“Weekend Effect” on 30-Day Readmissions among Stroke Survivors: An Analysis of the National Readmission Database

Hospitals and payers increasingly utilize 30-day readmission (30dRA) as a quality-of-care and cost reduction metric [1]. Readmission within a short time period after a stroke increases morbidity and mortality. Recent work showed that 12.1% of acute ischemic stroke (AIS) patients were readmitted within 30 days; 89.6% of the readmissions were considered unplanned and 12.9% were preventable [2].

Hospital services and quality of care that impact 30dRA may differ between weekends and weekdays. The discrepancy in discharge outcomes related to weekend versus weekday admission, termed the “weekend effect,” has been associated with adverse health outcomes for 23 of the 100 leading causes of death [3]. Previous studies indicated that weekend stroke admissions had a higher 30-day case fatality rate than admissions on week days [4-6]. A more recent study has shown similar 30-day case-fatality rates between week day and weekend admissions, but it was conducted in comprehensive stroke centers [7]. This may not be generalizable as dedicated stroke units that comply with comprehensive stroke unit principles may be protective against hospital case fatality due to the requirements needed for comprehensive stroke center certification [8]. Furthermore, the effect of weekend admission on 30dRA after stroke has not been assessed.

We hypothesized that stroke patients admitted at weekends would have higher odds of 30dRA than patients admitted on week days (i.e., Monday to Friday).

We utilized the Agency for Healthcare Research and Quality (AHRQ) Healthcare Utilization Project Nationwide Readmission Database (NRD) for 2013, a nationally representative, weighted probability sample of inpatient hospitalizations in the USA, regardless of insurance status. International Classification of Diseases Ninth Revision (ICD-9) codes (433.x1, 434.x1, and 436) were used to identify patients admitted with AIS. An AIS code in the first diagnostic position was classified as a primary stroke code. All patients with a primary AIS diagnosis of stroke were identified, and their initial (index) hospital admission was tagged for further investigation.

The exposure of interest, admission during a weekend, is a variable collected and reported in the NRD. The NRD defined weekend admissions as admission from 12:01 a.m. Saturday to 11:59 p.m. Sunday [9]. Given that the data in our analysis covered a 1-year time span, we stopped our analysis on 1 December, to ensure an entire 30-day period for a patient’s readmission to occur.

We defined 30dRA as any admission within 30 days of index hospitalization discharge. We excluded same-day events, readmissions exceeding 30 days from index hospitalization discharge, and in-hospital deaths. In a sensitivity analysis, we examined whether admission on the weekend was associated with 7-day and 60-day readmissions of AIS patients. Using previously validated methods, we classified readmission events as planned or unplanned, also identifying the number of readmissions deemed preventable due to ambulatory care-sensitive conditions [10]. Demographic characteristics such as age, sex, insurance type, number of chronic conditions, hospital teaching status, hospital bed size, and length of stay are saved in the NRD. Treatments including intravenous tissue-type plasminogen activator (tPA; ICD-9 code 99.1) or endovascular therapy (ICD-9 code 39.74) were determined by using procedure codes. AHRQ comorbidity measures using ICD-9-CM codes were assessed as potential covariates.

We computed proportions of patients with 30-day readmission overall, and unplanned and potentially preventable readmissions. Survey design logistic regression models were fit to estimate weighted crude odds ratios (OR) and 95% confidence intervals (CI) for the association between weekend admission and 30dRA. Considering the results were not significant, we did not adjust the models for any covariates of interest.

Of 319,317 patients with an index AIS, 38,625 (12.1%) experienced 30dRA. Patient demographics stratified by 30dRA are shown in Table 1. Approximately 26% of readmitted patients were initially a weekend admission. Age did not differ between those with and without 30dRA (71.8 vs. 70.6 years). Patients who experienced 30dRA also had more chronic conditions (mean 7.90.1) than those without 30dRA (mean 7.00.05). Index AIS patients with 30dRA also had a longer stay, increased All Patient Refined DRG severity (APDRG Severity), more cardiovascular-related comorbid conditions, less treatment with tPA, and higher mortality.

There was no association between weekend admission and all-cause 30dRA (OR 1.02; 95% CI 0.98–1.06) or between weekend admission and potentially preventable 30dRA (OR 1.03, 95% CI 0.99–1.08). There was no association of weekend admission on 7dRA (OR 1.01, 95% CI 0.95–1.07) or 60dRA (OR 1.01, 95% CI 0.98–1.05) (Table 2).

The final models were not adjusted for Table 1 elements due to no association between our primary predictor and primary outcome. Overall, the patients who were readmitted had a higher proportion of poor outcomes after their readmission hospitalization than the stroke patients who were not readmitted.

We found no relationship between weekend admissions and 30-day readmissions among AIS patients, indicating that the “weekend effect” previously described for other acute conditions may not extend to all post-stroke outcomes. Neither all-cause nor potentially preventable readmissions were associated with a weekend admission.

Our results were consistent with some other US studies suggesting that weekend AIS admissions have similar quality-of-care outcomes to those for week days [7, 11]. However, 30-day case fatality was the primary outcome of interest in those papers whereas we utilized 30-day readmissions in our study. The suggested decrease in stroke-related fatality indicates that other outcomes such as readmission should be investigated when exploring potential differences in quality of care. The absence of a weekend effect in our study could be due to an improved quality of care compared to in studies conducted previously. The discrepancy in discharge outcomes, particularly case fatality, related to weekend versus week day admissions, has been associated with adverse health outcomes for 23 of the 100 leading causes of death [3]. Studies investigating the weekend effect on mortality during admission around the world, e.g., in England, Australia, and Taiwan, show an increased risk of mortality from acute medical conditions such as myocardial infarctions during weekend admissions [12-14]. Hospital access and homogeneity of 24-h coverage 7 days a week may be influential factors to improved quality of care. Similar proportions of senior staffing 7 days a week who are capable of performing elective procedures, such as tPA, can potentially decrease adverse ischemic stroke outcomes. A reduced length of stay, a decreased risk of nosocomial infection, advanced neuroimaging, and early outpatient follow-up are factors that may have additionally improved quality of care over time. Other factors, such as whether the hospital is an academic center, or has a dedicated stroke unit, might influence post-stroke outcomes including readmission [7]. We explored whether these associations differed in different subgroups, but they remained the same in the subgroup analyses.

Our study is not without limitations. The NRD is an administrative dataset, and the ICD-9 CM codes are primarily used for billing purposes with a limited number of data elements. Since this study is based in the USA, the results may not extend to other health systems or countries. We had no information on severity of stroke, stroke etiology, stroke subtype, or individual functional ability prior to stroke, which could influence both weekend admissions and 30-day readmission after a stroke. Future studies should use either the US National Institutes of Health Stroke Score or the level of consciousness as a surrogate for stroke severity. We had no information on whether patients died after leaving the hospital, and no information on health care utilization patterns that might prevent readmission. Finally, these results are only generalizable to ischemic stroke hospital readmissions that take place in the USA.

These results demonstrate that 30-day readmissions due to ischemic stroke hospitalizations during the business week and at the weekend are similar. We inspected 7-day, 30-day, and 60-day readmissions to capture critical time intervals that would potentially allude to specific quality-of-care predictors. We used a large, nationally representative dataset that meant that a false-negative result was unlikely. While optimistic, these results require confirmation in future prospective studies.

The Institutional Review Board at our center waived review of this study due to the absence of identifiable patient data.

The authors declare that there are no conflicts of interest to report. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the NIH.

Dr. Boehme is supported by NINDS NIH R03 NS101417 and NINDS NIMHD R21 MD012451. Dr. Elkind serves on the National, Founders Affiliate, and New York City chapter boards of the American Heart Association/American Stroke Association, and receives royalties from UpToDate for chapters related to cryptogenic stroke.