- Email: [email protected]
Multicenter randomized clinical trial of horne uterine activity monitoring for detection of preterrn labor
Multicenter randomized clinical trial of horne uterine activity monitoring for detection of preterm labor Susan M. Mou, MD: Shirazali G. Sunderji, MD; Stanley Gall, MD,c Helen How, MD,c Vinu Patel, MD,b Mark Gray, MD: Herbert L. Kayne, PhD/ and Michael Corwin, MD" Kansas City, Missouri, Syracuse, New York, Chicago, Illinois, and Boston, Massachusetts Home uterine activity monitoring has been described as an effective means of detecting uterine contractions, but controversy exists whether it is home uterine activity monitoring or increased nursing support in conjunction with it that contributes to earlier detection of preterm labor. In this study 377 women at risk for preterm labor from three centers were prospectively, randomly assigned to high-risk prenatal care alone (not monitored) or to the same care with twice-daily home uterine activity monitoring without increased nursing support (monitored). The two groups were medically and demographically similar at entry into the study. Routine visits, nonroutine visits, and gestational age at diagnosis of preterm labor were similar in both groups. Preterm labor occurred in 41 of 198 monitored and 39 of 179 not monitored patients. Mean cervical dilatation was 1.4 cm in 41 monitored compared with 2.5 cm for 37 not monitored (p = 0.0006); 73.1% of monitored and 27.5% of not monitored had preterm labor detected before 2 cm dilatation (p = 0.00009). Neonatal outcome of singleton pregnancies showed greater birth weight, fewer days in the neonatal intensive care unit, and fewer babies requiring oxygen therapy and mechanical ventilation in the monitored group. The better outcomes are probably due to the increased likelihood of diagnosis of preterm labor before advanced cervical dilatation with home uterine activity monitoring, thus providing the clinician with a better chance to initiate tocolytic therapy directed at improving pregnancy outcome. (AM J OBSTET GVNECOL 1991 ;165:858-66.)
Key words: Home uterine activity monitoring, preterm labor Preterm delivery is the most significant contributor to perinatal morbidity and mortality in the United States.' Unfortunately, many episodes of preterm labor are not detected until it is too late to provide effective tocolytic therapy. Recent studies have shown that an increase in the number of uterine contractions precedes the onset of preterm labor and is present in early labor; however, women often are not aware of the increased number of contractions until labor is advanced."' 3 Studies showing that devices designed to monitor uterine activity offer a more reliable method of quantitating contractions suggest that home uterine activity monitoring may allow earlier detection of the onset of preterm labor!' 5 Although many studies examining its efficacy have suggested beneficial effects, most of these studies have included intensive perinatal nursing support in combination with the objective data obtained
From the Department of Obstetrics and Gynecology, Truman Medical Center, University of Missouri-Kansas City: the Perinatal Center, Health Science Center at Syracuse, State University of New York,' the University of Illinois Hospital,' the Boston University School of Public Health! and the Department of Pediatrics, Boston City Hospital and Boston University .' Presented at the Eleventh Annual Meeting of the Society of Perinatal Obstetricians, San Francisco, California, January 28-February 2, 1991. Reprint requests: Dr. Susan M. Mou, Department of Obstetrics and Gynecology, Truman Medical Center, Kansas City, MO 64108, 6/6/31430
858
with a device. 5- 7 Therefore it has not been possible to separate the effects of the detection of increased uterine activity. A prospective, randomized, blinded, multicenter trial of a home uterine activity monitoring device as the sole addition to current standard high-risk obstetric care was conducted to evaluate the effect of objective measurement of uterine activity in the diagnosis of preterm labor.
Methods Recruitment and eligibility. Prenatal charts at three study sites, Truman Medical Center, Kansas City, Mo., State University of New York Health Science Center, Syracuse, N.Y., and the University of Illinois Hospital, Chicago, were reviewed at the time of the initial prenatal visit, at 22 to 26 weeks' gestation, and at 32 weeks' gestation. Recruitment was conducted from September 1988 to August 1989 at Truman Medical Center and the State University of New York, Syracuse, but was terminated in June 1989 at the University of Illinois because of intended closure of the hospital. Subjects were eligible for recruitment if they were found to have an increased risk of preterm labor as judged by a Creasy risk score of ~ 10. 8 Women were excluded if they had significant psychiatric problems precluding compliance with the study protocol, did not speak English, or were >32 weeks' gestation. Eligible patients were approached for informed consent and participation in the
Home uterine activity monitoring to detect preterm labor 859
Volume 165 "'umber 4, Part I
2316 Screened
I 509 Eligible on t-initial screening I 377 Enrolled & Randomized t-198M 179NM
132 Eligible never enrolled
~
28 PTL not possible 16M 12NM
~
I
349 PTL possible 182M 167NM
~
15 PTL status unknown 8M 7NM
I--
252 No PTL after entry 131M 121NM
37 40 52 3
no telephone consent not requested refused unknown
13 PTL prior to entry (7M, 6NM) 7 medical hospitalization (5M, 2NM) 8 early fetal demise (4M, 4NM)
I 334 PTL status known 174M 160NM
I 82 PTL atter entry 43M 39NM Fig. 1. Study population flow chart. M, Monitored: NM, not monitored: PTL, pretenn labor.
study according to a protocol approved by institutional review boards at each study site. The 52 eligible women who refused randomization had Creasy risk scores similar to those of study participants. No other data are available on these women. The How of patients from 2316 initially screened for eligibility through the 377 who were enrolled and randomized is shown in Fig. 1. Randomization. All randomization and group assignment was performed by study personnel without direct patient care responsibilities. After enrollment, the group assignments were made by opening consecutively numbered envelopes that randomized patients by a table of random numbers. A different random number sequence was used for each study site, and each site had separate randomization sets for singletons and multiple gestations. No local study personnel had access to the group assignment. On the basis of these procedures, the 377 patients were assigned to either the monitored group (n = 198) or the not monitored group (n = 179) (Fig. 1). The distribution of monitored and not monitored patients by site was as followed: Truman Medical Center had 82 monitored and 82 not monitored patients; State University of New York had 74 monitored and68 not monitored patients; and University of Illinois had 42 monitored and 29 not monitored patients. The not monitored group received standard high-risk obstetric care, whereas the monitored group received the same standard high-risk obstetric care plus twice daily home uterine activity monitoring.
Obstetric care Education. All patients, regardless of group assignment, received education regarding why they were at risk for preterm labor and information about preterm labor precautions, including these signs and symptoms of preterm labor: (1) uterine contractions, (2) dull, low backache, (3) menstrual-like cramps, (4) diarrhea or indigestion, (5) pressure or heaviness in the lower abdomen, back, pelvis, or thighs, (6) vaginal discharge, and (7) vaginal drainage or leaking of amniotic Huid. They were also instructed to notify their physician or clinic if they suspected preterm labor. All patients received education about the monitoring protocol (see below) and uterine self-palpation. Monitoring protocol. The home uterine activity monitoring device used for this study was the Genesis System (Physiologic Diagnostic Service, Atlanta). The system consists of a guard ring type tocodynamometer connected to a portable battery-operated data recorder, a communication module allowing transmission of data over standard telephone lines, and a receiving computer with strip chart. Patients in the monitored group received instruction on use of the Genesis System at 24 to 26 weeks or at time of entry into the study if they were enrolled at 26 to 32 weeks. Patients were to monitor for 1 hour twice daily, 7 days per weeks. Monitoring began at 24 weeks and continued to 37 weeks or delivery, if delivery occurred at <37 weeks. If patients entered the study after 24 weeks' gestation, they began monitoring on enroll-
860
Mou et al.
ment. The patients were scheduled for twice-daily (in the morning and evening) uterine activity data transmission. No advice or instructions were given over the telephone. In addition, patients were instructed to use the system for 1 hour if they experienced any signs or symptoms of preterm labor. These unscheduled sessions were transmitted immediately. All monitoring was performed in the patient's home, and the uterine activity data were transmitted by telephone to a remote site. The receiver personnel served to facilitate the transmission of the uterine contraction data and report the number of contractions to the patient. Patients were previously informed that the personnel would not be able to answer questions, and the personnel were instructed to neither ask medical questions nor give medical advice. The investigators verified on an ongoing basis that the personnel provided no medical information to the monitored patients. Procedures for routine and nonroutine visits. For both routine and nonroutine care, women in both groups were seen by obstetric caretakers as part of the usual high-risk obstetric care provided at their respective perinatal sites. The minimum care scheduled was a visit every 4 weeks until 30 weeks' gestation, at least every 2 weeks between 30 and 36 weeks, and at least weekly thereafter. The actual schedule was based on each patient's medical problems, and, in general, subjects had more scheduled routine visits than the minimum required. Nonroutine visits were indicated if the patient had signs or symptoms of preterm labor as described above. Nonroutine visits were obtained at the obstetric care provider'S office or clinic during the day, and in hospital labor and delivery units, at night. The indications and procedures followed for nonroutine visits were the same for the two groups. In the case of suspected uterine contractions, the action to be taken by the patient was the same if the monitoring data showed increased contractions or if the patient had the sensation of increased contractions. It consisted of the following: (1) If, on the basis of patient sensation or monitoring, there were four or more contractions per hour lasting >40 seconds, the patient was to assume a left recumbent position, take oral hydration, empty the bladder, and then reassess the number of contractions for 1 hour. (2) If on reassessment there were four or more contractions per hour, then a nonroutine medical visit was indicated. (3) For twin gestations, six contractions per hour rather than four were used in this protocol. This was based on data showing higher baseline contraction frequencies at each week of gestation from 23 to 36 weeks in twin gestations as compared with singleton gestations. 5 • 9 All patients were allowed to arrange a
October 1991 Am J Obstet Gynecol
nonroutine visit for prompt evaluation if they felt it was necessary. When patients were seen with possible pre term labor, fetal monitoring and cervical examinations were performed by experienced examiners (senior residents or attending physicians) who were not aware of the subject's group assignment. Preterm labor was defined, before the beginning of the study, as four or more contractions per hour plus a change in cervical examination, or four or more contractions per hour and a cervical examination showing >2 cm dilatation and 75% effacement and/or ruptured membranes. In actuality, patients who were diagnosed with preterm labor all had a cervical change of effacement or dilatation as compared with their most recent cervical examination. Tocolytic therapy was prescribed at each of the study sites according to previously existing protocols. In general, treatment at all sites for established preterm labor included bed rest, hydration, evaluation for acute medical problems, and, if no contraindications for drug treatment existed, infusion of one of the common intravenous tocolytics (ritodrine hydrochloride, magnesium sulfate, or terbutaline). If labor stopped, this intravenous therapy was followed by oral tocolytic therapy. Blinding. All patients were instructed not to inform caretakers of group assignment. Caretakers were informed that they were seeing a study patient but not told the group assignment. If uterine contractions were suspected, caretakers were not informed if the contractions were detected by the monitor or the patient. If a potential examiner inadvertently learned a patient's group assignment, then the study cervical examination was performed by a different experienced caretaker who was not aware of the group assignment. Statistics Sample size. As part of planning this study, the sample size calculation was performed on the basis of cervical dilatation of <2 cm being the definition of early detection of preterm labor and assuming that 60% of monitored patients would have early detection compared with only 30% of not monitored patients. Alpha was 0.05 and f3 was 0.20 (power = 80%). Thus 48 patients with preterm labor per group would be required. It was further assumed that 30% of the enrolled highrisk patients would have preterm labor so that a total enrollment of 320 would be required. Statistical analysis. Two group comparisons were made with t tests or Wilcoxon rank sum tests, depending on frequency distributions for continuous variables. X2 tests or Fisher's exact probabilities (depending on number of categories and cell counts) were done for categorical variables. Three group comparisons were made
Home uterine activity monitoring to detect preterm labor
Volume 165 Number 4, Part 1
861
Table I. Characteristics of subjects at entry to the study All enrollees (n = 377) Monitored
(n = 82)
Not monitored
Not monitored (n 0: 39)
(n = 179)
Monitored (n = 43)
27.0 ± 5.1 53.5 73.9
26.3 ± 5.2 47.6 68.9
26.7 ± 4.9 45.2 73.8
26.7 ± 5.6 44.7 69.2
5.1 5.6 0.5 21.2
7.3 8.9
4.7 7.0 2.3 25.6
7.7 5.1 2.6 33.3
2.6 ± 1.5 1.7 ± 1.2 0.5 ± 0.8 0.3 ± 0.6 26.6 ± 2.8 53.5 20.9 11.6 11.6 15 10-37
3.0 ± 1.9 2.0 ± 1.3 0.6 ± 0.9 0.2 ± 0.4 25.8 ± 2.7 71.8 20.5 10.3 10.3 17 10-34
(n = 198)
Demographic Maternal age (yr, mean ± SD) White (%) High school graduate (%) Substance use Alcohol (%) Cocaine (%) Heroin (%) Cigarette use (%) Obstetric Prior pregnancies (No.) Live births (No.) Spontaneous abortions (No.) Elective abortions (No.) Weeks' gestation at entry Previous preterm birth (%) Twin gestation (%) 2nd Trimester abortion (%) Threatened labor (%) Creasy score, median Creasy score, extremes
Subgroup with preterm labor
2.6 ± 1.7 1.7 ± 1.2 0.5 ± 0.9 0.3 ± 0.7 26.6 ± 2.8 45.0 12.1 Il.l
16.6 15 10-41
with analysis of variance or X2 depending on the type of data. Two-tailed probability or :s0.05 was considered significant. No adjustments were made for multiple comparisons. Results
Comparison of groups at entry. The demographic and medical characteristics of the monitored and not monitored groups are summarized in Table I. At entry to the study no significant differences were observed between the two groups for any of these characteristics. This was true for both the 377 patients originally randomized and the subgroup of 82 who went on to have preterm labor. In addition, since randomization was stratified by site and twins, it was verified that within each randomization no significant differences were observed between the monitored and not monitored groups in any of their medical and demographic characteristics. Comparison of groups after entry. Several analyses were performed to verify that the two groups received similar care after entry to the study. There were no significant differences between the two groups. The number of routine visits per patient among the monitored and the not monitored patients was 7.1 and 6.7, respectively. For nonroutine visits the monitored patients averaged 2.0 visits and not monitored patients averaged 1.8 visits. Fig. 2 illustrates the number of routine and nonroutine visits for the two groups during
I
l.l
27.9 2.8 ± 1.8 1.8 ± 1.3 0.6 ± 1.0 0.3 ± 0.6 26.4 ± 2.9 54.8 7.8 12.8 11.1
15 10-32
I
each week of gestation that the patients participated in the study. The incidence and timing of pre term labor was also similar for the two groups. Pre term labor was detected in 43 of 174 (24.7%) monitored patients and in 39 of 160 (24.4%) not monitored patients (Fig. 1). The average gestational age at diagnosis of preterm labor was 32.9 weeks for both groups (Table II). Cervical dilatation at diagnosis of preterm labor: The primary end point for the study was the timing of detection of pre term labor as measured by the cervical dilatation at the time of diagnosis. Cervical dilatation data were available for 77 of the 82 subjects who had preterm labor. Dilatation data were not recorded at the time of diagnosis of preterm labor for two monitored and two not monitored subjects. Dilatation data for one not monitored subject were not used because a cerclage was in place at the time of the diagnosis of preterm labor. Women in the monitored group had significantly less cervical dilatation at the time of diagnosis of preterm labor (Table II). The mean dilatation for the monitored group was 1.4 cm compared with 2.5 cm in the not monitored patients (p = 0.0006). Some patients had cervical dilatation before the onset of preterm labor; therefore the change in dilatation from the cervical examination immediately prior to the onset of preterm labor also was studied. Women in the monitored group had a mean change in cervical dilatation of 1.1 cm
862
Mou et al.
October 1991 Am J Obstet Gynecol
1.0 0.8
E
••••• Routine Not Mon. (n= 179) 0--0 Routine Mon. (n=198) ...... NRV Not Mon. (n=179) 0--0 NRV Mon. (n=198)
0.6
.~
ca
a. ........
...•....•.
11 0.4
'00
:>
..........
~,
.'
.......
....
...........•... .
0.2
24 25 26
27
28
29
30
31
32 33
34 35
36
Gestational Age (wks) Fig. 2. Number of routine and nonroutine visits per patient per week of gestation for entire study population. Routine, Scheduled prenatal visits; NRV, nonroutine prenatal visits; not mon., not monitored patients; Mon., monitored patients.
100
A
100
80
60 p=.016
40 CD
E
40
20
20
37
0
0
0
0
100
t!-
80
80
60
60
40
40
~
20 0
p=.OOOO9
80
60 C
8
<1.0
C
100
p=.OO5
<2.0
0
20 86 <3.0
0
Cervical Dilation (cm)
o Monitored
(n=41)
~
p=.01
98 <4.0
Not Monitored (n=36)
Fig. 3. Percent of women with early detection of preterm labor with < 1.0 em (A), <2.0 em (8), <3.0 em (C), and <4.0 cm (D) used as cutoffs for early detection of preterm labor.
compared with 2.0 cm in not monitored women (P = 0.004). In addition to evaluating the overall distribution of cervical dilatation, the likelihood of early detection of preterm labor was also determined. The initial sample size calculation for the end point of cervical dilatation was based on the definition of early detection of preterm labor being a cervical dilatation of <2 cm, and it was assumed that 60% of monitored patients would have early detection compared with only 30% of not monitored patients. Detection of preterm labor before
2 cm cervical dilatation was actually accomplished in 73.1 % of monitored subjects compared with only 27.8% of not monitored subjects (p = 0.00009) (Fig. 3, B). Since clinicians may vary with regard to their definition of early detection of preterm labor, Fig. 3 shows the group comparisons with four different cutoff values (1, 2, 3, and 4 cm) for early detection. The monitored group had significantly earlier detection than the not monitored group for all four cutoff values. Since patients with pre term labor represent a subset of the original randomized groups, multiple regression
Home uterine activity monitoring to detect preterm labor
Volume 165 I\:umber 4, Part I
863
Table II. Obstetric outcomes for women with pre term labor
Gestational age at diagnosis of preterm labor (wk) Mean ± SD Cervical dilatation at diagnosis of pre term labor (cm)* Mean ± SD Median (extremes) Change in cervical diagnosis from prior examination (cm)* Mean ± SD Median (extremes) Duration of gestation after diagnosis of preterm labor (wk)t Mean ± SD Gestational age at delivery (wk)t Mean ± SD
Monitored (n = 43)
Not monitored (n = 39)
32,9 ± 3,2
32,9 ± 2,9
1.4 ± L3 LO (0-6.5)
2.5 ± L5 2.0 (0-6.0)
0.0006 0.0002
1.1 ± L2 LO (0-6.5)
2.0 ± 1.4 2.0 (0-5.0)
0.004 0.001
3.7 ± 3.8
2.0 ± 2.9
0.02
36.6 ± 2.4
34.9 ± 3.2
0.009
p Value
On the rows where data are summarized with mean, data analysis was with Student t tests. On the rows where data are summarized with median, data analysis was with Wilcoxon rank sum test. *Di!atation data were not available for two monitored and three not monitored subjects. tGestational age at delivery was not available for two monitored and one not monitored subjects.
Table III. Neonatal outcomes for singleton gestations with preterm labor*
Birth weight (gm)t Mean ± SD <2500 gm (%) <2000 gm (%) Neonatal intensive care No. of infants Total days Oxygen therapy No. of infants Total days Mechanical ventilation No. of infants Total days
Monitored (n = 33)
Not monitored (n = 30)
P Value
2934 ± 708 19 10
2329 ± 733 63 37
0.002 0.0007 0.02
5 50
13 324
0 0
7 68
0.004
0 0
5 54
0.02
0.02
*Medicaliy indicated pre term deliveries (one in each group) were excluded from these analyses. tBirth weight data were not available for two monitored infants who were born at term.
analysis was performed with dilatation at diagnosis of preterm labor as the dependent variable and numerous independent variables, such as obstetric history variables including substance use, study site, and monitoring versus not monitoring. The only variable that had a significant independent effect on dilatation at diagnosis was use of the monitor, which was associated with a 0.9 cm decrease in cervical dilatation (p = 0.01). Cohort analyses were done for women who entered the study at <25 weeks' gestation, at ?:27 weeks, and at ?:29 weeks' gestation. Mean cervical dilatation and cervical change at the diagnosis of preterm labor was less in the monitored group for all of these cohorts. Pregnancy and neonatal outcome. To evaluate the clinical importance of the early detection of preterm labor observed in the monitored group, duration of gestation after diagnosis of preterm labor (Table II), infant birth weight, and measures of neonatal morbid-
ity (Table III) were compared between the two groups. The mean duration of gestation after diagnosis of preterm labor was 3.7 weeks in 41 monitored patients (one patient did not have gestational age at delivery recorded and one was excluded because of medically indicated preterm delivery) compared with only 2.0 weeks in 38 not monitored patients (one excluded because of medically indicated preterm delivery [p = 0.02]) (Table III). Therefore the mean gestational age at delivery was 36.6 weeks in the monitored group compared with only 34.9 in the not monitored group (p = 0.009) (Table II). These gestational duration data do not significantly change if the two medically indicated pre term deliveries are included in the analysis. Multiple regression analysis also was performed with gestational age at delivery as the dependent variable and numerous independent variables including age, race, Creasy score, substance use variables, maternal
864
Mou et al.
history variables, study site, monitoring versus not monitoring, and cervical dilatation at diagnosis of preterm labor. The only variable that had a significant independent effect on gestational age at delivery was cervical dilatation at diagnosis of preterm labor, which was associated with a 0.6-week decrease in gestational age at delivery for each centimeter increase in dilatation (P = 0.03). Birth weight and neonatal morbidity variables were evaluated after exclusion of 17 (nine monitored and eight not monitored) twin gestations and two (one in each group) medically indicated preterm deliveries. In addition, there were two patients for whom birth weight data were not available. Twin gestations were excluded from these analyses because of the potential effects of twinning on birth weight and neonatal morbidity. Birth weight of singletons differed significantly between the monitored and not monitored groups. The mean birth weight of infants from the monitored grou p (n = 31) was 2934 gm compared with only 2329 gm for the not monitored group (n = 30) (P = 0.002). Only 19% of infants from the monitored group weighed <2500 gm versus 63% of the not monitored babies (p = 0.0007). The incidence of infants weighing <2000 gm was only 10% for the monitored group compared with 37% for the not monitored group (p = 0.02) (Table III). Singleton infants born to monitored women also suffered significantly less neonatal morbidity as measured by need for neonatal intensive care, oxygen therapy, and mechanical ventilation (Table III). Only five infants (50 total neonatal intensive care unit days) of the 33 monitored singleton gestations required admission to the neonatal intensive care unit (p = 0.02). None of the 33 infants who were products of monitored singleton gestations required oxygen therapy or mechanical ventilation compared with seven infants who required oxygen therapy, for a total of 68 days (p = 0.004), and five infants who required mechanical ventilation for a total of 54 days (p = 0.02) among the 30 products of not monitored singleton gestations (Table III). There were 252 patients who were randomized into monitored and not monitored groups who did not experience preteI'm labor outcome. Data were available on 250 patients. The mean gestational age at delivery of the controls without preteI'm labor was 39.1 weeks (± 2.3 weeks SD). The mean gestational age at delivery of the monitored group without preteI'm labor was 38.8 weeks (±2.1 weeks SD), p = 0.28. Likewise, the birth weights of the singletons in the control group without preteI'm labor was 3154 gm (± 654 gm SD) and the birth weights of singletons in the monitored group was 3269 gm (±604 gm SD), p = 0.18. Comment
This study demonstrates that objective assessment of uterine activity, by home uterine activity monitoring- in
October 1991 Am J Obstet Gynecol
women at high risk for preterm labor, allows detection of preteI'm labor at less advanced cervical dilatation. The clinical importance of this earlier detection of preterm labor is supported by the increased gestational duration after diagnosis of preterm labor and the resultant improved gestational age at delivery, increased birth weight, and decreased neonatal morbidity observed in the monitored group. This report represents the largest published, prospective, multicenter, randomized, and blinded evaluation of the efficacy of home uterine activity monitoring for earlier detection of preterm labor in high-risk pregnancies. Previously published studies have suggested beneficial effects of home uterine activity monitoring" 5, 6; however, the study designs used did not allow for the separation of the impact of the use of the monitor from the impact of the additional perinatal nursing support. 7,10 Some studies 3 ,6 did not provide comparison groups receiving the same perinatal center obstetric care. The studies that failed to show beneficial effects of home uterine activity monitoring have not focused on the value of the device when it is used without additional nursing care. 11,12 Also, the present study differs from others in using twice-daily monitoring and transmission of data, rather than once-daily transmission of monitor strips. When one is evaluating a new device, it is essential to separate clearly those effects that are a result of the information derived from the use of the device from those effects of other nonstandard interventions. Previous studies have evaluated daily perinatal nursing support used along with home uterine activity monitoring. The current report evaluates the impact of the addition of objective assessment of uterine activity to high-risk obstetric care on pregnancy outcome. In this study design, (1) no routine daily perinatal nursing support was provided to either group, (2) all study patients were provided with identical high-risk obstetric care and education in self-palpation of uterine contractions (as described by Herron et al. 13), (3) caretakers were blinded to subject's group assignment, (4) preteI'm labor was strictly defined, (5) all subjects were included in the data analysis on the basis of their original group assignment regardless of compliance, and (6) the primary end point was cervical dilatation at diagnosis of preteI'm labor, since other variables (e.g., incidence of preterm birth) are strongly related to the incidence of preteI'm labor in the study population and to the effectiveness of tocolytic therapy. Although caretakers were blinded to the subjects' group assignments, it is not possible to be absolutely certain that all examiners were indeed unaware of a patient'S group assignment. It is estimated that not more than six women were examined by caretakers who were aware of the patient's group assignment. The results of the study did not change when these patients were excluded from the data an
Volume 165 Number 4, Part 1
Home uterine activity monitoring to detect preterm labor 865
The study did not include a sham monitoring group, and therefore it is not possible to determine the independent effects of the four procedures that are an integral part of the monitoring protocol. These procedures, which represent the only known differences between the two groups, include (1) wearing a tocodynamometry belt during each recording session, (2) relative inactivity during each recording session, (3) data transmission after each recording session , and (4) patient notification if four or more contractions were observed. Although it is possible that any of these four procedures may have contributed to the results, detection and notification of increased contractions is the most reasonable explanation for the early detection of preterm labor observed in the monitored group. There were similar outcomes in the control and monitored group among patients who did not go into preterm labor, which also makes the twice daily monitoring unlikely to affect the course of the pregnancies except for diagnosing preterm labor at an earlier cervical dilatation. The benefit of monitoring appears to be in the earlier detection of preterm labor, before perception by the patient and before significant cervical dilatation has occurred. Self-palpation of uterine contractions had been shown to be less sensitive than tocodynamometry in detecting uterine activity.2. 14 Objective quantitation of contractions is a useful method of surveillance for preterm labor and detects contractions before women perceive them. 2 Since progression of labor occurs over hours rather than days or weeks one would not expect the gestational age at diagnosis of pre term labor to differ significantly between the groups as shown in this study, with mean gestational age at diagnosis of preterm labor at 32.9 weeks in the monitored and not monitored groups. The observation that there were no differences between the groups for the gestational age at the time of diagnosis of preterm labor and the incidence of preterm labor makes it unlikely that overdiagnosis of preterm labor or lead time bias could account for the results of the study. This protocol did not address specific tocolytic regimens. Once preterm labor was diagnosed, each center continued its usual protocols. Therefore it is not possible to assess the relative effects of various medical interventions on the outcomes. The fact that the vast majority of caretakers were unaware of the patient'S group assignment makes it unlikely that there were significant differences between the medical interventions received by the two groups. In addition, the two groups were medically and demographically similar at entry into the study and at the time of diagnosis of preterm labor. They had similar numbers of routine and non routine obstetric visits, incidence of preterm labor, similar gestational age at diagnosis of preterm labor, and were similar in the relationship between dilatation at diagnosis of preterm labor and subsequent duration of I!estation.
Zlatnik l5 and King et al. 16 have shown that long-term tocolysis with significant prolongation of gestation is possible if preterm labor is diagnosed before advanced cervical dilatation. Multiple regression analysis confirmed that earlier detection of preterm labor in the monitored group led to significant prolongation of gestation. Although there is disagreement as to the specific definition of advanced cervical dilatation, a significantly higher percentage of the monitored patients in this study had early detection of preterm labor, as measured by cervical dilatation, whether <1, 2, 3, or 4 cm dilatation was used as the cutoff for early detection. The result was improved gestational age at delivery and increased birth weight of infants in the monitored group. A medically and sociodemographically diverse population from three perinatal sites participated in this study. Subgroup analyses were performed with specific medical risk factors, gestational ages at entry, gestational ages at diagnosis of preterm labor, and compliance with study protocol. The results of these analyses showed that the information derived from home uterine activity monitoring appeared to be similarly beneficial for all patients who used the monitor. The subset of the monitored group (n = 12) who did not comply with the protocol had results that were similar to those of the not monitored grou p. Prior preterm delivery was the only medical risk factor for preterm labor for which there were a sufficient number of patients for separate analysis. These data were consistent with the data for the entire study population showing beneficial effects for all outcome variables. On the basis of these study data, the United States Food and Drug Administration has given approval for use of this device in women with a history of preterm birth. l7 Although none of the other individual medical risk factors had a sufficient sample size for separate analyses, when considered as one group (containing the 31 patients without a history of preterm birth), the monitored group showed significant beneficial effects similar in magnitude to the subset of women with a history of preterm birth. Women with other medical risk factors for preterm labor may therefore also benefit from home uterine activity monitoring. Future studies may further address the benefit in women with other specific maternal risk factors. Our study demonstrates that home uterine activity monitoring, when used in selected high-risk women as the sole addition to standard high-risk care, results in earlier detection of preterm labor. This led to an impressive mean 1.7 -week improvement in gestation compared with not monitored women, as well as a mean 605 gm greater birth weight. In summary, this study demonstrated that use of home uterine activity monitoring, in selected high-risk patients, significantly increases the likelihood of diagnosis of preterm labor before advanced cervical dilatation, thus providing the clinician with a better op-
866
Mou et al.
portunity to initiate tocolytic therapy directed at improving pregnancy outcome. We are indebted to David Simic, Anita Quinlan, Christine Falchi, and Laurie Dressman for their role in coordinating the data collection and to Dr. Howard Golub for his help in the statistical analyses. REFERENCES I. McCormick Me. The contribution of low birthweight to infant mortality and childhood morbidity. N Engl J Med 1985;312:82-90. 2. Newman RB, Campbell BA, Stramm S1. Objective tocodynamometry identifies labor onset earlier than subjective maternal perception. Obstet Gynecol 1990;76: 1089-92. 3. Katz M, Gill P, Newman RB. Detection of preterm labor by ambulatory monitoring of uterine activity: a preliminary report. Obstet Gynecol 1986;68:773-8. 4. Katz M, Newman RB, Gill PJ. Assessment of uterine activity in ambulatory patients at high risk of preterm labor and delivery. AMJ OBSTET GYNECOL 1986;154:44-7. 5. Morrison JC, Martin IN, Martin RW, Gookin RS, Wiser W1. Prevention of pre term birth by ambulatory assessment of uterine activity. A randomized study. AM J OBSTET GYNECOL 1987;156:536-43. 6. Hill WC, Fleming AD, Martin RW, et al. Home uterine activity monitoring is associated with a reduction in preterm birth. Obstet Gynecol 1990;76: 135-75. 7. Diagnostic and therapeutic technology assessment (DATTA) .. Home monitoring of uterine activity. JAMA 1989;261 :3027-9.
October 1991 Am.J Obstet Gynecol
8. Creasy RK, Gummer BA, Liggins GC. System for predicting spontaneous preterm birth. Obstet Gynecol 1980;55:692-5. 9. Newman RB, Gill PJ, Campion S, Katz M. The influence of fetal number on antepartum uterine activity. Obstet Gynecol 1989;73:695-9. 10. Martin RW, Gookin KS, Hill CW, et al. Uterine activity compared with symptomatology in the detection of preterm labor. Obstet Gynecol 1990;76:195S. 11. lams JD, Johnson FF, O'Shaughnessy RW, West Le. A prospective random trial of home uterine activity monitoring in pregnancies at increased risk of preterm labor. AMJ OBSTET GYNECOL 1987;157:638-43. 12. lams JD, Johnson FF, O'Shaughnessy RW. A prospective random trial of home uterine monitoring in pregnancies at increased risk of preterm labor. Part II. AM J OBSTET GYNECOL 1988; 159:595-603. 13. Herron MA, Katz M, Creasy RK. Evaluation of a pre term birth prevention program: preliminary report. Obstet Gynecol 1982;59:452. 14. Brutsman LE, Langer 0, Anyaegbunam A, Belle C, Merkatz IR. Education does not improve patient perception of preterm uterine contractility. Obstet Gynecol 1990;76:97S. 15. Ziatnik F. The applicability oflabar inhibition to the problem of prematurity. AM J OBSTET GYNECOL 1972;103: 704-6. 16. King JF, Grant A, Marc IN, Keirse C, Chalmers I. Betamime tics in preterm labour: an overview of the randomized controlled trials. Br J Obstet GynaecoI1988;95: 1988. 17. Fed Reg 1990 Oct 23;55(205):427-79.
Key words: Home uterine activity monitoring, preterm labor Preterm delivery is the most significant contributor to perinatal morbidity and mortality in the United States.' Unfortunately, many episodes of preterm labor are not detected until it is too late to provide effective tocolytic therapy. Recent studies have shown that an increase in the number of uterine contractions precedes the onset of preterm labor and is present in early labor; however, women often are not aware of the increased number of contractions until labor is advanced."' 3 Studies showing that devices designed to monitor uterine activity offer a more reliable method of quantitating contractions suggest that home uterine activity monitoring may allow earlier detection of the onset of preterm labor!' 5 Although many studies examining its efficacy have suggested beneficial effects, most of these studies have included intensive perinatal nursing support in combination with the objective data obtained
From the Department of Obstetrics and Gynecology, Truman Medical Center, University of Missouri-Kansas City: the Perinatal Center, Health Science Center at Syracuse, State University of New York,' the University of Illinois Hospital,' the Boston University School of Public Health! and the Department of Pediatrics, Boston City Hospital and Boston University .' Presented at the Eleventh Annual Meeting of the Society of Perinatal Obstetricians, San Francisco, California, January 28-February 2, 1991. Reprint requests: Dr. Susan M. Mou, Department of Obstetrics and Gynecology, Truman Medical Center, Kansas City, MO 64108, 6/6/31430
858
with a device. 5- 7 Therefore it has not been possible to separate the effects of the detection of increased uterine activity. A prospective, randomized, blinded, multicenter trial of a home uterine activity monitoring device as the sole addition to current standard high-risk obstetric care was conducted to evaluate the effect of objective measurement of uterine activity in the diagnosis of preterm labor.
Methods Recruitment and eligibility. Prenatal charts at three study sites, Truman Medical Center, Kansas City, Mo., State University of New York Health Science Center, Syracuse, N.Y., and the University of Illinois Hospital, Chicago, were reviewed at the time of the initial prenatal visit, at 22 to 26 weeks' gestation, and at 32 weeks' gestation. Recruitment was conducted from September 1988 to August 1989 at Truman Medical Center and the State University of New York, Syracuse, but was terminated in June 1989 at the University of Illinois because of intended closure of the hospital. Subjects were eligible for recruitment if they were found to have an increased risk of preterm labor as judged by a Creasy risk score of ~ 10. 8 Women were excluded if they had significant psychiatric problems precluding compliance with the study protocol, did not speak English, or were >32 weeks' gestation. Eligible patients were approached for informed consent and participation in the
Home uterine activity monitoring to detect preterm labor 859
Volume 165 "'umber 4, Part I
2316 Screened
I 509 Eligible on t-initial screening I 377 Enrolled & Randomized t-198M 179NM
132 Eligible never enrolled
~
28 PTL not possible 16M 12NM
~
I
349 PTL possible 182M 167NM
~
15 PTL status unknown 8M 7NM
I--
252 No PTL after entry 131M 121NM
37 40 52 3
no telephone consent not requested refused unknown
13 PTL prior to entry (7M, 6NM) 7 medical hospitalization (5M, 2NM) 8 early fetal demise (4M, 4NM)
I 334 PTL status known 174M 160NM
I 82 PTL atter entry 43M 39NM Fig. 1. Study population flow chart. M, Monitored: NM, not monitored: PTL, pretenn labor.
study according to a protocol approved by institutional review boards at each study site. The 52 eligible women who refused randomization had Creasy risk scores similar to those of study participants. No other data are available on these women. The How of patients from 2316 initially screened for eligibility through the 377 who were enrolled and randomized is shown in Fig. 1. Randomization. All randomization and group assignment was performed by study personnel without direct patient care responsibilities. After enrollment, the group assignments were made by opening consecutively numbered envelopes that randomized patients by a table of random numbers. A different random number sequence was used for each study site, and each site had separate randomization sets for singletons and multiple gestations. No local study personnel had access to the group assignment. On the basis of these procedures, the 377 patients were assigned to either the monitored group (n = 198) or the not monitored group (n = 179) (Fig. 1). The distribution of monitored and not monitored patients by site was as followed: Truman Medical Center had 82 monitored and 82 not monitored patients; State University of New York had 74 monitored and68 not monitored patients; and University of Illinois had 42 monitored and 29 not monitored patients. The not monitored group received standard high-risk obstetric care, whereas the monitored group received the same standard high-risk obstetric care plus twice daily home uterine activity monitoring.
Obstetric care Education. All patients, regardless of group assignment, received education regarding why they were at risk for preterm labor and information about preterm labor precautions, including these signs and symptoms of preterm labor: (1) uterine contractions, (2) dull, low backache, (3) menstrual-like cramps, (4) diarrhea or indigestion, (5) pressure or heaviness in the lower abdomen, back, pelvis, or thighs, (6) vaginal discharge, and (7) vaginal drainage or leaking of amniotic Huid. They were also instructed to notify their physician or clinic if they suspected preterm labor. All patients received education about the monitoring protocol (see below) and uterine self-palpation. Monitoring protocol. The home uterine activity monitoring device used for this study was the Genesis System (Physiologic Diagnostic Service, Atlanta). The system consists of a guard ring type tocodynamometer connected to a portable battery-operated data recorder, a communication module allowing transmission of data over standard telephone lines, and a receiving computer with strip chart. Patients in the monitored group received instruction on use of the Genesis System at 24 to 26 weeks or at time of entry into the study if they were enrolled at 26 to 32 weeks. Patients were to monitor for 1 hour twice daily, 7 days per weeks. Monitoring began at 24 weeks and continued to 37 weeks or delivery, if delivery occurred at <37 weeks. If patients entered the study after 24 weeks' gestation, they began monitoring on enroll-
860
Mou et al.
ment. The patients were scheduled for twice-daily (in the morning and evening) uterine activity data transmission. No advice or instructions were given over the telephone. In addition, patients were instructed to use the system for 1 hour if they experienced any signs or symptoms of preterm labor. These unscheduled sessions were transmitted immediately. All monitoring was performed in the patient's home, and the uterine activity data were transmitted by telephone to a remote site. The receiver personnel served to facilitate the transmission of the uterine contraction data and report the number of contractions to the patient. Patients were previously informed that the personnel would not be able to answer questions, and the personnel were instructed to neither ask medical questions nor give medical advice. The investigators verified on an ongoing basis that the personnel provided no medical information to the monitored patients. Procedures for routine and nonroutine visits. For both routine and nonroutine care, women in both groups were seen by obstetric caretakers as part of the usual high-risk obstetric care provided at their respective perinatal sites. The minimum care scheduled was a visit every 4 weeks until 30 weeks' gestation, at least every 2 weeks between 30 and 36 weeks, and at least weekly thereafter. The actual schedule was based on each patient's medical problems, and, in general, subjects had more scheduled routine visits than the minimum required. Nonroutine visits were indicated if the patient had signs or symptoms of preterm labor as described above. Nonroutine visits were obtained at the obstetric care provider'S office or clinic during the day, and in hospital labor and delivery units, at night. The indications and procedures followed for nonroutine visits were the same for the two groups. In the case of suspected uterine contractions, the action to be taken by the patient was the same if the monitoring data showed increased contractions or if the patient had the sensation of increased contractions. It consisted of the following: (1) If, on the basis of patient sensation or monitoring, there were four or more contractions per hour lasting >40 seconds, the patient was to assume a left recumbent position, take oral hydration, empty the bladder, and then reassess the number of contractions for 1 hour. (2) If on reassessment there were four or more contractions per hour, then a nonroutine medical visit was indicated. (3) For twin gestations, six contractions per hour rather than four were used in this protocol. This was based on data showing higher baseline contraction frequencies at each week of gestation from 23 to 36 weeks in twin gestations as compared with singleton gestations. 5 • 9 All patients were allowed to arrange a
October 1991 Am J Obstet Gynecol
nonroutine visit for prompt evaluation if they felt it was necessary. When patients were seen with possible pre term labor, fetal monitoring and cervical examinations were performed by experienced examiners (senior residents or attending physicians) who were not aware of the subject's group assignment. Preterm labor was defined, before the beginning of the study, as four or more contractions per hour plus a change in cervical examination, or four or more contractions per hour and a cervical examination showing >2 cm dilatation and 75% effacement and/or ruptured membranes. In actuality, patients who were diagnosed with preterm labor all had a cervical change of effacement or dilatation as compared with their most recent cervical examination. Tocolytic therapy was prescribed at each of the study sites according to previously existing protocols. In general, treatment at all sites for established preterm labor included bed rest, hydration, evaluation for acute medical problems, and, if no contraindications for drug treatment existed, infusion of one of the common intravenous tocolytics (ritodrine hydrochloride, magnesium sulfate, or terbutaline). If labor stopped, this intravenous therapy was followed by oral tocolytic therapy. Blinding. All patients were instructed not to inform caretakers of group assignment. Caretakers were informed that they were seeing a study patient but not told the group assignment. If uterine contractions were suspected, caretakers were not informed if the contractions were detected by the monitor or the patient. If a potential examiner inadvertently learned a patient's group assignment, then the study cervical examination was performed by a different experienced caretaker who was not aware of the group assignment. Statistics Sample size. As part of planning this study, the sample size calculation was performed on the basis of cervical dilatation of <2 cm being the definition of early detection of preterm labor and assuming that 60% of monitored patients would have early detection compared with only 30% of not monitored patients. Alpha was 0.05 and f3 was 0.20 (power = 80%). Thus 48 patients with preterm labor per group would be required. It was further assumed that 30% of the enrolled highrisk patients would have preterm labor so that a total enrollment of 320 would be required. Statistical analysis. Two group comparisons were made with t tests or Wilcoxon rank sum tests, depending on frequency distributions for continuous variables. X2 tests or Fisher's exact probabilities (depending on number of categories and cell counts) were done for categorical variables. Three group comparisons were made
Home uterine activity monitoring to detect preterm labor
Volume 165 Number 4, Part 1
861
Table I. Characteristics of subjects at entry to the study All enrollees (n = 377) Monitored
(n = 82)
Not monitored
Not monitored (n 0: 39)
(n = 179)
Monitored (n = 43)
27.0 ± 5.1 53.5 73.9
26.3 ± 5.2 47.6 68.9
26.7 ± 4.9 45.2 73.8
26.7 ± 5.6 44.7 69.2
5.1 5.6 0.5 21.2
7.3 8.9
4.7 7.0 2.3 25.6
7.7 5.1 2.6 33.3
2.6 ± 1.5 1.7 ± 1.2 0.5 ± 0.8 0.3 ± 0.6 26.6 ± 2.8 53.5 20.9 11.6 11.6 15 10-37
3.0 ± 1.9 2.0 ± 1.3 0.6 ± 0.9 0.2 ± 0.4 25.8 ± 2.7 71.8 20.5 10.3 10.3 17 10-34
(n = 198)
Demographic Maternal age (yr, mean ± SD) White (%) High school graduate (%) Substance use Alcohol (%) Cocaine (%) Heroin (%) Cigarette use (%) Obstetric Prior pregnancies (No.) Live births (No.) Spontaneous abortions (No.) Elective abortions (No.) Weeks' gestation at entry Previous preterm birth (%) Twin gestation (%) 2nd Trimester abortion (%) Threatened labor (%) Creasy score, median Creasy score, extremes
Subgroup with preterm labor
2.6 ± 1.7 1.7 ± 1.2 0.5 ± 0.9 0.3 ± 0.7 26.6 ± 2.8 45.0 12.1 Il.l
16.6 15 10-41
with analysis of variance or X2 depending on the type of data. Two-tailed probability or :s0.05 was considered significant. No adjustments were made for multiple comparisons. Results
Comparison of groups at entry. The demographic and medical characteristics of the monitored and not monitored groups are summarized in Table I. At entry to the study no significant differences were observed between the two groups for any of these characteristics. This was true for both the 377 patients originally randomized and the subgroup of 82 who went on to have preterm labor. In addition, since randomization was stratified by site and twins, it was verified that within each randomization no significant differences were observed between the monitored and not monitored groups in any of their medical and demographic characteristics. Comparison of groups after entry. Several analyses were performed to verify that the two groups received similar care after entry to the study. There were no significant differences between the two groups. The number of routine visits per patient among the monitored and the not monitored patients was 7.1 and 6.7, respectively. For nonroutine visits the monitored patients averaged 2.0 visits and not monitored patients averaged 1.8 visits. Fig. 2 illustrates the number of routine and nonroutine visits for the two groups during
I
l.l
27.9 2.8 ± 1.8 1.8 ± 1.3 0.6 ± 1.0 0.3 ± 0.6 26.4 ± 2.9 54.8 7.8 12.8 11.1
15 10-32
I
each week of gestation that the patients participated in the study. The incidence and timing of pre term labor was also similar for the two groups. Pre term labor was detected in 43 of 174 (24.7%) monitored patients and in 39 of 160 (24.4%) not monitored patients (Fig. 1). The average gestational age at diagnosis of preterm labor was 32.9 weeks for both groups (Table II). Cervical dilatation at diagnosis of preterm labor: The primary end point for the study was the timing of detection of pre term labor as measured by the cervical dilatation at the time of diagnosis. Cervical dilatation data were available for 77 of the 82 subjects who had preterm labor. Dilatation data were not recorded at the time of diagnosis of preterm labor for two monitored and two not monitored subjects. Dilatation data for one not monitored subject were not used because a cerclage was in place at the time of the diagnosis of preterm labor. Women in the monitored group had significantly less cervical dilatation at the time of diagnosis of preterm labor (Table II). The mean dilatation for the monitored group was 1.4 cm compared with 2.5 cm in the not monitored patients (p = 0.0006). Some patients had cervical dilatation before the onset of preterm labor; therefore the change in dilatation from the cervical examination immediately prior to the onset of preterm labor also was studied. Women in the monitored group had a mean change in cervical dilatation of 1.1 cm
862
Mou et al.
October 1991 Am J Obstet Gynecol
1.0 0.8
E
••••• Routine Not Mon. (n= 179) 0--0 Routine Mon. (n=198) ...... NRV Not Mon. (n=179) 0--0 NRV Mon. (n=198)
0.6
.~
ca
a. ........
...•....•.
11 0.4
'00
:>
..........
~,
.'
.......
....
...........•... .
0.2
24 25 26
27
28
29
30
31
32 33
34 35
36
Gestational Age (wks) Fig. 2. Number of routine and nonroutine visits per patient per week of gestation for entire study population. Routine, Scheduled prenatal visits; NRV, nonroutine prenatal visits; not mon., not monitored patients; Mon., monitored patients.
100
A
100
80
60 p=.016
40 CD
E
40
20
20
37
0
0
0
0
100
t!-
80
80
60
60
40
40
~
20 0
p=.OOOO9
80
60 C
8
<1.0
C
100
p=.OO5
<2.0
0
20 86 <3.0
0
Cervical Dilation (cm)
o Monitored
(n=41)
~
p=.01
98 <4.0
Not Monitored (n=36)
Fig. 3. Percent of women with early detection of preterm labor with < 1.0 em (A), <2.0 em (8), <3.0 em (C), and <4.0 cm (D) used as cutoffs for early detection of preterm labor.
compared with 2.0 cm in not monitored women (P = 0.004). In addition to evaluating the overall distribution of cervical dilatation, the likelihood of early detection of preterm labor was also determined. The initial sample size calculation for the end point of cervical dilatation was based on the definition of early detection of preterm labor being a cervical dilatation of <2 cm, and it was assumed that 60% of monitored patients would have early detection compared with only 30% of not monitored patients. Detection of preterm labor before
2 cm cervical dilatation was actually accomplished in 73.1 % of monitored subjects compared with only 27.8% of not monitored subjects (p = 0.00009) (Fig. 3, B). Since clinicians may vary with regard to their definition of early detection of preterm labor, Fig. 3 shows the group comparisons with four different cutoff values (1, 2, 3, and 4 cm) for early detection. The monitored group had significantly earlier detection than the not monitored group for all four cutoff values. Since patients with pre term labor represent a subset of the original randomized groups, multiple regression
Home uterine activity monitoring to detect preterm labor
Volume 165 I\:umber 4, Part I
863
Table II. Obstetric outcomes for women with pre term labor
Gestational age at diagnosis of preterm labor (wk) Mean ± SD Cervical dilatation at diagnosis of pre term labor (cm)* Mean ± SD Median (extremes) Change in cervical diagnosis from prior examination (cm)* Mean ± SD Median (extremes) Duration of gestation after diagnosis of preterm labor (wk)t Mean ± SD Gestational age at delivery (wk)t Mean ± SD
Monitored (n = 43)
Not monitored (n = 39)
32,9 ± 3,2
32,9 ± 2,9
1.4 ± L3 LO (0-6.5)
2.5 ± L5 2.0 (0-6.0)
0.0006 0.0002
1.1 ± L2 LO (0-6.5)
2.0 ± 1.4 2.0 (0-5.0)
0.004 0.001
3.7 ± 3.8
2.0 ± 2.9
0.02
36.6 ± 2.4
34.9 ± 3.2
0.009
p Value
On the rows where data are summarized with mean, data analysis was with Student t tests. On the rows where data are summarized with median, data analysis was with Wilcoxon rank sum test. *Di!atation data were not available for two monitored and three not monitored subjects. tGestational age at delivery was not available for two monitored and one not monitored subjects.
Table III. Neonatal outcomes for singleton gestations with preterm labor*
Birth weight (gm)t Mean ± SD <2500 gm (%) <2000 gm (%) Neonatal intensive care No. of infants Total days Oxygen therapy No. of infants Total days Mechanical ventilation No. of infants Total days
Monitored (n = 33)
Not monitored (n = 30)
P Value
2934 ± 708 19 10
2329 ± 733 63 37
0.002 0.0007 0.02
5 50
13 324
0 0
7 68
0.004
0 0
5 54
0.02
0.02
*Medicaliy indicated pre term deliveries (one in each group) were excluded from these analyses. tBirth weight data were not available for two monitored infants who were born at term.
analysis was performed with dilatation at diagnosis of preterm labor as the dependent variable and numerous independent variables, such as obstetric history variables including substance use, study site, and monitoring versus not monitoring. The only variable that had a significant independent effect on dilatation at diagnosis was use of the monitor, which was associated with a 0.9 cm decrease in cervical dilatation (p = 0.01). Cohort analyses were done for women who entered the study at <25 weeks' gestation, at ?:27 weeks, and at ?:29 weeks' gestation. Mean cervical dilatation and cervical change at the diagnosis of preterm labor was less in the monitored group for all of these cohorts. Pregnancy and neonatal outcome. To evaluate the clinical importance of the early detection of preterm labor observed in the monitored group, duration of gestation after diagnosis of preterm labor (Table II), infant birth weight, and measures of neonatal morbid-
ity (Table III) were compared between the two groups. The mean duration of gestation after diagnosis of preterm labor was 3.7 weeks in 41 monitored patients (one patient did not have gestational age at delivery recorded and one was excluded because of medically indicated preterm delivery) compared with only 2.0 weeks in 38 not monitored patients (one excluded because of medically indicated preterm delivery [p = 0.02]) (Table III). Therefore the mean gestational age at delivery was 36.6 weeks in the monitored group compared with only 34.9 in the not monitored group (p = 0.009) (Table II). These gestational duration data do not significantly change if the two medically indicated pre term deliveries are included in the analysis. Multiple regression analysis also was performed with gestational age at delivery as the dependent variable and numerous independent variables including age, race, Creasy score, substance use variables, maternal
864
Mou et al.
history variables, study site, monitoring versus not monitoring, and cervical dilatation at diagnosis of preterm labor. The only variable that had a significant independent effect on gestational age at delivery was cervical dilatation at diagnosis of preterm labor, which was associated with a 0.6-week decrease in gestational age at delivery for each centimeter increase in dilatation (P = 0.03). Birth weight and neonatal morbidity variables were evaluated after exclusion of 17 (nine monitored and eight not monitored) twin gestations and two (one in each group) medically indicated preterm deliveries. In addition, there were two patients for whom birth weight data were not available. Twin gestations were excluded from these analyses because of the potential effects of twinning on birth weight and neonatal morbidity. Birth weight of singletons differed significantly between the monitored and not monitored groups. The mean birth weight of infants from the monitored grou p (n = 31) was 2934 gm compared with only 2329 gm for the not monitored group (n = 30) (P = 0.002). Only 19% of infants from the monitored group weighed <2500 gm versus 63% of the not monitored babies (p = 0.0007). The incidence of infants weighing <2000 gm was only 10% for the monitored group compared with 37% for the not monitored group (p = 0.02) (Table III). Singleton infants born to monitored women also suffered significantly less neonatal morbidity as measured by need for neonatal intensive care, oxygen therapy, and mechanical ventilation (Table III). Only five infants (50 total neonatal intensive care unit days) of the 33 monitored singleton gestations required admission to the neonatal intensive care unit (p = 0.02). None of the 33 infants who were products of monitored singleton gestations required oxygen therapy or mechanical ventilation compared with seven infants who required oxygen therapy, for a total of 68 days (p = 0.004), and five infants who required mechanical ventilation for a total of 54 days (p = 0.02) among the 30 products of not monitored singleton gestations (Table III). There were 252 patients who were randomized into monitored and not monitored groups who did not experience preteI'm labor outcome. Data were available on 250 patients. The mean gestational age at delivery of the controls without preteI'm labor was 39.1 weeks (± 2.3 weeks SD). The mean gestational age at delivery of the monitored group without preteI'm labor was 38.8 weeks (±2.1 weeks SD), p = 0.28. Likewise, the birth weights of the singletons in the control group without preteI'm labor was 3154 gm (± 654 gm SD) and the birth weights of singletons in the monitored group was 3269 gm (±604 gm SD), p = 0.18. Comment
This study demonstrates that objective assessment of uterine activity, by home uterine activity monitoring- in
October 1991 Am J Obstet Gynecol
women at high risk for preterm labor, allows detection of preteI'm labor at less advanced cervical dilatation. The clinical importance of this earlier detection of preterm labor is supported by the increased gestational duration after diagnosis of preterm labor and the resultant improved gestational age at delivery, increased birth weight, and decreased neonatal morbidity observed in the monitored group. This report represents the largest published, prospective, multicenter, randomized, and blinded evaluation of the efficacy of home uterine activity monitoring for earlier detection of preterm labor in high-risk pregnancies. Previously published studies have suggested beneficial effects of home uterine activity monitoring" 5, 6; however, the study designs used did not allow for the separation of the impact of the use of the monitor from the impact of the additional perinatal nursing support. 7,10 Some studies 3 ,6 did not provide comparison groups receiving the same perinatal center obstetric care. The studies that failed to show beneficial effects of home uterine activity monitoring have not focused on the value of the device when it is used without additional nursing care. 11,12 Also, the present study differs from others in using twice-daily monitoring and transmission of data, rather than once-daily transmission of monitor strips. When one is evaluating a new device, it is essential to separate clearly those effects that are a result of the information derived from the use of the device from those effects of other nonstandard interventions. Previous studies have evaluated daily perinatal nursing support used along with home uterine activity monitoring. The current report evaluates the impact of the addition of objective assessment of uterine activity to high-risk obstetric care on pregnancy outcome. In this study design, (1) no routine daily perinatal nursing support was provided to either group, (2) all study patients were provided with identical high-risk obstetric care and education in self-palpation of uterine contractions (as described by Herron et al. 13), (3) caretakers were blinded to subject's group assignment, (4) preteI'm labor was strictly defined, (5) all subjects were included in the data analysis on the basis of their original group assignment regardless of compliance, and (6) the primary end point was cervical dilatation at diagnosis of preteI'm labor, since other variables (e.g., incidence of preterm birth) are strongly related to the incidence of preteI'm labor in the study population and to the effectiveness of tocolytic therapy. Although caretakers were blinded to the subjects' group assignments, it is not possible to be absolutely certain that all examiners were indeed unaware of a patient'S group assignment. It is estimated that not more than six women were examined by caretakers who were aware of the patient's group assignment. The results of the study did not change when these patients were excluded from the data an
Volume 165 Number 4, Part 1
Home uterine activity monitoring to detect preterm labor 865
The study did not include a sham monitoring group, and therefore it is not possible to determine the independent effects of the four procedures that are an integral part of the monitoring protocol. These procedures, which represent the only known differences between the two groups, include (1) wearing a tocodynamometry belt during each recording session, (2) relative inactivity during each recording session, (3) data transmission after each recording session , and (4) patient notification if four or more contractions were observed. Although it is possible that any of these four procedures may have contributed to the results, detection and notification of increased contractions is the most reasonable explanation for the early detection of preterm labor observed in the monitored group. There were similar outcomes in the control and monitored group among patients who did not go into preterm labor, which also makes the twice daily monitoring unlikely to affect the course of the pregnancies except for diagnosing preterm labor at an earlier cervical dilatation. The benefit of monitoring appears to be in the earlier detection of preterm labor, before perception by the patient and before significant cervical dilatation has occurred. Self-palpation of uterine contractions had been shown to be less sensitive than tocodynamometry in detecting uterine activity.2. 14 Objective quantitation of contractions is a useful method of surveillance for preterm labor and detects contractions before women perceive them. 2 Since progression of labor occurs over hours rather than days or weeks one would not expect the gestational age at diagnosis of pre term labor to differ significantly between the groups as shown in this study, with mean gestational age at diagnosis of preterm labor at 32.9 weeks in the monitored and not monitored groups. The observation that there were no differences between the groups for the gestational age at the time of diagnosis of preterm labor and the incidence of preterm labor makes it unlikely that overdiagnosis of preterm labor or lead time bias could account for the results of the study. This protocol did not address specific tocolytic regimens. Once preterm labor was diagnosed, each center continued its usual protocols. Therefore it is not possible to assess the relative effects of various medical interventions on the outcomes. The fact that the vast majority of caretakers were unaware of the patient'S group assignment makes it unlikely that there were significant differences between the medical interventions received by the two groups. In addition, the two groups were medically and demographically similar at entry into the study and at the time of diagnosis of preterm labor. They had similar numbers of routine and non routine obstetric visits, incidence of preterm labor, similar gestational age at diagnosis of preterm labor, and were similar in the relationship between dilatation at diagnosis of preterm labor and subsequent duration of I!estation.
Zlatnik l5 and King et al. 16 have shown that long-term tocolysis with significant prolongation of gestation is possible if preterm labor is diagnosed before advanced cervical dilatation. Multiple regression analysis confirmed that earlier detection of preterm labor in the monitored group led to significant prolongation of gestation. Although there is disagreement as to the specific definition of advanced cervical dilatation, a significantly higher percentage of the monitored patients in this study had early detection of preterm labor, as measured by cervical dilatation, whether <1, 2, 3, or 4 cm dilatation was used as the cutoff for early detection. The result was improved gestational age at delivery and increased birth weight of infants in the monitored group. A medically and sociodemographically diverse population from three perinatal sites participated in this study. Subgroup analyses were performed with specific medical risk factors, gestational ages at entry, gestational ages at diagnosis of preterm labor, and compliance with study protocol. The results of these analyses showed that the information derived from home uterine activity monitoring appeared to be similarly beneficial for all patients who used the monitor. The subset of the monitored group (n = 12) who did not comply with the protocol had results that were similar to those of the not monitored grou p. Prior preterm delivery was the only medical risk factor for preterm labor for which there were a sufficient number of patients for separate analysis. These data were consistent with the data for the entire study population showing beneficial effects for all outcome variables. On the basis of these study data, the United States Food and Drug Administration has given approval for use of this device in women with a history of preterm birth. l7 Although none of the other individual medical risk factors had a sufficient sample size for separate analyses, when considered as one group (containing the 31 patients without a history of preterm birth), the monitored group showed significant beneficial effects similar in magnitude to the subset of women with a history of preterm birth. Women with other medical risk factors for preterm labor may therefore also benefit from home uterine activity monitoring. Future studies may further address the benefit in women with other specific maternal risk factors. Our study demonstrates that home uterine activity monitoring, when used in selected high-risk women as the sole addition to standard high-risk care, results in earlier detection of preterm labor. This led to an impressive mean 1.7 -week improvement in gestation compared with not monitored women, as well as a mean 605 gm greater birth weight. In summary, this study demonstrated that use of home uterine activity monitoring, in selected high-risk patients, significantly increases the likelihood of diagnosis of preterm labor before advanced cervical dilatation, thus providing the clinician with a better op-
866
Mou et al.
portunity to initiate tocolytic therapy directed at improving pregnancy outcome. We are indebted to David Simic, Anita Quinlan, Christine Falchi, and Laurie Dressman for their role in coordinating the data collection and to Dr. Howard Golub for his help in the statistical analyses. REFERENCES I. McCormick Me. The contribution of low birthweight to infant mortality and childhood morbidity. N Engl J Med 1985;312:82-90. 2. Newman RB, Campbell BA, Stramm S1. Objective tocodynamometry identifies labor onset earlier than subjective maternal perception. Obstet Gynecol 1990;76: 1089-92. 3. Katz M, Gill P, Newman RB. Detection of preterm labor by ambulatory monitoring of uterine activity: a preliminary report. Obstet Gynecol 1986;68:773-8. 4. Katz M, Newman RB, Gill PJ. Assessment of uterine activity in ambulatory patients at high risk of preterm labor and delivery. AMJ OBSTET GYNECOL 1986;154:44-7. 5. Morrison JC, Martin IN, Martin RW, Gookin RS, Wiser W1. Prevention of pre term birth by ambulatory assessment of uterine activity. A randomized study. AM J OBSTET GYNECOL 1987;156:536-43. 6. Hill WC, Fleming AD, Martin RW, et al. Home uterine activity monitoring is associated with a reduction in preterm birth. Obstet Gynecol 1990;76: 135-75. 7. Diagnostic and therapeutic technology assessment (DATTA) .. Home monitoring of uterine activity. JAMA 1989;261 :3027-9.
October 1991 Am.J Obstet Gynecol
8. Creasy RK, Gummer BA, Liggins GC. System for predicting spontaneous preterm birth. Obstet Gynecol 1980;55:692-5. 9. Newman RB, Gill PJ, Campion S, Katz M. The influence of fetal number on antepartum uterine activity. Obstet Gynecol 1989;73:695-9. 10. Martin RW, Gookin KS, Hill CW, et al. Uterine activity compared with symptomatology in the detection of preterm labor. Obstet Gynecol 1990;76:195S. 11. lams JD, Johnson FF, O'Shaughnessy RW, West Le. A prospective random trial of home uterine activity monitoring in pregnancies at increased risk of preterm labor. AMJ OBSTET GYNECOL 1987;157:638-43. 12. lams JD, Johnson FF, O'Shaughnessy RW. A prospective random trial of home uterine monitoring in pregnancies at increased risk of preterm labor. Part II. AM J OBSTET GYNECOL 1988; 159:595-603. 13. Herron MA, Katz M, Creasy RK. Evaluation of a pre term birth prevention program: preliminary report. Obstet Gynecol 1982;59:452. 14. Brutsman LE, Langer 0, Anyaegbunam A, Belle C, Merkatz IR. Education does not improve patient perception of preterm uterine contractility. Obstet Gynecol 1990;76:97S. 15. Ziatnik F. The applicability oflabar inhibition to the problem of prematurity. AM J OBSTET GYNECOL 1972;103: 704-6. 16. King JF, Grant A, Marc IN, Keirse C, Chalmers I. Betamime tics in preterm labour: an overview of the randomized controlled trials. Br J Obstet GynaecoI1988;95: 1988. 17. Fed Reg 1990 Oct 23;55(205):427-79.