Differences in Institutional Cesarean Delivery Rates: The Role of Pain Management

Although the seemingly relentless increase in cesarean delivery rates in North America in the 1970s and 1980s has stabilized in recent years and even reversed slightly in some regions,¹ rates are still generally considered to be too high, and efforts to diminish them continue.² In spite of accumulating evidence about the preventable causes of cesarean delivery, individual institutions have had difficulty changing practice.³ At BC Women’s Hospital, the largest maternity facility in Canada with 7500 births annually, the cesarean delivery rate in 1995 was 22.9%.⁴ Although BC Women’s is the referral/tertiary center for the province of British Columbia, 80% to 85% of the births occur to low-risk women in the surrounding catchment area of Vancouver. Approximately 30 obstetricians and 100 family practice physicians have admitting privileges. At nearby Burnaby Hospital, a community general hospital providing primary- and secondary-level maternity care, cesarean delivery rates were 10.3%.⁴ At Burnaby, 3 obstetricians and 38 family practice physicians deliver 2000 babies annually. At the time of our study, neither hospital had practicing midwives. The discrepancy in cesarean delivery rates between hospitals only 20 minutes’ driving distance apart afforded us the opportunity to look at differences in practice between the institutions.

Methods

Phase I

In phase I we completed a retrospective cohort study using hospital records. Records were considered eligible if parturient women met the following criteria: nulliparous, maternal age 16 to 35 years, singleton gestation, and gestation of 37 completed weeks or longer. We excluded women with known pregnancy complications.

Charts were selected from a computer-generated list that incorporated inclusion and exclusion criteria and were then reviewed by a research nurse. Blinding as to hospital was not possible. Using this system, 430 nulliparous women from each hospital were selected consecutively. The sample size was calculated to be 430 per group to have 80% power with type I error set at 0.05, to determine a difference in cesarean delivery rate of 30% from a baseline of 23%. We examined the role of demographic and obstetric factors in the association between hospital and cesarean delivery rate. Maternal and fetal outcomes were measured, including postpartum hemorrhage and infection, as were APGAR scores at 5 minutes. We used logistic regression to simultaneously adjust for confounding factors.

Phase II

Since we were not able to randomly allocate women to hospitals, it was important to determine if women giving birth at BC Women’s compared with Burnaby selfselected to either of the hospitals because of differing expectations for pain management in labor. We also wanted to discern if there were aspects of the care given during the intrapartum period that might explain differing rates of use of epidural analgesia. In the phase II observation study, a research assistant was assigned to observe consecutive women giving birth 2 days per week at each hospital. Our research assistant was a medical student who had not practiced in either hospital. Structured observations were recorded along with times for each observation period and the stage of labor at which the observation took place. The same inclusion and exclusion criteria were used as in our phase I retrospective study to ensure that we were observing comparable women without preexisting risk factors for cesarean delivery at the onset of labor. The phase II study did, however, include multiparous women to maximize use of the time the student was available to assist with the study.

Analysis

We compared demographic characteristics and rates of selected outcomes between hospitals, using the chi-square statistic for categoric variables and the Student t test statistic for continuous variables. When expected cell frequencies were less than 2, we used the Fisher exact test for categoric variables instead of the chi-square statistic. The type I error, 2-sided a, was set at 0.05. Multivariate analysis was undertaken using unconditional logistic regression. We obtained maximum likelihood estimates of the odds ratios using the logistic model. We calculated 95% confidence intervals using the estimates of the standard error derived from the model.

Results

Phase I

Comparing cesarean delivery rates from our retrospective study of 857 births, the crude nulliparous rate at BC Women’s was 20.7%, while at Burnaby it was 6.7% (P <.0001). We examined a number of demographic and obstetric factors to determine if they would explain the reason for the increased rate at BC Women’s.

Women at Burnaby were younger, more likely to be single, and white (P <.0001, Table 1). A larger proportion of women at BC Women’s were Asian. Cesarean delivery rates did not differ between white and Asian women (9.1% vs 9.3%) at Burnaby, but there was a marked difference (18% vs 28.6%) at BC Women’s (data not shown). Women arrived earlier in labor at BC Women’s Table 2. The lengths of the first and second stages were longer at BC Women’s, as was the length of time membranes were ruptured. Time from admission to delivery was also significantly increased at BC Women’s.

Rates of obstetric interventions were compared between the 2 settings Table 2. Rate of induction of labor was not different between the 2 hospitals; however, augmentation of labor with oxytocin occurred significantly more frequently at Burnaby than at BC Women’s. Artificial rupture of membranes during labor was performed significantly more often at BC Women’s than at Burnaby. BC Women’s primarily used epidural analgesia, while Burnaby used intramuscular or intravenous administration of meperidine (Demerol). Use of Entonox (nitrous oxygen and oxygen) was also significantly different. Electronic fetal monitoring (EFM) was used for almost all patients at both hospitals; however, at Burnaby nurses were more likely to obtain only a baseline or admission fetal monitoring record, or to monitor intermittently. BC Women’s, on the other hand, was more likely to employ EFM continuously. Increased use of EFM both with and without epidural analgesia resulted in women being confined to bed more often, since remote methods (telemetry) of monitoring fetal heart rates were rarely used.

Overall rates of assisted vaginal deliveries at the 2 hospitals were not different, although forceps were used more frequently at BC Women’s and vacuum extraction at Burnaby. More frequent use of cesarean delivery at BC Women’s was not associated with an improvement in selected maternal/newborn outcomes (postpartum hemorrhage and 5-minute APGAR score <7). Postpartum infections were significantly more common at BC Women’s (P <.0001).

Data were analyzed for each hospital separately to determine all the factors contributing to cesarean delivery in each Table 3. After regressing all the demographic factors in a multiple regression model with cesarean delivery as the dependant variable for each hospital, only age remained significant. To evaluate the role of obstetric interventions for each hospital each intervention was entered separately, the most significant retained, and the process repeated with the next most significant retained until no other variables remained significantly associated with cesarean delivery. Age was retained in all the models.

In Burnaby, patient age, cervical dilatation on admission to the hospital, induction, epidural analgesia, and augmentation with oxytocin were important. At BC Women’s the same factors were important, as well as race/ethnicity (white vs nonwhite). Use of electronic fetal monitoring, Entonox, and intravenous or intramuscular narcotics did not predict cesarean delivery in either setting.

To compare the 2 hospitals in cesarean delivery rate we created a model combining data from both institutions. With both in the model, we could quantify the risk for cesarean delivery in one hospital compared with the other. First we examined the role of demographic factors in accounting for differences in cesarean delivery rate. Adjusting for age, the odds ratio (OR) for cesarean delivery at BC Women’s compared with Burnaby was 3.4 (95% confidence interval [CI], 2.1-5.4).

Cervical dilatation on admission, oxytocin augmentation, and epidural analgesia remained statistically significant in addition to age for predicting cesarean delivery. When age, augmentation, and cervical dilatation were retained in the model there remained a statistically significant difference between the 2 hospitals; the OR for BC Women’s versus Burnaby for cesarean delivery was 3.8 (95% CI, 2.3- 6.3). When we adjusted for epidural in addition, the OR for BC Women’s versus Burnaby dropped to 2.0 (95% CI, 1.09-3.5); the difference became only marginally statistically significant. The principal factor associated with the difference in cesarean delivery rate between the 2 hospitals, therefore, was use of epidural analgesia.

Phase II

Our sample consisted of 24 laboring women at Burnaby and 26 laboring women at BC Women’s. The groups were comparable in mean age, marital status, employment, and parity. There were more white women: 18 (75%) in the Burnaby group compared with 11 (42.3%) in the BC Women’s group.

Almost all the women were accompanied by support persons, 24 (100%) at Burnaby and 25 (96.2%) at BC Women’s. The majority at Burnaby had been to childbirth education classes (18 [78.3%] vs 14 [58.3%] at BC Women’s).

It was not possible to observe all the women throughout the entire length of labor. Given this reality and the fact that length of labor differed between women, observations related to management of labor were expressed as rates (ie, the number of minutes spent in bed divided by the total observation time in hours). Women in the 2 hospitals spent similar amounts of time in bed and in the shower. Women at Burnaby spent significantly more time walking, 12 minutes per hour compared with 4.7 minutes per hour at BC Women’s (P=.01, Table 4). There was no difference between the 2 hospital groups in the amount of physical contact with the laboring woman and her nurse or support person. At BC Women’s, women were more often accompanied by a nurse, physician, medical student, or resident and less often left alone than at Burnaby. Women at BC Women’s were exposed to greater numbers of different caregivers.

There were no differences between the groups in the number of couples who expressed specific expectations of labor Table 5. Ten women in Burnaby and 8 in BC Women’s were observed to ask for help with pain management, and responses to these requests included general encouragement or specific verbal advice as to what method they should use. Women were more frequently offered epidural analgesia at BC Women’s; there were no differences in rates of offers for other pain management options. This question related to a preemptive discussion about pain management. In terms of the pain management that actually took place, there was a significant difference in the number of doses of narcotic, in that no women at BC Women’s received narcotics. There was a trend toward more exposure to Entonox and epidural at BC Women’s, although these differences were not significant.

Discussion

Institutional differences in cesarean birth rates have previously been associated with differences in maternity care practices.⁵ The time during labor (state of cervical dilatation) at which women are admitted to the hospital has been associated with cesarean delivery rate.⁶

Age as a risk factor for cesarean delivery has also been well documented. The influence of age appears to be important independent of risk and remains important even among women without risk factors.^7-9 The reason for this is unknown but may point to an association with physician bias or age-related biological factors in the labor process that remain unmeasured.^10,11

The need for oxytocin augmentation has been associated with cesarean delivery.¹² Randomized controlled trials, however, have not demonstrated an association of early use of oxytocin augmentation with cesarean delivery reduction.¹³ Augmentation did not contribute significantly to the model examining differences in cesarean delivery between hospitals when an epidural was included.

A meta-analysis by Halpern and colleagues¹⁴ evaluated the association of epidural analgesia and cesarean delivery. In that report there was no increase in risk of cesarean delivery associated with use of epidural versus narcotic analgesia (OR=1.28; 95% CI, 0.55-2.93 for nulliparous women; OR=0.83; 95% CI, 0.22-3.15 for multiparous women). Findings from the intention-to-treat analysis that could not overcome the limitation of high rates of crossover or noncompliance among some of the trials^15-17 are in contrast with a protocol-compliant analysis in some of the studies that did show an effect of epidural on cesarean delivery rates.¹⁵ Although the protocol-compliant analysis provides important information, it does not necessarily indicate that epidural analgesia is the cause of higher cesarean rates. Subjects having more difficult labors may be more likely to cross over to epidural analgesia from the narcotic arm of the study, increasing the potential need for cesarean delivery in the epidural arm. In the meta-analysis the cesarean delivery rate among women receiving epidurals was 8.2%,¹⁴ much lower than at BC Women’s and more in line with those at Burnaby. The results, therefore, may not be generalizable to BC Women’s, at which high cesarean delivery rates would indicate greater opportunity for changes in intrapartum management, including use of epidural analgesia, to influence cesarean delivery rate.

An important aspect of our work addressed in phase II, our observation study, considered characteristics of women choosing delivery at either hospital that might also influence their choice to use an epidural. Although there were differences in ethnic background between the 2 study groups, these differences did not explain the difference in cesarean delivery rate between the 2 hospitals in our larger study. Although the differences were not statistically significant, more women at Burnaby appeared to have attended prenatal classes. The literature to date has not demonstrated an effect of prenatal education on cesarean delivery.¹⁸

Our observational study failed to suggest differences in client expectations between the 2 hospitals. Features of labor management did differ, however, in that women in Burnaby ambulated more, had fewer types and numbers of caregivers in labor and were less often offered an epidural. Nurses in Burnaby may have offered epidural less often because they were aware that anesthetists had to come to the delivery suite from elsewhere in the hospital or from outside the hospital. Alternatively, a cohort study has reported that nurses grouped according to cesarean rate quintiles differed in their recording of psychosocial data and other aspects of nursing care.¹⁹ Aspects of nursing practice may influence use of epidural analgesia. Our phase II study is limited by its small size, and other additional effects of nursing practice on labor cannot be ruled out. A large randomized controlled study is under way to examine the association of aspects of nursing care (including consistency of nursing caregiver) with cesarean delivery rates.²⁰ Ultimately, the decision to undertake a cesarean delivery resides with the obstetrician. It is possible that in the smaller community hospital physicians may practice with a greater degree of cohesion, perhaps influenced more readily by the philosophy of opinion leaders who advocate a more conservative approach to cesarean delivery. In addition, the presence of obstetric and family practice residency programs at BC Women’s may have encouraged use of interventions, including cesarean the delivery.

The role of ambulation in cesarean delivery remains controversial. A randomized trial failed to demonstrate an association of ambulation with cesarean delivery, but this study had a 22% crossover rate and enrolled women after they had attained a cervical dilatation rate of 3 to 5 cm, when early ambulation may have already exerted a positive effect.²¹ Among low-risk parturients cared for by midwives and not requiring either augmentation of labor or epidural analgesia, ambulation has been associated with a reduction in cesarean delivery rates.²²

Limitations

Our study is limited by lack of knowledge of women’s levels of anxiety and pain during their labor. This would have allowed us to gauge whether less frequent use of epidural analgesia at Burnaby was associated with a cost of diminished satisfaction with the childbirth experience.

We were fortunate in being able to study 2 institutions that were only 20 minutes’ driving distance apart and that served populations from which we were able to sample women who were demographically comparable. Our retrospective cohort analysis identified use of an epidural as the main predictive factor differentiating cesarean delivery rates between the hospitals. It is not possible to determine cause and effect from this retrospective study design, however, and use of epidural analgesia may be a proxy for other unmeasured variables, such as physician practice, anxiety level among patients, or level of education. Direct observation of intrapartum care in a follow-up study failed to differentiate clients attending either hospital in terms of preparation for or expectations of the labor experience. Differences in some aspects of caregiving, however, including more frequent offering of epidural for pain management, may explain the increased use of epidural at BC Women’s. Factors influencing use of epidural need to be studied more thoroughly to support its appropriate place in an overall strategy for pain management in labor.

Related Resources

• Society of Obstetricians and Gynaecologists of Canada (SOGC) http://www.sogc.org/SOGCnet
• College of Family Physicians of Canada www.cfpc.ca
• Canadian Paediatric Society www.cps.ca

Although the seemingly relentless increase in cesarean delivery rates in North America in the 1970s and 1980s has stabilized in recent years and even reversed slightly in some regions,¹ rates are still generally considered to be too high, and efforts to diminish them continue.² In spite of accumulating evidence about the preventable causes of cesarean delivery, individual institutions have had difficulty changing practice.³ At BC Women’s Hospital, the largest maternity facility in Canada with 7500 births annually, the cesarean delivery rate in 1995 was 22.9%.⁴ Although BC Women’s is the referral/tertiary center for the province of British Columbia, 80% to 85% of the births occur to low-risk women in the surrounding catchment area of Vancouver. Approximately 30 obstetricians and 100 family practice physicians have admitting privileges. At nearby Burnaby Hospital, a community general hospital providing primary- and secondary-level maternity care, cesarean delivery rates were 10.3%.⁴ At Burnaby, 3 obstetricians and 38 family practice physicians deliver 2000 babies annually. At the time of our study, neither hospital had practicing midwives. The discrepancy in cesarean delivery rates between hospitals only 20 minutes’ driving distance apart afforded us the opportunity to look at differences in practice between the institutions.

Methods

Phase I

In phase I we completed a retrospective cohort study using hospital records. Records were considered eligible if parturient women met the following criteria: nulliparous, maternal age 16 to 35 years, singleton gestation, and gestation of 37 completed weeks or longer. We excluded women with known pregnancy complications.

Charts were selected from a computer-generated list that incorporated inclusion and exclusion criteria and were then reviewed by a research nurse. Blinding as to hospital was not possible. Using this system, 430 nulliparous women from each hospital were selected consecutively. The sample size was calculated to be 430 per group to have 80% power with type I error set at 0.05, to determine a difference in cesarean delivery rate of 30% from a baseline of 23%. We examined the role of demographic and obstetric factors in the association between hospital and cesarean delivery rate. Maternal and fetal outcomes were measured, including postpartum hemorrhage and infection, as were APGAR scores at 5 minutes. We used logistic regression to simultaneously adjust for confounding factors.

Phase II

Since we were not able to randomly allocate women to hospitals, it was important to determine if women giving birth at BC Women’s compared with Burnaby selfselected to either of the hospitals because of differing expectations for pain management in labor. We also wanted to discern if there were aspects of the care given during the intrapartum period that might explain differing rates of use of epidural analgesia. In the phase II observation study, a research assistant was assigned to observe consecutive women giving birth 2 days per week at each hospital. Our research assistant was a medical student who had not practiced in either hospital. Structured observations were recorded along with times for each observation period and the stage of labor at which the observation took place. The same inclusion and exclusion criteria were used as in our phase I retrospective study to ensure that we were observing comparable women without preexisting risk factors for cesarean delivery at the onset of labor. The phase II study did, however, include multiparous women to maximize use of the time the student was available to assist with the study.

Analysis

We compared demographic characteristics and rates of selected outcomes between hospitals, using the chi-square statistic for categoric variables and the Student t test statistic for continuous variables. When expected cell frequencies were less than 2, we used the Fisher exact test for categoric variables instead of the chi-square statistic. The type I error, 2-sided a, was set at 0.05. Multivariate analysis was undertaken using unconditional logistic regression. We obtained maximum likelihood estimates of the odds ratios using the logistic model. We calculated 95% confidence intervals using the estimates of the standard error derived from the model.

Results

Phase I

Comparing cesarean delivery rates from our retrospective study of 857 births, the crude nulliparous rate at BC Women’s was 20.7%, while at Burnaby it was 6.7% (P <.0001). We examined a number of demographic and obstetric factors to determine if they would explain the reason for the increased rate at BC Women’s.

Women at Burnaby were younger, more likely to be single, and white (P <.0001, Table 1). A larger proportion of women at BC Women’s were Asian. Cesarean delivery rates did not differ between white and Asian women (9.1% vs 9.3%) at Burnaby, but there was a marked difference (18% vs 28.6%) at BC Women’s (data not shown). Women arrived earlier in labor at BC Women’s Table 2. The lengths of the first and second stages were longer at BC Women’s, as was the length of time membranes were ruptured. Time from admission to delivery was also significantly increased at BC Women’s.

Rates of obstetric interventions were compared between the 2 settings Table 2. Rate of induction of labor was not different between the 2 hospitals; however, augmentation of labor with oxytocin occurred significantly more frequently at Burnaby than at BC Women’s. Artificial rupture of membranes during labor was performed significantly more often at BC Women’s than at Burnaby. BC Women’s primarily used epidural analgesia, while Burnaby used intramuscular or intravenous administration of meperidine (Demerol). Use of Entonox (nitrous oxygen and oxygen) was also significantly different. Electronic fetal monitoring (EFM) was used for almost all patients at both hospitals; however, at Burnaby nurses were more likely to obtain only a baseline or admission fetal monitoring record, or to monitor intermittently. BC Women’s, on the other hand, was more likely to employ EFM continuously. Increased use of EFM both with and without epidural analgesia resulted in women being confined to bed more often, since remote methods (telemetry) of monitoring fetal heart rates were rarely used.

Overall rates of assisted vaginal deliveries at the 2 hospitals were not different, although forceps were used more frequently at BC Women’s and vacuum extraction at Burnaby. More frequent use of cesarean delivery at BC Women’s was not associated with an improvement in selected maternal/newborn outcomes (postpartum hemorrhage and 5-minute APGAR score <7). Postpartum infections were significantly more common at BC Women’s (P <.0001).

Data were analyzed for each hospital separately to determine all the factors contributing to cesarean delivery in each Table 3. After regressing all the demographic factors in a multiple regression model with cesarean delivery as the dependant variable for each hospital, only age remained significant. To evaluate the role of obstetric interventions for each hospital each intervention was entered separately, the most significant retained, and the process repeated with the next most significant retained until no other variables remained significantly associated with cesarean delivery. Age was retained in all the models.

In Burnaby, patient age, cervical dilatation on admission to the hospital, induction, epidural analgesia, and augmentation with oxytocin were important. At BC Women’s the same factors were important, as well as race/ethnicity (white vs nonwhite). Use of electronic fetal monitoring, Entonox, and intravenous or intramuscular narcotics did not predict cesarean delivery in either setting.

To compare the 2 hospitals in cesarean delivery rate we created a model combining data from both institutions. With both in the model, we could quantify the risk for cesarean delivery in one hospital compared with the other. First we examined the role of demographic factors in accounting for differences in cesarean delivery rate. Adjusting for age, the odds ratio (OR) for cesarean delivery at BC Women’s compared with Burnaby was 3.4 (95% confidence interval [CI], 2.1-5.4).

Cervical dilatation on admission, oxytocin augmentation, and epidural analgesia remained statistically significant in addition to age for predicting cesarean delivery. When age, augmentation, and cervical dilatation were retained in the model there remained a statistically significant difference between the 2 hospitals; the OR for BC Women’s versus Burnaby for cesarean delivery was 3.8 (95% CI, 2.3- 6.3). When we adjusted for epidural in addition, the OR for BC Women’s versus Burnaby dropped to 2.0 (95% CI, 1.09-3.5); the difference became only marginally statistically significant. The principal factor associated with the difference in cesarean delivery rate between the 2 hospitals, therefore, was use of epidural analgesia.

Phase II

Our sample consisted of 24 laboring women at Burnaby and 26 laboring women at BC Women’s. The groups were comparable in mean age, marital status, employment, and parity. There were more white women: 18 (75%) in the Burnaby group compared with 11 (42.3%) in the BC Women’s group.

Almost all the women were accompanied by support persons, 24 (100%) at Burnaby and 25 (96.2%) at BC Women’s. The majority at Burnaby had been to childbirth education classes (18 [78.3%] vs 14 [58.3%] at BC Women’s).

It was not possible to observe all the women throughout the entire length of labor. Given this reality and the fact that length of labor differed between women, observations related to management of labor were expressed as rates (ie, the number of minutes spent in bed divided by the total observation time in hours). Women in the 2 hospitals spent similar amounts of time in bed and in the shower. Women at Burnaby spent significantly more time walking, 12 minutes per hour compared with 4.7 minutes per hour at BC Women’s (P=.01, Table 4). There was no difference between the 2 hospital groups in the amount of physical contact with the laboring woman and her nurse or support person. At BC Women’s, women were more often accompanied by a nurse, physician, medical student, or resident and less often left alone than at Burnaby. Women at BC Women’s were exposed to greater numbers of different caregivers.

There were no differences between the groups in the number of couples who expressed specific expectations of labor Table 5. Ten women in Burnaby and 8 in BC Women’s were observed to ask for help with pain management, and responses to these requests included general encouragement or specific verbal advice as to what method they should use. Women were more frequently offered epidural analgesia at BC Women’s; there were no differences in rates of offers for other pain management options. This question related to a preemptive discussion about pain management. In terms of the pain management that actually took place, there was a significant difference in the number of doses of narcotic, in that no women at BC Women’s received narcotics. There was a trend toward more exposure to Entonox and epidural at BC Women’s, although these differences were not significant.

Discussion

Institutional differences in cesarean birth rates have previously been associated with differences in maternity care practices.⁵ The time during labor (state of cervical dilatation) at which women are admitted to the hospital has been associated with cesarean delivery rate.⁶

Age as a risk factor for cesarean delivery has also been well documented. The influence of age appears to be important independent of risk and remains important even among women without risk factors.^7-9 The reason for this is unknown but may point to an association with physician bias or age-related biological factors in the labor process that remain unmeasured.^10,11

The need for oxytocin augmentation has been associated with cesarean delivery.¹² Randomized controlled trials, however, have not demonstrated an association of early use of oxytocin augmentation with cesarean delivery reduction.¹³ Augmentation did not contribute significantly to the model examining differences in cesarean delivery between hospitals when an epidural was included.

A meta-analysis by Halpern and colleagues¹⁴ evaluated the association of epidural analgesia and cesarean delivery. In that report there was no increase in risk of cesarean delivery associated with use of epidural versus narcotic analgesia (OR=1.28; 95% CI, 0.55-2.93 for nulliparous women; OR=0.83; 95% CI, 0.22-3.15 for multiparous women). Findings from the intention-to-treat analysis that could not overcome the limitation of high rates of crossover or noncompliance among some of the trials^15-17 are in contrast with a protocol-compliant analysis in some of the studies that did show an effect of epidural on cesarean delivery rates.¹⁵ Although the protocol-compliant analysis provides important information, it does not necessarily indicate that epidural analgesia is the cause of higher cesarean rates. Subjects having more difficult labors may be more likely to cross over to epidural analgesia from the narcotic arm of the study, increasing the potential need for cesarean delivery in the epidural arm. In the meta-analysis the cesarean delivery rate among women receiving epidurals was 8.2%,¹⁴ much lower than at BC Women’s and more in line with those at Burnaby. The results, therefore, may not be generalizable to BC Women’s, at which high cesarean delivery rates would indicate greater opportunity for changes in intrapartum management, including use of epidural analgesia, to influence cesarean delivery rate.

An important aspect of our work addressed in phase II, our observation study, considered characteristics of women choosing delivery at either hospital that might also influence their choice to use an epidural. Although there were differences in ethnic background between the 2 study groups, these differences did not explain the difference in cesarean delivery rate between the 2 hospitals in our larger study. Although the differences were not statistically significant, more women at Burnaby appeared to have attended prenatal classes. The literature to date has not demonstrated an effect of prenatal education on cesarean delivery.¹⁸

Our observational study failed to suggest differences in client expectations between the 2 hospitals. Features of labor management did differ, however, in that women in Burnaby ambulated more, had fewer types and numbers of caregivers in labor and were less often offered an epidural. Nurses in Burnaby may have offered epidural less often because they were aware that anesthetists had to come to the delivery suite from elsewhere in the hospital or from outside the hospital. Alternatively, a cohort study has reported that nurses grouped according to cesarean rate quintiles differed in their recording of psychosocial data and other aspects of nursing care.¹⁹ Aspects of nursing practice may influence use of epidural analgesia. Our phase II study is limited by its small size, and other additional effects of nursing practice on labor cannot be ruled out. A large randomized controlled study is under way to examine the association of aspects of nursing care (including consistency of nursing caregiver) with cesarean delivery rates.²⁰ Ultimately, the decision to undertake a cesarean delivery resides with the obstetrician. It is possible that in the smaller community hospital physicians may practice with a greater degree of cohesion, perhaps influenced more readily by the philosophy of opinion leaders who advocate a more conservative approach to cesarean delivery. In addition, the presence of obstetric and family practice residency programs at BC Women’s may have encouraged use of interventions, including cesarean the delivery.

The role of ambulation in cesarean delivery remains controversial. A randomized trial failed to demonstrate an association of ambulation with cesarean delivery, but this study had a 22% crossover rate and enrolled women after they had attained a cervical dilatation rate of 3 to 5 cm, when early ambulation may have already exerted a positive effect.²¹ Among low-risk parturients cared for by midwives and not requiring either augmentation of labor or epidural analgesia, ambulation has been associated with a reduction in cesarean delivery rates.²²

Limitations

Our study is limited by lack of knowledge of women’s levels of anxiety and pain during their labor. This would have allowed us to gauge whether less frequent use of epidural analgesia at Burnaby was associated with a cost of diminished satisfaction with the childbirth experience.

We were fortunate in being able to study 2 institutions that were only 20 minutes’ driving distance apart and that served populations from which we were able to sample women who were demographically comparable. Our retrospective cohort analysis identified use of an epidural as the main predictive factor differentiating cesarean delivery rates between the hospitals. It is not possible to determine cause and effect from this retrospective study design, however, and use of epidural analgesia may be a proxy for other unmeasured variables, such as physician practice, anxiety level among patients, or level of education. Direct observation of intrapartum care in a follow-up study failed to differentiate clients attending either hospital in terms of preparation for or expectations of the labor experience. Differences in some aspects of caregiving, however, including more frequent offering of epidural for pain management, may explain the increased use of epidural at BC Women’s. Factors influencing use of epidural need to be studied more thoroughly to support its appropriate place in an overall strategy for pain management in labor.

Related Resources

• Society of Obstetricians and Gynaecologists of Canada (SOGC) http://www.sogc.org/SOGCnet
• College of Family Physicians of Canada www.cfpc.ca
• Canadian Paediatric Society www.cps.ca

Although the seemingly relentless increase in cesarean delivery rates in North America in the 1970s and 1980s has stabilized in recent years and even reversed slightly in some regions,¹ rates are still generally considered to be too high, and efforts to diminish them continue.² In spite of accumulating evidence about the preventable causes of cesarean delivery, individual institutions have had difficulty changing practice.³ At BC Women’s Hospital, the largest maternity facility in Canada with 7500 births annually, the cesarean delivery rate in 1995 was 22.9%.⁴ Although BC Women’s is the referral/tertiary center for the province of British Columbia, 80% to 85% of the births occur to low-risk women in the surrounding catchment area of Vancouver. Approximately 30 obstetricians and 100 family practice physicians have admitting privileges. At nearby Burnaby Hospital, a community general hospital providing primary- and secondary-level maternity care, cesarean delivery rates were 10.3%.⁴ At Burnaby, 3 obstetricians and 38 family practice physicians deliver 2000 babies annually. At the time of our study, neither hospital had practicing midwives. The discrepancy in cesarean delivery rates between hospitals only 20 minutes’ driving distance apart afforded us the opportunity to look at differences in practice between the institutions.

Methods

Phase I

In phase I we completed a retrospective cohort study using hospital records. Records were considered eligible if parturient women met the following criteria: nulliparous, maternal age 16 to 35 years, singleton gestation, and gestation of 37 completed weeks or longer. We excluded women with known pregnancy complications.

Charts were selected from a computer-generated list that incorporated inclusion and exclusion criteria and were then reviewed by a research nurse. Blinding as to hospital was not possible. Using this system, 430 nulliparous women from each hospital were selected consecutively. The sample size was calculated to be 430 per group to have 80% power with type I error set at 0.05, to determine a difference in cesarean delivery rate of 30% from a baseline of 23%. We examined the role of demographic and obstetric factors in the association between hospital and cesarean delivery rate. Maternal and fetal outcomes were measured, including postpartum hemorrhage and infection, as were APGAR scores at 5 minutes. We used logistic regression to simultaneously adjust for confounding factors.

Phase II

Since we were not able to randomly allocate women to hospitals, it was important to determine if women giving birth at BC Women’s compared with Burnaby selfselected to either of the hospitals because of differing expectations for pain management in labor. We also wanted to discern if there were aspects of the care given during the intrapartum period that might explain differing rates of use of epidural analgesia. In the phase II observation study, a research assistant was assigned to observe consecutive women giving birth 2 days per week at each hospital. Our research assistant was a medical student who had not practiced in either hospital. Structured observations were recorded along with times for each observation period and the stage of labor at which the observation took place. The same inclusion and exclusion criteria were used as in our phase I retrospective study to ensure that we were observing comparable women without preexisting risk factors for cesarean delivery at the onset of labor. The phase II study did, however, include multiparous women to maximize use of the time the student was available to assist with the study.

Analysis

We compared demographic characteristics and rates of selected outcomes between hospitals, using the chi-square statistic for categoric variables and the Student t test statistic for continuous variables. When expected cell frequencies were less than 2, we used the Fisher exact test for categoric variables instead of the chi-square statistic. The type I error, 2-sided a, was set at 0.05. Multivariate analysis was undertaken using unconditional logistic regression. We obtained maximum likelihood estimates of the odds ratios using the logistic model. We calculated 95% confidence intervals using the estimates of the standard error derived from the model.

Results

Phase I

Comparing cesarean delivery rates from our retrospective study of 857 births, the crude nulliparous rate at BC Women’s was 20.7%, while at Burnaby it was 6.7% (P <.0001). We examined a number of demographic and obstetric factors to determine if they would explain the reason for the increased rate at BC Women’s.

Women at Burnaby were younger, more likely to be single, and white (P <.0001, Table 1). A larger proportion of women at BC Women’s were Asian. Cesarean delivery rates did not differ between white and Asian women (9.1% vs 9.3%) at Burnaby, but there was a marked difference (18% vs 28.6%) at BC Women’s (data not shown). Women arrived earlier in labor at BC Women’s Table 2. The lengths of the first and second stages were longer at BC Women’s, as was the length of time membranes were ruptured. Time from admission to delivery was also significantly increased at BC Women’s.

Rates of obstetric interventions were compared between the 2 settings Table 2. Rate of induction of labor was not different between the 2 hospitals; however, augmentation of labor with oxytocin occurred significantly more frequently at Burnaby than at BC Women’s. Artificial rupture of membranes during labor was performed significantly more often at BC Women’s than at Burnaby. BC Women’s primarily used epidural analgesia, while Burnaby used intramuscular or intravenous administration of meperidine (Demerol). Use of Entonox (nitrous oxygen and oxygen) was also significantly different. Electronic fetal monitoring (EFM) was used for almost all patients at both hospitals; however, at Burnaby nurses were more likely to obtain only a baseline or admission fetal monitoring record, or to monitor intermittently. BC Women’s, on the other hand, was more likely to employ EFM continuously. Increased use of EFM both with and without epidural analgesia resulted in women being confined to bed more often, since remote methods (telemetry) of monitoring fetal heart rates were rarely used.

Overall rates of assisted vaginal deliveries at the 2 hospitals were not different, although forceps were used more frequently at BC Women’s and vacuum extraction at Burnaby. More frequent use of cesarean delivery at BC Women’s was not associated with an improvement in selected maternal/newborn outcomes (postpartum hemorrhage and 5-minute APGAR score <7). Postpartum infections were significantly more common at BC Women’s (P <.0001).

Data were analyzed for each hospital separately to determine all the factors contributing to cesarean delivery in each Table 3. After regressing all the demographic factors in a multiple regression model with cesarean delivery as the dependant variable for each hospital, only age remained significant. To evaluate the role of obstetric interventions for each hospital each intervention was entered separately, the most significant retained, and the process repeated with the next most significant retained until no other variables remained significantly associated with cesarean delivery. Age was retained in all the models.

In Burnaby, patient age, cervical dilatation on admission to the hospital, induction, epidural analgesia, and augmentation with oxytocin were important. At BC Women’s the same factors were important, as well as race/ethnicity (white vs nonwhite). Use of electronic fetal monitoring, Entonox, and intravenous or intramuscular narcotics did not predict cesarean delivery in either setting.

To compare the 2 hospitals in cesarean delivery rate we created a model combining data from both institutions. With both in the model, we could quantify the risk for cesarean delivery in one hospital compared with the other. First we examined the role of demographic factors in accounting for differences in cesarean delivery rate. Adjusting for age, the odds ratio (OR) for cesarean delivery at BC Women’s compared with Burnaby was 3.4 (95% confidence interval [CI], 2.1-5.4).

Cervical dilatation on admission, oxytocin augmentation, and epidural analgesia remained statistically significant in addition to age for predicting cesarean delivery. When age, augmentation, and cervical dilatation were retained in the model there remained a statistically significant difference between the 2 hospitals; the OR for BC Women’s versus Burnaby for cesarean delivery was 3.8 (95% CI, 2.3- 6.3). When we adjusted for epidural in addition, the OR for BC Women’s versus Burnaby dropped to 2.0 (95% CI, 1.09-3.5); the difference became only marginally statistically significant. The principal factor associated with the difference in cesarean delivery rate between the 2 hospitals, therefore, was use of epidural analgesia.

Phase II

Our sample consisted of 24 laboring women at Burnaby and 26 laboring women at BC Women’s. The groups were comparable in mean age, marital status, employment, and parity. There were more white women: 18 (75%) in the Burnaby group compared with 11 (42.3%) in the BC Women’s group.

Almost all the women were accompanied by support persons, 24 (100%) at Burnaby and 25 (96.2%) at BC Women’s. The majority at Burnaby had been to childbirth education classes (18 [78.3%] vs 14 [58.3%] at BC Women’s).

It was not possible to observe all the women throughout the entire length of labor. Given this reality and the fact that length of labor differed between women, observations related to management of labor were expressed as rates (ie, the number of minutes spent in bed divided by the total observation time in hours). Women in the 2 hospitals spent similar amounts of time in bed and in the shower. Women at Burnaby spent significantly more time walking, 12 minutes per hour compared with 4.7 minutes per hour at BC Women’s (P=.01, Table 4). There was no difference between the 2 hospital groups in the amount of physical contact with the laboring woman and her nurse or support person. At BC Women’s, women were more often accompanied by a nurse, physician, medical student, or resident and less often left alone than at Burnaby. Women at BC Women’s were exposed to greater numbers of different caregivers.

There were no differences between the groups in the number of couples who expressed specific expectations of labor Table 5. Ten women in Burnaby and 8 in BC Women’s were observed to ask for help with pain management, and responses to these requests included general encouragement or specific verbal advice as to what method they should use. Women were more frequently offered epidural analgesia at BC Women’s; there were no differences in rates of offers for other pain management options. This question related to a preemptive discussion about pain management. In terms of the pain management that actually took place, there was a significant difference in the number of doses of narcotic, in that no women at BC Women’s received narcotics. There was a trend toward more exposure to Entonox and epidural at BC Women’s, although these differences were not significant.

Discussion

Institutional differences in cesarean birth rates have previously been associated with differences in maternity care practices.⁵ The time during labor (state of cervical dilatation) at which women are admitted to the hospital has been associated with cesarean delivery rate.⁶

Age as a risk factor for cesarean delivery has also been well documented. The influence of age appears to be important independent of risk and remains important even among women without risk factors.^7-9 The reason for this is unknown but may point to an association with physician bias or age-related biological factors in the labor process that remain unmeasured.^10,11

The need for oxytocin augmentation has been associated with cesarean delivery.¹² Randomized controlled trials, however, have not demonstrated an association of early use of oxytocin augmentation with cesarean delivery reduction.¹³ Augmentation did not contribute significantly to the model examining differences in cesarean delivery between hospitals when an epidural was included.

A meta-analysis by Halpern and colleagues¹⁴ evaluated the association of epidural analgesia and cesarean delivery. In that report there was no increase in risk of cesarean delivery associated with use of epidural versus narcotic analgesia (OR=1.28; 95% CI, 0.55-2.93 for nulliparous women; OR=0.83; 95% CI, 0.22-3.15 for multiparous women). Findings from the intention-to-treat analysis that could not overcome the limitation of high rates of crossover or noncompliance among some of the trials^15-17 are in contrast with a protocol-compliant analysis in some of the studies that did show an effect of epidural on cesarean delivery rates.¹⁵ Although the protocol-compliant analysis provides important information, it does not necessarily indicate that epidural analgesia is the cause of higher cesarean rates. Subjects having more difficult labors may be more likely to cross over to epidural analgesia from the narcotic arm of the study, increasing the potential need for cesarean delivery in the epidural arm. In the meta-analysis the cesarean delivery rate among women receiving epidurals was 8.2%,¹⁴ much lower than at BC Women’s and more in line with those at Burnaby. The results, therefore, may not be generalizable to BC Women’s, at which high cesarean delivery rates would indicate greater opportunity for changes in intrapartum management, including use of epidural analgesia, to influence cesarean delivery rate.

An important aspect of our work addressed in phase II, our observation study, considered characteristics of women choosing delivery at either hospital that might also influence their choice to use an epidural. Although there were differences in ethnic background between the 2 study groups, these differences did not explain the difference in cesarean delivery rate between the 2 hospitals in our larger study. Although the differences were not statistically significant, more women at Burnaby appeared to have attended prenatal classes. The literature to date has not demonstrated an effect of prenatal education on cesarean delivery.¹⁸

Our observational study failed to suggest differences in client expectations between the 2 hospitals. Features of labor management did differ, however, in that women in Burnaby ambulated more, had fewer types and numbers of caregivers in labor and were less often offered an epidural. Nurses in Burnaby may have offered epidural less often because they were aware that anesthetists had to come to the delivery suite from elsewhere in the hospital or from outside the hospital. Alternatively, a cohort study has reported that nurses grouped according to cesarean rate quintiles differed in their recording of psychosocial data and other aspects of nursing care.¹⁹ Aspects of nursing practice may influence use of epidural analgesia. Our phase II study is limited by its small size, and other additional effects of nursing practice on labor cannot be ruled out. A large randomized controlled study is under way to examine the association of aspects of nursing care (including consistency of nursing caregiver) with cesarean delivery rates.²⁰ Ultimately, the decision to undertake a cesarean delivery resides with the obstetrician. It is possible that in the smaller community hospital physicians may practice with a greater degree of cohesion, perhaps influenced more readily by the philosophy of opinion leaders who advocate a more conservative approach to cesarean delivery. In addition, the presence of obstetric and family practice residency programs at BC Women’s may have encouraged use of interventions, including cesarean the delivery.

The role of ambulation in cesarean delivery remains controversial. A randomized trial failed to demonstrate an association of ambulation with cesarean delivery, but this study had a 22% crossover rate and enrolled women after they had attained a cervical dilatation rate of 3 to 5 cm, when early ambulation may have already exerted a positive effect.²¹ Among low-risk parturients cared for by midwives and not requiring either augmentation of labor or epidural analgesia, ambulation has been associated with a reduction in cesarean delivery rates.²²

Limitations

Our study is limited by lack of knowledge of women’s levels of anxiety and pain during their labor. This would have allowed us to gauge whether less frequent use of epidural analgesia at Burnaby was associated with a cost of diminished satisfaction with the childbirth experience.

We were fortunate in being able to study 2 institutions that were only 20 minutes’ driving distance apart and that served populations from which we were able to sample women who were demographically comparable. Our retrospective cohort analysis identified use of an epidural as the main predictive factor differentiating cesarean delivery rates between the hospitals. It is not possible to determine cause and effect from this retrospective study design, however, and use of epidural analgesia may be a proxy for other unmeasured variables, such as physician practice, anxiety level among patients, or level of education. Direct observation of intrapartum care in a follow-up study failed to differentiate clients attending either hospital in terms of preparation for or expectations of the labor experience. Differences in some aspects of caregiving, however, including more frequent offering of epidural for pain management, may explain the increased use of epidural at BC Women’s. Factors influencing use of epidural need to be studied more thoroughly to support its appropriate place in an overall strategy for pain management in labor.

Related Resources

• Society of Obstetricians and Gynaecologists of Canada (SOGC) http://www.sogc.org/SOGCnet
• College of Family Physicians of Canada www.cfpc.ca
• Canadian Paediatric Society www.cps.ca