- Email: [email protected]
Fetal fibronectin detection in preterm labor: Evaluation of a prototype bedside dipstick technique and cervical assessment
Fetal fibronectin detection in preterm labor: Evaluation of a prototype bedside dipstick technique and cervical assessment Matthew A.G. Coleman, MB, ChB,a Lesley M.E. McCowan, MD,a Neil S. Pattison, MD,a and Murray Mitchell, DPhil, DScb Auckland, New Zealand OBJECTIVE: The aims of this study were to evaluate a prototype bedside test for fetal fibronectin detection in women with symptoms of preterm labor, to compare the efficacy of obtaining fetal fibronectin swabs with and without a speculum, and to assess the value of combining the fetal fibronectin test with cervical dilatation for predicting delivery within 10 days. STUDY DESIGN: This investigation prospectively studied a cohort of women with symptoms of preterm labor (n = 121), gestational age between 24 and 336⁄7 weeks, and cervical dilatation ≤3 cm. Vaginal swabs for fibronectin testing with the bedside dipstick Fetal Fibronectin Membrane Immunoassay test (Adeza Biomedical, Sunnyvale, Calif) were obtained with and without a speculum. Results were not available to the clinicians. RESULTS: Twenty-two percent of the fetal fibronectin samples collected with a speculum (27/121) yielded positive results. The incidence of delivery within 10 days of testing was 14% (17/121). Fetal fibronectin detection predicted delivery within 10 days with sensitivity, specificity, and positive and negative predictive values of 65%, 85%, 41%, and 94%, respectively. The positive and negative likelihood ratios were 4.3 and 0.41, respectively. There was substantial association between samples obtained with and without a speculum (Cohen κ statistic 0.65, 95% confidence interval 0.44-0.87). Cervical dilatation ≥1 cm predicted delivery within 10 days with sensitivity, specificity, and positive and negative predictive values of 71%, 87%, 46%, and 95%, respectively, with positive and negative likelihood ratios of 5.5 and 0.33, respectively. Combination of fetal fibronectin testing with cervical dilatation ≥1 cm did not significantly improve the fetal fibronectin test characteristics. After exclusion of women with slight vaginal bleeding, cervical dilatation ≥1 cm and fetal fibronectin status remained the only independent variables associated with delivery within 10 days. CONCLUSION: Fetal fibronectin testing according to this prototype may have a limited role in clinical decision analysis. In this study a cervical dilatation ≥1 cm had predictive values equivalent to those of the fetal fibronectin test, and it should be evaluated further in a clinical setting. (Am J Obstet Gynecol 1998;179:1553-8.)
Key words: Cervical dilatation, delivery within 10 days, fetal fibronectin detection, preterm labor
Despite advances in perinatal and neonatal care, preterm labor and delivery remain the leading causes of neonatal morbidity and mortality.1 Perinatal outcome might be improved if women at genuine risk for preterm labor could be clearly distinguished. Incorrect or falsepositive diagnoses of preterm labor can result in unnecessary and potentially hazardous treatment.2 Equally, failure to correctly note true preterm labor may jeopardize fetal well-being and result in a missed opportunity to administer corticosteroids. The development of a diagnostic test that reliably differentiates between “true” and “false”
From the Departments of Obstetrics and Gynaecologya and Pharmacology and Clinical Pharmacology,b National Women’s Hospital. Supported by Auckland Healthcare and the Mercia Barnes Trust. Received for publication January 24, 1998; revised May 6, 1998; accepted May 13, 1998. Reprint requests: Lesley McCowan, MD, Department of Obstetrics and Gynaecology, National Women’s Hospital, Private Bag 92 189, Auckland 3, New Zealand. Copyright © 1998 by Mosby, Inc. 0002-9378/98 $5.00 + 0 6/1/91642
preterm labor could significantly reduce unnecessary hospitalization, medication, social disruption, and anxiety. The presence of fetal fibronectin in the cervicovaginal secretions of women with symptoms of preterm labor is associated with increased risk of subsequent preterm birth.3-16 The value of fetal fibronectin detection in the clinical management of women with preterm labor, however, remains to be determined. Four published studies have examined the reliability of bedside testing for the detection of fetal fibronectin and its correlation with subsequent preterm delivery in women with symptoms of preterm labor.11-14 All these studies used the ROM-Plus (Adeza Biomedical, Sunnyvale, Calif) vertical-flow immunoassay method (spot test with control ring) to detect fetal fibronectin. Three of these studies had small numbers,11, 12, 14 and the largest13 included women at as late as 36 completed weeks’ gestation. The prototype Fetal Fibronectin Membrane Immunoassay (Adeza Biomedical) dipstick test is easier to perform and more economical than the ROM-Plus kit. It has not been previously evaluated in clinical studies. In addition, if fetal fibronectin 1553
1554 Coleman et al
testing is to play a role in clinical practice, it may be more acceptable to women if samples can be obtained without a speculum. This has not previously been evaluated. The aim of this study was therefore to assess the diagnostic efficacy of a prototype bedside fetal fibronectin dipstick immunoassay test for the prediction of preterm delivery in women with symptoms of preterm labor. The following hypotheses were tested: (1) A negative fetal fibronectin test result in a woman with symptoms of threatened preterm labor would predict the absence of preterm delivery within 10 days of the test. (2) The diagnostic efficacy of fetal fibronectin samples collected with and without a speculum would be similar. (3) The combination of cervical dilatation and the fetal fibronectin test would improve prediction of delivery within 10 days. Material and methods This prospective study was conducted at National Women’s Hospital, a tertiary referral hospital, in Auckland, New Zealand, between May 1996 and June 1997. The protocol was approved by the regional ethics committee, and informed consent was obtained from all participating women. One researcher (M.C.) who was not part of the routine clinical team was contacted after the admission of women with symptoms suggestive of preterm labor and was responsible for recruitment, speculum and vaginal examinations, and fetal fibronectin swab interpretation. Consecutive women were recruited by this researcher, except during short periods of leave. The inclusion criteria were all the following: • Cervical dilatation ≤3 cm • Gestational age between 24 and 33 weeks 6 days • Either contractions more frequent than 1 contraction/10 min for ≥1 hour or transfer to National Women’s Hospital with a diagnosis of preterm labor and parenteral albuterol (Salbutamol) treatment The exclusion criteria were any of the following: • Moderate or heavy vaginal bleeding apparent on admission or speculum examination • Placenta previa • Previous recruitment for the study • Confirmed rupture of membranes Gestational age was established by menstrual history and confirmed by ultrasonography at <24 weeks’ gestation. If menstrual history was unreliable or there was discordance of >10 days between menstrual and ultrasonographic dates, the ultrasonographic date was used.17 Moderate or heavy vaginal bleeding was defined as any bleeding seen by the medical staff before vaginal examination or more than the slightest bleeding seen at speculum examination. Women in whom slight bleeding was detected or provoked by speculum examination were not excluded. Recent sexual intercourse was defined as sexual intercourse within 24 hours of fetal fibronectin test-
December 1998 Am J Obstet Gynecol
ing. Fetal Fibronectin Membrane Immunoassay test kits were supplied by Adeza Biomedical. Vaginal swabs were collected as follows. First a sample of the cervicovaginal secretions was obtained without a speculum. The labia were separated, and then a sterile polyethylene terephthalate fiber (Dacron) swab was inserted high into the vagina toward the posterior fornix, where it was gently rotated for ≥10 seconds, in accordance with the manufacturer’s instructions. A second sample was collected from the posterior fornix after insertion of a sterile speculum lubricated with a small amount of KY lubricating jelly (Johnson & Johnson, New Brunswick, NJ). Swabs were inserted into separate tubes containing extraction buffer and mixed vigorously in the buffer for 10 to 15 seconds. The swabs were then removed and the lower end of the dipstick was inserted into the buffer for 10 minutes, after which time the results were interpreted. A positive result was indicated by the presence of 2 lines, whereas a negative result was indicated by the presence of a single control line. The additional line that indicated a positive result could vary from faint to dark. After speculum examination the cervical dilatation was determined digitally. All clinicians and women were blinded to the fetal fibronectin test results. After completion of recruitment, we estimated the interobserver variation of this fetal fibronectin test among clinicians at National Women’s Hospital. Three different test strips were prepared with clearly positive, weakly positive, and clearly negative results. Clinicians were independently given the same instructions and also an example of a clearly positive test result. A brief clinical history accompanied each of the 3 separate test strips. Clinicians were then asked to categorize the results of these 3 fetal fibronectin test strips. The demographic, delivery, and outcome details were recorded for all women. The medical records of women with false-negative test results (negative fetal fibronectin test result but delivery within 10 days) were audited after completion of the study by a second researcher who was blinded to the reasons for review and the results of the fetal fibronectin test. The test characteristics, including likelihood ratios for this fetal fibronectin test and for differing cervical dilatations in predicting delivery within 10 days, were then calculated and compared. The test characteristics of the fetal fibronectin test were also calculated after removal of potential confounders (slight vaginal bleeding and recent sexual intercourse). Combinations of the fetal fibronectin test result and cervical dilatation were combined to determine the best marker for delivery within 10 days. Statistical methods. The Wilcoxon rank sum test was used for the analysis of nonparametric data. Categoric data were analyzed with the Fisher exact test. Likelihood ratios were calculated for the prediction of delivery within 10 days and of delivery before 34 weeks as an aid
Coleman et al 1555
Volume 179, Number 6, Part 1 Am J Obstet Gynecol
Table I. Demographic and clinical details of women stratified by fetal fibronectin status detected with a speculum Positive (n = 27)
Negative (n = 94)
No.
No.
%
Table II. Fetal fibronectin test characteristics and comparison of fetal fibronectin detected with and without a speculum for the prediction of delivery within 10 days All women (with a speculum, n = 121)
%
Age at delivery 28.8 ± 5.6 28.8 ± 6 (mean ± SD, y) European 15 56 60* 64* Gestational age at 312⁄7 24-335⁄7 304⁄7 241⁄7-336⁄7 recruitment† (wk) Interval between 35 1-90 51 1-113‡ recruitment and delivery† (d) Gestational age at 354⁄7 244⁄7-413⁄7 382⁄7‡ 264⁄7-424⁄7‡ delivery† (wk) Birth weight† (g) 2450 705-4125 3120‡ 800-4330‡ Cervical dilatation 14 52 12‡ 13‡ ≥1 cm at recruitment Tocolysis after 12 44 34 36 recruitment Minimal vaginal 10 37 3‡ 3‡ bleeding Sexual intercourse 7 26 5§ 5§ within 24 h *There was no significant difference between any racial group with respect to fetal fibronectin test result. †Median and range, rather than number and percentage. ‡P ≤ .01. §P = .02.
in evaluating the value of this test in our population. Multiple logistic regression analysis was also performed to measure the associations of the explanatory variables (fetal fibronectin status, interval from sampling to delivery, cervical dilatation, previous preterm delivery, minimal vaginal bleeding, uterine activity, gestational age at recruitment, and multiple gestation) with the response variable (fetal fibronectin status or delivery within 10 days). The agreement between the different methods for obtaining fetal fibronectin swabs was determined by the Cohen κ statistic. Statistical analysis was carried out with the SAS statistical package for Windows 6.11 (SAS Institute Inc, Cary, NC). Sample size calculation. A sample size of 130 was estimated to be necessary for a true negative predictive value of fetal fibronectin of 99.5%. The lower confidence limit was set at 95% and the power was set at 80%. P < .05 was considered significant. In assessing interobserver variability, we estimated that we would have to sample 30 of the estimated 50 hospital clinicians to detect a 5% interobserver difference in test interpretation. Results One hundred forty-two women were recruited for the study. After review, 21 women were excluded because of failure to meet the inclusion criteria. Eleven women had uterine contractions less frequently than 1 contraction/10 min, 9 women had fetal fibronectin swabs taken
% Delivery ≤10 d† Positive FFN test result† Sensitivity Specificity PPV NPV LR+ LR–
95% CI
14 (17) 22 (27) 65 85 41 94 4.3 0.41
42-88 78-92 22-60 89-99 2.4-7.6 0.2-0.8
Subgroup* (n = 107) With a speculum
Without a speculum
12 (14) 21 (22)
13 (14) 15 (16)
57 85 36 93 3.8 0.5
43 89 38 91 3.9 0.64
CI, Confidence interval; FFN, fetal fibronectin; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Women who had 2 concurrent samples collected for fibronectin, first without and then with a speculum. †Number in parentheses is number of women.
but were subsequently found to have cervical dilatation >3 cm, and preterm rupture of membranes could not be excluded in 1 case. The demographic and clinical details of the 121 remaining women, stratified according to fetal fibronectin status, are shown in Table I. All fetal fibronectin results refer to samples taken with a speculum unless otherwise stated. Eighty-seven percent of women (102/121) underwent a dating scan before 20 weeks’ gestation; only 1 woman did not undergo a scan before 24 weeks’ gestation. There were 14 sets of twins and 1 set of triplets. Eighty-two percent (99/121) received corticosteroids during the index admission. Two percent (2/121) were receiving intravenous albuterol at the time of sampling. Fourteen percent (17/121) were delivered within 10 days of testing, 16% (19/121) were delivered before 34 weeks’ gestation, and 40% (48/121) were delivered before 37 completed weeks’ gestation. Twenty-one percent (26/121) had cervical dilatation ≥1 cm, and 46% of these women (12/26) were delivered within 10 days. Twelve percent (14/121) had cervical dilatation ≥2 cm and 5% (6/121) had a cervical dilatation of 3 cm. Eleven percent (13/121) had either a history of recent vaginal bleeding (n = 4) or slight vaginal bleeding apparent on speculum examination (n = 9). Of those with slight vaginal bleeding at speculum examination, 89% (8/9) had a positive fetal fibronectin test result and 67% were delivered within 10 days. Two women with a history of bleeding (n = 4) but without blood visible at speculum examination had positive fetal fibronectin test results. Predictive value of fetal fibronectin testing. Twenty-two percent (27/121) of the samples collected with a specu-
1556 Coleman et al
December 1998 Am J Obstet Gynecol
Table III. Test characteristics for predicting delivery within 10 days by fetal fibronectin detection in different subgroups after removal of confounders
Exclusion group Total No. Delivery ≤10 d† Positive FFN test result† Sensitivity Specificity PPV NPV LR+ LR–
Recent intercourse (n = 12)
Slight bleeding (n = 13)
Slight bleeding and/or* recent intercourse (n = 22)
109 14 (15) 18 (20)
108 9 (10) 16 (17)
99 9 (9) 13 (13)
60 88 45 93 5 0.45
50 88 29 95 4.2 0.57
44 90 31 94 4.4 0.6
Table IV. Test characteristics of cervical dilatation ≥1 cm alone and in combination with fetal fibronectin status (with a speculum) for predicting delivery within 10 days Cervical dilatation ≥1 cm % Positive FFN test result* Sensitivity Specificity PPV NPV LR+ LR–
95% CI
21 (26) 71 87 46 95 5.5 0.33
Cervical dilatation ≥1 cm and/or positive FFN test result %
95% CI
32 (39) 49-93 81-93 27-65 91-99 3.1-9.8 0.16-0.69
82 76 36 96 3.4 0.24
64-100 68-84 21-51 92-100 2.2-5.2 0.10-0.59
FFN, Fetal fibronectin; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Three women had both minimal vaginal bleeding and recent sexual intercourse. †Number in parentheses is number of women.
FFN, Fetal fibronectin; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Number in parentheses is number of women.
lum yielded positive results for fetal fibronectin. The test characteristics for fetal fibronectin detection and the prediction of delivery within 10 days are shown in Table II. A positive fetal fibronectin test result distinguished 65% of those women who were delivered within 10 days of testing. Fetal fibronectin detection predicted delivery within 7 days with sensitivity of 67%, specificity of 84%, positive predictive value of 37%, and negative predictive value of 95%. The positive likelihood ratio was 4.2 and the negative likelihood ratio was 0.4. Detection of fetal fibronectin predicted delivery before 34 weeks’ gestation with sensitivity, specificity, and positive and negative predictive values of 53%, 83%, 37%, and 90%, respectively. The positive and negative likelihood ratios for delivery before 34 weeks’ gestation were 3.1 and 0.56, respectively. Factors independently associated with a positive fetal fibronectin result were minimal vaginal bleeding, (χ2 = 25, P = .0001), cervical dilatation ≥1 cm, (χ2 = 14.2, P = .0002), and recent sexual intercourse (χ2 = 5, P = .02). Neither the frequency of contractions at recruitment nor the interval between recruitment and delivery was independently associated with a positive fetal fibronectin test result. Comparison of fetal fibronectin detection with and without a speculum. One hundred seven women had 2 concurrent vaginal samples collected, first without and then with a speculum. Twenty-one percent of the samples taken with a speculum (22/107) yielded positive results and 15% of the samples taken without a speculum (16/107) yielded positive results (P = .28). Only 73% of the women (16/22) who had positive results for fetal fibronectin from a sample taken with a speculum also had positive results from sample that was taken without a speculum (Table II). When the subgroups were compared, there were no significant differences between any
of the test characteristics. The Cohen κ statistic for agreement between the 2 methods of collection was 0.65 (95% confidence interval 0.44-0.87). Confounding factors. Confounding factors associated with a positive fetal fibronectin test results were minimal vaginal bleeding and recent sexual intercourse. The removal of women with those confounding factors tended to reduce the performance of the test. The performance values of the fetal fibronectin test when these factors are removed are shown in Table III. Comparison of cervical dilatation and fetal fibronectin testing. The characteristics of differing cervical dilatations for the prediction of delivery within 10 days were tested. The data for cervical dilatation ≥1 cm alone and in combination with fetal fibronectin are shown in Table IV. Cervical dilatation ≥1 cm had the highest sensitivity and had a similar specificity to those of cervical dilatation ≥2 cm and cervical dilatation of 3 cm (data not shown). Combination of cervical dilatation ≥1 cm with the fetal fibronectin test did not significantly improve the test performance (Table IV). Factors independently associated with delivery within 10 days of sampling were cervical dilatation ≥1 cm (χ2 = 28.3, P = .0001) and slight vaginal bleeding (χ2 = 10.1, P = .001). When slight vaginal bleeding was removed from the logistic regression model, the only significant independent factors associated with delivery within 10 days were cervical dilatation ≥1 cm (χ2 = 28.3, P = .0001) and the detection of fetal fibronectin (χ2 = 6.9, P = .009). The frequency of uterine activity on admission was not related to delivery within 10 days. False-negative and false-positive fetal fibronectin test results. Five percent of women in the speculum group (6/121) and 7% of women in the subgroup who had samples obtained without a speculum (8/107) had false-
Coleman et al 1557
Volume 179, Number 6, Part 1 Am J Obstet Gynecol
Table V. Clinical details of women with false-negative fetal fibronectin test results Case
GA at rec. (wk)
Ongoing uterine activity
Cervical dilatation at rec (cm)
Interval of rec to delivery (d)
1 2 3 4 5
32 33 32 26 28
Yes Yes No Yes No
2 0 1.5 0 0
3 1 9 4 2
6
33
No
2
5
Delivery details Spontaneous labor and vaginal delivery CD at 3 cm dilatation (abnormal FHR pattern) CD before labor for SRM and maternal request CD before labor for SRM and chorioamnionitis CD before labor for concealed abruptio placentae and deteriorating fetal condition CD before labor for suspected chorioamnionitis
GA, Gestational age; rec., recruitment; CD, cesarean delivery; FHR, fetal heart rate; SRM, spontaneous rupture of membranes.
negative results. The clinical details of the false-negative results obtained with a speculum are shown in Table V. Thirteen percent of women (16/121) had false-positive fetal fibronectin tests. When the subgroup who were examined with and without a speculum were compared, there were no significant differences in the false-positive or false-negative rates between sampling types (P = .40 and P = .42, respectively). Forty-four percent of the women with false-positive test results obtained with a speculum (7/16) had either recent sexual intercourse (within 24 hours, n = 5), slight vaginal bleeding (n = 4), or both (n = 2). Interobserver variation. Interobserver variation in the assessment of 3 fetal fibronectin test strips was evaluated between 36 clinicians. Eighty-nine percent (32/36) agreed with the clearly positive strip result. One hundred percent agreed with the negative test strip result. For the faintly positive result, 53% (19/36) classified it as positive and 47% (17/36) classified it as negative. Comment This is the first large prospective study to examine the clinical utility of this prototype bedside fetal fibronectin membrane immunoassay test kit. It also compared the value of combining fetal fibronectin testing with cervical dilatation for predicting preterm delivery in women with preterm labor. No other study has reported on the diagnostic value of this fetal fibronectin test or on that of samples taken from the vagina without a speculum. The clinical value of a test in a population can be determined by the degree that the test result changes the pretest probability of the outcome of interest in that population. Likelihood ratios are considered an appropriate way of expressing the ability of a diagnostic test to adjust the pretest probability of disease presence.18, 19 Pooled likelihood ratios have been reported in a recent overview of the diagnostic accuracy of both laboratory and bedside fetal fibronectin tests.20 For the prediction of preterm delivery within 1 week of testing in women with symptoms of preterm labor, these were found to be 5 (95% confidence interval 3.8-6.4) for a positive test result and 0.2 (95% confidence interval 0.1-0.4) for a negative test result. The findings in our study are consistent with these results (positive and negative likelihood ratios for deliv-
ery within 10 days of 4.6 and 0.6, respectively). Positive likelihood ratios between 2 and 5 and negative likelihood ratios between 0.5 and 0.2, however, are described as only “poor to fair.”19 This indicates that this prototype fetal fibronectin test has only a small effect on the pretest probability of delivery within 10 days and may therefore have limited clinical value. The sensitivity of the fetal fibronectin test in our study (65%) was less than that reported in previous studies of the bedside test (89%-100%)11-14 and also than that of the laboratory enzyme-linked immunosorbent assay test for predicting delivery within 7 days (80%-93%).5, 6, 8 False-negative test results diminish the sensitivity of a test, and the reduction in sensitivity in our study may be explained in part by consideration of the false-negative results. Among the samples taken with a speculum, only 2 of the 6 false-negative results were from women who were delivered within 10 days after spontaneous labor and therefore may be considered genuinely false negative (Table V, cases 1 and 2). In other studies the primary outcome has been more precisely defined as spontaneous preterm birth (after premature rupture of membranes or spontaneous preterm labor).21 If these criteria had been used in this study, the negative predictive value would have been 96%. Differentiation of a weakly positive result from a negative result can be difficult with this fetal fibronectin test. Many of the positive results in this study were faint lines. Despite the use of a single experienced observer, incorrect interpretation may have contributed to some of the false-negative results in this study. Our data on interobserver variation suggest that a significant number (53%) of clinicians may classify a test result as negative despite the presence of a faint line on the dipstick (indicating a positive result). Interobserver variation has not been reported for the other bedside test of fetal fibronectin. In this study cervical dilatation was assessed by a single observer. Use of cervical dilatation ≥1 cm to predict preterm delivery generated a positive likelihood ratio of 5.5, similar to that of the fetal fibronectin test alone (positive likelihood ratio 4.3). Combination of the fetal fibronectin test with cervical dilatation ≥1 cm did not significantly improve the test performance. Four other studies of fetal fibronectin have also re-
1558 Coleman et al
ported on the value of cervical dilatation as a predictor of preterm delivery.4-6, 12 In contrast to our study, each of these reported that fetal fibronectin testing was equivalent to or better than cervical dilatation in predicting delivery within 7 days. This difference between the findings in our study and those of others may be explained by a number of factors. First, in our study the sensitivity of the fetal fibronectin test was less than has been previously reported. Second, in the other studies cervical assessment was performed by multiple observers, which would increase the observer variation. The pretest probability of delivery within 10 days for those women with cervical dilatation ≥1 cm was 46%, and for those with slight vaginal bleeding was 54%. In these clinical settings, the fetal fibronectin test did not aid in decision analysis. A previous study5 also confirmed the independent association of vaginal bleeding with preterm delivery among women with preterm labor. Screening women for fetal fibronectin without the discomfort that may be caused by speculum examination would be simpler and might be more acceptable. The Cohen κ value of 0.65 indicates substantial agreement between the techniques for obtaining fetal fibronectin samples. Further studies are required, however, to establish whether there is any significant difference between these 2 methods for obtaining fetal fibronectin samples. Exclusion of women who had recent sexual intercourse did not change the test qualities significantly. Although these women had a trend toward more positive fetal fibronectin test results, a third of the false-positive results also occurred among these women. Conclusion. The prototype Fetal Fibronectin Membrane Immunoassay test performed only fairly in the prediction of delivery within 10 days. This may be related in part to the design of the test. If a more distinctive indicator or color change were possible in the presence of a weakly positive result, this might improve the test performance. In our study cervical dilatation ≥1 cm (assessed by a single researcher) correctly distinguished a similar proportion of women who were delivered within 10 days, 46% compared with 41% distinguished with the fetal fibronectin test. These women and those with slight vaginal bleeding are at high risk for preterm delivery, and in this context the addition of this fetal fibronectin test does not appear to improve decision analysis. The value in a clinical setting of fetal fibronectin testing in combination with clinical predictors needs further evaluation. We thank the following: Adeza Biomedical, Baxter Healthcare, Kevin Townend, and Vivienne Topping. REFERENCES
1. Copper RL, Goldenberg RL, Creasy RK, DuBard MA, Davis RO, Entman SS, et al. A multicenter study of preterm birth weight and gestational age–specific neonatal mortality. Am J Obstet Gynecol 1993;168:78-84.
December 1998 Am J Obstet Gynecol
2. Anonymous. Treatment of preterm labor with the beta-adrenergic agonist ritodrine. The Canadian Preterm Labor Investigators Group. N Engl J Med 1992;327:308-12. 3. Lockwood CJ, Senyei AE, Dische MR, et al. Fetal fibronectin in cervical and vaginal secretions as a predictor of preterm delivery. N Engl J Med 1991;325:669-74. 4. Peaceman AM, Andrews W, Thorp JM, Cliver SP, Lukes A, Iams JD, et al. Fetal fibronectin as a predictor of preterm birth in patients with symptoms: a multicenter trial. Am J Obstet Gynecol 1997;177:13-8. 5. Iams JD, Casal D, McGregor JA, Goodwin TM, Kreadon US, Lowensohn R, et al. Fetal fibronectin improves the accuracy of diagnosis of preterm labor. Am J Obstet Gynecol 1995;173 141-5. 6. Bartnicki J, Casal D, Kreaden US, Saling E, Vetter K. Fetal fibronectin in vaginal specimens predicts preterm delivery and verylow-birth-weight infants. Am J Obstet Gynecol 1996;174:971-4. 7. Burrus DR, Ernest JM, Veille JC. Fetal fibronectin, interleukin-6, and C-reactive protein are useful in establishing prognostic subcategories of idiopathic preterm labor. Am J Obstet Gynecol 1995;173:1258-62. 8. Malak TM, Sizmur F, Bell SC, Taylor DJ. Fetal fibronectin in cervicovaginal secretions as a predictor of preterm birth. Br J Obstet Gynaecol 1996;103:648-53. 9. Morrison JC, Allbert JR, MCLaughlin BN, Whitworth NS, Roberts WE, Martin RW. Oncofetal fetal fibronectin in patients with false labor as a predictor of preterm delivery. Am J Obstet Gynecol 1993;168:538-42. 10. Langer B, Boudier E, Schlaeder G. Cervico-vaginal fetal fibronectin: predictive value during false labor. Acta Obstet Gynecol Scand 1997;76:218-21. 11. Parker J, Bell R, Brennecke S. Fetal fibronectin in the cervicovaginal fluid of women with threatened preterm labour as a predictor of delivery before 34 weeks’ gestation. Aust N Z J Obstet Gynaecol 1995;35:257-61. 12. Senden IP, Owen P. Comparison of cervical assessment, fetal fibronectin and fetal breathing in the diagnosis of preterm labour. Clin Exp Obstet Gynecol 1996;23:5-9. 13. Benattar C, Taieb J, Fernandez H, Lindendaum A, Frydman R, Ville Y. Rapid fetal fibronectin swab-test in preterm labor patients treated by betamimetics. Eur J Obstet Gynecol Reprod Biol 1997;72:131-5. 14. Owen P, Scott A. Can fetal fibronectin testing improve the management of preterm labour? Clin Exp Obstet Gynecol 1997;24:19-22. 15. Rizzo G, Capponi A, Arduini D, Lorido C, Romanini C. The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic examination of the uterine cervix in predicting premature delivery for patients with preterm labor and intact membranes. Am J Obstet Gynecol 1996;175:1146-51. 16. Rozenberg P, Goffinet F, Malagrida L, Giudicelli Y, Perdu M, Houssin I, et al. Evaluating the risk of preterm delivery: a comparison of fetal fibronectin and transvaginal ultrasonographic measurement of cervical length. Am J Obstet Gynecol 1997;176:196-9. 17. Gardosi J, Mul T, Francis A, Hall J, Fishel S. Comparison of second trimester biometry in singleton and twin pregnancies conceived with assisted reproductive techniques. Br J Obstet Gynaecol 1997;104:737-40. 18. Jaeschke R, Guyatt G, Sackett DL. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 1994;271:389-91. 19. Schechter MT, Sheps SB. Diagnostic testing revisited: pathways through uncertainty. Can Med Assoc J 1985;132:755-60. 20. Chien PF, Khan KS, Ogston S, Owen P. The diagnostic accuracy of cervico-vaginal fetal fibronectin in predicting preterm delivery: an overview. Br J Obstet Gynaecol 1997;104:436-44. 21. Goldenberg RL, Iams JD, Mercer BM, Mels PJ, Moawad AH, Copper RL, et al. The preterm prediction study: the value of new vs standard risk factors in predicting early and all spontaneous preterm births. Am J Public Health 1998;88:233-8.
Key words: Cervical dilatation, delivery within 10 days, fetal fibronectin detection, preterm labor
Despite advances in perinatal and neonatal care, preterm labor and delivery remain the leading causes of neonatal morbidity and mortality.1 Perinatal outcome might be improved if women at genuine risk for preterm labor could be clearly distinguished. Incorrect or falsepositive diagnoses of preterm labor can result in unnecessary and potentially hazardous treatment.2 Equally, failure to correctly note true preterm labor may jeopardize fetal well-being and result in a missed opportunity to administer corticosteroids. The development of a diagnostic test that reliably differentiates between “true” and “false”
From the Departments of Obstetrics and Gynaecologya and Pharmacology and Clinical Pharmacology,b National Women’s Hospital. Supported by Auckland Healthcare and the Mercia Barnes Trust. Received for publication January 24, 1998; revised May 6, 1998; accepted May 13, 1998. Reprint requests: Lesley McCowan, MD, Department of Obstetrics and Gynaecology, National Women’s Hospital, Private Bag 92 189, Auckland 3, New Zealand. Copyright © 1998 by Mosby, Inc. 0002-9378/98 $5.00 + 0 6/1/91642
preterm labor could significantly reduce unnecessary hospitalization, medication, social disruption, and anxiety. The presence of fetal fibronectin in the cervicovaginal secretions of women with symptoms of preterm labor is associated with increased risk of subsequent preterm birth.3-16 The value of fetal fibronectin detection in the clinical management of women with preterm labor, however, remains to be determined. Four published studies have examined the reliability of bedside testing for the detection of fetal fibronectin and its correlation with subsequent preterm delivery in women with symptoms of preterm labor.11-14 All these studies used the ROM-Plus (Adeza Biomedical, Sunnyvale, Calif) vertical-flow immunoassay method (spot test with control ring) to detect fetal fibronectin. Three of these studies had small numbers,11, 12, 14 and the largest13 included women at as late as 36 completed weeks’ gestation. The prototype Fetal Fibronectin Membrane Immunoassay (Adeza Biomedical) dipstick test is easier to perform and more economical than the ROM-Plus kit. It has not been previously evaluated in clinical studies. In addition, if fetal fibronectin 1553
1554 Coleman et al
testing is to play a role in clinical practice, it may be more acceptable to women if samples can be obtained without a speculum. This has not previously been evaluated. The aim of this study was therefore to assess the diagnostic efficacy of a prototype bedside fetal fibronectin dipstick immunoassay test for the prediction of preterm delivery in women with symptoms of preterm labor. The following hypotheses were tested: (1) A negative fetal fibronectin test result in a woman with symptoms of threatened preterm labor would predict the absence of preterm delivery within 10 days of the test. (2) The diagnostic efficacy of fetal fibronectin samples collected with and without a speculum would be similar. (3) The combination of cervical dilatation and the fetal fibronectin test would improve prediction of delivery within 10 days. Material and methods This prospective study was conducted at National Women’s Hospital, a tertiary referral hospital, in Auckland, New Zealand, between May 1996 and June 1997. The protocol was approved by the regional ethics committee, and informed consent was obtained from all participating women. One researcher (M.C.) who was not part of the routine clinical team was contacted after the admission of women with symptoms suggestive of preterm labor and was responsible for recruitment, speculum and vaginal examinations, and fetal fibronectin swab interpretation. Consecutive women were recruited by this researcher, except during short periods of leave. The inclusion criteria were all the following: • Cervical dilatation ≤3 cm • Gestational age between 24 and 33 weeks 6 days • Either contractions more frequent than 1 contraction/10 min for ≥1 hour or transfer to National Women’s Hospital with a diagnosis of preterm labor and parenteral albuterol (Salbutamol) treatment The exclusion criteria were any of the following: • Moderate or heavy vaginal bleeding apparent on admission or speculum examination • Placenta previa • Previous recruitment for the study • Confirmed rupture of membranes Gestational age was established by menstrual history and confirmed by ultrasonography at <24 weeks’ gestation. If menstrual history was unreliable or there was discordance of >10 days between menstrual and ultrasonographic dates, the ultrasonographic date was used.17 Moderate or heavy vaginal bleeding was defined as any bleeding seen by the medical staff before vaginal examination or more than the slightest bleeding seen at speculum examination. Women in whom slight bleeding was detected or provoked by speculum examination were not excluded. Recent sexual intercourse was defined as sexual intercourse within 24 hours of fetal fibronectin test-
December 1998 Am J Obstet Gynecol
ing. Fetal Fibronectin Membrane Immunoassay test kits were supplied by Adeza Biomedical. Vaginal swabs were collected as follows. First a sample of the cervicovaginal secretions was obtained without a speculum. The labia were separated, and then a sterile polyethylene terephthalate fiber (Dacron) swab was inserted high into the vagina toward the posterior fornix, where it was gently rotated for ≥10 seconds, in accordance with the manufacturer’s instructions. A second sample was collected from the posterior fornix after insertion of a sterile speculum lubricated with a small amount of KY lubricating jelly (Johnson & Johnson, New Brunswick, NJ). Swabs were inserted into separate tubes containing extraction buffer and mixed vigorously in the buffer for 10 to 15 seconds. The swabs were then removed and the lower end of the dipstick was inserted into the buffer for 10 minutes, after which time the results were interpreted. A positive result was indicated by the presence of 2 lines, whereas a negative result was indicated by the presence of a single control line. The additional line that indicated a positive result could vary from faint to dark. After speculum examination the cervical dilatation was determined digitally. All clinicians and women were blinded to the fetal fibronectin test results. After completion of recruitment, we estimated the interobserver variation of this fetal fibronectin test among clinicians at National Women’s Hospital. Three different test strips were prepared with clearly positive, weakly positive, and clearly negative results. Clinicians were independently given the same instructions and also an example of a clearly positive test result. A brief clinical history accompanied each of the 3 separate test strips. Clinicians were then asked to categorize the results of these 3 fetal fibronectin test strips. The demographic, delivery, and outcome details were recorded for all women. The medical records of women with false-negative test results (negative fetal fibronectin test result but delivery within 10 days) were audited after completion of the study by a second researcher who was blinded to the reasons for review and the results of the fetal fibronectin test. The test characteristics, including likelihood ratios for this fetal fibronectin test and for differing cervical dilatations in predicting delivery within 10 days, were then calculated and compared. The test characteristics of the fetal fibronectin test were also calculated after removal of potential confounders (slight vaginal bleeding and recent sexual intercourse). Combinations of the fetal fibronectin test result and cervical dilatation were combined to determine the best marker for delivery within 10 days. Statistical methods. The Wilcoxon rank sum test was used for the analysis of nonparametric data. Categoric data were analyzed with the Fisher exact test. Likelihood ratios were calculated for the prediction of delivery within 10 days and of delivery before 34 weeks as an aid
Coleman et al 1555
Volume 179, Number 6, Part 1 Am J Obstet Gynecol
Table I. Demographic and clinical details of women stratified by fetal fibronectin status detected with a speculum Positive (n = 27)
Negative (n = 94)
No.
No.
%
Table II. Fetal fibronectin test characteristics and comparison of fetal fibronectin detected with and without a speculum for the prediction of delivery within 10 days All women (with a speculum, n = 121)
%
Age at delivery 28.8 ± 5.6 28.8 ± 6 (mean ± SD, y) European 15 56 60* 64* Gestational age at 312⁄7 24-335⁄7 304⁄7 241⁄7-336⁄7 recruitment† (wk) Interval between 35 1-90 51 1-113‡ recruitment and delivery† (d) Gestational age at 354⁄7 244⁄7-413⁄7 382⁄7‡ 264⁄7-424⁄7‡ delivery† (wk) Birth weight† (g) 2450 705-4125 3120‡ 800-4330‡ Cervical dilatation 14 52 12‡ 13‡ ≥1 cm at recruitment Tocolysis after 12 44 34 36 recruitment Minimal vaginal 10 37 3‡ 3‡ bleeding Sexual intercourse 7 26 5§ 5§ within 24 h *There was no significant difference between any racial group with respect to fetal fibronectin test result. †Median and range, rather than number and percentage. ‡P ≤ .01. §P = .02.
in evaluating the value of this test in our population. Multiple logistic regression analysis was also performed to measure the associations of the explanatory variables (fetal fibronectin status, interval from sampling to delivery, cervical dilatation, previous preterm delivery, minimal vaginal bleeding, uterine activity, gestational age at recruitment, and multiple gestation) with the response variable (fetal fibronectin status or delivery within 10 days). The agreement between the different methods for obtaining fetal fibronectin swabs was determined by the Cohen κ statistic. Statistical analysis was carried out with the SAS statistical package for Windows 6.11 (SAS Institute Inc, Cary, NC). Sample size calculation. A sample size of 130 was estimated to be necessary for a true negative predictive value of fetal fibronectin of 99.5%. The lower confidence limit was set at 95% and the power was set at 80%. P < .05 was considered significant. In assessing interobserver variability, we estimated that we would have to sample 30 of the estimated 50 hospital clinicians to detect a 5% interobserver difference in test interpretation. Results One hundred forty-two women were recruited for the study. After review, 21 women were excluded because of failure to meet the inclusion criteria. Eleven women had uterine contractions less frequently than 1 contraction/10 min, 9 women had fetal fibronectin swabs taken
% Delivery ≤10 d† Positive FFN test result† Sensitivity Specificity PPV NPV LR+ LR–
95% CI
14 (17) 22 (27) 65 85 41 94 4.3 0.41
42-88 78-92 22-60 89-99 2.4-7.6 0.2-0.8
Subgroup* (n = 107) With a speculum
Without a speculum
12 (14) 21 (22)
13 (14) 15 (16)
57 85 36 93 3.8 0.5
43 89 38 91 3.9 0.64
CI, Confidence interval; FFN, fetal fibronectin; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Women who had 2 concurrent samples collected for fibronectin, first without and then with a speculum. †Number in parentheses is number of women.
but were subsequently found to have cervical dilatation >3 cm, and preterm rupture of membranes could not be excluded in 1 case. The demographic and clinical details of the 121 remaining women, stratified according to fetal fibronectin status, are shown in Table I. All fetal fibronectin results refer to samples taken with a speculum unless otherwise stated. Eighty-seven percent of women (102/121) underwent a dating scan before 20 weeks’ gestation; only 1 woman did not undergo a scan before 24 weeks’ gestation. There were 14 sets of twins and 1 set of triplets. Eighty-two percent (99/121) received corticosteroids during the index admission. Two percent (2/121) were receiving intravenous albuterol at the time of sampling. Fourteen percent (17/121) were delivered within 10 days of testing, 16% (19/121) were delivered before 34 weeks’ gestation, and 40% (48/121) were delivered before 37 completed weeks’ gestation. Twenty-one percent (26/121) had cervical dilatation ≥1 cm, and 46% of these women (12/26) were delivered within 10 days. Twelve percent (14/121) had cervical dilatation ≥2 cm and 5% (6/121) had a cervical dilatation of 3 cm. Eleven percent (13/121) had either a history of recent vaginal bleeding (n = 4) or slight vaginal bleeding apparent on speculum examination (n = 9). Of those with slight vaginal bleeding at speculum examination, 89% (8/9) had a positive fetal fibronectin test result and 67% were delivered within 10 days. Two women with a history of bleeding (n = 4) but without blood visible at speculum examination had positive fetal fibronectin test results. Predictive value of fetal fibronectin testing. Twenty-two percent (27/121) of the samples collected with a specu-
1556 Coleman et al
December 1998 Am J Obstet Gynecol
Table III. Test characteristics for predicting delivery within 10 days by fetal fibronectin detection in different subgroups after removal of confounders
Exclusion group Total No. Delivery ≤10 d† Positive FFN test result† Sensitivity Specificity PPV NPV LR+ LR–
Recent intercourse (n = 12)
Slight bleeding (n = 13)
Slight bleeding and/or* recent intercourse (n = 22)
109 14 (15) 18 (20)
108 9 (10) 16 (17)
99 9 (9) 13 (13)
60 88 45 93 5 0.45
50 88 29 95 4.2 0.57
44 90 31 94 4.4 0.6
Table IV. Test characteristics of cervical dilatation ≥1 cm alone and in combination with fetal fibronectin status (with a speculum) for predicting delivery within 10 days Cervical dilatation ≥1 cm % Positive FFN test result* Sensitivity Specificity PPV NPV LR+ LR–
95% CI
21 (26) 71 87 46 95 5.5 0.33
Cervical dilatation ≥1 cm and/or positive FFN test result %
95% CI
32 (39) 49-93 81-93 27-65 91-99 3.1-9.8 0.16-0.69
82 76 36 96 3.4 0.24
64-100 68-84 21-51 92-100 2.2-5.2 0.10-0.59
FFN, Fetal fibronectin; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Three women had both minimal vaginal bleeding and recent sexual intercourse. †Number in parentheses is number of women.
FFN, Fetal fibronectin; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio of a positive test result; LR–, likelihood ratio of a negative test result. *Number in parentheses is number of women.
lum yielded positive results for fetal fibronectin. The test characteristics for fetal fibronectin detection and the prediction of delivery within 10 days are shown in Table II. A positive fetal fibronectin test result distinguished 65% of those women who were delivered within 10 days of testing. Fetal fibronectin detection predicted delivery within 7 days with sensitivity of 67%, specificity of 84%, positive predictive value of 37%, and negative predictive value of 95%. The positive likelihood ratio was 4.2 and the negative likelihood ratio was 0.4. Detection of fetal fibronectin predicted delivery before 34 weeks’ gestation with sensitivity, specificity, and positive and negative predictive values of 53%, 83%, 37%, and 90%, respectively. The positive and negative likelihood ratios for delivery before 34 weeks’ gestation were 3.1 and 0.56, respectively. Factors independently associated with a positive fetal fibronectin result were minimal vaginal bleeding, (χ2 = 25, P = .0001), cervical dilatation ≥1 cm, (χ2 = 14.2, P = .0002), and recent sexual intercourse (χ2 = 5, P = .02). Neither the frequency of contractions at recruitment nor the interval between recruitment and delivery was independently associated with a positive fetal fibronectin test result. Comparison of fetal fibronectin detection with and without a speculum. One hundred seven women had 2 concurrent vaginal samples collected, first without and then with a speculum. Twenty-one percent of the samples taken with a speculum (22/107) yielded positive results and 15% of the samples taken without a speculum (16/107) yielded positive results (P = .28). Only 73% of the women (16/22) who had positive results for fetal fibronectin from a sample taken with a speculum also had positive results from sample that was taken without a speculum (Table II). When the subgroups were compared, there were no significant differences between any
of the test characteristics. The Cohen κ statistic for agreement between the 2 methods of collection was 0.65 (95% confidence interval 0.44-0.87). Confounding factors. Confounding factors associated with a positive fetal fibronectin test results were minimal vaginal bleeding and recent sexual intercourse. The removal of women with those confounding factors tended to reduce the performance of the test. The performance values of the fetal fibronectin test when these factors are removed are shown in Table III. Comparison of cervical dilatation and fetal fibronectin testing. The characteristics of differing cervical dilatations for the prediction of delivery within 10 days were tested. The data for cervical dilatation ≥1 cm alone and in combination with fetal fibronectin are shown in Table IV. Cervical dilatation ≥1 cm had the highest sensitivity and had a similar specificity to those of cervical dilatation ≥2 cm and cervical dilatation of 3 cm (data not shown). Combination of cervical dilatation ≥1 cm with the fetal fibronectin test did not significantly improve the test performance (Table IV). Factors independently associated with delivery within 10 days of sampling were cervical dilatation ≥1 cm (χ2 = 28.3, P = .0001) and slight vaginal bleeding (χ2 = 10.1, P = .001). When slight vaginal bleeding was removed from the logistic regression model, the only significant independent factors associated with delivery within 10 days were cervical dilatation ≥1 cm (χ2 = 28.3, P = .0001) and the detection of fetal fibronectin (χ2 = 6.9, P = .009). The frequency of uterine activity on admission was not related to delivery within 10 days. False-negative and false-positive fetal fibronectin test results. Five percent of women in the speculum group (6/121) and 7% of women in the subgroup who had samples obtained without a speculum (8/107) had false-
Coleman et al 1557
Volume 179, Number 6, Part 1 Am J Obstet Gynecol
Table V. Clinical details of women with false-negative fetal fibronectin test results Case
GA at rec. (wk)
Ongoing uterine activity
Cervical dilatation at rec (cm)
Interval of rec to delivery (d)
1 2 3 4 5
32 33 32 26 28
Yes Yes No Yes No
2 0 1.5 0 0
3 1 9 4 2
6
33
No
2
5
Delivery details Spontaneous labor and vaginal delivery CD at 3 cm dilatation (abnormal FHR pattern) CD before labor for SRM and maternal request CD before labor for SRM and chorioamnionitis CD before labor for concealed abruptio placentae and deteriorating fetal condition CD before labor for suspected chorioamnionitis
GA, Gestational age; rec., recruitment; CD, cesarean delivery; FHR, fetal heart rate; SRM, spontaneous rupture of membranes.
negative results. The clinical details of the false-negative results obtained with a speculum are shown in Table V. Thirteen percent of women (16/121) had false-positive fetal fibronectin tests. When the subgroup who were examined with and without a speculum were compared, there were no significant differences in the false-positive or false-negative rates between sampling types (P = .40 and P = .42, respectively). Forty-four percent of the women with false-positive test results obtained with a speculum (7/16) had either recent sexual intercourse (within 24 hours, n = 5), slight vaginal bleeding (n = 4), or both (n = 2). Interobserver variation. Interobserver variation in the assessment of 3 fetal fibronectin test strips was evaluated between 36 clinicians. Eighty-nine percent (32/36) agreed with the clearly positive strip result. One hundred percent agreed with the negative test strip result. For the faintly positive result, 53% (19/36) classified it as positive and 47% (17/36) classified it as negative. Comment This is the first large prospective study to examine the clinical utility of this prototype bedside fetal fibronectin membrane immunoassay test kit. It also compared the value of combining fetal fibronectin testing with cervical dilatation for predicting preterm delivery in women with preterm labor. No other study has reported on the diagnostic value of this fetal fibronectin test or on that of samples taken from the vagina without a speculum. The clinical value of a test in a population can be determined by the degree that the test result changes the pretest probability of the outcome of interest in that population. Likelihood ratios are considered an appropriate way of expressing the ability of a diagnostic test to adjust the pretest probability of disease presence.18, 19 Pooled likelihood ratios have been reported in a recent overview of the diagnostic accuracy of both laboratory and bedside fetal fibronectin tests.20 For the prediction of preterm delivery within 1 week of testing in women with symptoms of preterm labor, these were found to be 5 (95% confidence interval 3.8-6.4) for a positive test result and 0.2 (95% confidence interval 0.1-0.4) for a negative test result. The findings in our study are consistent with these results (positive and negative likelihood ratios for deliv-
ery within 10 days of 4.6 and 0.6, respectively). Positive likelihood ratios between 2 and 5 and negative likelihood ratios between 0.5 and 0.2, however, are described as only “poor to fair.”19 This indicates that this prototype fetal fibronectin test has only a small effect on the pretest probability of delivery within 10 days and may therefore have limited clinical value. The sensitivity of the fetal fibronectin test in our study (65%) was less than that reported in previous studies of the bedside test (89%-100%)11-14 and also than that of the laboratory enzyme-linked immunosorbent assay test for predicting delivery within 7 days (80%-93%).5, 6, 8 False-negative test results diminish the sensitivity of a test, and the reduction in sensitivity in our study may be explained in part by consideration of the false-negative results. Among the samples taken with a speculum, only 2 of the 6 false-negative results were from women who were delivered within 10 days after spontaneous labor and therefore may be considered genuinely false negative (Table V, cases 1 and 2). In other studies the primary outcome has been more precisely defined as spontaneous preterm birth (after premature rupture of membranes or spontaneous preterm labor).21 If these criteria had been used in this study, the negative predictive value would have been 96%. Differentiation of a weakly positive result from a negative result can be difficult with this fetal fibronectin test. Many of the positive results in this study were faint lines. Despite the use of a single experienced observer, incorrect interpretation may have contributed to some of the false-negative results in this study. Our data on interobserver variation suggest that a significant number (53%) of clinicians may classify a test result as negative despite the presence of a faint line on the dipstick (indicating a positive result). Interobserver variation has not been reported for the other bedside test of fetal fibronectin. In this study cervical dilatation was assessed by a single observer. Use of cervical dilatation ≥1 cm to predict preterm delivery generated a positive likelihood ratio of 5.5, similar to that of the fetal fibronectin test alone (positive likelihood ratio 4.3). Combination of the fetal fibronectin test with cervical dilatation ≥1 cm did not significantly improve the test performance. Four other studies of fetal fibronectin have also re-
1558 Coleman et al
ported on the value of cervical dilatation as a predictor of preterm delivery.4-6, 12 In contrast to our study, each of these reported that fetal fibronectin testing was equivalent to or better than cervical dilatation in predicting delivery within 7 days. This difference between the findings in our study and those of others may be explained by a number of factors. First, in our study the sensitivity of the fetal fibronectin test was less than has been previously reported. Second, in the other studies cervical assessment was performed by multiple observers, which would increase the observer variation. The pretest probability of delivery within 10 days for those women with cervical dilatation ≥1 cm was 46%, and for those with slight vaginal bleeding was 54%. In these clinical settings, the fetal fibronectin test did not aid in decision analysis. A previous study5 also confirmed the independent association of vaginal bleeding with preterm delivery among women with preterm labor. Screening women for fetal fibronectin without the discomfort that may be caused by speculum examination would be simpler and might be more acceptable. The Cohen κ value of 0.65 indicates substantial agreement between the techniques for obtaining fetal fibronectin samples. Further studies are required, however, to establish whether there is any significant difference between these 2 methods for obtaining fetal fibronectin samples. Exclusion of women who had recent sexual intercourse did not change the test qualities significantly. Although these women had a trend toward more positive fetal fibronectin test results, a third of the false-positive results also occurred among these women. Conclusion. The prototype Fetal Fibronectin Membrane Immunoassay test performed only fairly in the prediction of delivery within 10 days. This may be related in part to the design of the test. If a more distinctive indicator or color change were possible in the presence of a weakly positive result, this might improve the test performance. In our study cervical dilatation ≥1 cm (assessed by a single researcher) correctly distinguished a similar proportion of women who were delivered within 10 days, 46% compared with 41% distinguished with the fetal fibronectin test. These women and those with slight vaginal bleeding are at high risk for preterm delivery, and in this context the addition of this fetal fibronectin test does not appear to improve decision analysis. The value in a clinical setting of fetal fibronectin testing in combination with clinical predictors needs further evaluation. We thank the following: Adeza Biomedical, Baxter Healthcare, Kevin Townend, and Vivienne Topping. REFERENCES
1. Copper RL, Goldenberg RL, Creasy RK, DuBard MA, Davis RO, Entman SS, et al. A multicenter study of preterm birth weight and gestational age–specific neonatal mortality. Am J Obstet Gynecol 1993;168:78-84.
December 1998 Am J Obstet Gynecol
2. Anonymous. Treatment of preterm labor with the beta-adrenergic agonist ritodrine. The Canadian Preterm Labor Investigators Group. N Engl J Med 1992;327:308-12. 3. Lockwood CJ, Senyei AE, Dische MR, et al. Fetal fibronectin in cervical and vaginal secretions as a predictor of preterm delivery. N Engl J Med 1991;325:669-74. 4. Peaceman AM, Andrews W, Thorp JM, Cliver SP, Lukes A, Iams JD, et al. Fetal fibronectin as a predictor of preterm birth in patients with symptoms: a multicenter trial. Am J Obstet Gynecol 1997;177:13-8. 5. Iams JD, Casal D, McGregor JA, Goodwin TM, Kreadon US, Lowensohn R, et al. Fetal fibronectin improves the accuracy of diagnosis of preterm labor. Am J Obstet Gynecol 1995;173 141-5. 6. Bartnicki J, Casal D, Kreaden US, Saling E, Vetter K. Fetal fibronectin in vaginal specimens predicts preterm delivery and verylow-birth-weight infants. Am J Obstet Gynecol 1996;174:971-4. 7. Burrus DR, Ernest JM, Veille JC. Fetal fibronectin, interleukin-6, and C-reactive protein are useful in establishing prognostic subcategories of idiopathic preterm labor. Am J Obstet Gynecol 1995;173:1258-62. 8. Malak TM, Sizmur F, Bell SC, Taylor DJ. Fetal fibronectin in cervicovaginal secretions as a predictor of preterm birth. Br J Obstet Gynaecol 1996;103:648-53. 9. Morrison JC, Allbert JR, MCLaughlin BN, Whitworth NS, Roberts WE, Martin RW. Oncofetal fetal fibronectin in patients with false labor as a predictor of preterm delivery. Am J Obstet Gynecol 1993;168:538-42. 10. Langer B, Boudier E, Schlaeder G. Cervico-vaginal fetal fibronectin: predictive value during false labor. Acta Obstet Gynecol Scand 1997;76:218-21. 11. Parker J, Bell R, Brennecke S. Fetal fibronectin in the cervicovaginal fluid of women with threatened preterm labour as a predictor of delivery before 34 weeks’ gestation. Aust N Z J Obstet Gynaecol 1995;35:257-61. 12. Senden IP, Owen P. Comparison of cervical assessment, fetal fibronectin and fetal breathing in the diagnosis of preterm labour. Clin Exp Obstet Gynecol 1996;23:5-9. 13. Benattar C, Taieb J, Fernandez H, Lindendaum A, Frydman R, Ville Y. Rapid fetal fibronectin swab-test in preterm labor patients treated by betamimetics. Eur J Obstet Gynecol Reprod Biol 1997;72:131-5. 14. Owen P, Scott A. Can fetal fibronectin testing improve the management of preterm labour? Clin Exp Obstet Gynecol 1997;24:19-22. 15. Rizzo G, Capponi A, Arduini D, Lorido C, Romanini C. The value of fetal fibronectin in cervical and vaginal secretions and of ultrasonographic examination of the uterine cervix in predicting premature delivery for patients with preterm labor and intact membranes. Am J Obstet Gynecol 1996;175:1146-51. 16. Rozenberg P, Goffinet F, Malagrida L, Giudicelli Y, Perdu M, Houssin I, et al. Evaluating the risk of preterm delivery: a comparison of fetal fibronectin and transvaginal ultrasonographic measurement of cervical length. Am J Obstet Gynecol 1997;176:196-9. 17. Gardosi J, Mul T, Francis A, Hall J, Fishel S. Comparison of second trimester biometry in singleton and twin pregnancies conceived with assisted reproductive techniques. Br J Obstet Gynaecol 1997;104:737-40. 18. Jaeschke R, Guyatt G, Sackett DL. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? Evidence-Based Medicine Working Group. JAMA 1994;271:389-91. 19. Schechter MT, Sheps SB. Diagnostic testing revisited: pathways through uncertainty. Can Med Assoc J 1985;132:755-60. 20. Chien PF, Khan KS, Ogston S, Owen P. The diagnostic accuracy of cervico-vaginal fetal fibronectin in predicting preterm delivery: an overview. Br J Obstet Gynaecol 1997;104:436-44. 21. Goldenberg RL, Iams JD, Mercer BM, Mels PJ, Moawad AH, Copper RL, et al. The preterm prediction study: the value of new vs standard risk factors in predicting early and all spontaneous preterm births. Am J Public Health 1998;88:233-8.