American Journal of Obstetrics and Gynecology (2004) 191, 235e40

www.elsevier.com/locate/ajog

Does fetal fibronectin use in the diagnosis of preterm

labor affect physician behavior and health care costs?
A randomized trial
W. A. Grobman, MD, MBA,* E. E. Welshman, MSW, E. A. Calhoun, PhD
Sections of Maternal-Fetal Medicine, Northwestern University Medical School and the Institute for Social Research
and Health Policy Studies, Northwestern University, Chicago, Ill
Received for publication August 22, 2003; revised November 20, 2003; accepted November 26, 2003

KEY WORDS
Fetal bronectin
Preterm contraction

Objective: The purpose of this study was to determine whether a knowledge of fetal bronectin
results aects patient treatment and health care costs.
Study design: Women between 24 and 34 weeks of gestation with a singleton pregnancy and preterm uterine contractions were eligible for enrollment. Once informed consent was given, a fetal
bronectin specimen was obtained, and women were assigned randomly into 2 groups. In
1 group, results of the fetal bronectin test were available; in the other group, results were not
available. The use of inpatient and outpatient health care resources subsequent to enrollment
was ascertained through the use of medical records, hospital billing data, and patient interviews.
This study was powered to allow the detection in the fetal bronectin group of a 20% reduction in
total health careerelated costs.
Results: The 2 groups were similar with respect to maternal age, parity, race, cervical examination
at admission, and estimated gestational age at enrollment and at delivery. Women who did not
have fetal bronectin results available were no dierent than those women who did with respect
to initial length of labor and delivery observation (median, 4 hours vs 3 hours), hospital admission (28% vs 26%), tocolysis (18% vs 16%), cessation of work (27% vs 26%), or total health
careerelated costs (log mean G SD, 7.6 G 1.2 vs 7.5 G 1.1).
Conclusion: In this study population, the use of fetal bronectin did not aect physician behavior
or health care costs related to preterm contractions.
2004 Elsevier Inc. All rights reserved.

The presence of fetal bronectin in the posterior vaginal fornix of women with preterm uterine contractions

Supported by a grant from Adeza Biomedical, Sunnyvale, Calif.

* Reprint requests: William Grobman, MD, MBA, 333 East
Superior St, Suite 410, Chicago, IL 60611.
E-mail: w-grobman@northwestern.edu
0002-9378/$ - see front matter 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.ajog.2003.11.034

has been shown repeatedly to be associated with subsequent preterm delivery. The predictive capability of the
fetal bronectin test, however, is, in fact, greatest when
the test is negative. In the large cohort documented by
Peaceman et al,1 for example, women with a positive test
result had a 13% chance of being delivered within
1 week, and women with a negative test result had
a O99% chance that they would not be delivered within

236
1 week. Thus, the high negative predictive value of fetal
bronectin sampling supports less intervention for patients with this result.
The question remains, however, whether fetal bronectin use in clinical practice actually will lead to decreased interventions. Three studies have attempted to
elucidate the relationship between the use of the test
and consequent physician behavior. Giles et al2 in
New Zealand found that women with preterm contractions who had a negative fetal bronectin test result
were less likely to be transferred to a tertiary care center
than women with a positive test result. However, there
was no formal cost analysis or control group of women
who had not received the fetal bronectin test; therefore,
it was difcult to know the attributable impact of the
test itself. Joffe et al3 did compare a cohort of women
who had the fetal bronectin test with women who did
not and found that the use of the test resulted in lower
hospital costs, but this control group was historic, and
the biases that are generated by this type of control impair the validity of the conclusions. Plaut et al4 randomly assigned women with preterm contractions to
have fetal bronectin results either available or not
available. Their study showed no signicant differences
between the 2 groups, although there was not sufcient
power to evaluate the primary end point, namely, transport to a tertiary care facility.
Consequently, we felt that it was important to perform a prospective randomized trial to prove that fetal
bronectin could impact favorably on physician behaviors. The specic aim of this trial was to assess whether
the availability of fetal bronectin results for women
who undergo evaluation for preterm labor would affect
physician resource use and health care costs.

Methods
Patients who came to labor and delivery with complaints that were associated with preterm labor were
considered for participation in the study. Women who
were considered eligible for the trial were between 24
and 34 weeks of gestation with a singleton pregnancy,
had a primary complaint of uterine contractions, and
had O6 contractions per hour by external tocodynometry. Patients with vaginal bleeding, a cerclage, nonintact
amniotic membranes, R3 cm cervical dilation, or a vaginal examination or sexual intercourse within 24 hours
were excluded, because these conditions may affect the
reliability of the fetal bronectin test results. Patients
were also excluded if they had already received hospital
observation, admission, or treatment for preterm contractions.
Those patients who met the eligibility criteria and
who choose to participate signed written consent and
subsequently underwent a standardized evaluation by

Grobman, Welshman, and Calhoun

a physician. This evaluation included a focused history,
an assessment of fetal heart tones (with an external monitor) and uterine contraction frequency (with an external
monitor), and a physical examination. As part of the
physical examination, patients underwent an examination of the cervix with a speculum, at which time a Dacron swab was placed in the posterior vaginal fornix for
10 seconds to absorb cervicovaginal secretions. Immediately thereafter, a digital cervical examination was performed.
Patients were then assigned randomly either to
availability of fetal bronectin results or no availability of fetal bronectin results. Randomization was
performed through the use of computer-generated random assignment. Patient assignments were placed in
sequentially numbered opaque envelopes that were
maintained at labor and delivery. After a patient consented, an envelope, which was next in the sequential
ordering, was opened to reveal the patients group
assignment. The Dacron swab from the availability
group was sent immediately to the laboratory for an assessment of the fetal bronectin. The Dacron swab from
the no availability group was stored at
20(C. The
attending physicians were apprised of the patients history, physical examination, and randomized group assignment. The fetal bronectin test had not been used
in labor and delivery before the study, and during the
study, the test was only available within the study protocol.
The hospital laboratory performed the analysis of the
fetal bronectin specimen using the Rapid Fetal Fibronectin analyzer (the TLi cassette used with the TLi analyzer; Adeza Biomedical, Sunnyvale, Calif). Once
available, the results for the group of patients who were
assigned randomly to availability of fetal bronectin
results were communicated immediately to the resident
physician.
The resident communicated the results of the test to
the attending physician. To ensure that the attending
physician was aware of the test characteristics (sensitivity, specicity, and positive and negative predictive
value) of the fetal bronectin assay, 2 steps were taken.
First, before the start of the study, a letter was mailed to
all admitting obstetricians that detailed the study and
the test characteristics of the fetal bronectin assay. Additionally, each time an attending physician was called
with fetal bronectin results, those results were given
in the context of the following statement: If a patient
has a positive result, her risk of delivering in the next
7 days is approximately 13%; if she has a negative result, her risk of delivering in the next 7 days is less than
1%. Your patient had a positive/negative test. The
swabs from the group of patients who were assigned
randomly to no availability of fetal bronectin results
were analyzed as a single batch when all patients had
been assigned randomly and all follow-up data had been

237

Grobman, Welshman, and Calhoun

obtained. Laboratory personal who performed the fetal
bronectin test were blinded to patient characteristics
and outcomes.
All treatment decisions (with the exception of obtaining fetal bronectin results) for patients in each study
arm, both during and after admission to labor and delivery, were made by the attending physician. These decisions included the length of time of observation in
labor and delivery, a need for tocolysis and hospital admission, the hospital evaluations and admissions subsequent to the evaluation at the time of study entry, and
a need for bed rest. Patients who were seen for evaluation of preterm contractions after their initial random
assignment remained in their originally randomly assigned group and were not assigned randomly again.
The physicians of those patients who were assigned randomly to the results available arm had the option of
ordering the rapid fetal bronectin test again if the patient was seen O7 days after the previous fetal bronectin assay.
Patient demographic information was abstracted
from the charts at the time of the initial study recruitment. Once the patient had been delivered, the inpatient
and outpatient charts were abstracted to ascertain the
medical resources (eg, the number of hospital admissions, number of hospital days, hours on labor and delivery, and number of tocolytic treatments) that they
had required during their preterm antepartum care.
The medical billing department of the hospital provided
the total costs that were associated with this resource
use. Additionally, at delivery or at 36 to 37 weeks of gestation (whichever occurred rst), the patients were contacted either while in the hospital or by a phone call to
their homes. At this time, a research assistant administered a survey. This survey ascertained the contribution
of their admission for preterm contractions to the indirect costs of illness, such as lost income from time off
from work and need to hire additional home assistance.
This survey also elicited other noneconomic consequences (ie, loss of leisure time) that the patients had
experienced as a result of the preterm contractions and
the emotional content of patient experiences after their
discharge from the hospital.
Comparisons of continuous data were performed
with the Student t test or the Mann-Whitney U test,
as appropriate. The normality of data was assessed
with the Kolmogorov-Smirnov test, and data that were
not distributed normally were re-examined for normality after logarithmic transformation. Categoric variables
were analyzed with chi-squared analysis and Fishers exact test, as appropriate. Selected subgroup analysis was
performed using 2-way analysis of variance (ANOVA).
All tests were 2-tailed, and a probability value of !.05
was considered statistically signicant. Statistical analysis was performed with Minitab 13 (Minitab Inc, State
College, Pa).

This study was designed to determine whether differences in fetal bronectin availability could affect physician behavior and resource use. Correspondingly, the
primary end point was the total costs that were incurred,
as a consequence of preterm contractions, after study
enrollment. This total cost was determined by the combination of both the direct medical and nonmedical
costs that were related to obstetric care after the admission for preterm contractions. With an a score of .05
and a b score of .8, the determination was made that
50 patients must be enrolled in each study arm to demonstrate that fetal bronectin use could decrease the
total costs of $1500 by 20%.
This study was approved by the Institutional Review
Board of Northwestern University.

Results
Over a 12-month period, 112 patients were offered enrollment in the study. Of these, 12 patients declined
study participation; the remaining 100 patients were divided equally between the 2 groups by the randomization process. In the case of 1 woman who was
assigned randomly to the group with results immediately
available, the fetal bronectin results were not actually
available, because the rapid analyzer machine could
not generate a result. Nevertheless, the outcomes for this
patient were analyzed along with the other members of
the group to which she was assigned randomly.
Relevant characteristics of the study and control
group are listed in Table I. The patients in each group
did not differ in their gestational age or cervical dilation
at admission and had comparable socioeconomic backgrounds. Their pregnancy outcomes also were not signicantly different. Similar proportions of the patients
had positive fetal bronectin test results and preterm deliveries.
The predictive values of the fetal bronectin test were
similar to that reported in other studies.1,5 Of the 8
women with a positive test result, 1 woman was delivered within 1 week, and 5 women were delivered before
37 weeks of gestation, which gave a positive predictive
value for these outcomes of 12.5% and 62.5%, respectively. Of the 91 women with a negative test result, only
3 women were delivered within 1 week, and 17 women
were delivered before 37 weeks of gestation, which gave
negative predictive values of 96.7% and 81.3%, respectively.
As illustrated in Table II, there were no differences
between the 2 groups in medical resource usage. The
number of hours that patients spent in labor and delivery during their initial evaluation for preterm uterine
contractions was the same, regardless of whether physicians knew the results of the fetal bronectin test. Also,
the knowledge of these results did not change the

238

Grobman, Welshman, and Calhoun

Table I Demographic characteristics and pregnancy outcomes, stratified by study group

Characteristic/outcome
Maternal age (y)*
Estimated gestational
age at study
entry (wk)*
Nulliparity (n)
White (n)
Previous preterm
delivery (n)
Private insurance (n)
Dilation at admission
(cm)*
Fetal fibronectin
positive (n)
Preterm delivery (n)
Days from
study entry to delivery
(n)y
Estimated gestational
age at delivery
(wk)*

Fetal
fibronectin
results
available
(n = 50
women)

Fetal
fibronectin
results not
available
(n = 50
women)

Table II

Medical resource use, stratified by study group

Resource (n)
P value

29 G 6
29 G 3

29 G 6
29 G 3

.89
.96

28 (56%)
23 (46%)
4 (8%)

20 (40%)
21 (42%)
8 (16%)

.11
.69
.22

34 (64%)
0.5 G 0.7

29 (58%)
0.5 G 0.8

.30
.79

5 (10%)

3 (6%)

.45

10 (20%)
60 (47-74)

13 (26%)
60 (43-80)

.48
.87

Fetal
Fetal
fibronectin fibronectin
results not
results
available available
(n = 50
(n = 50
women) women) P value

Hours in labor
3
and delivery*
Tocolysis
8
Corticosteroids
8
Hospital admission
13
at time of study entry
Days*
2
6
Return to labor
and delivery for preterm
contractions after initial
discharge
5
Hospital admission
for preterm contractions
at any time
after study entry
Days during hospital admission 4

(2-5)

4 (2-5)

.44

(16%) 9 (18%) .79

(16%) 10 (20%) .60
(26%) 14 (28%) .82
(1-5)
2 (1-5) .83
(12 %) 12 (24 %) .10

(10%)

4 (8%)

.78

(2-7)

2 (1-11) .62

* Data are presented as median (interquartile range).

38 G 3

38 G 3

.81

* Data are presented as mean G SD.

y
Data are presented as median (interquartile range).

frequency of tocolysis, corticosteroid use, or conversion

to inpatient status immediately subsequent to the initial
evaluation. After patients were discharged from the hospital after their initial evaluation, the incidence of hospital readmission was similar in both groups.
Eighty-ve women completed the postpartum survey
at the scheduled time. The frequency of successful postpartum contact was no different between the women
who did and did not have the fetal bronectin results
available (44/50 vs 41/50 women, respectively; P O
.05). The proportion of women who were not working
before study enrollment also was similar between the 2
groups (15/44 and 18/41 women, respectively; P O
.05). In both the study and control groups, most patients
missed some work after their initial preterm delivery
evaluation (27/29 vs 22/23 women, respectively;
P O .05), and more than one quarter of the women in
each group never returned to work (7/27 vs 6/22 women,
respectively; P O .05). Most women in both groups felt
it necessary to avoid typical leisure activity (31/44 vs
25/41 women, respectively; P O .05) and arrange for increased help at home (33/44 vs 30/41 women, respectively; P O .05). There was no difference in the number
of those women who needed to pay for this increased
help (10/44 vs 7/41 women, respectively; P O .05).

Because knowledge about fetal bronectin results

may also help to allay a womans anxiety regarding preterm contractions, we asked women several questions
about their emotional state subsequent to their release
from the hospital after the initial preterm contraction
evaluation (Table III). In addition to the 15 women
who did not have results available for the questionnaire,
5 women were never discharged from the hospital after
their initial evaluation; consequently, results for these
questions are given for the remaining 80 women.
Women who had fetal bronectin results available were
as likely to feel nervous, frequently to think about their
uterine contractions, and to feel satised with their medical care.
Medical and nonmedical costs were calculated for the
women in each group. For all women, the mean direct
inpatient costs and mean indirect costs were $2807 and
$1013, respectively; the mean total expense was $3678.
These expenses were not normally distributed. To compare data with the use of parametric tests, we performed
a logarithmic conversion, after which the values were
distributed normally (Table IV). As shown, there were
no differences between the 2 groups with respect to
any cost variable, including the primary outcome of this
study, which were the total costs.
We examined how other factors might have inuenced our ndings. For example, perhaps the fetal bronectin test would be most useful at changing physician
behaviors for those patients who had cervical dilation
in addition to their preterm contractions. Patients were
subdivided into women with (n = 38) and without

239

Grobman, Welshman, and Calhoun

Table III Subjective responses after hospital discharge,
stratified by study group

Response (n)
Since admission, I have
been a very
nervous person.
Never
Sometimes
Most or all of the time
Since admission, I have
thought about preterm
contractions.
Never
Sometimes
Most or all of the time
Since admission, I have
been satisfied with
my medical care.
Never
Sometimes
Most or all of the time

Fetal
fibronectin
results
available
(n = 42
women)

Fetal
fibronectin
results not
available
(n = 38
women)

P value
.81

Table IV

Hospital and nonhospital costs, stratified by group

Cost

Fetal
fibronectin
results available
(n = 50 women)

Fetal
fibronectin
results not
available
(n = 50 women)

P value

Hospital
Nonhospital
Total

7.1 G 1.1
6.6 G 1.4
7.5 G 1.1

7.4 G 1.2
6.9 G 1.6
7.6 G 1.3

.16
.57
.51

Data are presented as log mean G SD ($).

8 (19%)
27 (64%)
7 (17%)

8 (21%)
22 (58%)
8 (21%)
.49

6 (14%)
13 (31%)
23 (55%)

4 (11%)
14 (37%)
20 (52%)
.31

0 (0%)
3 (7%)
39 (93%)

0 (0%)
7 (18%)
31 (82%)

(n = 32) cervical dilation on the rst pelvic examination,

and a 2-way ANOVA was performed. Neither the main
effect nor the interaction term of this ANOVA was signicant, which indicated that the fetal bronectin results
did not change physician behavior, regardless of cervical
dilation. We also hypothesized that physicians might experience a learning curve and react differently to the fetal bronectin results the longer that the test was in use.
We subdivided the patients into the rst and second 50
patients whose conditions were analyzed, but no such
learning curve could be demonstrated (P O .05, main effect and interaction term of 2-way ANOVA.)
Last, we imagined that patients with public aid
(n = 37), who were cared for primarily by house staff
as opposed to disparate private attending physicians
(n = 53), might have received different therapy when fetal bronectin results were known. However, no difference could be elucidated (P O .05, main effect and
interaction term of 2-way ANOVA).

Comment
The economic impact of preterm labor on society is substantial.6 The most obvious economic burden comes
from the intensive medical care that is required by preterm neonates. Another signicant component, however, comes from the over treatment of women who

are not actually in preterm labor. This over treatment

is prevalent because the signs and symptoms of women
in actual preterm labor are far from unique and overlap
considerably with those of women with preterm contractions only. Thus, the development of a test that can help
to identify those women who are not likely to deliver
preterm has been desired widely.
Fetal bronectin holds the promise of being such
a test. In several studies, it has been shown repeatedly
to have a high negative predictive value for delivery
within 1 week of testing.1,5,7 Because women with a negative fetal bronectin test result are at low risk for imminent delivery, physicians correspondingly should be less
likely to prescribe unnecessary treatment (eg, tocolytic
medications) and interventions that have long been used
but have never been shown to be benecial (eg, bed
rest).8 The reduction of these needless strategies could
help reduce medical costs and social disruption.
Yet, changes in physician behavior and the discarding
of established paradigms have proved difcult to engender in many aspects of medicine, and it is not inevitable
that the introduction of fetal bronectin will result in
improved efciencies. For example, if physicians are reluctant to abandon interventions (such as bed rest) in
the setting of a negative test result and begin to use increased interventions after a positive test (which has
a low predictive value), the number of medical and social resources that are consumed actually will increase
after the use of the test. Thus, the proposition that fetal
bronectin will decrease the consumption of these resources should be proved and not assumed.
There has been only 1 prospective, randomized study
to date that has examined whether fetal bronectin will
fulll its potential as a benecial modier of clinical
care. In this study, patients were enrolled in 4 different
community hospitals; although no difference in the primary outcome (number of transports to a tertiary care
facilities) was found, the study was underpowered because of lack of enrollment. We believe that Northwestern Memorial Hospital was a favorable setting in which
to evaluate more fully the usefulness of the fetal bronectin test, because the hospital had not introduced fetal
bronectin into clinical practice. Consequently, it was

240
not available outside of a study protocol, in which the
test was introduced in a randomized fashion to consenting patients who had preterm contractions. To ensure
that physicians understood the characteristics of the
newly available test results, an initial letter and a continued summary of these characteristics was provided to
the attending staff. However, we could nd no evidence
that the availability of the test results caused physicians
to change their clinical care. Medical interventions (such
as tocolytic and steroid administration and bed rest and
the corresponding decrement in work and leisure times)
were no different if the women had fetal bronectin results available. Economic consequences, whether medical or nonmedical, of the preterm contractions were
consequently no different.
This lack of impact of the fetal bronectin test does
not appear to be due to differences in the 2 groups under
study, because randomization resulted in comparable
demographic characteristics and pregnancy outcomes.
Also, subgroup analysis did not reveal that particular
circumstances (such as preterm contractions and a certain degree of cervical dilation) allowed the fetal bronectin test to provide optimal impact.
Subgroup analysis cannot rule out that fetal bronectin would have been more useful under particular circumstances, because this part of the analysis is not
randomized and is more prone to type II error. Additionally, although the subgroup analysis did not reveal
the existence of a learning curve, this study occurred
over the course of approximately 1 year, and it is possible that, if the test were used for a longer period, physicians might have become more comfortable with the
reliability of the test and changed their clinical responses. Also, this study was performed in a university
hospital that has 24-hour resident coverage and a relatively strict protocol for preterm labor management;
these conditions may not give insight into the value of
the test in different health settings.
Nevertheless, this study was unable to prove that the
knowledge of fetal bronectin results changes physician

Grobman, Welshman, and Calhoun

behavior, patient perceptions, or the consumption of
medical and economic resources. Because there has been
no other prospective randomized trial that has evaluated
the effect of fetal bronectin on physician behavior, the
supposition that it will inevitably enhance diagnosis and
improve resource use remains unproved. Further study
needs to be undertaken to better understand whether
and under what circumstances the predictive characteristics of the fetal bronectin test can be translated into
better clinical processes. Perhaps, for example, a standardized clinical pathway may need to be implemented
if the value of these tools is to be realized optimally.

References
1. Peaceman AM, Andrews WW, Thorp JM, Cliver SP, Lukes A,
Iams JD, et al. Fetal bronectin as a predictor of preterm birth in
patients with symptoms: a multicenter trial. Am J Obstet Gynecol
1997;177:13-8.
2. Giles W, Bisits A, Knox M, Madsen G, Smith R. The effect of fetal
bronectin testing on admissions to a tertiary maternal-fetal
medicine unit and cost savings. Am J Obstet Gynecol 2000;182:
439-42.
3. Joffe GM, Jacques D, Bemis-Heys R, Burton R, Skram B,
Shelburne P. Impact of the fetal bronectin assay on admissions
for preterm labor. Am J Obstet Gynecol 1999;180:581-6.
4. Plaut MM, Smith W, Kennedy K. Fetal bronectin: the impact
of a rapid test on the treatment of women with preterm labor
symptoms. Am J Obstet Gynecol 2003;188:1588-95.
5. Rinehart BK, Terrone DA, Isler CM, Barrilleaux PS, Bufkin L,
Morrison JC. Pregnancy outcome in women with preterm labor
symptoms without cervical change. Am J Obstet Gynecol 2001;
184:1004-7.
6. Nicholson WK, Frick KD, Powe NR. Economic burden of
hospitalizations for preterm labor in the United States. Obstet
Gynecol 2000;96:95-101.
7. Revah A, Hannah ME, Sue-A-Quan AK. Fetal bronectin as
a predictor of preterm birth: an overview. Am J Perinatol 1998;15:
613-21.
8. Goldenberg RL. The management of preterm labor. Obstet
Gynecol 2002;100:1020-37.