Women's Evaluations of Induction of Labor Versus Expectant Management for Prelabor Rupture of the Membranes at Term
Abstract
Background:
Induction of labor has become common practice in many Western countries, but few studies have assessed women's views.
Methods:
A randomized, controlled trial was conducted at 72 hospitals in six countries. Five thousand forty-one women meeting eligibility criteria, with no contraindications for induction of labor or expectant management, were randomly assigned to four groups: induction with intravenous oxytocin, induction with vaginal prostaglandin E2 gel, or expectant management followed by induction with either oxytocin or with prostaglandin E2 gel if complications developed. The three main outcome measures were evaluations of the treatment received, perceived control during childbirth, and evaluations of the experience of trial participation.
Results:
Questionnaires were completed by 81.9 percent of the sample. No significant differences occurred between the two induction groups. Compared with the expectant management groups, induced women were less likely to report there was nothing they liked about their treatment and less likely to report that the treatment caused additional worry. No between-group differences occurred in experienced control during childbirth. Women in the induction groups were more likely to be willing to participate in the study again and to feel reassured.
Conclusions:
Women's preferences should be considered when making decisions about their method of management when membranes rupture before labor. Obtaining participants' views is both feasible and worthwhile when evaluating forms of medical care. (BIRTH 24:4, December 1997)
Although induction of labor has become common practice in many Western countries and numerous studies of methods of induction have been conducted, few studies included assessments of patients' views (). Descriptive studies have suggested that induced labor may be more unpleasant, it may decrease women's perceptions of control during childbirth, and women who have experienced induction are likely to state a preference for spontaneous labor for future births (2–4). The studies did not control for any potentially influential factors, however, including the reason for induction and preexisting differences in the personal characteristics of women whose labors were induced compared with those who had spontaneous labor. An intrapartum medical intervention may not be as important a determinant of a woman's satisfaction with her childbirth experience as her belief that the intervention was necessary or desirable (5).
The induction method may also influence satisfaction with childbirth. Two small trials found that women rated vaginal prostaglandin preparations more favorably than intravenous oxytocin (6,7). Women preferred the prostaglandin preparation because it allowed freedom of movement, did not involve an intravenous infusion, and labor contractions were gradual in onset (7).
The Term Prelabour Rupture of the Membranes (TermPROM) Study was a multicenter international trial to evaluate a policy of induction of labor versus a policy of expectant management (waiting for labor to begin spontaneously with no evidence of fetal or maternal compromise) for women with prelabor rupture of the membranes at term (8). The primary outcome was the rate of serious neonatal infection, and the secondary outcome was the cesarean delivery rate. The TermPROM Study afforded a unique opportunity for a large-scale, unbiased comparison of women's experiences of induction versus spontaneous onset of labor, as well as their experiences of induction with intravenous oxytocin compared with induction with vaginal prostaglandin (PGE2) gel. After a brief summary of the study procedures and medical outcomes, the following report describes the aspect of the trial that concerned women's evaluations of their birth experiences. A detailed report of the study procedures and medical outcomes is provided elsewhere (8).
Summary of Study Procedures and Medical Outcomes
The TermPROM Study was approved by the research ethics committees at all participating centers, and all participants gave informed consent before enrollment. Women were eligible for the study if they were at more than 37 weeks' gestation, had ruptured membranes, and had a single fetus in a cephalic presentation. Women were excluded if they were in active labor, if a contraindication to induction or expectant management existed, or if a previous, failed attempt at induction had occurred.
Women were randomly assigned to one of four groups: immediate induction with intravenous oxytocin, immediate induction with vaginal PGE2 gel followed by intravenous oxytocin if necessary, expectant management followed by induction (if necessary) with intravenous oxytocin, or expectant management followed by induction (if necessary) with PGE2 gel. Because we anticipated the possibility of differences between the methods of induction (9), we planned three principal treatment comparisons among the four groups: induction with oxytocin versus expectant management (oxytocin), induction with PGE2 versus expectant management (PGE2), and induction with oxytocin versus induction with PGE2. The sample size was chosen to provide 80 percent power to detect a reduction of 50 percent or more, from more than 4 percent to less than 2 percent, in the rate of serious neonatal infection in each treatment comparison.
The TermPROM Study enrolled 5042 women between 1 January 1992, and 31 May 1995, at 72 hospitals in Canada, the United Kingdom, Australia, Israel, Sweden, and Denmark. Medical data were received for 5041 women, and no significant between-group differences were found in the baseline characteristics of the participants. The mean age was 28.4 ± 5.3 years, 59.5 percent were nulliparas, the mean gestational age was 38.9 ± 1.2 weeks, and 3.5 percent had a previous cesarean delivery. The median duration of membrane rupture at study entry was 5.3 hours (5th and 95th percentiles 1.3, 28.0).
Labor was induced in 89.0 percent of the induction with oxytocin group and in 89.7 percent in the induction with PGE2 group. Nearly all of the remaining women in the two induction groups began active labor before the induction could start; five women had cesarean deliveries without experiencing labor. Labor began spontaneously in 77.0 percent in the expectant management (oxytocin) group and in 78.5 percent in the expectant management (PGE2) group. Between-group comparisons (e.g., induction with oxytocin versus expectant management [oxytocin], induction with PGE2 versus expectant management [PGE2], and induction with oxytocin versus induction with PGE2) revealed no significant differences in the rates of serious neonatal infection and cesarean delivery.
Methods
Timing and Measures
Participants were asked to complete the evaluation questionnaires within the first few days postpartum, preferably before discharge. Although most women completed the questionnaires on their own, 19 percent required assistance from a family member or caregiver, usually because of low levels of literacy. No significant differences occurred between the groups in the numbers of women who required assistance. Three measures of satisfaction were included in the questionnaire: evaluations of the treatment received (expectant management or induction), perceived control during childbirth, and evaluations of the experience of participation in the study. The questions used to evaluate treatment and study participation are listed in Tables 1 to 3. Perceived control during childbirth was measured by the Labour Agentry Scale (LAS), a Likert-type summated rating scale with good evidence of reliability and validity (10,11). LAS scores can range from 29 (very low control) to a high of 203. Questionnaires were translated into the official language(s) of each participating country. English, French, Swedish, and Hebrew versions were pretested on between 28 and 50 nonstudy participants, and the Cronbach's alpha reliability coefficient for each version of the LAS was more than 0.86.
Statistical Analyses
All results were analyzed according to the intention-to-treat principle. Several questions included lists from which participants were asked to check all applicable responses. Percentages were calculated as the number of women who selected an item divided by all those who either did or did not select the item (either because it was not applicable or because they did not return a completed questionnaire). Four questions included an “other” category, in which women could write additional responses. One of the investigators (EH) used content analysis (12) to develop categories of responses, coded all responses, and calculated frequencies within each category. The categories were given to a research assistant, who independently coded responses. The results were compared and discussed until 100 percent agreement was achieved.
Two-sided p values are reported for all comparisons. Because multiple comparisons were made, the level of statistical significance for all but the primary and secondary outcomes was set at p<0.01. The between-group comparisons were analyzed by contingency table χ2 analyses for categorical variables, with odds ratios (OR) and associated 95 percent confidence intervals (CI) and Student's t tests for LAS scores.
Other variables that may affect women's evaluations (parity, country, maternal age, education, smoking, duration of ruptured membranes at entry into the trial, known unfavorable cervix, cesarean delivery) were entered into multiple logistic regression analyses for categorical variables (evaluations of treatment and of study participation) and multiple linear regression analyses for LAS scores.
Results
Questionnaire Response Rates
Completed questionnaires were obtained from 4129 women (81.9% of the sample). Response rates within treatment groups ranged between 81 and 83 percent, differences that were not statistically significant. The percentage of primiparas who completed questionnaires was identical to that of multiparas. When rates were examined by country, Canada and Israel had the lowest response rates (77.1 and 79.2%, respectively). Response rates in the United Kingdom, Australia, Sweden, and Denmark exceeded 85.0 percent.
Evaluations of Treatment Method
Women's evaluations of the method of labor management they received were more favorable in the induction groups than in the corresponding expectant groups (Table 1). Both induction groups were more likely to indicate there was nothing they disliked about their method of care, less likely to indicate there was nothing they liked, and less likely to indicate the method of care they received caused additional worry about their babies' or their own health. The two induction groups did not differ significantly in any likes or dislikes about their method of care, except for a nonsignificant trend in the induction with oxytocin group, suggesting greater dislike of having medication to induce labor (p= 0.02).
Between 9.1 and 20.2 percent in each group listed “other likes” about their method of care. Over 90 percent of the “other likes” were elaborations on previously selected items. Between 16.6 and 23.9 percent in each group listed “other dislikes” about their method of care. Pain (12%) and discomfort (15%) associated with waiting comprised the most common “other dislikes” in the expectant management (oxytocin) group and the expectant management (PGE2) group. The remaining “other dislikes” in both expectant management groups were elaborations on previously selected items, such as worry and having to wait for labor to begin. Over 60 percent of the “other dislikes” in both induction groups concerned aspects of the induction process, including rapid onset of painful contractions, discomfort in the application of the gel, and restricted mobility associated with intravenous infusions and electronic fetal heart rate monitoring.
Additional Independent Predictors of Negative Ratings of Method of Care
In response to the question, “What did you like?”, women with an unfavorable cervix on enrollment in the trial (OR = 1.33, 95% CI = 1.09, 1.64) and those who had a cesarean delivery (OR = 1.79, 95% CI = 1.35, 2.38) were more likely to indicate that “there was nothing I liked.” In response to the question, “What did you dislike?”, women who had a cesarean delivery were more likely to indicate worry about their own or their babies' health (OR = 1.74, 95% CI = 1.39, 2.18). Women in the United Kingdom (OR = 0.55, 95% CI= 0.46, 0.65), Sweden (OR = 0.35, 95% CI = 0.20, 0.64), and Israel (OR = 0.66, 95% CI = 0.48, 0.90) were less likely to indicate that their treatment caused them to worry compared with women in Canada.
Personal Control During Childbirth
LAS scores in the induction with oxytocin group (mean = 155.16 ± 29.13) were higher than scores in the expectant management (oxytocin) group (mean = 151.82 ± 30.91; t= 2.49, p= 0.01), but the mean difference was less than four points and of borderline statistical significance. LAS scores in the induction with PGE2 group (mean = 152.75 ± 30.65) were similar to the scores in the expectant management (PGE2) group (mean = 151.36 ± 30.54; t= 1.01, p= 0.31). LAS scores in the induction with oxytocin group versus the induction with PGE2 group were not significantly different (t= 1.81, p= 0.07). Multiple linear regression analyses of additional independent predictors of experienced control revealed that women who had a previous cesarean delivery and women who were smokers had lower levels of control (associated with a decrease of 12.0 and 4.9 points in LAS scores, respectively), whereas multiparity was associated with an increase of 6.5 points in LAS scores.
Experiences As Study Participants
When asked about their likes and dislikes regarding participation in the study, women in both induction groups were more likely to indicate they were reassured by the extra attention to their own and their babies' health, and they were less likely to indicate that participation had increased their anxiety when compared with their respective expectant management group (Table 2). The two induction groups did not differ significantly in any of their likes and dislikes about participation in the study. The small number of “other” responses were elaborations on previously selected responses.
Nearly two-thirds of each group indicated that they liked helping to answer an important research question. Approximately one-fourth of women indicated that they disliked being randomized. Fewer than 7 percent in each group listed “other likes,” nearly all of which were re-statements of previously selected responses. When asked if they would participate in the Term PROM Study if they had it to do over again, women in both induction groups were more likely to indicate “yes” rather than “no” or “unsure” when compared with their respective expectant groups (Table 3). The two induction groups did not differ significantly. Multiple logistic regression analyses revealed two additional independent predictors of willingness to participate: women who had a previous cesarean delivery were less likely to be willing to participate (OR = 0.49, 95% CI = 0.39, 0.62), and multiparous women were more likely to be willing to do so (OR = 1.26, 95% CI = 1.07, 1.48).
Discussion
Obtaining women's views of their care is both feasible and worthwhile in the context of a large international study. The additional cost at the participating hospitals was small, and the costs at the data coordinating center were also modest in relation to the total study costs.
To our knowledge the TermPROM Study provided the first unbiased comparisons of women's evaluations of induction versus expectant management. Key features of the study design and methods, including random allocation, large sample size, high response rate, and the inclusion of participants from six countries, strengthen the internal and external validity of the results.
All statistically significant differences (likes and dislikes about the treatment method and about participation in the study, and in willingness to participate in the study again) favored induction of labor over expectant management. These results contradict those of previous studies of women's evaluations of induction of labor. Although the results of a randomized trial tell one what happened, why it happened is open to interpretation and speculation. A partial explanation for the study results may relate to the additional reason for and time in which to worry that was experienced by those allocated to expectant management. As part of the consent process, participants were informed of the uncertain role of prolonged rupture of the membranes in the development of neonatal infection, and those in the expectant management groups had additional time in which to worry as they waited for active labor to begin. This increased worry, as suggested by these women's responses to both their evaluations of treatment method and of participation in the study, may have overshadowed any other concerns about induction. Furthermore, the experience of expectant management was not purely a matter of waiting for labor to begin. Content analyses of women's written comments revealed examples of the discomfort and fatigue associated with intermittent contractions over hours or days before the onset of active labor, and of the distress associated with being confined to the hospital or with trying to carry on normal activities at home.
The apparently lower levels of worry in participants from the United Kingdom, Sweden, and Israel, when compared with Canada, highlights the importance of societal norms in the development of women's attitudes about risk and normalcy in childbearing (15). The multiple regression analyses show another pattern that is consistent with previous research: more difficult labors (such as that experienced by nulliparous women or when the cervix is unfavorable) and labors that culminated in cesarean delivery associated with less positive evaluations of the childbirth experience (10,16). The method of care, and the fact that the method was randomly assigned, had no apparent effect on women's perceptions of control during labor. The mean LAS scores compare with the results of other studies (10,11).
The method of induction had no apparent effect on women's evaluations. These results may reflect the fact that 43.1 percent of the induction with PGE2 group had oxytocin during labor (1.2% had oxytocin only, 9.1% had oxytocin following PGE2 gel, 32.8% had oxytocin to augment labor), thereby confounding the comparisons of women's views of the two methods.
Within-group comparisons of women's likes and dislikes about study participation revealed a consistent pattern (Table 2). Women in each study group were more likely to select the form of care they received as something they liked than as something they disliked. Just under one-third of women in the induction groups reported that they liked being induced, compared with approximately 8 percent who reported they disliked being induced. Similarly, approximately 30 percent of women in the expectant management groups reported they liked waiting for labor to start on its own, compared with approximately 11 percent who indicated they disliked waiting. These results may reflect the “what is, must be best” phenomenon commonly found in women's evaluations of their care during pregnancy and birth (13). Lumley suggested that future birth preference is an indicator of satisfaction (14). The high rates of willingness to participate in the trial again suggest that most women were satisfied with their experiences, and particularly so if they were in the induction groups.
Altruism appeared to play an important role in the decision to enroll in the study, and may have influenced participants' evaluations of their experiences in it. Whereas one-fourth of the sample indicated they disliked being randomized, approximately two-thirds thought that helping to answer a research question was a benefit of participation in the study.
The TermPROM Study results have improved the quality of information we can provide to pregnant women about the risks and benefits of induction of labor versus expectant management. Whereas no significant differences in serious neonatal infection or cesarean delivery occurred, other results (including clinical chorioamnionitis, postpartum fever, labor length, and neonatal intensive care unit stay) favored the induction with oxytocin group (8); the women's evaluations of their experiences also favored induction. It would be inappropriate to conclude, however, that because participants (who could not foretell the trial results) favored induction, therefore women in the future will prefer to have labor induced. We conclude that women with prelabor rupture of the membranes at term should be fully informed about the TermPROM Study results, and that their preferences should be considered when making decisions about their method of management.
Acknowledgements
We gratefully acknowledge the contributions of the women who participated in the study and the collaborators at the participating centers. A list of the members of the Term Prelabour Rupture of the Membranes (TermPROM) Study Group has been published in Hannah ME, Ohlsson A, Farine D, Hewson SA, Hodnett ED, Myhr T, et al. Induction of labor compared with expectant management for prelabor rupture of the membranes at term. N Engl J Med 1996;334:687–690.
