies, the majority of studies were rated ‘unclear risk’ of bias (Figure 3).

FIGURE 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

FIGURE 3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Estimated gestational age at delivery

Three studies reported the mean gestational age at delivery in weeks. 47,49,50 No individual study found any significant difference in gestational age at delivery between the index test and control groups (see Table 2). Similarly, the pooled estimate showed no significant difference (Figure 4), indicating that the clinicians' knowledge of fFN test results did not affect gestational age at delivery. The study by Lowe et al. 48 was not included in the meta-analysis because they reported median gestational age at time of delivery; this study also found no significant difference between the test and control groups. 48

TABLE 2 - Incidence of PTB and gestational ages at delivery

IQR, interquartile range; NR, not reported; SD, standard deviation.

Mistake in paper (reported as days).

Study ID	Main outcomes	fFN test: fFN testing done	Comparator: fFN testing not done/not known	p-value
Dutta 201147	Gestational age at time delivery in weeks (mean) (± SD)a	38.07 (3.25) (n = 43)	38.09 (2.33) (n = 38)	0.970
	Incidence of PTB/N (%) within 7 days	NR	NR	NR
Grobman 200450	Gestational age at time delivery in weeks (mean) (± SD)	38 ± 3	38 ± 3	0.810
	Incidence of PTB/N (%)	10 (20)	13 (26)	0.480
Lowe 200448	Gestational age at time delivery in weeks (median) (IQR)	38.3 (36.0–38.9)	37.4 (35–39)	0.258
	Incidence of PTB/N (%)	NR	NR	NR
Osório 201051	Gestational age at time delivery in weeks (mean) (± SD)	NR	NR	NR
	Incidence of PTB/N (%)	NR	NR	NR
Plaut 200349	Gestational age at time delivery in weeks (mean) (± SD)	38.2 ± 2.6	37.7 ± 2.4	0.860
	Incidence of PTB/N (%) within 14 days	2 (4)	1 (2)	NR

FIGURE 4.

Forest plot of mean gestational ages at delivery in weeks. *The labels for the forest plot have been swapped because of the positive outcome.

Length of maternal hospital stay

Length of hospital stay was a common outcome, reported in all the included studies. None of the studies reported significant difference for this outcome among all the randomised patients (Table 3). Plaut et al. 49 compared the length of hospital stay in patients who tested negative for fFN testing and in whom the test result was known to clinicians (index test group) with test-negative patients for whom the result was not disclosed (control group); no significant difference was found. A subgroup analysis of women with negative fFN test observed for > 6 hours showed a significant reduction in the length of hospital stay when the test result was known to clinicians. The hospital stay was shortened by 40%, from 37.8 to 22.7 hours (p = 0.04) (see Table 3). However, the sample size for this analysis was very small and it was not clear whether or not the analysis had been planned a priori. The unit of measurement for length of hospital stay varied across studies; where possible, we standardised extracted data to number of days spent in hospital and the results of individual studies are presented in a forest plot (Figure 5). Grobman et al. 50 reported median length of stay and Osório et al. 51 reported a dichotomous outcome (number of women with hospital stays > 6 days); neither study found a significant difference between the index test and control groups.

TABLE 3 - Length of hospital stays (days)

IQR, interquartile range; NR, not reported.

Median days (IQR).

Dichotomous outcome.

Study ID	fFN testing done	fFN testing not done/not known	p-value
Dutta 201147	0.736 (± 1.05) days (mean ± SD) (n =40) (interval of decision to admit to decision to discharge)	0.699 (± 1.05) (mean ± SD) (n = 37)	0.878
Grobman 200450	2 days (1–5)a (during admission at study entry)	2 days (1–5)a	0.830
	4 days (2–7)a (during admission after study entry)	2 days (1–11)a	0.620
Lowe 200448	NR	NR	0.224
Osório 201051	33.3% (2/6) women stayed in hospital for > 6 daysb	44.4% (4/9) stayed in hospital for > 6 daysb	0.680
Plaut 200349	Women with negative fFN test: 0.28 (± 0.28) days mean (± SD) (n = 47) (observation period + any admissions)	Women with negative fFN test: 0.34 (± 0.28) days mean (± SD) (n = 51)	0.350
	Women with negative fFN test: observed for > 6 hours 0.95 (± 0.6) days mean (± SD) (n = 10)	Women with negative fFN test: observed for > 6 hours 1.58 (± 0.6) days mean (± SD) (n = 8)	0.040

FIGURE 5.

Forest plot of length of stay.

Incidence of hospital admissions

Four studies reported the number of hospital admissions before delivery (Table 4). 47,48,50,51 Grobman et al. 50 reported the number of hospital admissions at study entry and number of admissions for preterm contractions any time after study entry separately. Individual study results (see Figure 6 and Table 4) indicate a lower incidence of maternal admissions in the fFN test group than in the control group for three out of four studies;47,50,51 however, no study showed a statistically significant difference between groups. The study by Lowe et al. 48 was the only study that numerically favoured the no fFN testing group, but the authors reported that there were significantly fewer antepartum hospital admissions among the women with negative fFN test results (p = 0.032). The pooled RR for hospital admission showed no significant difference between the fFN test and control groups (RR 0.93%; CI 0.66% to 1.3%) (Figure 6). The study by Dutta and Norman47 was the only study evaluating the number of admissions to neonatal intensive care unit (NICU); although this study reported a higher number of NICU admissions in fFN test group, the difference was not statistically significant. 47 The conference abstract51 did not report the number of participants randomised to each group but after observing the data carefully we assumed that 33 women were allocated to each group. This assumption had to be made to calculate odds ratio for forest plot in Figure 6.

TABLE 4 - Number of hospital/NICU admissions

Study ID	Hospital or NICU admissions	fFN testing done	fFN testing not done/not known	p-value
Dutta 201147	Antepartum admissions (%)	21 (42.9)	22 (50)	0.490
	Incidence of NICU admission (%)	10 (33.3)	3 (10)	0.080
Grobman 200450	Hospital admissions at study entry (%)	13 (26)	14 (28)	0.440
	Hospital admission for preterm contractions at any time after study entry (%)	5 (10)	4 (8)	0.780
Lowe 200448	Antepartum admissions (%)	16 (35)	12 (24)	0.265
Osório 201051	Hospital admissions for PTB (%)	6/33 (18.2)	9/33 (27.3)	0.560

FIGURE 6.

Analysis on number of hospital admissions.

Treatments administered

Three studies reported the use of tocolytic agents and corticosteroids (Table 5). 47,48,50 The meta-analysis in Figure 7 indicates that there was no significant difference in usage of tocolytic agents between two groups (pooled RR 1.0; 95% CI 0.69 to 1.44). Similarly, the meta-analysis in Figure 8 indicates that there was no significant difference in usage of corticosteroids between the two groups (pooled RR 0.93; 95% CI 0.68 to 1.27). Lowe et al. 48 also reported administration of an antibiotic therapy with no significant difference in usage between the groups. Plaut et al. 49 reported the administration of aggressive therapy which included use of tocolysis, corticosteroids and transfer to a tertiary care facility. Fourteen women were administered aggressive tocolytic therapy, of whom three delivered within 14 days and remaining 11 delivered after 14 days. 49 However, there was no significant difference in use of aggressive therapy between the two groups (see Table 5).

TABLE 5 - Treatment outcomes

Study ID	Treatments administered	fFN testing done	fFN testing not done/not known	p-value
Dutta 201147	Tocolysis (%)	3 (6.5)	4 (8.7)	1.000
	Corticosteroids (%)	17 (37)	21 (45.7)	0.397
Grobman 200450	Tocolysis (%)	8 (16)	9 (18)	0.790
	Corticosteroids (%)	8 (16)	10 (20)	0.600
Lowe 200448	Tocolysis (%)	22 (48)	23 (45)	0.840
	Corticosteroids (%)	23 (50)	22 (43)	0.545
	Antibiotics (%)	17 (37)	21 (41)	0.683
Plaut 200349	Aggressive therapy (%)	8 (16)	6 (11)	0.430

FIGURE 7.

Analysis on usage of tocolytic agents.

FIGURE 8.

Analysis on usage of corticosteroids.

Test accuracy

The 15 DTA studies identified by our update searches included a total of 2379 participants (range 38–516 participants). The majority of these studies reported data for more than one outcome (preterm delivery at various gestational ages and times from testing); 10 studies reported data for PTB within 7–10 days of testing,32,56–63 seven studies reported data for PTB before 34 weeks' gestation55,57,60,63–66 and seven studies reported data for PTB before 37 weeks' gestation. 54,55,60,63–65,67 In addition, four studies reported data for PTB before 35 weeks' gestation which were grouped with the 34 weeks category,54,58,59,61 and one study reported data for PTB before 38 weeks' gestation, which were grouped with the 37 weeks category. 61 Four studies included only women with singleton pregnancies,32,54,59,61 and the remainder either included both singleton and multiple pregnancies or did not report any inclusion/exclusion criteria for this factor. Eight studies reported the use of Adeza Biomedical Corporation fFN test kits,55–57,59,60,62,63,66 two studies used QuickCheck™ (Adeza Biomedical, Sunnyvale, CA, USA)/FullTerm fFN testing kits54,64 and one study used both. 58 Two studies did not specify the brand name of the kit used for fFN testing. 61,67 There were three non-English-language articles. For one Spanish article67 the data extraction and quality assessment was done by a native speaker. For remaining two articles, one Turkish68 and one Italian,65 limited data (results only) were taken from the English-language abstract; these studies are not included in Appendices 4 and 6. Further details of the inclusion/exclusion criteria, characteristics of study participants and details of the index test (fFN) are reported in the data extraction tables presented in Appendix 4.

The main risk of bias for these studies related to the ‘patient selection’ domain of our modified version of QUADAS-2; only three studies reported prospective, consecutive recruitment of participants. 54–56 The nature of the intervention meant that most included studies used commercial test kits, minimising the potential for bias arising from the conduct of the index test. Finally, the majority of included studies reported data for all participants. The results of QUADAS-2 assessment are summarised in Table 6 and full assessments for each study are provided in Appendix 6.

In addition to the 15 new studies described, data from 39 DTA studies included in the appendix of a previously published systematic review of fFN testing for the prediction of PTB were included in our meta-analysis. 35 Seventeen of these studies reported data for PTB within 7–10 days of testing, eight studies reported data for PTB before 34 weeks' gestation and 31 reported data for PTB before 37 weeks' gestation. Sixteen of the 39 studies included only women with singleton pregnancies and the remainder either included both singleton and multiple pregnancies or did not report any inclusion/exclusion criteria for this factor. The review from which these studies were taken used the authors' own, topic-specific, quality assessment tool; however, it was possible to determine from the data extraction tables that 10 of the 39 studies had reported prospective, consecutive recruitment of participants. The results and main characteristics of these studies are summarised in Appendix 7.

Accuracy of fetal fibronectin for the prediction of preterm birth within 7–10 days of testing

A total of 27 studies reported data on the accuracy of fFN testing to predict preterm delivery within 7–10 days of testing. Ten studies were identified by our update searches and 17 were taken from the previous systematic review, as described above. The results of the 10 new studies are summarised in Table 7. The pooled estimates of sensitivity and specificity, derived from these data using a bivariate model, were 76.7% (95% CI 70.4% to 82.0%) and 82.7% (95% CI 79.4% to 85.5%), respectively. The I² statistic indicated low between-study heterogeneity in the estimates of sensitivity (I² = 24.8%) and high between-study heterogeneity in the estimates of specificity (I² = 84.5%). Figure 9 shows individual studies, along with the summary estimate, plotted in receiver operating characteristic (ROC) space. Subgroup analyses, using a bivariate model, also showed similar estimates of test performance for studies that included only women with singleton pregnancies compared with unselected populations, and for studies identified by our update searches compared with studies from the previously published review (Table 8).

TABLE 7 - Accuracy of fFN for the prediction of PTB within 7–10 days of testing

NR, not reported.

Calculated values.

Turkish-language study, data extracted from English-language abstract.

Study ID	Description of arm and diagnostic threshold	TP	FN	FP	TN	Sensitivity (95% CI)	Specificity (95% CI)	Adverse events
Desjardins 200858	≥ 50 ng/ml	6	4	23	328	60.0% (26.2% to 87.8%)a	93.4% (90.3% to 95.8%)a	NR
Diaz 200954	≥ 50 ng/ml	18	6	34	122	75.0% (52.9% to 89.4%)	78.2% (70.7% to 84.2%)	Women with positive fFN had higher rate of adverse neonatal outcomes
Eroglu 200759	≥ 50 ng/ml (from kit manual)	5	1	9	36	83.3% (35.9% to 99.6%)a	80.0% (65.4% to 90.4%)a	Bacterial vaginosis
Groom 200660	≥ 50 ng/ml	7	3	24	145	70.0% (34.8% to 93.3%)a	85.8% (79.6% to 90.7%)a	NR
Henrich 201061	NR	5	0	17	59	100.0% (47.8% to 100.0%)a	77.6% (66.6% to 86.4%)a	NR
MacDonald 200762	≥ 50 ng/ml (from kit manual)	4	0	3	31	100.0% (39.8% to 100.0%)a	91.2% (76.3% to 98.1%)a	NR
Skoll 200657	≥ 50 ng/ml	12	3	20	114	80.0% (51.4% to 94.7%)	85.1% (77.6% to 90.4%)	NR
Swamy 200563	> 50 ng/ml	14	7	31	352	66.7% (43.0% to 85.4%)a,b	91.9% (88.7% to 94.4%)a,b	NR
Sümer 201068	≥ 50 ng/ml	1	4	7	55	20.0% (5.0% to 71.6%)a	88.7% (78.1% to 95.3%)a	NR
Tsoi 200656	≥ 50 ng/ml (from kit manual)	18	1	67	109	94.7% (74.0% to 99.9%)a	61.9% (54.3% to 69.1%)a	NR

FIGURE 9.

Receiver operating characteristic space plot of studies of fFN for the prediction of PTB within 7–10 days of testing.

TABLE 8 - Subgroup analyses of accuracy of fFN for the prediction of PTB within 7–10 days of testing

Data set	Sensitivity (95% CI)	Specificity (95% CI)
All studies (n = 27)	76.7% (70.4% to 82.0%)	82.7% (79.4% to 85.5%)
Studies of singleton pregnancies (n = 12)	75.8% (63.2% to 85.1%)	81.1% (75.8% to 85.6%)
Studies of unselected populations (n = 15)	76.4% (68.6% to 82.8%)	83.6% (79.6% to 87.0%)
Studies identified by update searches (n = 10)	76.3% (63.8% to 85.4%)	85.0% (78.8% to 89.6%)
Studies taken from previous systematic review10 (n = 17)	77.1% (69.4% to 83.4%)	81.7% (78.3% to 84.7%)

Accuracy of fetal fibronectin for the prediction of preterm birth at < 34 weeks' gestation

A total of 19 studies reported data on the accuracy of fFN testing to predict preterm delivery at < 34 weeks' gestation. Eleven studies were identified by our update searches and eight were taken from the previous systematic review, as described above. The results of the 11 new studies are summarised in Table 9. The pooled estimates of sensitivity and specificity, derived from these data using a bivariate model, were 69.1% (95% CI 58.6% to 77.9%) and 84.4% (95% CI 79.8% to 88.2%), respectively. The I² statistic indicated low between-study heterogeneity in the estimates of sensitivity (I² = 75.5%) and high between-study heterogeneity in the estimates of specificity (I² = 85.4%). Figure 10 shows individual studies, along with the summary estimate, plotted in ROC space. Subgroup analyses, using a bivariate model, also showed similar estimates of test performance for studies that included only women with singleton pregnancies versus unselected populations, and for studies identified by our update searches compared with studies from the previously published review (Table 10). A sensitivity analysis was carried out, which excluded four studies with a reference standard of PTB at < 35 weeks' gestation;54,58,59,61 there was no significant change in the results when these four studies were excluded.

TABLE 9 - Accuracy of fFN for the prediction of PTB at < 34 weeks' gestation

NR, not reported.

Calculated values.

Reference standard PTB (< 35 weeks' gestational age).

Study ID	Description of arm and diagnostic threshold	TP	FN	FP	TN	Sensitivity (95% CI)	Specificity (95% CI)	Adverse events
Reference standard outcome: preterm delivery (< 34 weeks' gestation)
Asakura 200955	> 50 ng/ml	10	6	11	81	62.5% (35.4% to 84.8%)a	88.0% (79.6% to 93.9%)a	NR
Audibert 201064	≥ 50 ng/ml	7	7	7	41	50.0% (23.0% to 77.0%)	85.0% (72.0% to 94.0%)	NR
bDesjardins 200858	≥ 50 ng/ml	14	22	15	310	38.9% (23.1% to 56.5%)a	95.4% (92.5% to 97.4%)a	NR
bDiaz 200954	≥ 50 ng/ml	12	0	40	128	100.0% (69.9% to 100.0%)	76.2% (68.9% to 82.3%)	NR
Driul 200965	≥ 50 ng/ml	11	4	31	36	73.3% (44.9% to 92.2%)a	53.7% (41.1% to 66.0%)a	NR
bEroglu 200759	NR	7	3	7	34	70.0% (34.8% to 93.3%)a	82.9% (67.9% to 92.8%)a	Bacterial vaginosis
Groom 200660	≥ 50 ng/ml	13	1	18	147	92.9% (66.1% to 99.8%)a	89.1% (83.3% to 93.4%)a	NR
bHenrich 201061	NR	10	2	12	57	83.3% (51.6% to 97.9%)a	82.6% (71.6% to 90.7%)a	NR
Singer 200766	≥ 50 ng/ml	19	21	61	415	47.5% (31.5% to 63.9%)a	87.2% (83.8% to 90.1%)a	NR
Skoll 200657	≥ 50 ng/ml	17	10	15	107	63.0% (42.4% to 79.9%)	87.6% (80.1% to 92.7%)	NR
Swamy 200563	> 50 ng/ml	27	38	20	319	41.5% (29.4% to 54.4%)a	94.1% (91.0% to 96.4%)a	NR

FIGURE 10.

Receiver operating characteristic space plot of studies of fFN for the prediction of PTB at < 34 weeks' gestation.

TABLE 10 - A sensitivity and subgroup analyses of accuracy of fFN for the prediction of PTB at < 34 weeks' gestation

Data set	Sensitivity (95% CI)	Specificity (95% CI)
All studies (n = 19)	69.1% (58.6% to 77.9%)	84.4% (79.8% to 88.2%)
Studies of singleton pregnancies (n = 9)	76.4% (57.7% to 88.5%)	82.4% (78.9% to 85.3%)
Studies of unselected populations (n = 10)	62.7% (49.6% to 74.2%)	85.0% (75.8% to 91.1%)
Studies identified by up-date searches (n = 11)	65.2% (51.8% to 76.5%)	86.3% (80.1% to 90.8%)
Studies taken from previous systematic review10 (n = 8)	74.0% (56.1% to 86.3%)	82.1% (77.6% to 85.8%)
Sensitivity analysis excluding studies with reference standards < 35 weeks' gestation (n = 15)	67.4% (56.3% to 76.8%)	83.8% (78.5% to 88.0%)

Accuracy of fetal fibronectin for the prediction of preterm birth at < 37 weeks' gestation

A total of 39 studies reported data on the accuracy of fFN testing to predict preterm delivery at < 37 weeks' gestation. Eight studies were identified by our update searches and 31 were taken from the previous systematic review, as described above. The results of the eight new studies are summarised in Table 11. The pooled estimates of sensitivity and specificity, derived from these data using a bivariate model, were 60.8% (95% CI 53.7% to 67.6%) and 85.3% (95% CI 82.5% to 87.7%), respectively. The I² statistic indicated low between-study heterogeneity in the estimates of sensitivity (I² = 83.7%) and high between-study heterogeneity in the estimates of specificity (I² = 72.9%). Figure 11 shows individual studies, along with the summary estimate, plotted in ROC space. Subgroup analyses, using a bivariate model, also showed similar estimates of test performance for studies that included only women with singleton pregnancies versus unselected populations, and for studies identified by our update searches compared with studies from the previously published review (Table 12). A sensitivity analysis was performed, which excluded one study with a reference standard of PTB at < 38 weeks' gestation;61 there was no significant change in the results excluding this study.

TABLE 11 - Accuracy of fFN for the prediction of PTB at < 37 weeks' gestation

NR, not reported.

Calculated values.

Reference standard PTB (< 38 weeks' gestational age).

Study ID	Description of arm and diagnostic threshold	TP	FN	FP	TN	Sensitivity (95% CI)	Specificity (95% CI)	Adverse events
Reference standard outcome: preterm delivery (< 37 weeks' gestation)
Asakura 200955	> 50 ng/ml	14	26	7	61	35.0% (20.6% to 51.7%)a	89.7% (79.9% to 95.8%)a	NR
Audibert 201064	≥ 50 ng/ml	11	12	3	36	48.0% (35.0% to 60.0%)	92.0% (86.0% to 99.0%)	NR
Diaz 200954	≥ 50 ng/ml	38	12	14	116	76.0% (61.5% to 86.5%)	89.2% (82.3% to 93.8%)	NR
Driul 200965	≥ 50 ng/ml	25	14	17	26	64.1% (47.2% to 78.8%)a	60.5% (44.4% to 75.0%)a	NR
Farfan 201167	≥ 50 ng/ml (from kit manual)	25	2	5	34	92.6% (75.7% to 99.1%)a	87.2% (72.2% to 95.7%)a	NR
Groom 200660	≥ 50 ng/ml	18	17	13	131	51.4% (34.0% to 68.6%)a	91.0% (85.1% to 95.1%)a	NR
bHenrich 201061	NR	17	12	5	47	58.6% (38.9% to 76.5%)a	90.4% (79.0% to 96.8%)a	NR
Swamy 200563	> 50 ng/ml	30	90	17	267	25.0% (17.5% to 33.7%)a	94.0% (90.6% to 96.5%)a	NR

FIGURE 11.

Receiver operating characteristic space plot of studies of fFN for the prediction of PTB at < 37 weeks' gestation.

TABLE 12 - A sensitivity and subgroup analyses of accuracy of fFN for the prediction of PTB at < 37 weeks' gestation

Data set	Sensitivity (95% CI)	Specificity (95% CI)
All studies (n = 39)	60.8% (53.7% to 67.6%)	85.3% (82.5% to 87.7%)
Studies of singleton pregnancies (n = 16)	66.4% (53.7% to 77.2%)	85.6% (80.3% to 89.7%)
Studies of unselected populations (n = 23)	57.3% (48.9% to 64.8%)	85.0% (81.7% to 87.8%)
Studies identified by up-date searches (n = 8)	57.1% (40.4% to 72.3%)	88.7% (82.7% to 92.8%)
Studies taken from previous systematic review10 (n = 31)	61.7% (53.9% to 69.0%)	84.2% (81.1% to 86.9%)
Sensitivity analysis excluding studies with reference standards < 38 weeks' gestation (n = 38)	60.9% (53.5% to 67.8%)	85.4% (82.4% to 88.0%)

Search strategy

Focused searches were undertaken to identify economic evaluations of the fFN test. No date limits were applied to these searches. The following resources were searched:

• MEDLINE (OvidSP): 1946 to January week 4 2012

• MEDLINE In-Process & Other Non-Indexed Citations (OvidSP): up to 2 February 2012

• MEDLINE Daily Update (OvidSP): up to 2 February 2012

• EMBASE (OvidSP): 1980 to week 4 2012

• DARE (Wiley): up to Issue 1 2012

• HTA Database (Wiley): up to Issue 1 2012

• NHS Economic Evaluation Database (NHS EED) (Wiley): up to Issue 1 2012

• Paediatric Economic Database Evaluation (PEDE) (URL: http://pede.ccb.sickkids.ca/pede/search.jsp): 1980–2010.

Appendix 1 gives a full specification of search strategies.

All references were downloaded in EndNote X5 software and were further screened for inclusion.

Review of economic analyses on fibronectin

The objective of the review of extant economic evaluations was to summarise methods and findings of existing peer-reviewed studies. A total of 88 titles and abstracts were screened, from which we selected 12 studies. After a further full-text screening, only two studies were kept. These studies matched our criteria of a full economic analysis in which the fFN test was compared with an alternative option for predicting preterm labour. Studies that did not include ICERs were excluded from the review. A summary of the studies and the quality assessments is provided in Appendix 8.

Mozurkewich et al. 69 developed a decision-analytic model to compare nine different treatment strategies for the management of women presented with threatened preterm labour (i.e. regular uterine contractions at 24–34 weeks, no cervical dilation, and intact uterine membranes). The treatment consisted of the administration of one of the fibronectin tests (fFN or fibronectin rapid test) and parenteral corticosteroids and/or tocolytics for the prevention of RDS. The strategies compared were:

1. Treat all women with corticosteroids and tocolytics.

2. Treat all with tocolytics and corticosteroids until the results of fFN tests were available and discharge those with negative results.

3. Discharge only women with a cervical length measure > 26 mm on the (vaginal or transperineal) ultrasonography.

4. Discharge only women with a negative result on the rapid fibronectin test.

5. Discharge only women that have a negative rapid fibronectin test or a cervical length > 26 mm.

6. Do not treat any women with corticosteroids or tocolytics.

7. Treat all women with outpatient corticosteroids but not with tocolytics.

8. Treat all women with corticosteroids but administer tocolytics to those with abnormal results on rapid fibronectin test.

9. Treat all women with corticosteroids but give tocolytics only to women with an abnormal measure of the cervical length.

The health outcomes considered were neonatal death and RDS and time horizon was set until the time of hospital discharge. Accuracy data for fFN tests were obtained from Revah et al. 70 The cost data for the study came from statistical data of University of Michigan Hospital and the literature review. Total costing for each of the strategies was calculated by adding up the costs for outpatient treatments, fibronectin testing, cervical length measurement, hospitalisations and treatment, maternal delivery, and neonatal care (until death or discharge). The most cost-effective strategy (extended dominance) in terms of costs per neonatal death prevented were strategy 8 (rapid fibronectin plus corticosteroids and tocolysis only in those with abnormal fFN results), with an average cost of $13,000 (1999 prices in Canadian dollars) and 39 deaths/1000, and strategy 2 (treating all until results of fFN tests were available and discharge those with negative results) with an average cost of $13,600 (1999 prices in Canadian dollars) and 39 deaths/1000. The most cost-effective strategy for the prevention of RDS was strategy 2, with an average cost of $13,600 (1999 prices in Canadian dollars) and 53 RDS cases/1000.

Tsourapas et al. 71 developed a decision-analysing model to assess the cost-effectiveness of alternative ‘test-and-treat’ strategies in the prevention of PTB before weeks 34 and 37. The study compared the results of six models defined according to population and outcome as follows.

• Model 1: Symptomatic women (with a viable PTB experiencing preterm labour) – giving birth before 37 weeks.

• Model 2: Symptomatic women – giving birth before 34 weeks.

• Model 3: Symptomatic women – giving birth within 7 days of treatment.

• Model 4: Symptomatic women – giving birth within 48 hours of treatment.

• Model 5: Asymptotic women – having threatened preterm labour before 37 weeks.

• Model 6: Asymptotic women – having threatened preterm labour before 34 weeks.

The models combined all possible ‘test-and-treat’ combinations. The alternative tests consisted of fFN testing, highly phosphorylated insulin-like growth factor-binding proteins (phIGFBPs), C-reactive protein (CRP), absence of fetal breathing and previous history of PTB. The treatment consisted of administering progestational agents (i.e. atosiban, indomethacin, calcium channel blockers, magnesium sulphate and terbutaline). The comparators were no screening testing and no intervention.

The cost data for the study came from literature reviews and statistical data of the Birmingham Women's Hospital, which were adjusted to 2006 prices (pounds sterling). Test accuracy and treatment effectiveness data came from a meta-analysis of the systematic literature review carried out by the authors. 10

The results of model 1 show that the least expensive strategy is to conduct the fFN test and administer indomethacin to those who test positive. The cost of this strategy amounts to £2053. However, this is not the most cost-effective strategy. The ‘no test/administer indomethacin to all’ strategy costs £2609 but saves 34 cases of spontaneous PTB per 1000 women. The ICER for this strategy was estimated to be £16,336 per additional case of spontaneous PTB, making it the most cost-effective. Other noteworthy cost-effective strategies were:

• In model 2 (avoiding premature births before week 34 for symptomatic women), testing with the amniotic fluid interleukin 6 test and providing hydration to those who tested positive (ICER £4976 per additional threatened PTB avoided).

• In model 3 (avoiding premature births before 7 days of hospitalisation for symptomatic women), using the cervical length measurement < 15 mm test and administer indomethacin to those who tested positive (ICER £1703 per additional threatened PTB avoided).

• In model 4 (avoiding premature births before 48 days of hospitalisation for symptomatic women) using the cervical length measurement < 15 mm test and administer indomethacin to those who tested positive (ICER £5268 per additional threatened PTB avoided).

Conclusions of review

Both of these studies showed that fibronectin testing could be cost-effective, depending on its place in the care pathway. However, whether or not testing was cost-effective depended on there being a difference in birth timing, which was not supported by the trial evidence from the systematic review. Therefore, we conducted a de novo analysis (see next section).

Model structure and methodology

Full cost-effectiveness modelling was not feasible, as evidence from the systematic review indicated that fFN testing had no significant effect on outcome in terms of live births/PTBs/gestational age at delivery. Although the upper boundary of the CI of the pooled gestational age at delivery was positive (0.83, see Figure 4), it was decided that this was not clinically relevant as the studies were not designed or powered to detect a difference or equivalence in gestational age. In fact, they were all explicitly intended to detect a reduction in unnecessary treatment and/or hospital stay. It might have been possible to construct a model in which fFN testing was placed in the care pathway with all other tests (essentially history and examination). However, this would have required a review of the accuracy of all of these tests, which was beyond the scope of this project. Furthermore, we consider that the trial results, particularly from the Dutta and Norman study,47 do represent outcome given actual clinical practice. Therefore, it was decided to adjust the model structure, focusing on a reduction in admissions and costs. The decision tree is shown in Figure 12. This model cannot estimate an ICER as there is no measure of effectiveness included. Although a hospital admission could be interpreted as an outcome, it is mainly relevant from a cost perspective. Therefore, the outcome is cost difference of the fFN-testing strategy compared with the control strategy.

FIGURE 12.

Adjusted structure decision tree.

Structural assumptions:

• The fFN-testing strategy, which is a combination of test-guided and clinical-guided care, will do no harm compared with the standard care strategy, which is based on clinical signs and symptoms alone. This means that it is assumed that in the fFN-testing strategy, there will be no patients deprived of care on the basis of test results when clinical signs and symptoms would indicate a hospital admission.

• The time horizon of the model includes hospital admission for observational purposes, but not the delivery itself, since costs and consequences of delivery are considered not to be affected by the testing.

Model parameters

The study by Dutta and Norman47 was the only one that was performed in the UK. Therefore, the results from this study were used for the base-case input parameters, as, from an NHS perspective, they were considered more appropriate than the results from the non-UK-based studies. Results from the non-UK-based studies were used in sensitivity analyses.

Base-case analysis

In the base-case analysis, the costs of hospitalisations were higher in the control strategy. The difference was, however, partly offset by the costs of testing in the fFN-testing strategy. Total average costs of the control strategy were £599.53, whereas the costs of the fFN-testing strategy were £575.65. Therefore, the fFN-testing strategy saves £23.88.

Additional analyses

Sensitivity analysis for price range

As it was difficult to obtain a reliable price for the rapid fFN test, in the base case we used a price which was derived from Honest et al. 10 In addition, we calculated incremental costs of the fFN-testing strategy for a range of prices for the test itself. The results of this analysis are shown in Figure 13. It is obvious that, all other costs held equal, the relation between test price and incremental costs is linear. Testing is cost-neutral at a test price of slightly over £45.

FIGURE 13.

Incremental costs of the fFN-testing strategy compared with control strategy at varying test prices.

Clinical effectiveness and test accuracy findings

Cost-effectiveness findings

The cost-effectiveness analysis did not include an effectiveness measure, since no clear indication of improved effectiveness given fFN testing was found in the trial data. Instead, the decision tree aimed to give an assessment of the costs associated with fFN testing. The base-case analysis showed a small cost advantage (i.e. £23.88) for the fFN-testing strategy. This result is however surrounded by quite some uncertainty, as it leaned very heavily on data from the study by Dutta and Norman47 As we were aware of this uncertainty, a number of additional analyses were done. For instance, the base-case analysis was re-run with data from alternative studies. When applying data from the study by Lowe et al. ,48 which reported a higher admission rate in the fFN-testing strategy, running the model resulted in an incremental cost of £170.71 of fFN testing compared with control. When using data from the study by Grobman et al. ,50 results were more or less comparable with the base case. Plaut et al. 49 reported a statistically significant shorter length of stay for the fFN-testing strategy. When this fact was applied to the decision tree, fFN testing led to a cost saving of £213.33.

Performing a probabilistic sensitivity analysis resulted in an average cost saving of £25.58. The 2.5th and 97.5th percentiles of the simulated results were –£305 and £240, respectively, which indicates that although the absolute difference in costs between strategies may be modest, there is uncertainty about whether fFN testing will come at a cost or generate savings. All other additional analyses show results in the same range.

As there was also uncertainty about the actual price of the rapid fFN test itself, incremental costs of fFN testing compared with usual care was calculated for a large range of possible prices. A price of £45 turned out to be the point where costs between strategies break even. Although the base-case price, which was derived from Honest et al. ,10 is below this ‘threshold’ of £45, it is difficult to say whether or not this is a realistic price. An Australian report dating from 2006 calculated prices for two types of rapid tests, of which the cheapest amounted to around $100 (Australian dollars), which converted with exchange rates of 2011 would mean £68. 29

Strengths, limitations and uncertainties of the systematic review

The systematic review conducted for this assessment represents a step forward on previously published systematic reviews,6,10,17,22,28,29 in that we have included both up-to-date data on the accuracy of fFN testing for the prediction of PTB and data from clinical trials assessing the effectiveness of including fFN testing in the clinical decision-making process. We have synthesised evidence from both study types in an attempt to provide a more complete picture of how fFN testing might be used in clinical practice. Our assessment of test accuracy uses a bivariate modelling approach, as recommended by current methodological guidance. 31,39 In addition, we have used a combination of subgroup analyses, regression analyses and sensitivity analyses to explore the potential effects on test accuracy of selected population and study design characteristics, as well as publication date.

Extensive literature searches were conducted in an attempt to maximise retrieval of relevant studies. These included electronic searches of a variety of bibliographic databases, as well as screening of clinical trials registers and conference abstracts to identify unpublished studies. Because of the known difficulties in identifying test accuracy studies using study design-related search terms, search strategies were developed to maximise sensitivity at the expense of reduced specificity. 32 Thus, large numbers of citations were identified and screened, many of which did not meet the inclusion criteria of the review. However, it should be noted that our review of test accuracy studies was an update of the Honest et al. 10 review, which had a methodological search filter; it is therefore possible that some relevant studies published before 2005 may not have been included.

Clear inclusion criteria were specified in the protocol for this review. Eligibility of studies for inclusion is therefore transparent. In addition, we have provided specific reasons for excluding any of the studies considered potentially relevant at initial citation screening (see Appendix 9). The review process followed recommended methods to minimise the potential for error and/or bias;31 studies were independently screened for inclusion by two reviewers and data extraction and quality assessment were done by one reviewer and checked by a second. Any disagreements were resolved by consensus.

The studies included in the review were RCTs and DTAs. Methodological quality of the RCTs was assessed using Cochrane risk of bias tool and for DTAs the assessment was done using a modified version of QUADAS-2. The QUADAS tool is recommended for assessing the methodological quality of test accuracy studies,31,39 widely adopted by researchers and key organisations such as the Cochrane Collaboration, the National Institute for Health and Care Excellence (NICE) in the UK, and Institut für Qualität und Wirtschaftlichkeit im Gesundheitswesen (IQWiG) in Germany. The revised version of QUADAS (QUADAS-2) has recently been published. 43 We consider QUADAS-2 to be the most appropriate tool currently available for the quality assessment of test accuracy studies. However, the applicability of QUADAS-2 to the current assessment was somewhat limited. It was considered that the inclusion criteria matched the review question and that questions of applicability were, therefore, not relevant. In addition, because the reference standard was the occurrence of preterm or term birth in all studies, we considered that there were no issues of bias relating to the adequacy or application of the reference standard and the ‘reference standard’ domain of QUADAS-2 was omitted; this study design also meant that many of the signalling questions for the ‘flow and timing’ domain were not considered relevant. The usefulness of quality assessment was further limited by poor reporting of primary study methods. The review-specific guidance used in our QUADAS-2 assessment is reported in Appendix 5. The results of the risk of bias assessment are reported, in full, for all included studies (see Appendix 6) and in summary in Chapter 3, Results and Table 6. The extent to which we were able to explore the impact of the remaining, relevant components of study quality on test accuracy was also constrained by the extent to which comparable data were reported in the previous systematic review from which some of our data were drawn. 10

The possibility of publication bias remains a potential problem for all systematic reviews. Considerations may differ for systematic reviews of test accuracy studies. It is relatively simple to define a positive result for studies of treatment (e.g. a significant difference between the treatment and control groups which favours treatment). This is not the case for test accuracy studies, which measure agreement between index test and reference standard. It would seem likely that studies finding greater agreement (high estimates of sensitivity and specificity) will be published more often. In addition, test accuracy data are often collected as part of routine clinical practice, or by retrospective review of records; test accuracy studies are not subject to the formal registration procedures applied to RCTs and are therefore more easily discarded when results appear unfavourable. The extent to which publication bias occurs in studies of test accuracy remains unclear; however, simulation studies have indicated that the effect of publication bias on meta-analytic estimates of test accuracy is minimal. 79 Formal assessment of publication bias in systematic reviews of test accuracy studies remains problematic and reliability is limited. 79 We did not undertake a statistical assessment of publication bias for accuracy studies or effectiveness studies in this review. However, our search strategy included a variety of routes to identify unpublished studies and resulted in the inclusion of a number of conference abstracts.

Despite efforts to include evidence from both clinical trials and test accuracy studies, it should be noted that few RCTs were identified and the RCTs included in this review are of generally poor quality and are likely to be underpowered. All the included RCTs were rated to be at ‘unclear risk’ of bias except for one study which was judged to be at low risk of bias across the key domains of Cochrane risk of bias assessment tool. 48 The main methodological issue for the included RCTs was the lack of appropriate power calculations and hence the potential for underpowered studies. Power calculations were generally poorly reported and failed to take account of the proportion of patients in the study that had a negative fFN test result (those whose test results have the potential to change management). In addition, one of the studies was stopped early because of the low enrolments and did not achieve the desired sample size. 47 Hence, studies may have been inadequately powered to detect possible benefits of fFN testing. As has been previously discussed in a systematic review of the methodological issues associated with clinical trials in this field, no study used a discordancy design. 30 A discordancy study design aims to randomise only those patients whose test results indicate a different management strategy to that based on usual assessment, in this case symptomatic women who have a negative fFN test result (without testing all symptomatic women are assumed to receive treatment, e.g. tocolysis, antenatal corticosteroids, hospitalisation). Thus, symptomatic women with a negative fFN result would be randomised to treatment (management decision based on usual assessment) or no treatment (management decision based on fFN test results). Comparison of clinical outcomes between these randomised groups shows the effects of deciding not to treat on the basis of a negative fFN result. A further significant methodological problem was that none of the included studies used a fixed management protocol in women with known fFN test results; treatment was generally at the clinicians' discretion. Plaut et al. 49 and Grobman et al. 53 provided clinicians with information about the positive and negative predictive values of the fFN test, but did not mandate treatment protocols. Studies reported outcomes by randomised group, but not stratified by fFN test result. Thus, the extent to which clinicians incorporated the knowledge of test results in their decision making remained largely unclear and potential effects of fFN testing may have been missed because clinicians did not use the test results in their decision-making. This issue was also highlighted by the previous systematic review of methodology. 30 Grobman et al. 50 reported that if the test were used for longer period of time physicians would think that the test was more reliable.

Information of the effects of fFN testing on neonatal outcomes was particularly scarce; only one study reported the neonatal outcomes (number of NICU admissions). 47 This study had a very small sample size and found no significant effects of testing.

Strengths, limitations and uncertainties of the cost-effectiveness analysis

One of the main strengths of the model is that the evidence we used to inform the parameters was relevant for the UK and as up-to-date and high quality as possible. Where evidence was not available from published studies or databases, we used the most likely and plausible ranges based on expert opinion.

An important limitation, however, also lies in the parameters. As there was only one UK-based study (Dutta and Norman47), this was the main provider of the data. Dutta and Norman47 reported that the admission rate in the fFN-testing strategy was lower than the admission rate in the control strategy. If, in clinical practice, testing for fFN would indeed lead to lowered admission rates, then the costs of testing could quite easily be offset. 47 However, although Grobman et al. 50 and Osório et al. 51 also found a slightly lower admission rate in fFN-tested patients, and Plaut et al. 49 reported a length of stay for fFN testing that was statistically significantly shorter than in the control group, there is also evidence to the contrary, as Lowe et al. 48 found that patients in the fFN-testing strategy group were more likely to be admitted than patients in the control strategy group. Given the fact that almost all costs in the model (e.g. hospital stay, tocolysis, corticosteroids, ultrasound examinations and hospital transfers) are admission driven, being uncertain about the admission rate has major implications for the uncertainty of the model outcome as a whole.

Another limitation lies in the fact that we could not incorporate effectiveness in the model. However, assuming that fFN testing is mainly an instrument to safely select those patients who do not need treatment, one would not necessarily expect it to have an impact on pregnancy outcome. Also, we cannot be sure that fFN testing would not be useful for preventing PTBs. Although, of course, it would have been technically possible to include pregnancy outcome in the model, this would have required reliance on accuracy data both for the fFN test and for all other tests (essentially history and examination), which was beyond the scope of this project. Moreover, we would also argue that the evidence from trials, in particular that by Dutta and Norman,47 given that it was conducted in the UK, would, on balance, provide more reliable data than data from a combination of accuracy studies of probable low quality. We are informed that a large study assessing fFN and resource use, conducted at Guy's and St Thomas' NHS Foundation Trust, is currently being prepared for publication [Assessment of Fetal Fibronectin Testing to Improve Preterm Management (AFFIRM) study].

MEDLINE (OvidSP): 2000 to September week 1 2011

Searched 16 September 2011.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (410)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (112)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (149)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (3)

5. or/1-4 (545)

6. fibronectins/ (19,207)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (31,130)

8. or/6-7 (31,130)

9. exp Obstetric Labor, Premature/ (14,939)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (40,748)

11. (PROM or PROM or PTB).ti,ab,ot. (3180)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (3111)

13. or/9-12 (45,249)

14. 5 or (8 and 13) (643)

15. randomized controlled trial.pt. (316,345)

16. controlled clinical trial.pt. (83,446)

17. randomized.ab. (221,935)

18. placebo.ab. (128,225)

19. drug therapy.fs. (1,495,908)

20. randomly.ab. (160,028)

21. trial.ab. (229,549)

22. groups.ab. (1,062,145)

23. meta-analysis.mp,pt. or review.pt. or search:.tw. (1,809,563)

24. or/15-23 (4,234,913)

25. animals/ not (animals/ and humans/) (3,584,947)

26. 24 not 25 (3,686,178)

27. 26 and 14 (224)

28. limit 27 to yr="2000 -Current" (153)

Systematic reviews filter:

Montori VM, Wilczynski NL, Morgan D, Haynes RB. Optimal search strategies for retrieving systematic reviews from MEDLINE: analytical survey (top strategy minimising the difference between sensitivity and specificity). BMJ 2005;330(7482):68.

Randomised controlled trials filter:

Lefebvre C, Manheimer E, Glanville J. Chapter 6: searching for studies. Box 6.4.c: Cochrane Highly sensitive search strategy for identifying randomized controlled trials in Medline: sensitivity-maximizing version (2008 version); OVID format. In Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions version 5.1.0. The Cochrane Collaboration; 2011. URL: www.cochrane-handbook.org

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP): 2000 to 15 September 2011, MEDLINE Daily Update (OvidSP): 2000 to 15 September 2011

Searched 16 September 2011.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (10)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (3)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (10)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (0)

5. or/1-4 (18)

6. fibronectins/ (8)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (633)

8. or/6-7 (633)

9. exp Obstetric Labor, Premature/ (18)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (1347)

11. (PROM or PROM or PTB).ti,ab,ot. (193)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (105)

13. or/9-12 (1564)

14. 5 or (8 and 13) (19)

15. randomized controlled trial.pt. (737)

16. controlled clinical trial.pt. (43)

17. randomized.ab. (10,394)

18. placebo.ab. (4201)

19. drug therapy.fs. (1213)

20. randomly.ab. (10,445)

21. trial.ab. (11147)

22. groups.ab. (60,889)

23. meta-analysis.mp,pt. or review.pt. or search:.tw. (16,499)

24. or/15-23 (94,506)

25. animals/ not (animals/ and humans/) (1599)

26. 24 not 25 (94,179)

27. 26 and 14 (4)

28. limit 27 to yr="2000 -Current" (4)

Systematic reviews filter:

Montori VM, Wilczynski NL, Morgan D, Haynes RB. Optimal search strategies for retrieving systematic reviews from MEDLINE: analytical survey (top strategy minimising the difference between sensitivity and specificity). BMJ 2005;330(7482):68.

Randomised controlled trials filter:

Lefebvre C, Manheimer E, Glanville J. Chapter 6: searching for studies. Box 6.4.c: Cochrane Highly sensitive search strategy for identifying randomized controlled trials in Medline: sensitivity-maximizing version (2008 version); OVID format. In Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions version 5.1.0. The Cochrane Collaboration; 2011. URL: www.cochrane-handbook.org

EMBASE (OvidSP): 2000 to week 36 2011

Searched 16 September 2011.

1. (f?etal adj2 fibronectin$).mp. (524)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).mp. (126)

3. (ffn or onfn or fdc-6).mp. (212)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).mp. (4)

5. or/1-4 (681)

6. Fibronectin/ (27,775)

7. (86088-83-7 or fibronectin$).mp. (35,981)

8. or/6-7 (35,981)

9. exp "immature and premature labor"/ (76,650)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).mp. (49,049)

11. (PROM or PROM or PTB).mp. (4030)

12. ((Short$ or reduced or multiple) adj4 gestation$).mp. (4435)

13. or/9-12 (101,617)

14. 5 or (8 and 13) (938)

15. Random$.tw. or clinical trial$.mp. or exp health care quality/ (2,536,239)

16. meta-analys:.mp. or search:.tw. or review.pt. (1,887,122)

17. or/15-16 (4,013,206)

18. 14 and 17 (361)

19. animal/ or animal experiment/ (3,076,644)

20. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).mp. (4,760,116)

21. or/19-20 (4,760,116)

22. exp human/ or human experiment/ (12,493,844)

23. 21 not (21 and 22) (3,823,462)

24. 18 not 23 (359)

25. limit 24 to (embase and yr="2000 -Current") (240)

Systematic reviews filter (best sensitivity and specificity) from:

Wilczynski NL, Haynes RB, the Hedges Team. EMBASE search strategies achieved high sensitivity and specificity for retrieving methodologically sound systematic reviews. J Clin Epidemiol 2007;60:29–33.

Randomised controlled trials (best sensitivity) from:

Wong SS, Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound treatment studies in EMBASE. J Med Libr Assoc 2006;94(1):41–7.

Cochrane Database of Systematic Reviews: 2000 to Issue 9 2011, Cochrane Central Register of Controlled Trials: 2000 to Issue 3 2011, Database of Abstracts of Reviews of Effects (Wiley): 2000 to Issue 3 2011, Health Technology Assessment Database (Wiley): 2000 to Issue 3 2011, NHS Economic Evaluation Database (Wiley): 2000 to Issue 3 2011, The Cochrane Library

Searched 19 September 2011.

The CDSR search retrieved 10 records.

The CENTRAL search retrieved 35 records.

The DARE search retrieved eight records.

The HTA search retrieved nine records.

The NHS EED search retrieved three records.

Science Citation Index Expanded (Web of Knowledge): 2000 to 19 September 2011

Searched 19 September 2011.

Date limit (time span) = 2000–11.

Cumulative Index to Nursing and Allied Health Literature (EBSCOhost): 2000 to 9 September 2011

Searched 19 September 2011.

Systematic reviews and randomised controlled trials filters based on:

Wong SS, Wilczynski NL, Haynes RB. Optimal CINAHL search strategies for identifying therapy studies and review articles. J Nurs Scholarsh 2006;38(2):194–9.

Maternity and Infant Care (OvidSP): 2000 to August 2011

Searched 19 September 2011.

1. (f?etal adj2 fibronectin$).af. (255)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).af. (10)

3. (ffn or onfn or fdc-6).af. (58)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).af. (2)

5. or/1-4 (262)

6. (86088-83-7 or fibronectin$).af. (342)

7. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).af. (13,953)

8. (PROM or PROM or PTB).af. (540)

9. ((Short$ or reduced or multiple) adj4 gestation$).af. (1105)

10. or/7-9 (14,663)

11. 5 or (6 and 10) (288)

12. (random$ or RCT or trial$ or systematic or placebo or groups or search).af. (33,697)

13. 11 and 12 (79)

14. limit 13 to yr="2000 -Current" (57)

National Institutes of Health ClinicalTrials.gov (Internet)

URL: www.clinicaltrials.gov

Searched 19 September 2011.

Advanced search option

Total = 27 references.

International Clinical Trials Registry Platform (Internet)

URL: http://apps.who.int/trialsearch/

Searched 19 September 2011.

86088-83-7 OR fibronectin OR fibronectins OR fFN 34

Current Controlled Trials: metaRegister of Controlled Trials (Internet)

URL: www.controlled-trials.com/mrct/

Searched 19 September 2011.

86088-83-7 OR fibronectin OR fibronectins OR fFN 23

EU Clinical Trials Register (Internet): 2000 to 11 September 2011

URL: www.clinicaltrialsregister.eu/

Searched 19 September 2011.

86088-83-7 OR fibronectin OR fibronectins OR fFN 6

MEDLINE (OvidSP): 2005 to November week 3 2011

Searched 29 November 2011.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (412)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (114)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (150)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (3)

5. or/1-4 (549)

6. fibronectins/ (19,529)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (31,753)

8. or/6-7 (31,753)

9. exp Obstetric Labor, Premature/ (15,259)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (41,550)

11. (PROM or PROM or PTB).ti,ab,ot. (3294)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (3162)

13. or/9-12 (46,172)

14. 5 or (8 and 13) (648)

15. animals/ not (animals/ and humans/) (3,630,436)

16. 14 not 15 (606)

17. limit 16 to yr="2005 -Current" (180)

18. remove duplicates from 17 (170)

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP): 2005 to 28 November 2011, MEDLINE Daily Update (OvidSP): 2005 to 16 November 2011

Searched 28 November 2011.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (13)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (4)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (11)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (0)

5. or/1-4 (22)

6. fibronectins/ (5)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (648)

8. or/6-7 (648)

9. exp Obstetric Labor, Premature/ (15)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (1511)

11. (PROM or PROM or PTB).ti,ab,ot. (204)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (110)

13. or/9-12 (1738)

14. 5 or (8 and 13) (22)

15. animals/ not (animals/ and humans/) (1559)

16. 14 not 15 (22)

17. limit 16 to yr="2005 -Current" (18)

18. remove duplicates from 17 (18)

EMBASE (OvidSP): 2005 to week 47 2011

Searched 29 November 2011.

1. (f?etal adj2 fibronectin$).mp. (531)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).mp. (130)

3. (ffn or onfn or fdc-6).mp. (216)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).mp. (4)

5. or/1-4 (692)

6. Fibronectin/ (28,249)

7. (86088-83-7 or fibronectin$).mp. (36,580)

8. or/6-7 (36,580)

9. exp "immature and premature labor"/ (78,183)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).mp. (50,214)

11. (PROM or PROM or PTB).mp. (4162)

12. ((Short$ or reduced or multiple) adj4 gestation$).mp. (4532)

13. or/9-12 (103,743)

14. 5 or (8 and 13) (951)

15. animal/ or animal experiment/ (3,123,853)

16. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).mp. (4,830,062)

17. or/15-16 (4,830,062)

18. exp human/ or human experiment/ (12,738,143)

19. 17 not (17 and 18) (3,870,815)

20. 14 not 19 (898)

21. limit 20 to (embase and yr="2005 -Current") (330)

22. remove duplicates from 21 (328)

Cochrane Database of Systematic Reviews (Wiley): 2005 to Issue 11 2011, Cochrane Central Register of Controlled Trials (Wiley): 2005 to Issue 4 2011, Database of Abstracts of Reviews of Effects (Wiley): 2005 to Issue 4 2011, Health technology Assessment Database (Wiley): 2005 to Issue 4 2011, NHS Economic Evaluation Database (Wiley): 2005 to Issue 4 2011, The Cochrane Library

Searched 29 November 2011.

The CDSR search retrieved 12 records.

The CENTRAL search retrieved 21 records.

The DARE search retrieved five records.

The HTA search retrieved six records.

The NHS EED search retrieved two records.

Cumulative Index to Nursing and Allied Health Literature (EBSCOhost): 2005 to 29 November 2011

Searched 29 November 2011.

Science Citation Index Expanded (Web of Knowledge): 2005 to 29 November 2011

Searched 29 November 2011.

Time span = 2005–11.

# 12 414 #11 OR #5

# 11 245 #10 AND #6

# 10 19,147 #9 OR #8 OR #7

# 9 1,520 TS=((Short* or reduced or multiple) near/4 gestation*)

# 8 2,140 TS=(PROM or PROM or PTB)

# 7 16,339 TS=(("Pre term" or preterm or premature or early or immature) near/5 (labo*r or birth* or childbirth* or deliver* or partu* or ruptur*))

# 6 9,532 TS=(86088-83-7 or fibronectin*)

# 5 383 #4 OR #3 OR #2 OR #1

# 4 32 TS=(tli system* or "tli iq" or tliiq or quikcheck)

# 3 94 TS=(ffn or onfn or fdc-6)

# 2 60 TS=((oncofetal or oncofoetal) near/2 fibronectin*)

# 1 263 TS=((fetal or foetal) near/2 fibronectin*)

National Institutes of Health ClinicalTrials.gov (Internet)

URL: www.clinicaltrials.gov

Searched 29 November 2011.

Advanced search option

Total = 26 references.

MEDLINE (OvidSP): 1946 to January week 4 2012

Searched 3 February 2012.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (407)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (113)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (150)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (3)

5. or/1-4 (543)

6. fibronectins/ (18,968)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (30,769)

8. or/6-7 (30,769)

9. exp Obstetric Labor, Premature/ (15,100)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (40,957)

11. (PROM or PROM or PTB).ti,ab,ot. (3185)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (3121)

13. or/9-12 (45,449)

14. 5 or (8 and 13) (637)

15. economics/ (26,147)

16. exp "costs and cost analysis"/ (160,560)

17. economics, dental/ (1834)

18. exp "economics, hospital"/ (17,605)

19. economics, medical/ (8423)

20. economics, nursing/ (3853)

21. economics, pharmaceutical/ (2283)

22. (economic$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (348,948)

23. (expenditure$ not energy).ti,ab. (14,668)

24. (value adj1 money).ti,ab. (17)

25. budget$.ti,ab. (14,919)

26. or/15-25 (463,385)

27. ((energy or oxygen) adj cost).ti,ab. (2361)

28. (metabolic adj cost).ti,ab. (623)

29. ((energy or oxygen) adj expenditure).ti,ab. (13,506)

30. or/27-29 (15,862)

31. 26 not 30 (459,766)

32. letter.pt. (729,121)

33. editorial.pt. (287,409)

34. historical article.pt. (279,013)

35. or/32-34 (1,282,430)

36. 31 not 35 (434,835)

37. animals/ not (animals/ and humans/) (3,556,824)

38. 36 not 37 (409,646)

39. 14 and 38 (39)

40. remove duplicates from 39 (39)

Economics filter:

Centre for Reviews and Dissemination. NHS EED economics filter: MEDLINE (Ovid) monthly search. York: Centre for Reviews and Dissemination; 2010. URL: www.york.ac.uk/inst/crd/intertasc/nhs_eed_strategies.html (cited 28 September 2010).

MEDLINE In-Process & Other Non-Indexed Citations (OvidSP): up to 2 February 2012, MEDLINE Daily Update (OvidSP): up to 2 February 2012

Searched 3 February 2012.

1. (f?etal adj2 fibronectin$).ti,ab,ot,hw. (11)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).ti,ab,ot,hw. (3)

3. (ffn or onfn or fdc-6).ti,ab,ot,hw. (11)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).ti,ab,ot,hw. (0)

5. or/1-4 (19)

6. fibronectins/ (7)

7. (86088-83-7 or fibronectin$).ti,ab,ot,rn. (629)

8. or/6-7 (629)

9. exp Obstetric Labor, Premature/ (22)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).ti,ab,ot,hw. (1533)

11. (PROM or PROM or PTB).ti,ab,ot. (196)

12. ((Short$ or reduced or multiple) adj4 gestation$).ti,ab,ot. (108)

13. or/9-12 (1751)

14. 5 or (8 and 13) (21)

15. economics/ (2)

16. exp "costs and cost analysis"/ (97)

17. economics, dental/ (0)

18. exp "economics, hospital"/ (7)

19. economics, medical/ (0)

20. economics, nursing/ (0)

21. economics, pharmaceutical/ (3)

22. (economic$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (26449)

23. (expenditure$ not energy).ti,ab. (729)

24. (value adj1 money).ti,ab. (3)

25. budget$.ti,ab. (1408)

26. or/15-25 (27,914)

27. ((energy or oxygen) adj cost).ti,ab. (148)

28. (metabolic adj cost).ti,ab. (37)

29. ((energy or oxygen) adj expenditure).ti,ab. (628)

30. or/27-29 (797)

31. 26 not 30 (27,704)

32. letter.pt. (17,492)

33. editorial.pt. (11,069)

34. historical article.pt. (117)

35. or/32-34 (28,663)

36. 31 not 35 (27,367)

37. animals/ not (animals/ and humans/) (1499)

38. 36 not 37 (27,325)

39. 14 and 38 (2)

40. remove duplicates from 39 (2)

Economics filter:

Centre for Reviews and Dissemination. NHS EED economics filter: MEDLINE (Ovid) monthly search. York: Centre for Reviews and Dissemination; 2010. URL: www.york.ac.uk/inst/crd/intertasc/nhs_eed_strategies.html (cited 28 September 2010).

EMBASE (OvidSP): 1980 to week 4 2012

Searched 3 February 2012.

1. (f?etal adj2 fibronectin$).mp. (535)

2. ((oncofetal or oncofoetal) adj2 fibronectin$).mp. (130)

3. (ffn or onfn or fdc-6).mp. (220)

4. (tli system$ or (tli adj iq) or tliiq or quikcheck).mp. (5)

5. or/1-4 (698)

6. Fibronectin/ (28,581)

7. (86088-83-7 or fibronectin$).mp. (36,957)

8. or/6-7 (36,957)

9. exp "immature and premature labor"/ (79,241)

10. ((Pre term or preterm or premature or early or immature) adj5 (labo?r or birth$ or childbirth$ or deliver$ or partu$ or ruptur$)).mp. (51,102)

11. (PROM or PROM or PTB).mp. (4300)

12. ((Short$ or reduced or multiple) adj4 gestation$).mp. (4624)

13. or/9-12 (105,221)

14. 5 or (8 and 13) (959)

15. health-economics/ (30,861)

16. exp economic-evaluation/ (176,566)

17. exp health-care-cost/ (169,283)

18. exp pharmacoeconomics/ (143,042)

19. or/15-18 (403,439)

20. (econom$ or cost or costs or costly or costing or price or prices or pricing or pharmacoeconomic$).ti,ab. (461,907)

21. (expenditure$ not energy).ti,ab. (18,337)

22. (value adj2 money).ti,ab. (1019)

23. budget$.ti,ab. (19,361)

24. or/20-23 (481,340)

25. 19 or 24 (718,447)

26. letter.pt. (753,716)

27. editorial.pt. (390,154)

28. note.pt. (463,410)

29. or/26-28 (1,607,280)

30. 25 not 29 (644,541)

31. (metabolic adj cost).ti,ab. (685)

32. ((energy or oxygen) adj cost).ti,ab. (2628)

33. ((energy or oxygen) adj expenditure).ti,ab. (15,957)

34. or/31-33 (18,582)

35. 30 not 34 (640,392)

36. 14 and 35 (64)

37. animal/ or animal experiment/ (3,141,087)

38. (rat or rats or mouse or mice or murine or rodent or rodents or hamster or hamsters or pig or pigs or porcine or rabbit or rabbits or animal or animals or dogs or dog or cats or cow or bovine or sheep or ovine or monkey or monkeys).mp. (4,869,368)

39. or/37-38 (4,869,368)

40. exp human/ or human experiment/ (12,876,181)

41. 39 not (39 and 40) (3,897,091)

42. 36 not 41 (63)

43. limit 42 to embase (52)

44. remove duplicates from 43 (52)

Economics filter:

Centre for Reviews and Dissemination. NHS EED economics filter: EMBASE (Ovid) weekly search. York: Centre for Reviews and Dissemination; 2010. URL: www.york.ac.uk/inst/crd/intertasc/nhs_eed_strategies.html (cited 17 March 2011).

Database of Abstracts of Reviews of Effects (Wiley): up to Issue 1 2012, Health technology Assessment Database (Wiley): up to Issue 1 2012, NHS Economic Evaluation Database (Wiley): up to Issue 1 2012

Searched 3 February 2012.

DARE search retrieved 12 records.

HTA search retrieved 10 records.

NHS EED search retrieved three records.

Paediatric Economic Database Evaluation (Internet): 1980–2010

URL: http://pede.ccb.sickkids.ca/pede/search.jsp

Searched 3 February 2012.

Searched ‘Title, Abstract, or Keywords’

Domain 1: patient selection

Question 1: Was a consecutive or random sample of patients enrolled?

• ‘yes’ → low risk of bias

• ‘unclear’ → unclear risk of bias

• ‘no’ → high risk of bias.

Question 2: Did the study avoid inappropriate exclusions?

• ‘no’ for < 10% of patients or ‘yes’ → low risk of bias

• ‘unclear’ → unclear risk of bias

• ‘no’ for ≥ 10% of patients → high risk of bias.

Domain 2: index test

Did the study prespecify the threshold for a positive result?

• ‘yes’ → low risk of bias

• ‘unclear’ → unclear risk of bias

• ’no’ → high risk of bias.

Domain 3: flow and timing

Were all patients included in the analysis?

• ‘no’ but for < 10% of patients or ‘yes’ → low risk of bias

• ‘unclear’ → unclear risk of bias

• ‘no’ for ≥ 10% of patients → high risk of bias.

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 1: patient selection

Domain 2: index test(s)

Domain 3: flow and timing

Domain 2: index test(s)

Domain 3: flow and timing

References and reasons

1. Adeniji AO, Olayemi O, Odukogbe AA, Oladokun A, Adeniji OI, Egbewale BE, et al. Cervico-vaginal foetal fibronectin: a predictor of cervical response at pre-induction cervical ripening. West Afr J Med 2005;24:334–7. (Intervention)

2. Akers A, Jarzembowski JA, Johnson CT, Lieberman RW, Dalton VK. Examining the relationship between positive mid-gestational fetal fibronectin assays and histological evidence of acute placental inflammation. J Perinat Med 2007;35:36–42. (Study design)

3. Andersen HF. Use of fetal fibronectin in women at risk for preterm delivery. Clin Obstet Gynecol 2000;43:746–58. (Study design)

4. Andrews WW, Goldenberg RL, National Institute of Child Health, Human Development Maternal-Fetal Medicine Units Network. What we have learned from an antibiotic trial in fetal fibronectin positive women. Semin Perinatol 2003;27:231–8. (Population)

5. Burwick RM, Zork NM, Lee GT, Ross MG, Kjos SL. Cervilenz assessment of cervical length compared to fetal fibronectin in the prediction of preterm delivery in women with threatened preterm labor. J Matern Fetal Neonatal Med 2011;24:127–31. (Intervention)

6. Conde-Agudelo A, Romero R. Fetal fibronectin as a predictor of spontaneous preterm delivery in multiple gestations: a systematic review and meta-analysis. Am J Obstet Gynecol 2009;201:S196. (Study design)

7. Crane J. The use of fetal fibronectin in predicting successful labor induction: a systematic review. Am J Obstet Gynecol 2005;193:S40. (Outcomes)

8. Elliott JP, Miller HS, Coleman S, Rhea D, Abril D, Hallbauer K, et al. A randomized multicenter study to determine the efficacy of activity restriction for preterm labor management in patients testing negative for fetal fibronectin. J Perinatol 2005;25:626–30. (Intervention)

9. Goldenberg RL, Andrews WW, Hoffman I, Fawzi W, Valentine M, Young A. Fetal fibronectin and adverse infant outcomes in a predominantly human immunodeficiency virus-infected African population: a Randomized controlled trial. Obstet Gynecol 2007;110:936. (Study design)

10. Goldenberg RL, Andrews WW, Hoffman I, Fawzi W, Valentine M, Young A, et al. Fetal fibronectin and adverse infant outcomes in a predominantly human immunodeficiency virus-infected African population: a randomized controlled trial. Obstet Gynecol 2007;109(2 Part 1):392–401. (Intervention)

11. Gomez R, Romero R, Medina L, Nien JK, Chaiworapongsa T, Carstens M, et al. Cervicovaginal fibronectin improves the prediction of preterm delivery based on sonographic cervical length in patients with preterm uterine contractions and intact membranes. Am J Obstet Gynecol 2005;192:350–9. (Study design)

12. Gomez-Bravo Topete E, Castillo-Lechuga C, Villegas-Su A, Briones-Garduno JC. [Predictive value of fetal fibronectin for preterm labor.] Cir Cir 2004;72:491–4. (Study design)

13. Hayes, Inc. Fetal fibronectin test in women with symptoms of preterm labor. Lansdale, PA: Hayes, Inc.; 2006. (Study design)

14. Imai M, Tani A, Saito M, Saito K, Amano K, Nisijima M. Significance of fetal fibronectin and cytokine measurement in the cervicovaginal secretions of women at term in predicting term labor and post-term pregnancy. Eur J Obstet Gynecol Reprod Biol 2001;97:53–8. (Study design)

15. Institute for Clinical Systems Improvement (ICSI). Fetal fibronectin for the prediction of preterm labor. Bloomington, IN: ICSI; 2000. (Study design)

16. Institute of Health Economics (IHE). Using fetal fibronectin to diagnose pre-term labour. Edmonton, AB: IHE; 2008. p. 79. (Study design)

17. Institute of Health Economics (IHE). Actim Partus Test BS TLi System as rapid response diagnostic tests. Edmonton, AB: IHE; 2008. (Study design)

18. Keeler SM, Roman AS, Coletta JM, Kiefer DG, Feuerman M, Rust OA. Fetal fibronectin testing in patients with short cervix in the midtrimester: can it identify optimal candidates for ultrasound-indicated cerclage? Am J Obstet Gynecol 2009;200:158.e1–6. (Study design)

19. Koenn ME. Fetal fibronectin. Clin Lab Sci 2002;15:96–8. (Study design)

20. Kurtzman J, Chandiramani M, Briley A, Poston L, Das A, Shennan A. Quantitative fetal fibronectin screening in asymptomatic high-risk patients and the spectrum of risk for recurrent preterm delivery. Am J Obstet Gynecol 2009;200:263.e1–6. (Population)

21. Lopez RL, Francis JA, Garite TJ, Dubyak JM. Fetal fibronectin detection as a predictor of preterm birth in actual clinical practice. Am J Obstet Gynecol 2000;182:1103–6. (Study design)

22. Mateus J, Pereira L, Baxter J, Berghella V, Tolosa J. Effectiveness of fetal fibronectin testing compared with digital cervical assessment of women with preterm contractions. Am J Perinatol 2007;24:381–5. (Study design)

23. Mercorio F, Mercorio A, Votino C, Di Spiezio Sardo A, Barba GV, et al. Fetal fibronectin as predictor of successful induction of mid-trimester abortion. Acta Obstet Gynecol Scand 2005;84:390–4. (Study design)

24. Mozurkewich EL, Naglie G, Krahn MD, Hayashi RH. Predicting preterm birth: a cost-effectiveness analysis. Am J Obstet Gynecol 2000;182:1589–97. (Study design)

25. Ness A, Visintine J, Ricci E, Berghella V. Does knowledge of cervical length and fetal fibronectin affect management of women with threatened preterm labor? A randomized trial. Am J Obstet Gynecol 2007;197:426.e1–7. (Intervention)

26. Ness A, Visintine J, Ricci E, Boyle K, Berghella V. Use of fetal fibronectin and transvaginal ultrasound cervical length to triage women with suspected preterm labor: a randomized trial. Am J Obstet Gynecol 2006;195:S67. (Intervention)

27. Nguyen TCQ. The cost-effectiveness of fetal fibronectin testing in suspected preterm labor: a randomized trial. Obstet Gynecol 2002;99(Suppl. 4):97S. (Outcomes)

28. Smith V, Devane D, Begley CM, Clarke M, Higgins S. A systematic review and quality assessment of systematic reviews of fetal fibronectin and transvaginal length for predicting preterm birth. Eur J Obstet Gynecol Reprod Biol 2007;133:134–42. (Study design)

29. Tsourapas A, Roberts TE, Barton PM, Honest H, Forbes C, Hyde CJ, et al. An economic evaluation of alternative test-intervention strategies to prevent spontaneous pre-term birth in singleton pregnancies. Acta Obstet Gynecol Scand 2009;88:1319–30. (Population)

30. Vis JY, Wilms FF, Oudijk MA, Porath MM, Scheepers HCJ, Bloemenkamp KWM, et al. Cost-effectiveness of fibronectin testing in a triage in women with threatened preterm labor: alleviation of pregnancy outcome by suspending tocolysis in early labor (APOSTEL-I trial). BMC Pregnancy Childbirth 2009;9(38). (Intervention)

Reference and reasons

1. fFN testing not always so accurate. Contemp Ob Gyn 2009;54:18. (Population)

2. Abenhaim HA, Morin L, Benjamin A. Does availability of fetal fibronectin testing in the management of threatened preterm labour affect the utilization of hospital resources? J Obstet Gynaecol Can 2005;27:689–94. (Study design)

3. Akers A, Jarzembowski JA, Johnson CT, Lieberman RW, Dalton VK. Examining the relationship between positive mid-gestational fetal fibronectin assays and histological evidence of acute placental inflammation. J Perinat Med 2007;35:36–42. (Outcomes)

4. Ayala-Mendez JA, Kurtzman J, Rosales-Ortiz S, Martinez-Alvarez O, Das A, Jimenez-Solis G. Fetal fibronectin status markedly modifies the risk of preterm delivery in low risk patients with symptomatic preterm labor and cervical shortening. Am J Obstet Gynecol 2008;199:S235. (Outcomes)

5. Bahado-Singh RO, Argoti P, Wilson L, Kruger M, Sorokin Y. Simultaneous evaluation of epidemiologic, psychosocial, biochemical and sonographic markers for prediction of spontaneous preterm birth. Paper presented at 57th Annual Scientific Meeting of the Society for Gynecologic Investigation, Orlando, FL, 24–27 March 2010. Reprod Sci 2010;17(Suppl. 1):277A. (Intervention)

6. Berghella V. MFM consult. When to use fetal fibronectin. Contemp Ob Gyn 2009;54:26. (Study design)

7. Bittar RE, Zugaib M. [Risk predictors for preterm birth.] Rev 2009;31:203–9. (Study design)

8. Chandiramani M, Di Renzo GC, Gottschalk E, Helmer H, Henrich W, Hoesli I, et al. Fetal fibronectin as a predictor of spontaneous preterm birth: a European perspective. J Matern Fetal Neonatal Med 2011;24:330–6. (Study design)

9. Chandiramani M, Shennan AH. Cervical insufficiency: prediction, diagnosis and prevention. Obstet Gynaecol 2008;10:99–106. (Intervention)

10. Facchinetti F, Paganelli S, Venturini P, Dante G, Palama L. [Biochemical mediators of the cervical modifications in the preterm birth.] G Ital Ostet Ginecol 2006;28:11–15. (Intervention)

11. Fox N, Saltzman D, Klauser C, Peress D, Gutierrez C, Rebarber A. Prediction of spontaneous preterm birth and very preterm birth in twin pregnancies using serial fetal fibronectin and cervical length. Am J Obstet Gynecol 2008;199:S25. (Population)

12. Franco R, Shaw K, Williams C, Hickok D. Risk of preterm delivery with a positive fetal fibronectin test result and a normal cervical length. Obstet Gynecol 2005;105:114S. (Outcomes)

13. Fuchs I, Dudenhausen JW. [Use of foetal fibronectin for prediction of a premature birth.] Gynakol Prax 2008;32:646–8. (Outcomes)

14. Ghidini A, Poggi SH, Korker V. Performance of vaginal fetal fibronectin as predictor of preterm delivery in a community hospital. J Soc Gynecol Investig 2006;13:338A. (Outcomes)

15. Giles W. Fetal fibronectin use in the management of threatened preterm labour. O & G Magazine 2006;8:39. (Outcomes)

16. Gottschalk EM, Wenzel S, Salomon NS, Dudenhausen JW, Henrich W. Importance of cervical length measurement and fetal fibronectin (fFN) to the prediction of lower premature birth rate. Geburtshilfe Frauenheilkunde 2008;68:S25. (Outcomes)

17. Hee L. Likelihood ratios for the prediction of preterm delivery with biomarkers. Acta Obstet Gynecol Scand 2011;90:1189–99. (Intervention)

18. Herbst A, Nilsson C. Diagnosis of early preterm labour. BJOG 2006;113(Suppl. 3):60–7. (Intervention)

19. Holmgren C, Lacoursiere DY, Esplin MS. Clinical predictors of a false negative fetal fibronectin (FFN). Am J Obstet Gynecol 2007;197:S204. (Outcomes)

20. Incerti M, Ghidini A, Korker V, Pezzullo JC. Performance of cervicovaginal fetal fibronectin in a community hospital setting. Arch Gynecol Obstet 2007;275:347–51. (Outcomes)

21. Jenkins SM, Kurtzman JT, Osann K. Dynamic cervical change: is real-time sonographic cervical shortening predictive of preterm delivery in patients with symptoms of preterm labor? Ultrasound Obstet Gynecol 2006;27:373–6. (Intervention)

22. Johnson CT, Dalton VK, Akers AY, Jarzembrowski JA, Lieberman RW, Johnson TRB. Fetal fibronectin assay results and placental histopathology in multiple gestation pregnancies. Obstet Gynecol 2006;107:80S. (Intervention)

23. Kang JH, Lee SE, Park CW, Jun JK, Romero R, Yoon BH. Cervical fetal fibronectin: an index of intra-amniotic inflammation, histologic chorioamnionitis and impending preterm delivery in patients with preterm labor and intact membranes. Am J Obstet Gynecol 2007;197:S47. (Outcomes)

24. Khan KS. Systematic reviews of diagnostic tests: a guide to methods and application. Best Pract Res Clin Obstet Gynaecol 2005;19:37–46. (Study design)

25. Knee A, Belisle E, Markenson G, Malshe A, Plevyak M, Bsat F, et al. Do pregnancies complicated by preterm births with a negative fetal fibronectin screen have different characteristics compared to those that have positive fibronectin results? Paper presented at 31st Annual Meeting of the Society for Maternal-Fetal Medicine: The Pregnancy Meeting, San Francisco, CA, 7–12 Feb 2011. Am J Obstet Gynecol 2011;204(Suppl. 1):S190. (Population)

26. Kosec V, Herman R. [Diagnosis of preterm birth.] Gynaecol Perinatol Suppl 2008;17:S38–41. (Outcomes)

27. Kuin RA, Vis JY, Mol BW. Fetal fibronectin as a short-term predictor of preterm birth in symptomatic patients. Obstet Gynecol 2010;115:186–7. (Study design)

28. Leitich H. Controversies in diagnosis of preterm labour. BJOG 2005;112(Suppl. 1):61–3. (Outcomes)

29. Markenson G. Predicting preterm labor. Female Patient 2008;33:19–20. (Study design)

30. Menon R, Torloni MR, Voltolini C, Torricelli M, Merialdi M, Betran AP, et al. Biomarkers of spontaneous preterm birth: an overview of the literature in the last four decades. Reprod Sci 2011;18:1046–70. (Intervention)

31. Muller A, McCullough M, Obican S, Norton H, Carter J, Gonzalez-Quintero VH. ‘False’ positive fetal fibronectin and rates of adverse obstetrical outcomes. Am J Obstet Gynecol 2006;195:S233. (Intervention)

32. Myint H, Singh V, Roberts NJ. Introducing fetal fibronectin testing to a district general hospital in women with symptoms of preterm labour. Paper presented at Perinatal Medicine conference; 15–17 Jun 2011; Harrogate, UK. Arch Dis Child Fetal Neonatal Ed 2011;96:Fa83. (Intervention)

33. O’Sullivan M. Fetal fibronectin as a tool to reduce preterm labour admissions. Aust Midwifery News 2006;6:14–15. (Study design)

34. Passuello V, Puhl AG, Seufert R, Fischl F, Kobl H. Fetal fibronectin: is (still actual) a marker of premature birth? Geburtshilfe Frauenheilkunde 2008;68:S50. (Intervention)

35. Pizzo A, Ardita FV, Oteri F, Caruso C, La Spada R, Accardo FM. [Fibronectin and preterm labour.] Gaz Med Ital Archiv Sci Med 2006;165:175–9. (Outcomes)

36. Ponting J, Tomlin M. Diagnosis of preterm labour: introducing the fetal fibronectin test. Brit J Midwifery 2011;19:24–31. (Study design)

37. Pucillo K, Munneke S. Fetal fibronectin as predictor of preterm delivery in women with symptoms of preterm labor: women with negative FFN test results are unlikely to deliver in the following 2 weeks; observation is still warranted, but expensive interventions may be unnecessary. Evid Based Pract 2008;11:11–12. (Outcomes)

38. Ramirez Pineda M, Duenas Diez JL, Sala Turrens J, Polo Padillo J, Bedoya Bergua C. [Analysis of two strategies for the management of threatened preterm labor.] Prog Obstet Ginecol 2010;53:261–6. (Study design)

39. Riboni F, Vitulo A, Dell’avanzo M, Plebani M, Battagliarin G, Paternoster D. Biochemical markers predicting pre-term delivery in symptomatic patients: phosphorylated insulin-like growth factor binding protein-1 and fetal fibronectin. Arch Gynecol Obstet 2011;284:1325–9. (Intervention)

40. Rouse DJ. Improved management in threatened preterm labor with rapid fetal fibronectin testing: commentary. Obstet Gynecol Surv 2006;61:688–9. (Study design)

41. Sanchez-Ramos L, Zamora J, Kaunitz AM. Fetal fibronectin as a short-term predictor of preterm birth in symptomatic patients reply. Obstet Gynecol 2010;115:187. (Study design)

42. Schmitz T, Maillard F, Bessard-Bacquaert S, Kayem G, Fulla Y, Cabrol D, et al. Selective use of fetal fibronectin detection after cervical length measurement to predict spontaneous preterm delivery in women with preterm labor. Am J Obstet Gynecol 2006;194:138–43. (Intervention)

43. Tanir HM, Sener T, Yildiz Z. Cervicovaginal fetal fibronectin (FFN) for prediction of preterm delivery in symptomatic cases: a prospective study. Clin Exp Obstet Gynecol 2008;35:61–4. (Outcomes)

44. Tekesin I, Eberhart LHJ, Schaefer V, Wallwiener D, Schmidt S. Evaluation and validation of a new risk score (CLEOPATRA score) to predict the probability of premature delivery for patients with threatened preterm labor. Ultrasound Obstet Gynecol 2005;26:699–706. (Outcomes)

45. Tekesin I, Marek S, Hellmeyer L, Reitz D, Schmidt S. Assessment of rapid fetal fibronectin in predicting preterm delivery. Obstet Gynecol 2005;105:280–4. (Study design)

46. Ting H-S, Chin P-S, Yeo GSH, Kwek K. Comparison of bedside test kits for prediction of preterm delivery: phosphorylated insulin-like growth factor binding protein-1 (pIGFBP-1) test and fetal fibronectin test. Ann Acad Med Singapore 2007;36:399–402. (Outcomes)

47. Vintzileos AM. Predicting preterm births in high-risk patients. Female Patient 2005;30:10–13. (Study design)

48. Wilms FF, van Stralen G, Porath MM, Papatsonis DNM, Oei SGG, Mol B-W, et al. [Predicating imminent preterm labour based on a determination of foetal fibronectin in a vaginal smear.] Ned Tijdschr Geneeskd 2009;153:B398. (Intervention)

49. Yoneda S, Sakai M, Sasaki Y, Shiozaki A, Hidaka T, Saito S. Interleukin-8 and glucose in amniotic fluid, fetal fibronectin in vaginal secretions and preterm labor index based on clinical variables are optimal predictive markers for preterm delivery in patients with intact membranes. J Obstet Gynaecol Res 2007;33:38–44. (Intervention)

4.1 Aims & objectives:

4.2 Intervention:

Fetal fibronectin is an extracellular matrix glycoprotein produced by amniocytes and by cytotrophoblast. 1 It is thought to be present mainly in the choriodecidual interface, which is a union between maternal and fetal tissues. 15 Normally, fFN is present in the cervicovaginal secretions of pregnant women until 22 weeks’ gestation. However, the level of fFN in cervicovaginal secretions drops after 22 weeks’ gestation (< 50 ng/mL). If the pregnancy is not normal, the level of fFN found in a cervicovaginal swab may be high (≥ 50 ng/mL) at or after 22 weeks’ gestation; elevated levels of fFN may indicate early onset of labour. 1

The fFN test can be used to assess the risk of preterm birth, within 7 to 14 days of testing, in symptomatic women. The fetal fibronectin test is available in two formats: a quantitative solid-phase enzyme-linked immunosorbent assay (ELISA) or a qualitative membrane immunosorbent assay (Rapid fFN for the TLi™ System, which recently changed to FullTerm™). 16–18 Rapid fFN testing is a more practical approach for diagnosing preterm labour as it gives the results instantly (30 min) unlike the ELISA assay which delivers the results 4 to 48 hours after sample collection. 17 However, there is limited clinical evidence on the use of rapid fFN to detect preterm labour as majority of evidence is based on ELISA.

Rapid fFN testing is a lateral flow, solid-phase immunosorbent assay designed to perform a qualitative detection (positive/negative) of fFN in cervicovaginal specimens collected in the Adeza Biomedical Collection Kit. 17 The cervicovaginal specimen (vaginal swab) is mixed with a liquid buffer in a collection tube and a portion of this sample is pipetted to the lateral flow, rapid fFN cassette in the TLi™ IQ Analyser. 17 The assay takes about approximately 30 min to process the sample and deliver the results. The TLi™ automatically prints and displays positive or negative results along with patient details (an fFN level of ≥ 50 ng/mL is positive result and an fFN level of < 50 ng/mL is negative result). 17

The intervention considered in this review is rapid fFN testing in addition to usual care.

5.1 Inclusion and Exclusion criteria:

5.2 Search strategy

Search strategies will be based on target condition and intervention, as recommended in the Centre for Reviews and Dissemination (CRD) guidance for undertaking reviews in health care and the Cochrane Handbook for Diagnostic Test Accuracy Reviews. 30–32