Abstract
Background:
Understanding and improving obstetric quality and safety is an important goal of professional societies, and many interventions such as checklists, safety bundles, educational interventions, or other culture changes have been attempted to improve the quality of care provided to obstetric patients. Although many factors contribute to delivery decisions, less work has addressed how provider issues such as fatigue or behaviors surrounding impending changes in shift may influence delivery mode and outcomes.
Objective:
The objective was assess whether intrapartum obstetric interventions and adverse outcomes differ based on temporal proximity of delivery to attending shift change.
Methods:
This was a secondary analysis from a multicenter obstetric cohort in which all patients with cephalic, singleton gestations who attempted vaginal birth were eligible for inclusion. The primary exposure used to quantify the relationship between the proximity of provider to their shift change and a delivery intervention was a ratio of (time from most recent attending shift change to vaginal delivery or decision for cesarean delivery) over (total length of shift). Ratios were used to represent the proportion of time completed in the shift, while standardizing for varying shift lengths. A sensitivity analysis restricted to patients delivered by physicians working 12-hour shifts was performed. Outcomes chosen included cesarean delivery, episiotomy, 3rd or 4th degree perineal laceration, 5-minute Apgar score <4, and neonatal intensive care unit admission. Chi-squared tests were used to evaluate outcomes based on proportion of attending shift completed. Adjusted and unadjusted logistic models fitting a cubic spline (when indicated) were used to determine whether the frequency of outcomes throughout the shift occurred in a statistically significant non-linear pattern
Results:
Of 82,851 patients eligible for inclusion, 47,262 (57%) had available ratio data and constituted the analyzable sample. Deliveries were evenly distributed throughout shifts, with 50.6% taking place in the first half of shifts. There were no statistically significant differences in the frequency of cesarean delivery, episiotomy, 3rd or4th degree perineal laceration, or 5-minute Apgar score <4 based on proportion of shift completed. Findings were unchanged when evaluated with a cubic spline in unadjusted and adjusted logistic models. Sensitivity analyses performed on the 22.2% of patients who were delivered by a physician completing a 12-hour shift showed similar findings. There was a small increase in the frequency of neonatal intensive care unit admission with greater proportion of shift completed (adjusted p=0.009), but findings did not persist in sensitivity analysis.
Conclusions:
Clinically significant differences in obstetric interventions and outcomes do not appear to exist based on temporal proximity to attending physician shift change. Future work should attempt to directly study unit culture and provider fatigue in order to further investigate opportunities to improve obstetric quality of care, and additional studies are needed to corroborate these findings in community settings.
Keywords: adverse perinatal outcomes, obstetrical interventions, provider fatigue, quality of care, quality improvement, shift change
INTRODUCTION
Understanding and improving obstetric quality and safety is an important goal of professional societies, and many interventions such as checklists, safety bundles, educational interventions, or other culture changes have been attempted to improve the quality of care provided to obstetric patients.1,2 Maternal safety bundles support, for example, building a maternity unit culture that promotes spontaneous progress of labor and vaginal birth as well as protocols for standardized management of labor.2 Such efforts are thought to be important steps towards improving maternal outcomes and experiences. One aspect of “unit culture” that warrants exploration in the effort to promote quality is provider decision making.
Provider decision making in the obstetric context remains an understudied aspect of obstetric quality of care. Some literature has suggested provider call system may be associated with changes in practice patterns, with data showing some obstetric interventions decrease and outcomes improve when providers are on a night float system, suggesting fatigue or scheduling may be associated with decision making.3,4 Other work has demonstrated that physician cognitive factors are associated with their patients’ delivery outcomes.5,6 In contrast, findings from the Maternal-Fetal Medicine Units (MFMU) Network Cesarean Registry identified no major differences in morbidity for cesarean deliveries performed during physician change of shift.7 Less work has addressed how provider behaviors surrounding impending changes in shift may influence delivery mode and outcomes.
Other existing research on the intersections of outcomes, quality of care, and delivery timing largely focuses on timing of delivery with regard to day versus night.8–11 Although this approach addresses issues of resource availability and utilization, it does not fully capture other aspects of quality of care, such as provider behavior or fatigue, as it cannot incorporate concepts of duration of work. We hypothesized that differences in provider behavior may occur at periods closer to shift change, resulting in differential frequency of interventions or adverse outcomes. Thus, the objective of this analysis was to assess differences in intrapartum obstetric interventions and adverse outcomes based on temporal proximity of delivery to the attending provider’s shift change in a large and diverse cohort of patients undergoing trial of labor. This analysis considered the proportion of time completed by the provider within their working shift as the exposure of interest.
METHODS
This is a secondary analysis of data from the Assessment of Perinatal Excellence (APEX) study. APEX was a large, multicenter observational study performed at 25 hospitals in the United States that consisted of 115,502 patients and their neonates who delivered at participating hospitals (2008–11). The cohort included patients with diverse sociodemographic characteristics who were delivered by an array of academic and community general obstetricians, family medicine physicians, midwives and maternal-fetal medicine physicians.12,13 Detailed demographic and clinical data for patients who were at least 23 weeks’ gestation and who had arrived at the hospital with a live fetus were collected by trained research personnel at each site. Maternal characteristics, details of the medical and obstetric history, intrapartum care, obstetric outcome, and indications for operative procedures were collected. The data collected at each site included the time of delivery and time of provider shift changes at each site, including whether such shift changes differed by weekend versus weekday status. Details of the study methodology have been reported previously.12,13
For this analysis, we included patients with cephalic, singleton births at any gestational age who underwent a trial of labor. Patients who had an intrauterine fetal demise, known fetal anomalies, multiple gestation, contraindications to trial of labor, prior cesarean delivery, or a scheduled cesarean delivery were excluded. Additionally, patients who delivered within 15 minutes of arrival to the labor and delivery unit were excluded given the scant amount of time that they were managed by the delivering attending provider.
In the present analysis, we selected the outcomes to include cesarean delivery, episiotomy, major perineal lacerations (3rd or 4th degree perineal laceration), 5-minute Apgar score <4, and neonatal intensive care unit (NICU) admission. Cesarean delivery and episiotomy were selected because they are obstetric outcomes that were hypothesized to potentially be associated with the time in shift.3 For example, an attending preparing to end a shift may have a greater propensity to recommend cesarean delivery or episiotomy due to the desire to not “sign out” a patient with protracted labor or prolonged second stage to the oncoming provider; under some compensation models, such a desire may be driven by financial implications if that provider will receive less revenue by signing out the delivery. Similarly, major perineal lacerations, low 5-minute Apgar score, and NICU admission were selected as adverse outcomes that may plausibly be associated with time in shift for similar reasons related to the propensity to intervene in the second stage (e.g. via performing forceps-assisted vaginal delivery) to hasten delivery before shift change or to not intervene (e.g. due to provider fatigue, deciding not to act on a potentially non-reassuring fetal heart tracing or labor dystocia in anticipation of another provider arriving to assume care) in order to await an upcoming shift change. While the former example could be associated with differential patterns in interventions or outcomes at the end of the shift, the latter example could be associated with clusters of interventions or outcomes at the beginning of the shift. For cesarean delivery, the chosen time was time of the decision; for all other outcomes, the assigned time was the time of delivery.
The primary exposure for this analysis was the proportion of time completed by a given provider within their working shift. Because provider shift lengths varied by institution as well as by day of the week, we quantified the proportion of time in the shift (i.e., the proximity to shift change) as a ratio of (time from most recent attending shift change to vaginal delivery or decision for cesarean delivery) over (total length of shift). Ratios represent the proportion of time completed in the shift, standardized for varying shift lengths. For bivariable analyses, the proportion of time-in-shift completed was analyzed in 5% blocks; blocks of 5% were chosen in order to capture the small changes that can occur throughout the shift. With this method, for example, the time period of “20–25%” indicates that 20% of the shift has occurred and 80% of the shift remains. The “50%” block represents the mid-point of the shift. For multivariable models, the proportion of time completed in the shift was analyzed as a continuous variable with splines to take into account the non-linear relationship possibility. Given the variability in shift lengths and the potential differences that may exist based on shift length differences, a sensitivity analysis was performed in which only patients delivered by physicians working 12-hour shifts was performed. Although resident physicians were involved in some deliveries in this cohort, for this analysis, the exposure of proportion of shift completed was restricted to the shift of the attending obstetrician at delivery. Patients delivered by midwives were not included, as the attending physicians are considered the attending of record for operative delivery and thus the inclusion of midwives could contribute to bias.
Baseline characteristics of the population were reported descriptively and compared using chi-square tests or Wilcoxon tests based on whether delivery was in the first half versus second half of the shift. The chi-square test and Cochran-Armitage test for trend were used to evaluate outcomes based on proportion of attending shift completed. We then performed logistic models fitting cubic splines to determine whether the frequency of outcomes throughout the shift occurred in a statistically significant non-linear pattern. Cubic splines were used in order to model the non-linear data and find a smooth function for the outcomes where the assumption of logistic models that the independent variable be linearly related to the log odds was not met. When outcomes did meet that assumption, logistic models without a cubic spline were fit. Models were also evaluated adjusting for the baseline covariates shown in Table 1, which were chosen a priori based on hypothesized associations with the outcomes. Imputation for missing data was not performed. SAS software (SAS Institute, Cary, NC) was used for the analyses. All tests were two-tailed. For descriptive analysis, a p-value of <0.05 was used to define statistical significance. For analyses evaluating outcomes, a p-value of <0.01 was used to account for multiple comparisons. Institutional review board approval was obtained from all participating centers under a waiver of informed consent.
Table 1:
Demographic and clinical characteristics of patients by provider shift completion
| Characteristics | Overall 47,262 | <50% shift completed n=23,935 | ≥50% shift completed n=23,327 | pvaluea |
|---|---|---|---|---|
| Age (y) | 27.2±6.2 | 27.2±6.2 | 27.2±6.2 | .91 |
| Race and ethnicity | ||||
| Non-Hispanic white | 18,566 (39.3) | 9368 (39.1) | 9198 (39.4) | |
| Non-Hispanic black | 11,896 (25.2) | 5962 (24.9) | 5934 (25.4) | .31 |
| Non-Hispanic Asian | 2643 (5.6) | 1373 (5.7) | 1270 (5.4) | |
| Hispanic | 11,999 (25.4) | 6118 (25.6) | 5881 (25.2) | |
| Other or unknown | 2158 (4.6) | 1114 (4.7) | 1044 (4.5) | |
| Insurance status | ||||
| Private | 20,710 (43.8) | 10,484 (43.8) | 10,226 (43.8) | |
| Government-assisted | 20,433 (43.2) | 10,337 (43.2) | 10,096 (43.3) | .92 |
| Uninsured | 6119(13.0) | 3114(13.0) | 3005(12.9) | |
| BMI at delivery (kg/m2) | 31.1±6.2 | 31.0±6.2 | 31.1±6.3 | .31 |
| Gestational age at delivery (wk) | ||||
| 23.0–27.6 | 178 (0.4) | 82 (0.3) | 96 (0.4) | .02 |
| 28.0–33.6 | 872 (1.9) | 439 (1.8) | 433 (1.9) | |
| 34.0–36.6 | 2981 (6.3) | 1530 (6.4) | 1451 (6.2) | |
| 37.0–38.6 | 12,370 (26.2) | 6399 (26.7) | 5971 (25.6) | |
| 39.0–40.6 | 25,889 (54.8) | 13,043 (54.5) | 12,846 (55.1) | |
| ≥41.0 | 4972 (10.5) | 2442 (10.2) | 2530 (10.9) | |
| Nulliparous | 22,322 (47.2) | 11,418 (47.7) | 10,904 (46.7) | .04 |
| Induction of labor | 17,480(37.0) | 8726 (36.5) | 8754 (37.5) | .02 |
Data are presented as n (%) or mean±standard deviation unless otherwise noted.
BMI, body mass index.
Yee et al. Obstetrical care and provider shift change. Am J Obster Gynecol 2021.
RESULTS
Of 115,502 in APEX, 80,842 patients were eligible for inclusion. In this cohort, 47,262 (58.5%) had available ratio data for time completed in shift and constituted the analyzable sample. Missing ratio data was largely due to lack of reporting of shift change time at 7 sites. Deliveries were evenly distributed throughout shifts, with 50.6% taking place in the first half of shifts (Figure 1). Almost one quarter (22.2%) of patients were delivered by physicians who worked 12-hour shifts, whereas 13.8% of patients were delivered by physicians who worked 24-hour shifts; the remainder had shifts lasting between 8 and 16 hours, but that were not 12-hour (Figure 2). At the included sites, the median number of residents present on labor and delivery was 4 on weekdays and 3 on weekends. The median hospital volume of births per year was 4,430 (range 2,332–14,032).
Figure 1.
Distribution of deliveries across shifts
Figure 2.
Histogram of total shift time across all eligible deliveries
The mean age of individuals in this cohort was 27.2 years (standard deviation 6.2), and 39.3% identified as non-Hispanic White (Table 1). There were no differences in age, race and ethnicity, insurance status, or body mass index at delivery based on whether the delivery occurred in the first or second half of the attending shift. There was a small but statistically significant difference in the distribution of gestational age at birth based on time completed in shift (p=0.02), with patients who delivered in the first half less likely to have very preterm births. Patients who delivered in the first half of the shift were also slightly more likely to be nulliparous and less likely to have undergone induction of labor (Table 1).
Most of the outcomes did not significantly vary by time in shift (Table 2). For cesarean delivery, the overall range was small, with peaks in cesarean frequency of 15.9% at 65–70% of completed shift and 15.6% at 95–100% of completed shift (i.e., immediately before shift change); the nadir cesarean frequency was 13.2% at 80–85% of completed shift. These frequencies did not vary significantly (p=0.61) with the shift proportion. Similarly, episiotomy frequency did not vary significantly (p=0.44), ranging from a nadir of 8.3% (60–65% of shift completed) to a peak of 10.7% (30–35% of shift completed). The frequency of major perineal laceration ranged from 2.2% to 3.5% and did not vary by time in shift (p=0.57) and 5-minute Apgar score <4 varied from 0.04% to 0.47% frequency, also not varying by time in shift (p=0.13) (Table 2). The frequency of NICU admission also varied little over the shift (nadir 9.3% at 0–5% of shift completed, peak 11.7% at 75–80% of shift completed), although this pattern did achieve statistical significance (p=0.01).
Table 2.
Frequency of obstetric interventions and adverse outcomes by proportion of time in shift completed
| % completed in shift | Cesarean | Episiotomy | Major perineal laceration | 5-minute Apgar < 4 | Neonatal intensive care unit admission |
|---|---|---|---|---|---|
|
| |||||
| 0–5% | 335 (14.1) | 226 (9.5) | 75 (3.2) | 2 (0.08) | 222 (9.3) |
| 5–10% | 351 (14.4) | 223 (9.1) | 63 (2.6) | 1 (0.04) | 235 (9.6) |
| 10–15% | 362 (14.7) | 242 (9.8) | 73 (3.0) | 9 (0.37) | 254 (10.3) |
| 15–20% | 330 (14.2) | 211 (9.1) | 75 (3.2) | 7 (0.30) | 235 (10.1) |
| 20–25% | 376 (15.3) | 235 (9.6) | 66 (2.7) | 1 (0.04) | 232 (9.4) |
| 25–30% | 342 (14.7) | 215 (9.2) | 69 (3.0) | 7 (0.30) | 254 (10.9) |
| 30–35% | 373 (15.8) | 253 (10.7) | 70 (3.0) | 11 (0.47) | 243 (10.3) |
| 35–40% | 341 (14.2) | 243 (10.1) | 66 (2.7) | 5 (0.21) | 234 (9.7) |
| 40–45% | 344 (14.0) | 237 (9.6) | 85 (3.5) | 8 (0.32) | 266 (10.8) |
| 45–50% | 328 (14.2) | 246 (10.6) | 67 (2.9) | 4(0.17) | 241 (10.4) |
| 50–55% | 357 (14.5) | 232 (9.4) | 70 (2.8) | 7 (0.28) | 261 (10.6) |
| 55–60% | 320 (13.9) | 215 (9.3) | 75 (3.3) | 6 (0.26) | 218 (9.5) |
| 60–65% | 340 (14.8) | 190 (8.3) | 51 (2.2) | 6 (0.26) | 250 (10.9) |
| 65–70% | 367 (15.9) | 212 (9.2) | 73 (3.2) | 2 (0.09) | 225 (9.8) |
| 70–75% | 335 (14.5) | 199 (8.6) | 68 (2.9) | 5 (0.22) | 248 (10.7) |
| 75–80% | 326 (14.1) | 206 (8.9) | 58 (2.5) | 7 (0.30) | 270 (11.7) |
| 80–85% | 307 (13.2) | 216 (9.3) | 56 (2.4) | 4(0.17) | 251 (10.8) |
| 85–90% | 359 (14.8) | 223 (9.2) | 72 (3.0) | 8 (0.33) | 258 (10.6) |
| 90–95% | 337 (14.8) | 211 (9.2) | 56 (2.5) | 4(0.18) | 256 (11.2) |
| 95–100% | 359 (15.6) | 198 (8.6) | 57 (2.5) | 9 (0.39) | 237 (10.3) |
| Overall | 6889 (14.6) | 4433 (9.4) | 1345 (2.8) | 113 (0.24) | 4890 (10.4) |
| p-value (chi-sq) | 0.61 | 0.44 | 0.57 | 0.13 | 0.43 |
| p-value (trend) | 0.74 | 0.07 | 0.10 | 0.27 | 0.01 |
Date are n (%) unless otherwise noted
Logistic models with cubic splines did not show statistically significant non-linear patterns for cesarean delivery, episiotomy, major perineal laceration, or 5-minute Apgar score <4 (Figures 3a–d) but did demonstrate a significant pattern for the logistic model (not using splines) for NICU admission (p=0.01; Figure 3e). After adjustment for the covariates noted in Table 1, the p-values were similar to unadjusted analysis (cesarean delivery p=0.52, episiotomy p=0.19, major perineal laceration p=0.42, Apgar score p=0.12, and NICU admission p=0.009).
Figure 3.
Frequency of outcomes by amount of attending shift completed
3a. Cesarean delivery
3b. Episiotomy
3c. Major perineal laceration
3d. 5-minute Apgar score <4
3e. Neonatal intensive care unit admission
Blue line represents smoothed frequency of the outcome by proportion of attending physician’s time completed in the working shift. The shaded area represents 95% confidence band. The probabilities and confidence limits were calculated by generalized additive models. P-value represents probability value for unadjusted models.
The sensitivity analyses performed with the sample restricted to patients whose physicians worked 12-hour shifts revealed similar findings for all outcomes except for NICU; note that statistical comparisons could not be performed for 5-minute Apgar score <4 due to small sample size (N=15). Specifically, the frequency of cesarean delivery continued to be invariant by time in shift (p=0.14), ranging from 9.0% (70–75% shift completed) to 15.4% (25–30% shift completed). Episiotomy, major perineal laceration, 5-minute Apgar <4, and NICU admission also did not vary significantly (Table 3). In both the unadjusted and adjusted logistic models of this restricted population, there remained no statistically significant non-linear patterns in frequencies of any outcomes (Figure 4).
Table 3:
Frequency of obstetric interventions and adverse outcomes by proportion of time in shift completed among patients delivered by physicians working 12-hour shifts
| % completed in shift | Cesarean | Episiotomy | Major perineal laceration | 5-minute Apgar < 4 | Neonatal intensive care unit admission |
|---|---|---|---|---|---|
|
| |||||
| 0–5% | 61 (12.0) | 68 (13.4) | 21 (4.1) | 0 (0.00) | 54 (10.7) |
| 5–10% | 64 (11.5) | 74 (13.3) | 18 (3.2) | 0 (0.00) | 61 (11.0) |
| 10–15% | 56 (11.0) | 79 (15.5) | 20 (3.9) | 1 (0.20) | 49 (9.6) |
| 15–20% | 61 (11.3) | 76 (14.0) | 18 (3.3) | 0 (0.00) | 54 (10.0) |
| 20–25% | 64 (11.6) | 76 (13.8) | 16 (2.9) | 0 (0.00) | 56 (10.1) |
| 25–30% | 74 (15.4) | 74 (15.4) | 19 (4.0) | 2 (0.42) | 47 (9.8) |
| 30–35% | 74 (13.7) | 87 (16.1) | 31 (5.7) | 2 (0.37) | 60 (11.1) |
| 35–40% | 61 (11.7) | 83 (16.0) | 19 (3.7) | 0 (0.00) | 61 (11.7) |
| 40–45% | 60 (11.9) | 77 (15.3) | 20 (4.0) | 2 (0.40) | 45 (8.9) |
| 45–50% | 51 (9.6) | 93 (17.4) | 17 (3.2) | 0 (0.00) | 45 (8.4) |
| 50–55% | 61 (11.2) | 77 (14.1) | 25 (4.6) | 1 (0.18) | 56 (10.3) |
| 55–60% | 59 (11.3) | 76 (14.6) | 23 (4.4) | 2 (0.38) | 40 (7.7) |
| 60–65% | 54 (10.8) | 63 (12.6) | 14 (2.8) | 0 (0.00) | 59 (11.8) |
| 65–70% | 59 (11.7) | 77 (15.2) | 21 (4.2) | 0 (0.00) | 46 (9.1) |
| 70–75% | 46 (9.0) | 80 (15.6) | 20 (3.9) | 0 (0.00) | 51 (9.9) |
| 75–80% | 74 (13.5) | 85 (15.5) | 23 (4.2) | 0 (0.00) | 57 (10.4) |
| 80–85% | 56 (10.3) | 88 (16.1) | 23 (4.2) | 1 (0.18) | 62 (11.4) |
| 85–90% | 81 (15.0) | 85 (15.7) | 13 (2.4) | 0 (0.00) | 65 (12.0) |
| 90–95% | 72 (13.8) | 83 (15.9) | 13 (2.5) | 1 (0.19) | 60 (11.5) |
| 95–100% | 61 (12.4) | 74 (15.0) | 18 (3.7) | 3 (0.61) | 52 (10.6) |
| Overall | 1249 (11.9) | 1575 (15.0) | 392 (3.7) | 15 (0.14) | 1080 (10.3) |
| p-value (chi-sq) | 0.14 | 0.93 | 0.56 | NA | 0.75 |
| p-value (trend) | 0.58 | 0.21 | 0.47 | NA | 0.47 |
Date are n (%) unless otherwise noted
Figure 4.
Frequency of outcomes by amount of attending shift completed in the sensitivity analysis cohort of obstetricians working 12-hour shifts
4a. Cesarean delivery
4b. Episiotomy
4c. Major perineal laceration
4d. 5-minute Apgar score <4
4e. Neonatal intensive care unit admission
Sensitivity analysis is restricted to patients delivered by attending obstetricians working 12-hour shifts. Blue line represents smoothed frequency of the outcome by proportion of attending physician’s time completed in the working shift. The shaded area represents 95% confidence band. The probabilities and confidence limits were calculated by generalized additive models. P-value represents probability value for unadjusted models.
COMMENT
Principal Results
Understanding how practice patterns, such as timing of delivery within a physician’s work shift, are associated with delivery interventions and outcomes is one potentially important aspect of promoting quality of care in obstetrics. In this large and diverse obstetric cohort, we sought to identify whether there were differences in obstetric interventions or adverse outcomes based on the proportion of the attending obstetrician’s shift completed. We identified no statistically significant or clinically meaningful differences in cesarean delivery, episiotomy, major perineal lacerations, or 5-minute Apgar score <4 based on proximity to attending shift change. Furthermore, as attending shift durations varied widely, we performed a sensitivity analysis restricted to individuals delivered only by physicians who were working 12-hour shifts; findings from the sensitivity analyses mirrored the primary findings. Although we identified a small increase in NICU admission frequency with greater proportion of shift completed, the absolute risk difference was very small, and the findings were not confirmed on the sensitivity analysis, suggesting the clinical relevance of this finding, if even truly extant, is likely small.
Results in Context
The concept of obstetrician decision making in the context of work schedule has been previously explored with regard to call systems.3,4 Our findings differ from some of this literature, which identified changes in practice patterns associated with call schedule. For example, Barber et al identified that when one large group practice transitioned from a traditional call schedule to a night-float system, patients were less likely to undergo induction of labor or receive an episiotomy; in addition, major perineal lacerations and umbilical artery pH <7.10 were less frequent.3 Other work identified difference in vaginal birth after cesarean based on provider call system.4 It is plausible that the type of call schedule may influence practice patterns, however our data suggest that within a given shift the actual practice patterns may not vary significantly.
Additionally, existing literature regarding whether obstetric outcomes differ based on time of day of delivery or shift change have had conflicting results. Recent work from the MFMU identified that nighttime delivery was not associated with differences in postpartum hemorrhage-related management and morbidity.10 Similarly, data from the MFMU Cesarean Registry found that delivery during a shift change was not associated with differences in morbidity7; our findings from a more recent obstetric cohort are consistent with these data from 1999–2000. Other MFMU work also identified no differences in cesarean-associated morbidity based on time of delivery.9 However, these data are in contrast to other literature suggesting some outcomes may be worse at night8,14 or that the likelihood of performing an intrapartum cesarean delivery may vary by time of day.11 In our analysis, night and day shifts are treated as a stable exposure because the goal was to assess factors such as provider fatigue and related behavior when approaching shift change, which we hypothesized may function as separate mechanisms contributing to outcomes than the time of day. However, this topic may warrant further attention in future analyses.
Clinical and Research Implications
Labor and delivery unit organizations are complex and vary widely in their composition, operations, and culture. Aside from differences in call systems, units have family medicine faculty, midwives, laborists, or other personnel to support the provision of care. Such variability in organization and culture may be highly influential with regard to decision making, intervention timing, and obstetric outcomes. The complexity of obstetric units suggests future work is needed. For example, we focused on attending obstetrician shift change, but future work may consider whether differences exist based on resident or nurse shifts, as has been suggested in prior work.7 Second, although we studied time worked in shift as a proxy for other cognitive or behavioral factors, we could not directly assess issues such as fatigue, desire to go home, desire to not sign out certain types of patients, financial pressures to expedite delivery, differential interpretation of patient status by a new provider starting a shift, or other components of a unit culture. Future work must prospectively evaluate how such behavioral or attitudinal factors affect decisions. Third, although community hospitals were included in the cohort, a majority of study sites were academic centers, which may differ from other centers because of call systems, the presence of residents, or other systems in place to mitigate the effects of fatigue. Future work should investigate whether these findings persist in the community setting. In addition, given the need for a choice with regard to the exposure and outcomes that were of interest and could reasonably be related to one another, we chose a priori in this analysis to examine the time completed in shift in 5% blocks and as a continuous variable. We acknowledge that although no approach is clearly more appropriate a priori, different approaches could be used; evaluating different approaches may be of interest for future studies. Finally, the finding of a small change in NICU admission based on proportion of shift completed represents an area of future investigation, as NICU admission is influenced by multiple factors beyond the obstetrical team, which was the focus of this investigation.
Strengths and Limitations
Strengths of this analysis include data derived from a large cohort with detailed clinical and demographic information. Few existing data sets have available detailed information on issues such as length and timing of provider shifts, and because of the availability of these data, we were able to precisely determine when in a shift a delivery occurred and were able to account for such differences as nighttime or weekends. Data were also collected by trained research personnel at clinically and geographically diverse settings.
However, several limitations also warrant discussion. First, these data are observational and cannot demonstrate causation. Second, as noted, these deliveries largely occurred at academic medical centers and thus findings may not be generalizable. Although some centers were academically-affiliated community practices without residents, these sites were few and thus it is not possible to perform subgroup analyses by practice environment. Generalizability may also be somewhat impaired because of the exclusion of sites in which shift change was not reported. Third, we were unable to assess other rare outcomes, such as maternal death, given the infrequency of such events. Similarly, the frequency of 5-minute Apgar <4 was low, limiting power to detect potentially clinically important differences for this outcome. Finally, it was not possible to examine whether the delivering attending differed from the individual managing labor.
Additionally, these data were collected on deliveries performed between 2008 and 2011, making the most recent deliveries nearly 10 years ago. In that time, other quality improvement initiatives may have occurred, and changes in obstetrician work culture may have also changed how labor floor coverage occurs. Other changes that may have occurred include introduction of relative value unit compensation models or other productivity-based financial incentives, or innovations in medical record reporting that may have influenced the quality of data capture.
Conclusions
In summary, anecdotes from clinical practice suggest there may be differences in care provision when providers are fatigued or close to shift change, yet the analysis from this large and well-characterized cohort suggests there are no meaningful associations between the occurrence of intrapartum interventions or adverse birth outcomes and the proportion of attending shift completed. Future work should attempt to directly study unit culture and provider fatigue in order to further investigate opportunities to improve obstetric quality of care, as well as determine if these findings persist in community hospital settings.
AJOG AT A GLANCE:
This study was conducted to assess differences in intrapartum obstetric interventions and adverse outcomes based on temporal proximity of delivery to the attending provider’s shift change.
Differences in obstetric interventions and outcomes do not appear to exist based on temporal proximity to attending physician shift change.
There are no meaningful associations between the occurrence of intrapartum interventions or adverse birth outcomes and the proportion of attending shift completed. Future work should study unit culture and provider fatigue in order to further investigate opportunities to improve obstetric quality of care.
ACKNOWLEDGMENTS:
The authors thank William A. Grobman, M.D., M.B.A., Elizabeth Thom, Ph.D., Madeline M. Rice, Ph.D., Brian M. Mercer, M.D., and Catherine Y. Spong, M.D. for protocol development and oversight.
FUNDING: Supported by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD21410, HD27869, HD27915, HD27917, HD34116, HD34208, U10HD36801, HD40500, HD40512, HD40544, HD40545, HD40560, HD40485, HD53097, HD53118) and the National Center for Research Resources (UL1 RR024989; 5UL1 RR025764). Additionally, LMY was supported by 2K12 HD050121 at the time of the study. Comments and views of the authors do not necessarily represent views of the National Institutes of Health.
APPENDIX
In addition to the authors, other members of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network are as follows:
Northwestern University, Chicago, IL –G. Mallett, W. Grobman, M. Ramos-Brinson, A. Roy, L. Stein, P. Campbell, C. Collins, N. Jackson, M. Dinsmoor (NorthShore University HealthSystem), J. Senka (NorthShore University HealthSystem), K. Paychek (NorthShore University HealthSystem), A. Peaceman
Columbia University, New York, NY – M. Talucci, M. Zylfijaj, Z. Reid (Drexel U.), R. Leed (Drexel U.), J. Benson (Christiana H.), S. Forester (Christiana H.), C. Kitto (Christiana H.), S. Davis (St. Peter’s UH.), M. Falk (St. Peter’s UH.), C. Perez (St. Peter’s UH)
University of Utah Health Sciences Center, Salt Lake City, UT – K. Hill, A. Sowles, J. Postma (LDS Hospital), S. Alexander (LDS Hospital), G. Andersen (LDS Hospital), V. Scott (McKay-Dee), V. Morby (McKay-Dee), K. Jolley (UVRMC), J. Miller (UVRMC), B. Berg (UVRMC)
University of North Carolina at Chapel Hill, Chapel Hill, NC – K. Dorman, J. Mitchell, E. Kaluta, K. Clark (WakeMed), K. Spicer (WakeMed), S. Timlin (Rex), K. Wilson (Rex)
University of Texas Southwestern Medical Center, Dallas, TX – L. Moseley, K. Leveno (deceased), M. Santillan, J. Price, K. Buentipo, V. Bludau, T. Thomas, L. Fay, C. Melton, J. Kingsbery, R. Benezue
University of Pittsburgh, Pittsburgh, PA – H. Simhan, M. Bickus, D. Fischer, T. Kamon (deceased), D. DeAngelis
MetroHealth Medical Center-Case Western Reserve University, Cleveland, OH – B. Mercer, C. Milluzzi, W. Dalton, T. Dotson, P. McDonald, C. Brezine, A. McGrail
The Ohio State University, Columbus, OH – C. Latimer, L. Guzzo (St. Ann’s), F. Johnson, L. Gerwig (St. Ann’s), S. Fyffe, D. Loux (St. Ann’s), S. Frantz, D. Cline, S. Wylie, J. lams
University of Alabama at Birmingham, Birmingham, AL – M. Wallace, A. Northen, J. Grant, C. Colquitt, D. Rouse, W. Andrews
University of Texas Medical Branch, Galveston, TX – J. Moss, A. Salazar, A. Acosta, G. Hankins
Wayne State University, Detroit, MI – N. Hauff, L. Palmer, P. Lockhart, D. Driscoll, L. Wynn, C. Sudz, D. Dengate, C. Girard, S. Field
Brown University, Providence, RI – P. Breault, F. Smith, N. Annunziata, D. Allard, J. Silva, M. Gamage, J. Hunt, J. Tillinghast, N. Corcoran, M. Jimenez
The University of Texas Health Science Center at Houston, McGovern Medical School-Children’s Memorial Hermann Hospital, Houston, TX– F. Ortiz, P. Givens, B. Rech, C. Moran, M. Hutchinson, Z. Spears, C. Carreno, B. Heaps, G. Zamora
Oregon Health & Science University, Portland, OR – J. Seguin, M. Rincon, J. Snyder, C. Farrar, E. Lairson, C. Bonino, W. Smith (Kaiser Permanente), K. Beach (Kaiser Permanente), S. Van Dyke (Kaiser Permanente), S. Butcher (Kaiser Permanente)
The George Washington University Biostatistics Center, Washington, D.C. – E. Thom, M. Rice, Y. Zhao, V. Momirova, R. Palugod, B. Reamer, M. Larsen
Eunice Kennedy Shriver National Institute of Child Health and Human Development, Bethesda, MD – C. Spong, S. Tolivaisa
MFMU Network Steering Committee Chair (Medical University of South Carolina, Charleston, SC) – J. P. VanDorsten, M.D.
Footnotes
DISCLOSURE: The authors report no conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Lynn M. Yee, Departments of Obstetrics and Gynecology of Northwestern University, Chicago, IL
Paula McGee, George Washington University Biostatistics Center, Washington, DC
Jennifer L. Bailit, MetroHealth Medical Center-Case Western Reserve University, Cleveland, OH
Ronald J. Wapner, Columbia University, New York, NY
Michael W. Varner, University of Utah Health Sciences Center, Salt Lake City, UT
John M. Thorp, Jr., University of North Carolina at Chapel Hill, Chapel Hill, NC
Steve N. Caritis, University of Pittsburgh, Pittsburgh, PA
Mona Prasad, The Ohio State University, Columbus, OH
Alan T. N. Tita, University of Alabama at Birmingham, Birmingham, AL
George R. Saade, University of Texas Medical Branch, Galveston, TX
Yoram Sorokin, Wayne State University, Detroit, MI
Dwight J. Rouse, Brown University, Providence, RI
Sean C. Blackwell, University of Texas Health Science Center at Houston, McGovern Medical School-Children’s Memorial Hermann Hospital, Houston, TX
Jorge E. Tolosa, Oregon Health & Science University, Portland, OR
REFERENCES
- 1.Patient safety and quality improvement. 2019. (Accessed November 12, 2019, at www.smfm.org/safety.)
- 2.Safe reduction of primary cesarean birth (+AIM) bundle. 2015. (Accessed November 12, 2019, at https://safehealthcareforeverywoman.org/patient-safety-bundles/safe-reduction-of-primary-cesarean-birth/.)
- 3.Barber EL, Eisenberg DL, Grobman WA. Type of attending obstetrician call schedule and changes in labor management and outcome. Obstetrics and Gynecology 2011;118:1371–6. [DOI] [PubMed] [Google Scholar]
- 4.Yee LM, Liu LY, Grobman WA. Obstetrician call schedule and obstetric outcomes among women eligible for a trial of labor after cesarean. Am J Obstet Gynecol 2017;216:75.e1–.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Yee L, Liu LY, Grobman WA. Relationship between obstetricians’ cognitive and affective traits and delivery outcomes among women with a prior cesarean. Am J Obstet Gynecol 2015;213:413.e1-7. [DOI] [PubMed] [Google Scholar]
- 6.Yee LM, Liu LY, Grobman WA. The relationship between obstetricians’ cognitive and affective traits and their patients’ delivery outcomes. American journal of obstetrics and gynecology 2014;211:e1–6. [DOI] [PubMed] [Google Scholar]
- 7.Bailit JL, Landon MB, Lai Y, et al. Maternal-Fetal Medicine Units Network cesarean registry: impact of shift change on cesarean complications. Am J Obstet Gynecol 2008;198:173.e1-5. [DOI] [PubMed] [Google Scholar]
- 8.Peled Y, Melamed N, Chen R, Pardo J, Ben-Shitrit G, Yogev Y. The effect of time of day on outcome of unscheduled cesarean deliveries. Journal of Maternal-Fetal & Neonatal Medicine 2011;24:1051–4. [DOI] [PubMed] [Google Scholar]
- 9.Bailit JL, Landon MB, Thom E, et al. The MFMU Cesarean Registry: impact of time of day on cesarean complications. Am J Obstet Gynecol 2006;195:1132–7. [DOI] [PubMed] [Google Scholar]
- 10.Yee LM, McGee P, Bailit JL, et al. Daytime Compared With Nighttime Differences in Management and Outcomes of Postpartum Hemorrhage. Obstet Gynecol 2019;133:155–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Son M, Lai Y, Baiit J, et al. Association between time of day and the decision for an intrapartum cesarean. Obstet Gynecol 2020;135:535–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bailit J, Grobman W, Rice M, et al. Risk-adjusted models for adverse obstetric outcomes and variation in risk-adjusted outcomes across hospitals. Am J Obstet Gynecol 2013;209:446.e1-30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Grobman W, Bailit J, Rice M, et al. Can differences in obstetric outcomes be explained by differences in the care provided? The MFMU Network APEX study. Am J Obstet Gynecol 2014;211:e1-147.e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lyndon A, Lee H, Gay C, GIlbert W, Gould J, Lee K. Effect of time of birth on maternal morbidity during childbirth hospitalization in California. Am J Obstet Gynecol 2015;213:705.e1-11. [DOI] [PMC free article] [PubMed] [Google Scholar]