| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Clinical Laboratory, Department of Pathology, Brigham and Women’s Hospital, 75 Francis St., Boston, MA 02115
aauthor for correspondence: fax 617-731-4872, e-mail effoxman{at}partners.org
Early delivery attributable to preterm labor remains a significant cause of neonatal morbidity and mortality. Obstetricians often face the challenge of deciding whether to admit a woman to the hospital when she presents with possible preterm labor. In recent years, there has been much emphasis on developing laboratory tests that can help clinicians predict the likelihood of "true" preterm labor. The fetal fibronectin (fFN) assay has attracted interest because of its reported high negative predictive value for preterm labor. We implemented the Rapid Fetal Fibronectin lateral immunodiffusion assay on the TLiIQTM System (Adeza Biomedical) in the Brigham and Women’s Hospital clinical laboratory in October 2001. To determine how the test results would influence clinical decision-making in our hospital, we created a survey to be completed by clinicians at the time of test requisition. In this report, we compare the survey results, which reflect the pretest intention of the ordering clinicians, to the clinical course of tested patients.
fFN, a uniquely glycosylated form of the abundant plasma protein fibronectin, is found in the placenta and amniotic fluid, primarily at the uteroplacental junction, and is released into the upper vagina near the onset of labor (1). Although the presence of fFN in vaginal fluids does not necessarily indicate the onset of labor (positive predictive value, 15–25%), its absence rules out labor within 7 days with a very high negative predictive value (97–99.5%) (1)(2)(3). Therefore, a negative result indicates a low likelihood of delivery, but a positive test should not be interpreted as an indication of labor or a reason for admission.
From January to June 2002, all physicians requesting a fFN test at Brigham and Women’s Hospital were asked to complete a brief survey with the test requisition. On this survey, the physician was asked to record the gestational age of the fetus and the number of fetuses and to answer the following question by circling any or all of four possible responses: "I am sending a fFN test on this patient, because a negative result will help me to avoid one or more of the following interventions that I am currently considering: (1) antenatal corticosteroids, (2) admit to observation, (3) tocolysis, and (4) other".
Institutional Review Board approval was obtained from the Human Research Committee at Brigham and Women’s Hospital to review the medical records of patients who received a fFN test during the study period. A total of 245 tests were performed during this time period. Gestational age was between 22 and 34 weeks for all of the patients, with a mean maternal age of 26 years. We were able to obtain relevant clinical data for 175 tests. Only patients with singleton gestations (152 of 175 tests; 139 patients) were included in subsequent analyses. Twelve of these patients received more than one fFN test during the study period; only the first fFN test result was included in the analysis. Of the 139 patients studied, we were able to obtain survey results for 58 (42%).
Positive test rates were similar in the overall patient population, in the subset of patients with clinical data, and in the subset of patients with data and survey results (
22%; Table 1A
). The proportion of patients who delivered within 7 days and the test performance were similar in the patients with clinical data (n = 139) and in the subset with data plus survey results (n = 58). These observations suggest that the smaller patient subset with survey results is representative of the larger subset. Data analysis is presented by patient subset in Table 1
for comparison purposes. Exact 95% confidence intervals around proportions were calculated by the Blyth–Still–Casella method. Proportions were compared using the Fisher exact test. StatXact software (Ver. 5) was used (Cytel Software Corporation).
|
Although assessing test performance was not the primary objective of our study, we did calculate test performance for comparison with previously published studies. When we used as endpoint delivery within 7 days, fFN test performance at our hospital was similar to that reported previously (Table 1B
) (1)(2)(3). As expected, the fFN test had a high negative predictive value and low positive predictive value. The high negative predictive value reflects both test performance and the low overall prevalence of preterm delivery in our patient population. For a more population-independent assessment of test performance, we calculated the likelihood ratios for positive results (4.53; 95% confidence interval, 2.84–7.20) and negative results (0.18; 95% confidence interval, 0.03–1.08) for all patients with clinical data. The 95% confidence intervals around these ratios were quite wide, but the point estimates of the likelihood ratios we obtained were very similar to those reported in a much larger metaanalysis of fFN performance in which the ratios showed narrow confidence intervals (4).
The major goal of this study was to determine the effect of implementing the fFN test on hospital admission rates. We hypothesized that if clinicians were using a negative fFN result to rule out preterm labor, test implementation would lead to a decrease in admissions for preterm labor. One previous study showed reduced hospital admissions for preterm labor as a result of fFN testing by comparing admission rates before and after test introduction (5). Because of changes in our hospital’s referral base in early 2002, we did not believe we could accurately assess the impact of fFN testing by comparing admission rates before and after test introduction. We therefore assessed the effect of implementing the fFN test as follows. We first compared the rate of hospital admission for >12 h for fFN-negative patients to the admission rate for fFN-positive patients. We found that fFN-negative patients were less likely to be admitted than fFN-positive patients (2% of fFN-negative patients vs 19% of fFN-positive patients were admitted; see Table 1C
). This difference in admission rates was statistically significant (P = 0.002). For the smaller subset of patients with survey results (n = 58), the trend was similar although not statistically significant (2% vs 14%; P = 0.14). These data suggest that clinicians are incorporating the fFN result in their decision-making process. However, it is also possible that the differential admission rates reflect differences in the pretest characteristics of fFN-negative and -positive patients, rather than the impact of test results on clinical management.
To help distinguish between these possibilities, we examined the results of the pretest survey that accompanied each test requisition. We determined which patients had a high pretest probability of admission by noting whether the clinician circled "admit to observation" as a possibility on the pretest survey. If this option was circled, we counted the patient as "admission considered" (Table 1D
). Although we cannot know the frequency with which these patients would have been admitted without the fFN result, we reasoned that clinicians’ responses to this survey question should reflect any pretest differences between the fFN-positive and -negative patient populations. Overall, 57% of patients were considered for admission. Interestingly, the proportions of fFN-positive and -negative patients considered for admission were very similar (64% and 55%, respectively; P = 0.56). This finding suggests that the fFN result itself, rather than a pretest characteristic of the two patient populations, was the major reason for the differential admission rates in fFN-negative vs fFN-positive patients.
A secondary point regarding test utilization should be emphasized as well. When a test with a low positive predictive value is introduced, there is always a concern that clinicians will overinterpret a positive test result. For the fFN test, a positive test result does not indicate a likely preterm delivery. However, if clinicians do not fully understand this limitation of the test, they may place too much weight on a positive test result and be more inclined to admit patients who test positive than they would have been based on clinical criteria only. In this setting, test implementation could "backfire" and lead to an increase, rather than a decrease, in admission rates. Although our results show that a disproportionate number of patients with positive test results were admitted, this represented only 19% of fFN-positive patients. This suggests that clinicians were not overinterpreting positive test results and indeed were comfortable foregoing admission on fFN-positive patients.
In summary, we describe selected aspects of 6 months of clinical experience with the fFN test at Brigham and Women’s Hospital. fFN-negative patients were less likely to be admitted than fFN-positive patients, and this difference appeared to be attributable to clinical use of the fFN test result rather than to pretest factors. A limitation of our study is the small number of patients for whom we have pretest survey data. Furthermore, our study does not directly address how changes in admission practices affect overall patient outcomes. Although future studies will be needed, the data presented here represent one of the first direct analyses of fFN test utilization and support the conclusion that use of fFN testing reduces hospital admissions for preterm labor.
Acknowledgments
We thank Drs. Errol Norwitz and James W. Winkelman for helpful advice and assistance in implementing the survey, Dr. Tom McElrath for many helpful discussions and comments on the clinical use of the fFN test, and Dr. Shelley Hurwitz for helpful assistance with statistical analyses.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
