Advertisement

Randomized clinical trials are not always the best way to assess diagnostic tests: the case of fetal fibronectin testing

      To the Editors:
      Recently, Berghella and Saccone
      • Berghella V.
      • Saccone G.
      Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
      published in this journal a meta-analysis on fetal fibronectin as predictor of preterm birth. In this meta-analysis, the authors concluded that women assigned randomly to the knowledge of fibronectin results did not have reduced preterm birth rates compared to a control group. In addition, both groups had similar rates of hospitalization, and use of tocolytics and steroids. Mean hospital costs were even slightly higher in the fetal fibronectin group. This conclusion was based on 6 randomized clinical trials (RCTs) reporting on 546 women with threatened preterm labor; 13% of the included delivered within 1 week after presentation.
      Based on this meta-analysis Macones
      • Macones G.A.
      Fetal fibronectin testing in threatened preterm labor: time to stop.
      stated in an accompanying editorial that the continued use of fetal fibronectin testing in women with threatened preterm labor cannot be justified. We raise concerns about the authors' methods and the conclusions drawn from their results. In doing so, we would like to focus on the main clinical question concerning the fetal fibronectin test, the possible clinical outcomes, the study designs that could be applied, and the accompanying pitfalls.
      Most of the women presenting with symptoms of preterm labor do not deliver on a short term.
      • Berghella V.
      • Saccone G.
      Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
      The value of the fetal fibronectin test is mostly in its high negative predictive value, which allows identification of women who are unlikely to deliver in due course, especially within 7-10 days (>98%).
      • Honest H.
      • Bachmann L.M.
      • Gupta J.K.
      • Kleijnen J.
      • Khan K.S.
      Accuracy of cervicovaginal fetal fibronectin test in predicting risk of spontaneous preterm birth: systematic review.
      In continuation of this test characteristic, clinical research should focus on the question of whether addition of the fibronectin to the established workup in threatened preterm labor, including the cervical length (CL) measurement, could improve the identification of women who will not deliver on a short term. Such improvement reduces unnecessary treatments with corticosteroids and tocolytics, referrals, the number of hospitalization days, and health care cost, without compromising neonatal health outcomes. These clinical outcomes, including neonatal health, should therefore be the primary outcomes in fibronectin studies.
      In their meta-analysis the authors focus on preterm delivery <37 weeks gestation as primary outcome.
      • Berghella V.
      • Saccone G.
      Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
      Unsurprisingly, no difference was found between the 2 groups because of the following arguments. A test by itself will never prevent women from delivering prematurely. A subsequent therapy could, but is very unlikely in case of the fibronectin test and tocolysis. First, as mentioned before, the fibronectin test has its value in ruling out premature delivery and withholding women from unnecessary therapy. Secondly, it is unknown if tocolytics, by prolonging pregnancy, indeed improve neonatal outcome. According to recent World Health Organization recommendations there is no convincing evidence that tocolytics do.
      WHO
      Recommendations on interventions to improve preterm birth outcomes.
      Since corticosteroids aim to improve neonatal outcomes we think that neonatal outcomes should be addressed in studies evaluating fetal fibronectin. These were only reported in 2 of the included studies (n = 164).
      • Dutta D.
      • Norman J.E.
      Pilot study into the efficacy of fetal fibronectin testing in minimizing hospital admissions in women presenting with symptoms of preterm labor: a randomized controlled trial of obstetric and neonatal outcomes.
      • Lee G.T.
      • Burwick R.
      • Zork N.
      • Kjos S.
      Does the use of fetal fibronectin in an algorithm for preterm labor reduce triage evaluation times?.
      The final question is why the authors did not find any differences in the secondary outcome, such as maternal hospitalization and administration of tocolytics and corticosteroids (n = 438)
      • Dutta D.
      • Norman J.E.
      Pilot study into the efficacy of fetal fibronectin testing in minimizing hospital admissions in women presenting with symptoms of preterm labor: a randomized controlled trial of obstetric and neonatal outcomes.
      • Lee G.T.
      • Burwick R.
      • Zork N.
      • Kjos S.
      Does the use of fetal fibronectin in an algorithm for preterm labor reduce triage evaluation times?.
      • Grobman W.A.
      • Welshman E.E.
      • Calhoun E.A.
      Does fetal fibronectin use in the diagnosis of preterm labor affect physician behavior and health care costs? A randomized trial.
      • Lowe M.P.
      • Zimmerman B.
      • Hansen W.
      Prospective randomized controlled trial of fetal fibronectin on preterm labor management in a tertiary care center.
      • Nguyen T.T.
      • Toy E.C.
      • Baker B.
      The cost-effectiveness of fetal fibronectin testing in suspected preterm labor: randomized trial.
      and hospitalization costs (n = 77).
      • Nguyen T.T.
      • Toy E.C.
      • Baker B.
      The cost-effectiveness of fetal fibronectin testing in suspected preterm labor: randomized trial.
      The explanation for these negative findings lies within the number of included women and in the study design. Different study designs could be applied to examine whether the fibronectin test could reduce unnecessary treatments, referrals, hospitalization, and health care costs. In the meta-analysis the authors limited their search to RCTs, in which the new test is added to the established workup in threatened preterm labor. A difference in any outcome would be the effect of the added fibronectin test. However, if RCTs that assessed a new diagnostic test fail to show any difference one should be aware of potential pitfalls. These pitfalls occur most frequently within the relations between testing and subsequent treatment, which can be intricate and may obscure evaluations of the potential effect of the test itself.
      • Bossuyt P.M.
      • Lijmer J.G.
      • Mol B.W.J.
      Randomized comparisons of medical tests: sometimes invalid, not always efficient.
      • Vis J.Y.
      • Wilms F.F.
      • Oudijk M.A.
      • Bossuyt P.M.
      • et al.
      Why were the results of randomized trials on the clinical utility of fetal fibronectin negative? A systematic review of their study designs.
      When introducing a new test to the established workup it might be unclear how to interpret the results of this test in light of the results of other tests. For example, how should we interpret a negative fibronectin test in women with a CL <15 mm or a CL <25 mm? Or how should we interpret a positive fibronectin test in women with CL >30 mm? Consequently, clinicians could ignore the fibronectin result in their subsequent management, and patient outcomes will not be different as a result of testing, regardless of test accuracy. In the meta-analysis, 3 of the 6 included studies did not embed the mandatory use of a fixed protocol in relation to fibronectin results.
      • Grobman W.A.
      • Welshman E.E.
      • Calhoun E.A.
      Does fetal fibronectin use in the diagnosis of preterm labor affect physician behavior and health care costs? A randomized trial.
      • Lowe M.P.
      • Zimmerman B.
      • Hansen W.
      Prospective randomized controlled trial of fetal fibronectin on preterm labor management in a tertiary care center.
      • Plaut M.M.
      • Smith W.
      • Kennedy K.
      Fetal fibronectin: the impact of a rapid test on the treatment of women with preterm labor symptoms.
      In view of the concerns raised above, we advocate that a prospective cohort design should in some clinical questions be acceptable for the evaluation of diagnostic and prognostic tests, and we think in the case of fibronectin and CL this is justified. From these studies, decision rules for test interpretation could be extracted. The fibronectin test for example could be applied as triage test, an add-on test, or could even completely replace the existing CL measurement.
      • Bossuyt P.M.
      • Irwig L.
      • Craig J.
      • Glasziou P.
      Comparative accuracy: assessing new tests against existing diagnostic pathways.
      The Apostel-1 study group performed a cohort study in 665 women with threatened preterm labor in The Netherlands. From the analyses it appeared that women with a CL <15 mm or with CL between 15-30 mm with a positive fibronectin test were at increased risk for preterm delivery within 7 days. Adding the fibronectin in this specific group of women would result in a 10% reduction in transfers, maternal admission, and treatment.
      • van Baaren G.J.
      • Vis J.Y.
      • Wilms F.F.
      • et al.
      Predictive value of cervical length measurement and fibronectin testing in threatened preterm labor.
      This information did not need an RCT, and limitation of the meta-analysis of Berghella and Saccone
      • Berghella V.
      • Saccone G.
      Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
      to RCTs only does not justify the conclusion that fetal fibronectin testing in women with threatened preterm labor cannot be justified.
      Based on the aforementioned issues, we are not convinced that the meta-analysis of Berghella and Saccone
      • Berghella V.
      • Saccone G.
      Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
      of RCTs justifies the abandonment of the fibronectin test in symptomatic women, as proposed in the editorial comment. Although RCTs are considered the gold standard in evidence synthesis and provide the best answer in many research questions, they do not necessarily do so in the evaluation of diagnostic tests.

      References

        • Berghella V.
        • Saccone G.
        Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials.
        Am J Obstet Gynecol. 2016; 215: 431-438
        • Macones G.A.
        Fetal fibronectin testing in threatened preterm labor: time to stop.
        Am J Obstet Gynecol. 2016; 215: 405
        • Honest H.
        • Bachmann L.M.
        • Gupta J.K.
        • Kleijnen J.
        • Khan K.S.
        Accuracy of cervicovaginal fetal fibronectin test in predicting risk of spontaneous preterm birth: systematic review.
        BMJ. 2002; 325: 301
        • WHO
        Recommendations on interventions to improve preterm birth outcomes.
        World Health Organization, Geneva2015
        • Dutta D.
        • Norman J.E.
        Pilot study into the efficacy of fetal fibronectin testing in minimizing hospital admissions in women presenting with symptoms of preterm labor: a randomized controlled trial of obstetric and neonatal outcomes.
        Arch Gynecol Obstet. 2011; 284: 559-565
        • Lee G.T.
        • Burwick R.
        • Zork N.
        • Kjos S.
        Does the use of fetal fibronectin in an algorithm for preterm labor reduce triage evaluation times?.
        J Matern Fetal Neonatal Med. 2013; 26: 706-709
        • Grobman W.A.
        • Welshman E.E.
        • Calhoun E.A.
        Does fetal fibronectin use in the diagnosis of preterm labor affect physician behavior and health care costs? A randomized trial.
        Am J Obstet Gynecol. 2004; 191: 235-240
        • Lowe M.P.
        • Zimmerman B.
        • Hansen W.
        Prospective randomized controlled trial of fetal fibronectin on preterm labor management in a tertiary care center.
        Am J Obstet Gynecol. 2004; 190 ([see comment]): 358-362
        • Nguyen T.T.
        • Toy E.C.
        • Baker B.
        The cost-effectiveness of fetal fibronectin testing in suspected preterm labor: randomized trial.
        Obstet Gynecol. 2002; 99: 97S
        • Bossuyt P.M.
        • Lijmer J.G.
        • Mol B.W.J.
        Randomized comparisons of medical tests: sometimes invalid, not always efficient.
        Lancet. 2000; 356: 1844-1847
        • Vis J.Y.
        • Wilms F.F.
        • Oudijk M.A.
        • Bossuyt P.M.
        • et al.
        Why were the results of randomized trials on the clinical utility of fetal fibronectin negative? A systematic review of their study designs.
        Am J Perinatol. 2011; 28: 145-150
        • Plaut M.M.
        • Smith W.
        • Kennedy K.
        Fetal fibronectin: the impact of a rapid test on the treatment of women with preterm labor symptoms.
        Am J Obstet Gynecol. 2003; 188: 1588-1593
        • Bossuyt P.M.
        • Irwig L.
        • Craig J.
        • Glasziou P.
        Comparative accuracy: assessing new tests against existing diagnostic pathways.
        BMJ. 2006; 332: 1089-1092
        • van Baaren G.J.
        • Vis J.Y.
        • Wilms F.F.
        • et al.
        Predictive value of cervical length measurement and fibronectin testing in threatened preterm labor.
        Obstet Gynecol. 2014; 123: 1185-1192

      Linked Article

      • Fetal fibronectin testing for prevention of preterm birth in singleton pregnancies with threatened preterm labor: a systematic review and metaanalysis of randomized controlled trials
        American Journal of Obstetrics & GynecologyVol. 215Issue 4
        • Preview
          Fetal fibronectin is an extracellular matrix glycoprotein that is produced by amniocytes and cytotrophoblasts and has been shown to predict spontaneous preterm birth.
        • Full-Text
        • PDF
      • Reply
        American Journal of Obstetrics & GynecologyVol. 218Issue 1
        • Preview
          I thank Dr van Baaren and colleagues for their letter concerning the meta-analysis and editorial of fetal fibronectin testing in women with symptoms of preterm labor.1,2 I believe the purpose of their letter is to point out that observational studies can also be used to assess the performance of diagnostic tests. To support this argument, the authors cite their own work, which was an observational study of how fetal fibronectin and cervical length could be used to risk-stratify women with symptoms of preterm labor.
        • Full-Text
        • PDF
      • Reply
        American Journal of Obstetrics & GynecologyVol. 218Issue 1
        • Preview
          We thank van Baaren et al for their letter and their interest in our study on fetal fibronectin (FFN) for management of women with preterm labor (PTL) at around 23-34 weeks.1 We congratulate their group for having done important work in this area, and confirmed themselves the good prediction of transvaginal ultrasound (TVU) cervical length (CL) in combination with FFN for prediction of preterm birth (PTB) in women with threatened PTL.2-5
        • Full-Text
        • PDF