Association Between Weekend Discharge and Hospital Readmission Rates Following Major Surgery

Importance  Although evidence suggests worse outcomes for patients admitted to the hospital on a weekend, to our knowledge, no previous studies have investigated the effects of weekend discharge.

Objective  To determine whether weekend discharge would be associated with an increased rate of 30- and 90-day hospital readmission.

Design, Setting, and Participants  Retrospective review of discharge abstracts from the California Office of State Health Planning and Development from 2012 identifying all patients who underwent abdominal aortic aneurysm (AAA) repair, colectomy, total hip arthroplasty, and pancreatectomy. This study was conducted from January to December 2012.

Main Outcomes and Measures  Thirty- and 90-day readmission rates were compared between patients discharged on a weekend vs weekday.

Results  Of 128 057 patients, 5225 patients (4.1%) underwent AAA repair; 29 388 (22.9%), colectomy; 91 168 (71.2%), hip replacement; and 2276 (1.8%), pancreatectomy. Overall, 29 883 (23.3%) were discharged on a weekend. Although there were no significant differences with respect to sex, age, race/ethnicity, insurance status, or type of admission, patients discharged on a weekend had shorter length of stays and were less often discharged to a skilled nursing facility. Overall, the 30-day readmission rate was 9.4% after AAA repair, 13.6% after colectomy, 7.5% after hip replacement, and 16.3% after pancreatectomy. Hospital readmission rates were similar for those discharged on a weekend vs weekday after AAA repair (8.8% vs 9.3%; P = .55) and pancreatectomy (17.5% vs 15.9%; P = .40). However, weekend discharge was associated with a lower 30-day readmission rate for patients undergoing colectomy (12.1% vs 14.1%; P < .001) and hip replacement (6.9% vs 7.7%; P < .001). On multivariable analysis, weekend discharge was inversely associated with readmission after colectomy (odds ratio [OR], 0.86; 95% CI, 0.79-0.93) but not AAA repair (OR, 0.93; 95% CI, 0.73-1.19), hip replacement (OR, 0.97; 95% CI, 0.91-1.03), or pancreatectomy (OR, 1.02; 95% CI, 0.76-1.36). Finally, a substantial percentage of 30-day readmissions occurred at a different hospital (AAA repair: 40.5%; colectomy: 25.8%; hip replacement: 32.5%; and pancreatectomy: 19.7%) compared with the index hospitalization. Similar results were seen for 90-day readmissions.

Conclusions and Relevance  Weekend discharge after major surgery is not associated with higher 30- or 90-day readmission rates.

Hospital readmission, defined as admission to a hospital within 30 days of discharge from an acute care facility, is associated with increased morbidity, mortality, and health care expenditures.¹ In 2013, as part of the Patient Protection and Affordable Care Act, Medicare began assigning financial penalties to those hospitals with the highest readmission rates for common medical conditions.² In its third year, the program announced penalties for 2610 hospitals that will total more than $428 million.³ Beginning in 2015, the Centers for Medicare and Medicaid Services will expand to include penalties for unplanned readmissions after at least 1 surgical procedure.⁴ For these reasons, identifying risk factors that contribute to hospital readmissions for patients undergoing surgery have important quality and health policy implications.

Hospital discharge after major surgery is a complex process that aims to achieve the safe transition of care of a hospitalized patient to another setting (eg, home or skilled nursing facility [SNF]). Successful discharge planning requires the coordination of multiple teams across inpatient and outpatient settings. It depends on the adequate staffing of physicians, nurses, case managers, social workers, and pharmacists, as well as clear communication among patients, caretakers, and health care professionals. Although considerable evidence has highlighted worse outcomes for patients admitted to the hospital^5-7 or undergoing surgery^8-10 on a weekend, to our knowledge, the impact of weekend discharge on hospital readmission rates has not been well explored.

There are several reasons to suspect that weekend discharge could lead to worse outcomes after a major operation. First, hospital staffing may be reduced on the weekend. In addition, physicians and other practitioners often cross-cover for one another other on the weekend, so patients are more likely to be seen by on-call staff immediately prior to discharge. Loss of health care professional continuity may negatively impact on the timeliness of discharge, accuracy of discharge instructions, and medication reconciliation.¹¹ Outpatient offices are typically closed on the weekend and therefore timely follow-up appointments may be omitted. Reduced availability of outpatient pharmacy on weekends may result in delays in patients receiving their medications,¹² which is associated with medication noncompliance and poor outcomes during the transition to the outpatient setting.¹³ A single-institution study found that weekend discharge after kidney transplantation was associated with increased readmission rates.¹⁴

Based on this rationale, the current study was designed to investigate the association between weekend discharge and 30- and 90-day readmissions in patients undergoing major surgery. To do so, a large-state, all-payer discharge database with individual patient record linkage numbers was selected to capture all readmissions, even those to a different hospital, for patients undergoing abdominal aortic aneurysm (AAA) repair, colectomy, pancreatectomy, and total hip arthroplasty. We hypothesized that patients discharged on a weekend would have higher readmission rates compared with those discharged on a weekday following major surgery.

Approval for this study conducted from January to December 2012 was obtained from both the California Committee for the Protection of Human Subjects and the Stanford University institutional review board; obtaining patient consent was not applicable as the database used features a complex method of excluding identifiable personal information from individual-level data. The California Office of State Health Planning and Development (OSHPD) 2012 Patient Discharge Data (PDD) was used for this study. The OSHPD-PDD contains records for patients discharged from every general, acute, nonfederal hospital within the state. Demographic variables contained in the data set include age, sex, race/ethnicity, and a unique patient-level record linkage number to associate discharged patients with subsequent hospitalizations. Clinical information contained in the PDD includes principle diagnosis and up to 24 additional diagnoses with an indicator as to whether the condition was present on admission. The Charlson Comorbidity Index score was calculated based on present-on-admission comorbidities. The set also contains coding for the principle procedure and up to 19 secondary procedures performed during the index admission. Details about the admission include dates of admission and discharge, admission type (scheduled or unscheduled), expected payer/insurance, and disposition. A unique hospital identifier was used to identify the location of care for both index and subsequent readmission. Complications recorded included urinary tract infection, wound infection, myocardial infarction, deep venous thrombosis, pulmonary embolism, pneumonia, hemorrhage, and sepsis. Complications were assessed by using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes with concomitant coding that the diagnosis was not present on admission.

The ICD-9-CM codes were used to identify all patients undergoing the following primary procedures in 2012: AAA repair (38.34, 38.44, 38.64, and 39.71), colectomy or proctectomy (herein referred to as colectomy; 45.71-45.8, 48.41, 48.49, 48.5, 48.61-48.65, and 48.69), hip replacement (79.25, 79.35, 81.51, and 81.52), and pancreatectomy (52.51-52.6). All patients 18 years or older were included. Exclusion criteria included in-hospital mortality or transfer to a different inpatient facility. Patients were separated into 2 groups based on the day of discharge. Weekday was defined as Monday through Friday, while weekend was defined as Saturday and Sunday. Hospital readmission was defined as the first inpatient hospitalization for any reason at either 30 or 90 days following discharge from an index acute care hospitalization.

Demographic data, hospital variables, and readmission rates were directly compared for patients discharged on a weekend vs weekday after admission for each surgery type. The χ² test was used to compare mean values between groups, with statistical significance set as P < .05. The ICD-9-CM codes for the principal readmission diagnosis were assessed to identify the most common reasons for readmission. The hospital identifier was used to determine whether the readmission occurred at the index (discharging) hospital or a different facility. Univariate and multivariable logistic regression models were built to estimate the odds of hospital readmission based on weekend vs weekday discharge after controlling for other patient factors. All statistical analyses were 2 tailed and performed using SAS version 9.3 for windows (SAS Corp). Odds ratios (ORs) were considered significant when the 95% CI did not include 1 and the P value was less than .05.

In 2012, there were 128 057 patients hospitalized after undergoing AAA repair, colectomy, hip replacement, or pancreatectomy in nonfederal California hospitals. The cohort consisted of 5225 patients (4.1%) undergoing AAA repair; 29 388 (22.9%), colectomy; 91 168 (71.2%), hip replacement; and 2276 (1.8%) pancreatectomy. A total of 29 883 patients (23.3%) were discharged on a weekend and 98 174 (76.7%) discharged on a weekday, which was distributed similarly in each of the 4 procedures. Demographic data for each procedure are listed in Table 1. No significant differences were found between weekend and weekday discharge groups with respect to sex, age, race/ethnicity, insurance status, or type of admission.

Patients discharged on a weekend had a shorter hospital length of stay for each procedure (mean [SD], AAA repair: 4.3 [5.8] days vs 5.0 [8.0] days; colectomy: 7.3 [8.7] days vs 8.8 [11.2] days; hip replacement: 4.4 [5.4] days vs 5.2 [7.5] days; and pancreatectomy: 9.0 [10.6] days vs 9.4 [9.6] days). In addition, patients discharged on a weekend were more likely to be discharged to home compared with weekday discharges (AAA repair: 73.1% vs 69.0%; colectomy: 70.2% vs 62.2%; and pancreatectomy: 71.9% vs 64.9%) except for those undergoing hip replacement (25.9% vs 26.7%). Furthermore, fewer proportions of patients discharged on a weekend were sent to an SNF compared with weekday discharges (AAA repair: 6.4% vs 8.3%; colectomy: 6.5% vs 11.7%; hip replacement: 31.6% vs 34.8%; and pancreatectomy: 3.0% vs 7.1%).

Table 2 shows the unadjusted 30- and 90-day readmission rates associated for weekend and weekday discharges, stratified by procedure. Overall, the 30-day readmission rate was 9.4% after AAA repair, 13.6% after colectomy, 7.5% after hip replacement, and 16.3% after pancreatectomy. When comparing weekend and weekday discharges, there was no difference in the readmission rate after AAA repair (8.8% vs 9.3%; P = .55) or pancreatectomy (17.5% vs 15.9%; P = .40). However, weekend discharge was associated with a lower 30-day readmission rate for patients undergoing colectomy (12.1% vs 14.1%; P < .001) and hip replacement surgery (6.9% vs 7.7%; P < .001). Results for 90-day readmission rates revealed similar patterns (Table 2).

The most common reasons for readmission are shown in Figure 1. Wound-related complications were the most common reasons for readmission following colectomy (25.2%) and pancreatectomy (20.5%). A substantial percentage of readmissions occurred at a different hospital (AAA repair: 40.5%; colectomy: 25.8%; hip replacement: 32.5%; and pancreatectomy: 19.7%) than the index hospitalization (Figure 2).

Logistic regression analyses predicting the odds of readmission are presented in Table 3. On multivariable analysis, weekend discharge was inversely associated with readmission after colectomy (OR, 0.86; 95% CI, 0.79-0.93) but not AAA repair (OR, 0.93; 95% CI, 0.73-1.19), hip replacement (OR, 0.97; 95% CI, 0.91-1.03), or pancreatectomy (OR, 1.02; 95% CI, 0.76-1.36). Similar results were seen on univariate and multivariate logistic regression analysis predicting the odds of 90-day readmissions; therefore, the data are not shown.

We found significant associations between select patient factors (older age, male sex, black race, comorbidity index, and payer status), discharge disposition (discharge to SNF or acute rehabilitation facility), and hospital factors (presence of any complication, length of stay, and unplanned initial admission). These associations are also shown in Table 3.

Using a California statewide discharge database linking individual patient records from all nonfederal hospitals, we examined hospital readmission rates after AAA repair, colectomy, total hip replacement, and pancreatectomy. Contrary to our hypothesis, weekend discharge was not associated with an increased risk for hospital readmission. In fact, weekend discharge was inversely associated with readmission after colectomy. We did find other factors that predicted hospital readmissions, including black race, insurance status, occurrence of a postoperative complication, admission type (eg, emergency), and discharge disposition. We also found that a large proportion of patients was readmitted to a hospital different from the discharging hospital.

The correlation between select patient factors and readmissions after surgery are consistent with prior studies. We found an association between race and postoperative readmissions, as black individuals were more likely to be readmitted, similar to findings of previous studies.^15-17 Furthermore, while the role of insurance status in predicting hospital readmissions is not well understood,^18,19 our study found that Medicare and Medicaid insurance were associated with readmission, while lack of insurance was predictive against readmission. In addition, our results support previous findings that complications^20,21 and unplanned or urgent procedures²² are predictors of worse surgical outcomes. After controlling for confounding variables, we found that patients with any complication were almost 4 times more likely to be readmitted, and unplanned admissions were up to 12 times more likely to be readmitted compared with elective cases. The effect of discharge disposition remains less understood.^23,24 We found that SNF discharge was associated with the need for readmission after AAA repair and colectomy; however, on the contrary, it predicted against readmission in cases of hip replacement and pancreatectomy.

An important finding in our study was that a significant percentage of patients was readmitted to a different hospital than the discharging hospital. By using a unique method with individual record linkage numbers, only patients readmitted to a federal hospital or a hospital in a different state would be missed in this type of population-based analysis. To our knowledge, few prior studies have investigated the rate of readmission to alternative hospitals.^25-28 One study used OSHPD data with the California Cancer Registry to track readmissions following pancreaticoduodenectomy and found that 47% of readmissions occurred at a secondary hospital.²⁷ Because single-institution studies on hospital readmission underreport total readmission rates, future studies on hospital readmissions should use a population-based approach to accurately capture all readmissions following discharge.

One explanation for why weekend discharge was not associated with higher readmission rates in our study may be that current systems in place already do an effective job. For example, those patients considered at high risk for readmission may already be selected for weekday discharge. In our study, patients discharged to an SNF, an inherently high-risk group for readmission, in general were less likely to be discharged on a weekend. In addition, patients discharged on a weekend had shorter hospital length of stay and were more likely to be sent home, suggesting these patients may have been less medically complicated. Anecdotally, it is not uncommon for physicians to defer discharging high-risk patients until after the weekend. Despite fears of reduced staffing on the weekend, family members may be more available to help with the transition home. Finally, the weekend may also afford more opportunity for health care professionals and nursing staff, free from the time pressures of the weekday, to educate patients and caretakers in preparation for discharge. It is unclear why our results varied across procedures studied. It may be that a selection bias, toward discharging high-risk patients on a weekday, is even more pronounced after colectomy and hip replacement. Alternatively, these operations may have more established postoperative fast-track pathways that enable safe timely discharge regardless of the day of the week.^29,30

Some limitations to our study should be acknowledged. The use of administrative data has well-known limitations and the possibility of coding inaccuracy cannot be excluded. Specifically, readmission diagnoses based on administrative data should be cautiously interpreted. Sacks et al³¹ studied the reason for readmission following noncardiac surgery based on clinical medical record review compared with administrative coding and found a 30.8% discordant rate. In addition, details on the discharge process including differences, if any, between weekend and weekday cases were not available. Finally, the effect of weekend discharge on clinical outcomes, such as mortality, was not available.

Among patients hospitalized in California after major surgery, discharge on a weekend is not associated with increased hospital readmission. In hospitals with adequate staffing, appropriate coordination of care protocols and discharge systems in place, health care professionals may consider discharging postoperative patients on a weekend without concern for an increased need for readmission.

Corresponding Author: Jordan M. Cloyd, MD, Department of Surgery, Stanford University, 300 Pasteur Dr, MC5641, Stanford, CA 94305 (jcloyd@stanford.edu).

Accepted for Publication: March 14, 2015.

Published Online: June 3, 2015. doi:10.1001/jamasurg.2015.1087.

Author Contributions: Mr Ma and Dr Rhoads had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Cloyd, Rhoads.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Cloyd, Chen, Ma.

Critical revision of the manuscript for important intellectual content: Cloyd, Rhoads.

Statistical analysis: Ma, Rhoads.

Study supervision: Cloyd, Rhoads.

Conflict of Interest Disclosures: None reported.