Continuity of Nursing Care and Its Link to Cesarean Birth Rate
This study was partially supported by the Fonds de la recherche en santé du Québec through a career award to Anita J. Gagnon and a student bursary to Katharine M. Meier from the Faculty of Medicine of McGill University, Montreal, Quebec.
Anita J. Gagnon, RN, MPH, PhD, School of Nursing, McGill University, 3506 University Street, Montreal, Quebec, Canada H3A 2A7.
ABSTRACT: Background: High cesarean birth rates are an international concern. The role of patterns of nursing care responsibility in preventing or contributing to cesarean births has been understudied. Our study sought to identify and describe indicators of continuity of nursing care responsibility during labor and to explore whether any association between these indicators and risk of cesarean birth could be identified empirically using an existing data set. Methods: We obtained a representative sample of low-risk women giving birth in an intrapartum unit at a university hospital in Quebec, Canada, with approximately 3,700 births per year. To be considered for inclusion, women needed to have been primiparous, carrying singletons in vertex position, and at 37 weeks’ gestation or more. All women giving birth over a 13-month period were assessed for eligibility using the hospital’s birth log. Data were extracted from the medical records of every second eligible birth, including information related to patterns of nursing care responsibility, maternal and infant characteristics, obstetric procedures, non–health-related risk factors, and type of birth. Results: Data on all variables of interest were available for 467 women. These women were cared for by 1–17 nurses, care responsibility changed hands for them from 1 to 28 times, and the mean length of labor for which the same nurse was responsible for a woman ranged from 10 to 1,045 minutes. After controlling for length of labor, maternal age, maternal height, infant weight, gestational age, induction, type of rupture, and epidural analgesia, the odds ratio for cesarean birth due to number of nurses was 1.17 (95% CI 1.04, 1.32); 1 or more nurses switch per 2 hours (i.e., number of times care responsibilities changed hands), 1.04 (95% CI 0.62, 1.74); and 33 percent or more of the labor attended by the same nurse, 0.74 (95% CI 0.42, 1.30). Conclusions: An association was observed between number of nurses caring for a laboring woman and risk of cesarean delivery. Estimates of the association of other patterns of nursing care responsibility on cesarean birth were not sufficiently precise to draw conclusions. (BIRTH 34:1 March 2007)
High cesarean birth rates are an international concern (1,2). Contributors to these high rates include biological and administrative factors, but the roles of nurses and of patterns of nursing care delivery in risk of cesarean birth are less clear and have been understudied. The nurse’s role in labor and birth varies according to institutional policy, unit practice, and nurse preference. In the last 2 decades, the use of continuous electronic fetal monitoring, epidural analgesia, and artificial rupture of membranes has dramatically increased. These practice changes have a tremendous impact on nursing care.
The amount of support actually provided by nurses to laboring women has been empirically studied using work sampling techniques, in which randomly selected periods of time were chosen for observation and nurses’ activities at those times were recorded (3,4). These studies showed that nurses spent about 7–9 percent of their time providing supportive care to women in labor. Limited emphasis on supportive care in the educational formation of nurses and unit cultures that favor technological expertise may be contributors to small amounts of personal contact. In fact, only one fifth of nurses’ time was spent in the room with the women (3). This fact stands in stark contrast to the continuous nature of care inherent in “one-to-one” support during labor, which has been shown to be effective in decreasing the cesarean birth rate, among other benefits (5).
The idea for the current study came from our having seen the management of laboring women change when nursing care responsibility changed hands. For example, we had noticed that the lunch breaks of certain nurses who had been providing one-to-one nursing care were times when intravenous lines were inserted, stimulation was begun, and amniotomies were performed, to name a few. We believed that these interventions were related to the replacement nurses’ lack of knowledge about how the women were progressing through their labors, since the replacement nurses had not cared for them before the break. There is a dearth of studies on the role of nursing care as an independent predictor of, or protector against, cesarean birth. Only one study directly examined the likelihood that a nurse would influence the delivery outcome of laboring women (6). Births were attributed to a nurse if she had been present at the time of birth as noted in the obstetric log. Nurses who had attended over 15 births were ranked into quintiles of high, middle, and low cesarean birth groups. Results based on 338 births showed that large differences in the cesarean birth rates of nurses in the lowest and highest quintiles could not be explained by differences in maternal age and gravity, attendance at childbirth class, insurance status, reliance on public assistance, physician attendant at labor, use of epidural analgesia, augmentation of labor, cervical dilation when the nurse assumed care, infant weight, or gestational age. The study concluded that, “the nurses’ care during labor is an important factor influencing cesarean birth rates”(6). The strength of these conclusions could not be assessed due to the possibility of analytical bias (the same study data may have been used to both define and test the association, i.e., post hoc analyses).
Unplanned cesarean sections for dystocia in nulliparous women, better and safer surgical techniques, and sophisticated techniques for fetal evaluation have accounted for many cesarean births (7). Extremes in maternal age (8), short stature, very low or very high infant birthweight (9), higher fetal age, and length of labor under 16 hours are positively associated with cesarean section (10).
Several obstetric interventions may increase the risk of cesarean births, such as amniotomy to shorten spontaneous labor (11) and induction of labor with oxytocin (12). Epidural analgesia is associated with greater pain relief than nonepidural methods, with several untoward effects, and may be associated with an increase in cesarean section rates (although this factor remains uncertain because completed randomized controlled trials are too small or do not permit a clear interpretation of their results) (13).
The diagnoses of dystocia and fetal distress leading to cesarean delivery can be subjective (14). Rates of cesarean delivery for these diagnoses in similar populations differ by physicians (15–19). Cesarean sections have been found to be more likely to be performed on Fridays, between 6 am and 6 pm(10), and by physicians with high cesarean birth rates the previous year (15). The decision to operate is also weighted differently at different sites (20) and by different types of practitioners (21).
One effective approach to reduce cesarean births through physician behavior change has been to incorporate key data into a perinatal database, which is subsequently used to provide feedback to the individual physician about his or her practice (22,23). Databases in perinatal care are being, and can be, developed, which include minimal amounts of nursing data that could ultimately serve the same purpose for nurses. Before suggesting that such data be included in computerized perinatal databases, we were interested in determining whether some routinely available “administrative” data could be useful to highlight associations between patterns of nursing care and cesarean birth rates. A data set created for an earlier study provided an opportunity for us to explore this issue.
Research objectives for this study were twofold: (1) to identify and describe indicators of continuity of nursing care during labor from data usually gathered for administrative purposes and (2) to explore whether any association between these indicators and risk of cesarean birth could be identified empirically using an existing data set.
The data set was created from a study conducted in an intrapartum unit at a university teaching hospital in Quebec, Canada, with approximately 3,700 births per year and a cesarean birth rate of 24 percent. Data were gathered retrospectively by medical record review. Approval to conduct the study was received from the hospital Research Ethics Committee before beginning.
The unit provides care to both low- and high-risk women. Most attendants at birth are obstetricians, with fewer general practitioners, no midwives, and few doulas or labor support persons in attendance. Nurses are generally assigned 2 women, one in early labor and one in late labor. During the second stage of labor, care is on a one-to-one basis. Variations in the nurse/patient ratio are dependent on the patient census and on the number of inductions scheduled. Inductions and stimulation of labor are frequent and lead to close monitoring by nursing staff. Continuous electronic fetal monitoring is frequent and epidural analgesia is the most common form of pain relief, although in early labor Demerol may be offered to primiparous women. Nonmedical methods of pain relief such as walking, massage, and changing position are suggested for early labor. Showers are not used. Family members are welcome on the unit. Women who arrive on the unit in the latent phase of labor are encouraged to walk around if the membranes are intact. If the couple is willing, they may be encouraged to return home. As contractions become more regular, bed rest and continuous electronic fetal monitoring tend to be the usual care.
Women meeting the following criteria were included in the study: admission to the intrapartum unit, nulliparous, singleton, vertex presentation, and 37 weeks or more gestational age. Data from the records of 550 women, every other woman giving birth, and meeting initial eligibility criteria between March 1, 1997, and March 31, 1998, were collected. The researchers thought that selection of every other birth would provide data on a reasonable number of births while avoiding systematic bias. This procedure was selected to ensure representation from the entire time period under study.
Administrative data retrieved from the record consisted of the name of each nurse responsible for the woman and the date and time associated with each signature of each nurse. These signatures were recorded approximately every 30 minutes as per unit policy. From this information we were able to create variables that we thought reflected continuity of nursing care responsibility. Continuity of care itself could not be estimated from these data since no records were available of what care occurred between signatures. However, the signing nurse was the person responsible for knowing about the women in whose chart she was recording and therefore would have played a key role in the health care team with respect to decisions about her care. Continuity of nursing care responsibility variables that could be created included the number of nurses caring for each woman during labor, the number of times care responsibilities changed hands (switches), and the length of time each nurse was responsible for the laboring woman. Together with data on length of labor, we were able to calculate the proportion of the total labor that the same nurse was responsible for the laboring woman.
We retrieved data on known predictors of cesarean birth, including total length of labor, infant birthweight, day of the week of birth, maternal age, gestational age, method of delivery, physician attendant during pregnancy and birth, mother’s height, type of rupture of membranes, epidural analgesia administration, stimulation, and induction. Records not included in the study sample were used for training, during which time data retrieval by 2 of the authors (K.M.M. and K.W.) was performed until 100 percent agreement was consistently reached. For study participants, data were collected by K.M.M. and repeated by K.W. on a 5 percent random sample of the records. Clarification on nurse signatures was sought from the nurse manager and/or unit clerk. All data were entered into the Microsoft Access database program (24).
Descriptive analyses were performed on all variables, outliers examined, and variables categorized when needed to conform with requirements for analytical assumptions. Bivariate analyses with cesarean birth were performed to reduce the number of variables included in the maximum logistic regression models (25).
The 3 variables describing nursing care responsibility (i.e., number of times responsibility for care changed hands, or “switches,”≥1 per 2 hours; number of nurses per labor; same nurse responsible for ≥33% of labor) were incorporated into 3 separate logistic regression models with other known predictors of cesarean birth (8–13). These predictors included maternal and infant characteristics (labor length > 9 hr, maternal age > 35 yr, maternal height < 157.5 cm, infant weight > 4,000 g, gestational age > 40 wk) and obstetric intervention risk factors (induction, type of rupture of membranes, epidural analgesia). In each of the maximum models, 1 of the 3 variables describing care responsibility patterns was forced into the model.
All variables were defined to be clinically meaningful and to maximize statistical efficiency by ensuring adequate cell sizes. Variables were removed from the maximum models when their removal resulted in a change in the beta coefficient for the nurse variable of less than 10 percent, otherwise they remained in the model to control for confounding (25). Model sizes were based on requiring 10 women for each variable in the model plus the intercept (26). Given a total of 85 women who experienced a cesarean section, we attempted to define our maximum models with between 6 and 7 predictors plus the intercept. However, we were successfully able to work with 9 variables in our maximum models and to reduce them to models with between 6 and 7 predictors. The assumption of linearity in the logit was assessed as required for logistic regression analyses.
Five hundred ninety-six maternal records were identified from the log, 550 records were reviewed, and 467 were confirmed to have met study inclusion criteria and were not missing data. Overall, the mean number (SD) of nurses per woman was 5.4 (2.4), range 1–17; the mean number of switches (SD) of nurses per woman was 7.4 (4.5), range 1–31; and the mean length of labor (SD) that the same nurse stayed with a woman was 343 minutes (126), range 10–1,045. A description of the population can be seen in Table 1. Data on these women formed the basis for all logistic regression analyses.
Table 1. Women Included in All Final Models: Descriptors (n = 467)
|Maternal age (yr)||27.5 (5.3)|
|Maternal height (cm)||162.3 (7.3)|
|Gestational age (wk)||39.7 (1.2)|
|Stimulated with oxytocin||45.2%|
|Artificial rupture of the membranes||51.2%|
|Length of labor (hr)||14.0 (7.7)|
|Birthweight (g)||3,378 (471.8)|
Results of final logistic regression models are shown in Table 2. An independent association of the number of nurses caring for a woman in labor with cesarean birth was found even after controlling for factors known to be strong predictors of cesarean birth (length of labor, older maternal age, short maternal stature, large-for-gestational-age infant, and induction). Each additional nurse caring for the same laboring woman was associated with a 17 percent greater risk of cesarean section, with a range of risk of 4–32 percent greater. Associations of other patterns of care variables with cesarean birth were not sufficiently precise to be informative.
Table 2. Association of Patterns of Nursing Care Responsibility and Risk of Cesarean Birth: Results of Final Logistic Regression Models (n = 467)
|Switches ≥1 per 2 hr||1.04||0.62, 1.74|
|Number of nurses*||1.17||1.04, 1.32|
|Same nurse responsible for ≥33% of labor||0.74||0.42, 1.30|
In this exploratory retrospective study of the association of patterns of nursing care responsibility with cesarean birth, each additional nurse responsible for a woman in labor increased the woman’s risk of cesarean birth from 4 to 32 percent. These results were observed while controlling for maternal, infant, and obstetric intervention risks for cesarean section. Indicators of continuity of care were readily created from routinely collected administrative data. They suggest that a high degree of fragmentation of nursing care responsibility during labor exists.
The current study adds to the literature in the field because it offers both a different approach to examining how nursing practice affects health outcomes and a strategy that could potentially be used to provide nurses with feedback on their practice and its health effects through the use of routinely collected data. We tried unsuccessfully to identify additional literature on continuity of nursing care and obstetric outcome. Small studies examining maternal perceptions of nursing care during labor in the United States have been conducted, but none examined continuity of care responsibility directly (27,28). The report of a large North American trial of nursing care in labor did not describe nursing care characteristics in either arm of the trial (29).
The fact that only 1 of the 3 indicators of patterns of nursing care responsibility was significantly associated with risk of cesarean birth is intriguing. Given that the 3 indicators are very closely related, someone might have thought that if one showed an effect, so too would the others. The lack of effect of the “number of switches,” however, might be explained by the fact that several switches could have occurred among few nurses (e.g., nurses who are paired) such that a relatively high level of continuity of responsibility remained. Having the “same nurse responsible for ≥33 percent of the labor” likely provided some of the information already included in the “number of nurses,” whereas the latter included additional information.
Methodological strengths of our study included ensuring an adequate sample size, using medical record sampling techniques that maximized generalizability of results, obtaining data on covariates known to be associated with cesarean birth risk (except race and ethnicity), and conducting appropriate statistical procedures with attention to the verification of underlying statistical assumptions.
Certain limitations to this study are present. Nurses were identified by their signatures on the record; this identified the nurse as being responsible for care during a certain time period, not what was done during that time. We were also unable to include race and ethnicity as a predictor of cesarean birth.
Nurse managers may wish to support creation of a computerized administrative database that can be used to inform uses of their practice and to permit ongoing reinforcement of administrative trends and patient outcomes. The level of fragmentation of care responsibility in this university hospital setting identified through such a database suggests that greater attention should be paid to patterns of care responsibility; it is difficult to envision any benefit of reduced continuity. Nurse managers could begin by systematically determining the extent of continuity of care responsibility on their units and contributing factors. If high levels of continuity of responsibility are identified, positive reinforcement would be warranted. Solutions to fragmentation (if found), developed in conjunction with staff, would be appropriate. It may be most appropriate if nurses are paired with one another so that they “cover” each other during times that they are off the unit; positive reinforcement could be given for maintaining the assigned pairing. Nurses may wish to inform women about their expected availability during the labor, and required care could be planned jointly for times when the nurse is expected to be absent. An adequate quantity of supplies might be kept near patient areas to reduce the need for being off the unit. Flexible scheduling, which could enable nurses to remain with women for the duration of their labor, might be explored.
This study highlights the feasibility and importance of examining the association of nursing care responsibility patterns with health outcomes. More research on the effects of specific nursing care activities in their “usual” context needs to be conducted. Research in settings where “primary nursing” (i.e., case based) is in place may facilitate such an examination. Incorporation of routinely collected administrative data into computerized databases is a reasonable point of departure for exploring this issue.
An association was observed between number of nurses caring for a laboring woman and risk of cesarean delivery. Estimates of the association of other patterns of nursing care responsibility on cesarean birth were not sufficiently precise to draw conclusions.
The authors would like to thank Tamara Rader for her support in conducting the study and Yongjun Gao for conducting the analyses.