Association between postoperative length of stay after cesarean delivery and postpartum maternal complications

Abstract

Introduction

There is no standard for how long a patient's postpartum stay should be after cesarean delivery. There are known patient‐ and system‐level benefits of shortening the postoperative length of stay, and data are mixed on the relationship between maternal length of stay after cesarean deliveries and post‐discharge complications. Our objective was to examine whether postoperative length of stay after cesarean delivery was associated with postpartum maternal complications.

Methods

We conducted a retrospective cohort study of all cesarean deliveries in 2019 in the National Surgical Quality Improvement Program database. Patients with a postoperative length of stay over 4 days postpartum were excluded, which was used as the cutoff given insurance mandate minimums. Patients were categorized based on postpartum (delivery to discharge) length of stay: 0–2, 3, and 4 days. Our primary outcome was a composite of maternal complications defined by hospital readmission, reoperation, and death up to 30 days postpartum. Outcomes were compared between the exposure groups, and multivariable logistic regression was used to estimate the association between postoperative length of stay and study outcomes using 4 days as the referent.

Results

Of the 13,540 patients with a cesarean delivery, 8325 (61.6%) had a postoperative length of stay of 0–2 days, 4173 (30.8%) of 3 days, and 1042 (7.7%) of 4 days. Patients with a postoperative length of stay of 2 or fewer days tended to be younger and have lower rates of cigarette use, diabetes, unscheduled cesarean delivery (vs. scheduled), hypertension, general anesthesia, and American Society of Anesthesiologists class 3 or 4. Compared with a 4‐day postoperative length of stay, patients with 0–2 days postoperative length of stay had significantly decreased odds of the primary composite outcome, as well as readmissions and reoperations. Compared with a 4‐day postoperative length of stay, patients with a 3‐day postoperative length of stay had similar odds of the primary outcome.

Conclusion

In this national patient sample, early postpartum discharge after cesarean was not associated with increased odds of a composite of postpartum maternal complications including maternal death, readmissions, or reoperations. Patient who were discharged early did not have an increased odds of maternal complications; this may be due to being intrinsically lower risk or due to being deemed appropriate by their caring physicians. Further studies using databases including obstetric‐specific data are needed.

1. INTRODUCTION

Cesarean deliveries are among the most common surgeries performed in the United States, accounting for nearly 1.2 million surgeries in 2019 and for 31.7% of all deliveries [1]. Despite its frequency, no standard exists as to how long a patient's postpartum stay should be after delivery. The Neonatal and Maternal Health Protection Act (NMHPA) was introduced in 1996, as a response to a trend of insurers limiting postpartum care coverage [2]. This act mandated that certain healthcare insurers could not restrict benefits for a hospital stay to less than “96 hours following a delivery by cesarean section” [3]. Though this is not a mandate for all hospitals, some individual states have implemented minimums on hospital postoperative length of stay (LOS) following deliveries, citing the NMHPA guidelines [4]. Though these laws are similar in intent, there is wide variance in each state's law including whether there is a guaranteed minimum duration of postoperative stay, whether the mandate is specifically to prevent the restriction of benefits by insurers, and whether each state chose to apply the same guidance to patients with Medicaid [4].

Shortening the postoperative LOS and immediate postoperative recovery period for both obstetric and gynecologic procedures is a priority—spurring initiatives such as Enhanced Recovery After Surgery (ERAS)—as length of hospital stay has been used as a quality marker of hospital resource utilization and has been associated with quality of care [5, 6, 7]. Starting in the field of colorectal and expanding to other surgical subspecialties, ERAS protocols have been shown to reduce hospital postoperative LOSs by up to 50% without increasing readmissions or postoperative complications [8]. In obstetrics, studies have shown that enhanced recovery pathways for cesarean deliveries reduce postpartum LOS, as well as total direct costs of clinical care [9, 10].

While there are known patient‐ and system‐level benefits of shortening the postoperative LOS, the data are mixed on the relationship between maternal postoperative LOS after cesarean deliveries and post‐discharge complications. This relationship between maternal LOS after cesarean deliveries and post‐discharge complications is particularly important given both the aforementioned state and national legislation guiding an appropriate postoperative LOS. As such, our objective was to evaluate whether maternal postoperative LOS after cesarean was associated with maternal complications after discharge including readmission, reoperations, and deaths, using a nationally representative patient sample.

2. METHODS

We performed a retrospective cohort study of all cesarean deliveries using the 2019 American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP) database. The NSQIP is a national database of patients undergoing surgical procedures with blinded, risk‐adjusted data [11]. The data included preoperative risk factors (e.g., age, diabetes mellitus, hypertension, smoking history, functional status, chronic obstructive pulmonary disease (COPD), cancer history), intraoperative data (e.g., operation time, transfusion of blood products), 30‐day postoperative morbidity and mortality (e.g., sepsis, pneumonia, myocardial infarction, unplanned intubation, urinary tract infection), and other adverse surgical outcomes (e.g., reoperation, readmission) from participating institutions across the United States. Data also included demographics, medical comorbidities, laboratory abnormalities, hospital and postoperative LOSs, venous thromboembolisms, estimated surgical blood loss, surgical and wound infections, and pneumonia. NSQIP utilizes trained and certified surgical clinical reviewers at each institution to abstract the data. Included cases are selected based on a systematic sampling process. Institutional Review Board approval was not obtained for this study as it was deemed to be exempt as all data are deidentified.

For this analysis, patients were included if they underwent a cesarean delivery, defined by the appropriate Current Procedural Terminology (CPT) code (CPT 59510, 59514, 59515, 59618, 59620, 59622). Patients with a postoperative LOS > 4 days were excluded (n = 307), given the assumption that extended LOS beyond 4 days was most likely a result of a medical, surgical, or postpartum complication. Patients were also excluded if they had missing age (n = 1) or were over 60 years of age (n = 4); American Society of Anesthesiologists (ASA) class level 5 (n = 2) or not assigned (n = 98); LOS missing (n = 1); unknown functional status (n = 88).

The primary exposure was LOS. Patients were categorized into postoperative LOS groups: 0–2 days, 3 days, and 4 days. These groupings were selected to align with existing literature, in particular, meta‐analyses using the 48 and 96 h marks to assess the safety of postpartum discharge timing [12]. The LOS variable in the NSQIP database was calculated from the date of operation to the date of discharge, with days as the unit.

The primary outcome was a composite of postpartum maternal complications defined as hospital readmission, reoperation, or death, and was ascertained through 30 days after the cesarean delivery. Secondary outcomes were the components of the composite of maternal complications.

Demographic and clinical characteristics were compared among postoperative LOS groups using Chi‐Square tests for categorical variables and analysis of variance or Kruskal–Wallis tests as appropriate. Primary and secondary outcomes were similarly compared between the postoperative LOS exposure groups, and crude odds ratios with 95% confidence intervals (95% CIs) using 4 days as the reference category were calculated. LOS of 4 days was selected as the referent due to the NMHPA insurance mandate and laws in some states guaranteeing postoperative coverage of up to 96 h. Multivariable logistic regression models, using characteristics that differed between groups with p value < 0.05, estimated adjusted odds ratios (aORs) with 95% CIs for the outcomes. Linear regression was performed to calculate the percent increase in the odds of the individual and composite outcomes with each unit increase in days from 0–2 days to 3 days to 4 days, adjusting for age, smoking, diabetes, emergency case, hypertension, anesthesia, and ASA class. A secondary analysis was planned to evaluate the association between postoperative LOS as a continuous variable and post‐discharge outcomes.

SAS software Version 9.4 (SAS Institute, Inc.) was used to analyze the data. Findings were considered statistically significant at a p value of less than 0.05.

3. RESULTS

Of the 1,076,441 patients in the 2019 NSQIP database, 14,029 were identified as having a cesarean delivery and were considered for inclusion in this analysis. Of these, 489 (3.5%) were excluded, leaving a total of 13,540 patients for analysis. Overall, 8325 (61.6%) had a postoperative LOS of 0–2 days, 4173 (30.8%) had a postoperative LOS of 3 days, and 1042 (7.7%) had a postoperative LOS of 4 days.

Baseline demographics and characteristics are presented in Table 1. Patients with a postoperative LOS of 2 or fewer days tended to be younger and have lower rates of cigarette use, insulin‐dependent diabetes, unscheduled cesarean delivery (vs. scheduled), hypertension, general anesthesia, and ASA class 3 or 4. For patients with a postoperative LOS of 4 days, 11.4% identified as Non‐Hispanic Black, compared to 12.1% for patients with a postoperative LOS of 3 days, and 6.9% for patients with a postoperative LOS of 2 or fewer days. Additionally, patients with a postoperative LOS of 2 or fewer days had a significantly lower proportion of comorbidities including diabetes (7.0%), hypertension (3.1%), and cigarette use (8.6%), and a higher ASA class of 3 or 4 (15.0%) when compared with patients who had a postoperative LOS of 3 days or 4 days (Table 1). Patients with a postoperative LOS of 2 or fewer days also had a lower frequency of general anesthesia (5.7%) compared with patients who had a postoperative LOS of 3 days (7.1%) or 4 days (8.3%).

TABLE 1.

Demographic and clinical characteristics by length of stay after cesarean delivery.

		Length of stay
Characteristics	Total ( N  = 13540)	0–2 Days ( N  = 8335)	3 Days ( N  = 4173)	4 Days ( N  = 1042)	p value
Mean age, years (SD)	30.8 (5.6)	30.8 (5.4)	30.7 (5.8)	31 (6)	0.512
Race/ethnicity					<0.001
Hispanic	1769 (13.1)	1068 (12.8)	538 (12.9)	163 (15.6)
Non‐Hispanic Black	1197 (8.8)	571 (6.9)	507 (12.1)	119 (11.4)
Non‐Hispanic White	4571 (33.8)	2476 (29.7)	1698 (40.7)	397 (38.1)
None of the above	1004 (7.4)	513 (6.2)	366 (8.8)	125 (12.0)
Unknown/Not reported	4999 (36.9)	3697 (44.4)	1064 (25.5)	238 (22.8)
Median BMI [IQR]	32.7 [8.7]	32.5 [8.5]	33.1 [9]	33.2 [10.5]	<0.001
Cigarette use	1285 (9.5)	717 (8.6)	446 (10.7)	122 (11.7)	<0.001
Diabetes mellitus					<0.001
Insulin dependent	745 (5.5)	413 (5.0)	243 (5.8)	89 (8.5)
Non‐insulin dependent	295 (2.2)	159 (1.9)	103 (2.5)	33 (3.2)
No	12,500 (92.3)	7753 (93.0)	3827 (91.7)	920 (88.3)
Hypertension, on meds	552 (4.1)	256 (3.1)	206 (4.9)	90 (8.6)	<0.001
Emergency case	4086 (30.2)	2415 (29.0)	1290 (30.9)	381 (36.6)	<0.001
Anesthesia method					<0.001
General	858 (6.3)	473 (5.7)	298 (7.1)	87 (8.3)
Neuraxial	12,247 (90.5)	7547 (90.5)	3773 (90.4)	927 (89.0)
Other/unknown	435 (3.2)	305 (3.7)	102 (2.4)	28 (2.7)
ASA classification					<0.001
1‐No Disturb	1061 (7.8)	782 (9.4)	238 (5.7)	41 (3.9)
2‐Mild Disturb	10,118 (74.7)	6294 (75.5)	3124 (74.9)	700 (67.2)
3/4‐Severe/Life Threat	2361 (17.4)	1249 (15.0)	811 (19.4)	301 (28.9)

Note: Data are presented as n (%) unless otherwise specified. Chi Square or Fisher's exact test for count data, unless otherwise noted; ANOVA F test for mean (SD) data; Kruskal–Wallis Chi Square test for median [IQR] data. The following variables were included as covariates in the calculation of the adjusted odds ratio: age, cigarette use, diabetes mellitus, emergency case, hypertension, anesthesia, and ASA class.

Abbreviations: ASA, American Society of Anesthesiologists; ANOVA, analysis of variance; BMI, body mass index; IQR, interquartile range; SD, standard deviation.

Patients with a postoperative LOS of 2 or fewer days had significantly decreased odds of the primary composite maternal complication when compared with patients with a postoperative LOS of 4 days (1.9% vs. 4.4%; aOR, 0.48; 95% CI, 0.34–0.68), representing a 52% reduction in odds of maternal complications in patients with a postoperative LOS of 2 days or fewer (Table 2). Regarding each individual component, patients with a postoperative LOS of 2 days or fewer also had significantly lower odds of readmissions (1.6% vs. 3.2%; aOR, 0.59; 95% CI, 0.40–0.87) and reoperations (0.5% vs. 1.8%; aOR, 0.30; 95% CI, 0.17–0.52) when compared with those with an LOS of 4 days. Patients with postoperative LOS of 3 days did not have significantly decreased odds of the primary composite maternal complication when compared with patients with a postoperative LOS of 4 days (3.1% vs. 4.4%; aOR, 0.76; 95% CI, 0.54–1.07). Patients with an LOS of 3 days did not have lower odds of readmissions (2.6% vs. 3.2%; aOR, 0.90; 95% CI, 0.60–1.34), but LOS of 3 days was associated with lower odds of reoperations (0.8% vs. 1.8%; aOR, 0.46; 95% CI, 0.26–0.81). Maternal death was rare, and there was no detectable difference observed in maternal deaths across groups.

TABLE 2.

Unadjusted and adjusted association between post‐cesarean length of stay measured as a categorical exposure and maternal postpartum complications.

Postoperative length of stay	Outcome	Unadjusted	Adjusted a
	N (%)	OR (95% CI)	OR (95% CI)
Composite maternal complications
2 Days or fewer	159 (1.9)	0.42 (0.30–0.59)	0.48 (0.34–0.68)
3 Days	130 (3.1)	0.70 (0.49–0.98)	0.76 (0.54–1.07)
4 Days	46 (4.4)	Ref	Ref
Maternal readmission
2 Days or fewer	135 (1.6)	0.50 (0.34–0.74)	0.59 (0.40–0.87)
3 Days	109 (2.6)	0.82 (0.55–1.22)	0.90 (0.60–1.34)
4 Days	33 (3.2)	Ref	Ref
Maternal reoperation
2 Days or fewer	44 (0.5)	0.29 (0.17–0.49)	–
3 Days	34 (0.8)	0.44 (0.25–0.78)	–
4 Days	19 (1.8)	Ref	–

Note: The following variables were included as covariates in the calculation of the adjusted odds ratio: age, cigarette use, diabetes mellitus, emergency case, hypertension, anesthesia, and ASA class.

Abbreviations: ASA, American Society of Anesthesiologists; CI, confidence interval; OR, odds ratio.

^aAn adjusted odds ratio was not calculated for maternal reoperation given the potential for overfitting this model.

When evaluating postoperative LOS as a continuous outcome, the results were consistent with the primary outcome (data not shown).

4. DISCUSSION

In a nationally representative sample of patients undergoing cesarean delivery, a postoperative LOS of 0–2 days was associated with significantly lower odds of composite postoperative maternal complications including readmissions and reoperation when compared with a postoperative LOS of 4 days. This association was not observed when comparing patients with an LOS of 3 days with those who had an LOS of 4 days. Patients who were discharged earlier (shorter LOS) were notably less medically complex, as evidenced by lower rates of diabetes, hypertension, and a higher ASA class of 3 or 4. While we postulate that the observed association between shorter LOS and lower postpartum complications reflects the decreased rates of complex medical comorbidities, the odds ratio remained significant after adjusting for the clinically significant demographic variables that differed between groups.

Multiple studies have examined postoperative LOS after cesarean and its influence on maternal outcomes—with mixed conclusions. A study using a national readmissions database showed that longer postoperative LOS after cesarean delivery was associated with decreased risk of postpartum readmissions specifically for hypertension‐related causes [13]. Thus, a task force on gestational hypertension and preeclampsia convened by the American College of Obstetricians and Gynecologists (ACOG) recommended at least 72 h of blood pressure monitoring for all deliveries [14]. Conversely, a meta‐analysis conducted in 1997 examining the safety of early postpartum discharge showed no increase in maternal or neonatal morbidity after discharge; however, the authors could not support or refute a practice of early discharge in the general population, stating that it “appears safe in carefully selected, consenting patients” [12]. A 2021 systematic review concluded that early postpartum discharge “probably” results no difference in maternal readmission; however, there was substantial heterogeneity in the studies, including the definition of early discharge that ranged from 6 h to 5 days [15].

In 2015, a study in the United Kingdom examined early discharge after cesarean delivery after implementation of ERAS protocols and saw an increase in postoperative Day 1 and 2 discharges without significant changes in readmission rates [5]. Other studies completed prior to the use of ERAS in Denmark and Italy also demonstrated a reduction in cesarean delivery postoperative LOS rates from the early 2000s to the mid‐2010s [16, 17]. Limitations of previous studies are the inclusion of a patient population different from ours in the United States, non‐contemporary cohorts, heterogeneity in discharge data, and limited evaluation of postpartum outcomes.

Our results are supported by a more recent US study examining discharge practices for cesarean deliveries prior to and during the COVID‐19 pandemic [18]. This study found a nearly threefold higher adjusted odds of a short postoperative LOS after all deliveries, including cesarean deliveries, which was defined by less than 3 midnights, during COVID‐19 pandemic without a difference in the adjusted odds of readmission [18]. The authors hypothesized that the pandemic may have led to shortened hospitalizations resulting from patient desires to reduce SARS‐CoV‐2 exposure, a change in hospital delivery protocols, and a higher threshold for readmissions [18]. Our study supports these findings in a large, diverse, national sample, prior to COVID‐19, and after the widespread implementation of ERAS.

These results suggest that patients who were discharged early did not have an increased odds of maternal complications; this may be due to being intrinsically lower risk or due to being deemed appropriate by their caring physicians. Mandated minimum requirements for postoperative LOS may not be clinically required. Further analysis should evaluate obstetric‐specific variables to identify key areas that put patients at risk for prolonged LOS after a cesarean delivery.

One strength of this study was the use of the NSQIP database. This nationally verified database includes data abstracted by trained clinical researchers using standardized processes. As a quality control measure, NSQIP undergoes frequent audits, which consistently show that the data are reliable [19]. Additionally, the national distribution of the participating hospitals, with geographic and ethnic diversity, makes these conclusions generalizable across institutions. We excluded those with an LOS of more than 4 days in an attempt to remove patients with complications that may require a prolonged hospitalization after a cesarean delivery.

The study has limitations. As this was a national surgical outcomes database, it lacked certain details such as postoperative outcomes (e.g., pain scores), baseline obstetric characteristics (e.g., parity, type of labor, indication for cesarean delivery), and pregnancy‐related medical conditions (e.g., preeclampsia, endometritis, and obstetric hemorrhage). Their absence raises the possibility of unmeasured confounding; however, we adjusted for covariates that differed in the bivariate analysis. These other medical and obstetric comorbidities could have additionally increased the postoperative LOS, which would raise the possibility of confounding by indication. Inherent to the retrospective design, the association between LOS and postpartum complications in this study does not imply causation. The same covariates were used for all outcomes for consistency in our analyses—including in the analysis of uncommon events such as reoperation. As such, we did not calculate aORs for this outcome, acknowledging the potential for overfitting this model. Alternative methods of adjustment, such as propensity score matching, could be utilized; however, these still fail to account for unmeasured confounding and are unlikely to yield different results due to our sample size being large and our number of events not being low [20, 21]. Patients readmitted for any obstetric complications up to 30 days after initial discharge would be identified in hospital readmissions, but it is unknown why these patients required readmission or reoperation. By excluding patients who had LOSs over 4 days, we exclude those patients who may have had an issue that needed addressing, thus requiring a longer LOS. Evaluation for readmission or reoperation was limited to 30 days after cesarean delivery—and could be subject to misclassification bias. Lastly, the database was limited by not having the exact time of deliveries, therefore postoperative LOS is an estimation in days rather than hours, as calculated by date of admission to date of discharge.

Further studies on postoperative LOS after cesareans should include obstetric‐specific characteristics, pregnancy‐related outcomes, and be extended to the full 6‐week postpartum period to fully account for complications in the postpartum window.

5. CONCLUSION

In summary, patients with a postoperative LOS of 2 or fewer days, when compared with 4 days, were found to have lower odds of readmissions, reoperations, and overall maternal composite of postpartum complications. In certain populations (e.g., fewer comorbidities, uncomplicated deliveries), early discharge may not be associated with greater complications and thereby may be appropriate and safe in appropriate clinical situations based on clinical discretion.

CONFLICT OF INTEREST STATEMENT

K.S.S., K.D.M., A.N.B., and R.J.K. report no financial or other conflicts of interest to disclose. A.S. disclosed that money was paid to her institution from the NIH, has received payment for legal consultations, and has received payment from Novocuff. Each author has confirmed compliance with the journal's requirements for authorship.

FUNDING INFORMATION

The authors received no specific funding for this work.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from NSQIP. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://www.facs.org/quality‐programs/data‐and‐registries/acs‐nsqip/ with the permission of NSQIP.