Effect of Weekend Compared With Weekday Stroke Admission on Thrombolytic Use, In-Hospital Mortality, Discharge Disposition, Hospital Charges, and Length of Stay in the Nationwide Inpatient Sample Database, 2002 to 2007
Background and Purpose—A stroke “weekend effect” on mortality has been demonstrated in other countries with a possible slight effect in the United States. We studied patients with stroke in the Nationwide Inpatient Sample database for a weekend effect on thrombolytic use, in-hospital mortality, discharge disposition, hospital charges, and length of stay.
Methods—The Nationwide Inpatient Sample 2002 to 2007 was searched for all emergency room admissions for International Classification of Diseases, 9th Revision codes corresponding to ischemic stroke. Generalized estimated equations for generalized linear models were performed, adjusting for gender, age, race, season, median income level, payer, comorbidity score, hospital region, hospital location, teaching status, bed size, and hospital annual stroke case volume to compare weekend versus weekday stroke admission incidence of thrombolytic use, in-hospital mortality, discharge disposition, hospital charges, and length of stay. The same analysis was performed using the International Classification of Diseases, 9th Revision codes for ischemic stroke AND transient cerebral ischemia to check internal validity for coding irregularities that may occur in differentiating stroke from transient ischemic attack.
Results—There were 599 087 emergency room admissions for ischemic stroke: 159 906 weekend admissions and 439 181 weekday admissions. Generalized estimated equation for generalized linear model analysis was performed and demonstrated weekend compared with weekday patients with stroke were slightly more likely to receive thrombolytics (OR=1.114; 95% CI=1.039 to 1.194; P=0.003); incur slightly higher total hospital charges (effect ratio=1.011; 95% CI=1.006 to 1.017; P<0.001); and have slightly longer lengths of stay (effect ratio=1.021; 95% CI=1.015 to 1.027; P<0.001). There was no difference in in-hospital mortality or discharge disposition.
Conclusions—There is a slight stroke weekend effect on thrombolytic use, total hospital charges, and length of stay, but no difference in in-hospital mortality or discharge disposition.
Hospital admission for ischemic stroke over the weekend compared with a weekday is associated with higher mortality in Canada,1 Sweden,2 Japan,3 and Taiwan.4 In England, Wales, and Northern Ireland, patients with stroke admitted on weekends wait longer to be admitted into a stroke unit and are less likely to have a brain scan within 24 hours.5 In the United Kingdom, weekend patients with stroke are less likely to receive thrombolytics,6 whereas in Germany, they are more likely.7
Studies of the stroke “weekend effect” in the United States, however, have been inconsistent. In the Get With the Guidelines–Stroke Program, off-hour ischemic stroke presentation (weekends and weeknights) is associated with slightly higher in-hospital mortality.8 The difference was small (0.6%), however, with a number of 166 needed to harm.8 In North Carolina, patients with acute ischemic stroke admitted on weekends wait longer to undergo a CT scan.9
In Virginia, weekend patients with stroke are more likely to receive tissue plasminogen activator, albeit with no improvement in mortality.10 In a study of 2 comprehensive stroke centers, there was no difference in discharge disposition, discharge, or 90-day modified Rankin score or 90-day mortality between patients with stroke admitted on weekends compared with weekdays.11
We studied the stroke “weekend effect” on thrombolytic use, in-hospital mortality, discharge disposition, hospital charges, and length of stay using the Nationwide Inpatient Sample (NIS) database for the years 2002 to 2007. The NIS is the largest all-payer hospital inpatient database in the United States and contains data approximating a 20% stratified sample of US hospitals. The NIS includes data for approximately 8 million hospital admissions each year, approximately one fifth of all inpatient admissions to US nonfederal hospitals.
We obtained the NIS database from the Agency for Healthcare Quality and Research’s Healthcare Cost and Utilization Project (Rockville, Md). For more information regarding the NIS database, see www.hcup-us.ahrq.gov/nisoverview.jsp.
Hospitalizations for ischemic stroke from 2002 to 2007 were collected from the NIS by International Classification of Diseases, 9th Revision codes 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 437.1. International Classification of Diseases, 9th Revision code 436 was not included because it is defined as “excludes: cerebrovascular accident, CVA (ischemic), embolic, hemorrhagic, thrombotic, postoperative cerebrovascular accident, stroke (ischemic), embolic, hemorrhagic, thrombotic.” A secondary analysis was performed adding hospitalizations with International Classification of Diseases, 9th Revision codes for transient cerebral ischemia 435.0, 435.1, 435.2, 435.3, 435.8, and 435.9 to check internal validity for coding irregularities that may occur in differentiating stroke from transient ischemic attack.
The NIS contains data on which hospitalizations were weekend admissions (Saturday to Sunday). Only admissions through the emergency room were included, because NIS data for transferred patients do not include the admission day to the initial facility. Weekend stroke admissions were compared with weekday stroke admissions for 5 end points: hospital mortality, discharge disposition (home or short-term facility versus long-term facility versus death), thrombolytic use, total hospital charges, and length of hospitalization.
The NIS contains data on discharge disposition, which we grouped into the 3 following outcomes: (1) home or short-term facility (routine, short-term hospital, against medical advice, home intravenous provider, another rehabilitation facility, another institution for outpatient services, this institution for outpatient services, discharged alive, destination unknown) versus (2) long-term facility (skilled nursing facility, intermediate care, hospice home, hospice medical facility, long-term care hospital, certified nursing facility) versus (3) death (died in the hospital, died at home, died in a medical facility, died in place unknown).
The analysis was adjusted for the following patient-specific factors, which are coded in the NIS database: gender, age, race, season (winter, spring, summer, fall), median income level in patient’s postal (zip) code (<$36 000, $36 000 to $44 999, ≥$45 000), payer (Medicare, Medicaid, private insurance, self-pay, or no charge), and a comorbidity score defined as the summation of the number of comorbidities entered for each patient in the NIS.12 The models also accounted for hospital-level factors: hospital region (Northeast, Midwest, South, West), hospital location (rural, urban nonteaching, urban teaching), bed size (small, medium, large), and hospital annual case volume of stroke admissions.
For each sampled hospital, all inpatient admissions for the year are contained in the NIS, permitting annual case volumes for hospitals to be calculated. Hospital annual case volume of stroke admissions was determined by the ranking of total stroke admissions across all sampled hospitals. For each year from 2002 to 2007, each individual hospital’s case volume of stroke admissions was defined as high if the number of total stroke admissions equaled or exceeded the third quartile of total stroke admissions across all sampled hospitals and defined as low if otherwise.
To simultaneously account for hospital- and patient-level variation in each of the 5 end points, generalized linear models with the use of generalized estimated equations were chosen to compare and make inferences about the differences between charges for weekend and weekday stroke admissions. In multivariate models such as the ones we used, it is possible to detect separate effects at the individual hospital level (eg, that certain hospitals provide expensive or inexpensive care across all patients treated) and at the level of specific procedures across different hospitals. The models accounted for data correlations by assuming exchangeability among admissions from the same hospital. Management (thrombolytic use, total hospital charges, or length of hospitalization) and outcomes (discharge disposition or in-hospital death) of patients with stroke admitted to the same hospital were likely to be correlated, and the correlation was assumed to be a constant within a hospital. Given a mixed scale of measurements for the 5 focused end points, generalized linear models used an integrated approach to facilitate comparison between weekday and weekend stroke admissions at the same time as adjusting for hospital- and patient-level characteristics.
For each stroke admission, 5 models (thrombolytic use, in-hospital mortality, discharge disposition, total hospital charges, and length of stay) were fitted. Bonferroni correction for multiple tests resulted in a probability value <0.01 (0.05/5) being considered statistically significant. In comparing total hospital charges across the 6 years from 2002 to 2007, we assumed a 3% annual inflation rate for each year and used the adjusted charges in the generalized estimated equation for generalized linear models to evaluate differences between procedures. To meet the distributional requirements of a generalized linear model, we used the logarithm of length of hospital stay and the logarithm of total inflation-adjusted charges as targeted outcomes in analyses. For patients who stayed in the hospital for <1 day, a 1-day stay was assumed.
A search in the NIS database years 2002 to 2007 yielded a total of 599 087 emergency room admissions for ischemic stroke: 159 906 weekend admissions and 439 181 weekday admissions. The patient demographic and hospital characteristics of the weekend and weekday patients with ischemic stroke are shown in Table 1.
Descriptive statistics for the incidence of thrombolytic use, in-hospital mortality, discharge disposition, hospital charges, and length of hospital stay for weekend compared with weekday stroke admission are shown in Table 2. These demonstrate no crude difference between weekend and weekday stroke admissions for the end points studied; however, these comparisons are unadjusted for hospital-specific and patient-specific factors and may not be inferential given the potential substructures of weekend and weekday stroke populations, which can be introduced by hospital and patient characteristics.
To make valid inferences about the association between weekend versus weekday stroke admission and the targeted end points, we used generalized estimated equation for generalized linear models to account for both hospital- and patient-level variation (Table 3) and found that weekend compared with weekday stroke admission patients were slightly more likely to receive thrombolytics (OR=1.114; 95% CI=1.039 to 1.194; P=0.003); incur slightly higher total hospital charges (effect ratio=1.011; 95% CI=1.006 to 1.017; P<0.001); and have slightly longer lengths of stay (effect ratio=1.021; 95% CI=1.015 to 1.027; P<0.001). The mean hospital charge for weekend admissions was higher by 1.1% of the mean hospital charge for weekday admissions, whereas the mean length of stay for weekend admissions was higher by 2.1% of the mean length of stay for weekday admissions. There was no difference in in-hospital mortality or discharge disposition.
There is a concern that coding irregularities may occur in differentiating ischemic stroke from transient ischemic attack in the database. Therefore, to check the internal validity of the results found, a secondary analysis was performed adding hospitalizations with International Classification of Diseases, 9th Revision codes for transient cerebral ischemia: 435.0, 435.1, 435.2, 435.3, 435.8, and 435.9. This secondary analysis demonstrated the same effects and magnitude of effects as the primary analysis on thrombolytic use, total hospital charges, lengths of stay, in-hospital mortality, and discharge disposition (Table 4).
Stroke is the third leading cause of death and the leading cause of disability in the United States. There are 795 000 people who have a stroke in the United States each year; every 40 seconds someone has a stroke.13 Outcomes from stroke treatment are thought to be time-dependent like myocardial infarction, in which “door-to-balloon” time is critical. Recent analysis has demonstrated that weekend admission for myocardial infarction is associated with higher mortality and that this is due to lower weekend access to cardiac invasive procedures.14 This has been described as the “weekend effect.”
A number of studies have been performed analyzing a weekend effect for stroke in other countries. A recent study of the Hospital Morbidity Database of Canada found in 26 676 patients admitted to 606 hospitals in Canada for ischemic stroke from April 2003 to March 2004 that weekend admissions (6609 [24.8%]) had significantly higher 7-day mortality than their age-, gender-, comorbidity-, and major medical complication-adjusted weekday counterparts.1
Similarly, weekend versus weekday patients with stroke were studied in the Swedish Hospital Discharge Register for the time periods 1968 to 1979 (6048 weekend versus 23 323 weekday patients); 1980 to 1989 (15 278 versus 52 226); 1990 to 1999 (39 033 versus 122 924); and 2000 to 2005 (27 179 versus 81 390).2 Age- and gender-adjusted weekend patients had significantly higher mortality and were significantly less likely to be discharged to their same place of residence.
In a prospective study of 1134 patients with stroke across 10 centers in Japan, weekday admission was an independent negative predictor of case fatality and a positive predictor of favorable outcome from acute stroke units; and for patients in rehabilitative therapy, weekday admission was associated with favorable outcome.3 The authors attributed this to the fact that in most acute stroke units in Japan, staffing level is lower on weekends and holidays and provision of rehabilitative services occurs only on weekdays.
In a study of 34 347 patients with ischemic stroke admitted to 245 hospitals in Taiwan in the Taiwanese National Health Insurance Research Database in 2005, weekday patients with stroke had decreased 30-day mortality than their gender-, age-, comorbidity-, surgery-, physician age-, physician specialty-, hospital ownership-, accreditation level-, teaching status-, geographic location-, regional resources-, and competition-adjusted weekend counterparts.4
Data from the 2004 National Stroke Audit from 246 hospitals in England, Wales, and Northern Ireland demonstrated that patients with stroke admitted on weekends waited longer to be admitted into a stroke unit and were less likely to have a brain scan within 24 hours.5 Data from the United Kingdom centers in the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST) project were analyzed and demonstrated weekend patients with stroke were less likely to receive thrombolytics.6
A few studies have demonstrated no weekend effect of stroke. Four year-data of 37 396 patients with stroke (2003 to 2006), from a prospective, hospital-based stroke registry for the federal state Hesse, Germany, demonstrated that patients admitted during nonoffice hours (weekend or nighttime) did not have different outcome or mortality than patients admitted during office hours after adjustment for clinical state and admission latency.7 Nonoffice hour patients, however, were more likely to receive thrombolytics. This is similar to the findings in the present study.
An analysis of 1578 patients with stroke in the Takashima Stroke Registry from 1988 to 2003 demonstrated that weekend day of admission, not the day of stroke onset, affected stroke fatality rates (although this did not reach statistical significance), suggesting that the “weekend effect” of stroke deaths is an artifact of referral bias.15 Our study eliminated this artifact by only including emergency room admissions and excluding hospital transfers.
Most of the studies of the stroke weekend effect, however, were conducted in other countries where the healthcare delivery system is different than the United States. There have been relatively few studies conducted in the United States. One such study analyzed 187 669 acute ischemic stroke admissions to 857 hospitals participating in The Get With the Guidelines–Stroke Program from 2003 to 2007 and demonstrated that off-hour admissions (weekends and weeknights) were associated with slightly higher in-hospital mortality than office-hour admissions: 5.8% versus 5.2%, respectively (P<0.001). The small 0.6% absolute difference translated into 166 number needed to harm.8
A multiple linear regression analysis of 20 374 patients with stroke enrolled in the North Carolina Collaborative Stroke Registry from January 2005 to April 2008 to study predictors of delay time from hospital arrival until CT scan found that among other factors, weekend and evening time arrival were associated with delay in CT scan.9
A study of 78 657 patients with acute ischemic stroke in the Virginia Patient Data System from January 1, 1998, to June 30, 2006, demonstrated weekend patients were more likely to receive tissue plasminogen activator but had no difference in mortality than their weekday counterparts.10 Prospective data on 2211 patients with stroke admitted to 2 comprehensive stroke centers were analyzed and showed that weekend patients had no difference in discharge disposition, discharge, or 90-day modified Rankin score or 90-day mortality than weekday patients.11
In this present study, we found no weekend effect on in-hospital mortality or discharge disposition, the same findings as the Albright study,11 and similar to The Get With the Guidelines–Stroke study, which only found a slight difference.8 We only included emergency room admissions for stroke to eliminate the artifact from patients transferred from other hospitals on days other than their day of admission that was found in the Takashima Stroke Registry study.15 We chose to only study ischemic stroke, because there have been previous studies using the NIS database to analyze a weekend effect on intracerebral hemorrhage16 and subarachnoid hemorrhage.17
Interestingly, in this present study, we found that weekend patients with stroke were slightly more likely to receive thrombolytics but there was no difference in in-hospital mortality, which is similar to the findings from the Virginia Patient Data System.10 There are several possible explanations for the increased likelihood for thrombolytic treatment on weekends. One possibility suggested by Kazley et al is that weekend patients with acute ischemic stroke present to hospitals earlier because they are unencumbered by work or traffic issues. This may translate to a higher proportion of these patients presenting within the 3-hour window for administration of intravenous tissue plasminogen activator.10 Another possibility suggested by Kazley et al is that elective surgical procedures at hospitals on weekends are rare; therefore, weekend patients with stroke may have quicker access to diagnostic imaging and tests and possibly quicker evaluation and determination of treatment.10 Still, another possibility is that physicians may be busy on weekdays with clinics and other clinical obligations, whereas on weekends, they may be more readily available to treat patients with acute ischemic stroke.10 An alternative explanation is that weekend patients with stroke have more severe strokes and that physicians are more willing to use thrombolytics for patients with severe strokes. A recent study of patients with acute coronary syndrome found that weekend patients tended to have a higher rate of ST-elevation myocardial infarction.18 Still, that thrombolytic use is greater on the weekend than the weekday might not intuitively make sense, and this finding may be affected by the study limitations of administrative data coding for severity of stroke and accurate reporting of thrombolytic use. Further studies such as prospective stroke registries or other large nationwide hospital data sets are needed to confirm this finding.
Although there was no weekend effect on in-hospital mortality and discharge disposition, there was a slight weekend effect on total hospital charges and length of hospitalization. One explanation is that although hospitals may be able to administer acute treatments such as thrombolytics or intensive care management on weekends, other services such as physical therapy, occupational therapy, nutritional services, speech/swallow therapy, and case management and discharge disposition services may be shorter staffed on weekends, contributing to longer hospitalizations and thus higher total charges. Another explanation is that weekend patients were more likely to receive thrombolytics, adding to hospital charges and potentially to length of hospitalization.
There are several limitations to the present study. The first limitation is the retrospective nature of the study, which has significant potential for selection bias. However, by nature, a study of the weekend effect cannot be a prospective randomized controlled trial; one cannot randomize a patient to weekend versus weekday. The second limitation is the risks inherent to coding error. In a large database such as the NIS, there is significant potential for coding error or variability in coding. We attempted to address one possible coding artifact by performing a secondary analysis adding the International Classification of Diseases, 9th Revision codes for transient ischemic attack to check for internal validity with coding irregularities that may occur in differentiating ischemic stroke from transient ischemic attack and found the statistical inference results to be the same. Another possibility for coding error is potential underreporting of recombinant tissue plasminogen activator use as was reported by Kleindorfer et al in an analysis of the Medicare Provider and Analysis Review data set and The Premier Hospital data set.19 This group estimated that the true rate of recombinant tissue plasminogen activator use in the United States is 1.8% to 2.1%, which is slightly higher than the 1.5% to 1.6% we found in our analysis. Nevertheless, if recombinant tissue plasminogen activator use is underreported, it should be equally underreported for weekday versus weekend admission patients with stroke and therefore should not affect our analysis of differences between weekday versus weekend stroke admission. The third limitation is that the NIS does not contain data on the neurological condition of patients; therefore, the analysis could not be adjusted for severity of stroke, a source for significant bias. A fourth limitation is that the NIS contains data on day of the week admission, which allows for an analysis of weekday versus weekend admission but does not allow an analysis of statutory holidays. Statutory holidays only account for 10 days of a 365-day year (2.7%), so the effect of this artifact is likely to be small.
The findings of this present study will hopefully lead to further analysis of the differences in diagnostic evaluation, imaging, and treatment provided on weekends compared with weekdays for patients with ischemic stroke. A careful analysis may reveal that stroke treatment across healthcare centers will need to more closely follow current stroke guidelines whether during the week or weekend, leading to overall improvement in stroke care and outcomes.
In the NIS database, 2002 to 2007, weekend patients with stroke were slightly more likely to receive thrombolytics, had slightly higher total hospital charges, and slightly longer length of hospitalization but no difference in in-hospital mortality or discharge disposition than weekday patients.
B.L.H. received honorarium from Codman Neurovascular and Actelion Pharmaceuticals, was the principal investigator of a sponsored registry, Micrus Endovascular, and received support for research from the National Institutes of Health, the Brain Aneurysm Foundation, and the Thomas H. Maren Foundation. J.M. received an unrestricted educational grant from Codman Neurovascular and is a consultant for Actelion Pharmaceuticals, Nfocus, and Lazarus Effect. M.F.W. received support for research from the National Institutes of Health.
- Received May 19, 2010.
- Revision received June 19, 2010.
- Accepted June 30, 2010.
Saposnik G, Baibergenova A, Bayer N, Hachinski V. Weekends: a dangerous time for having a stroke? Stroke. 2007; 38: 1211–1215.
Janszky I, Ahnve S, Ljung R. Weekend versus weekday admission and stroke outcome in Sweden from 1968 to 2005. Stroke. 2007; 38: e94.
Rudd AG, Hoffman A, Down C, Pearson M, Lowe D. Access to stroke care in England, Wales and Northern Ireland: the effect of age, gender and weekend admission. Age Ageing. 2007; 36: 247–255.
Lees KR, Ford GA, Muir KW, Ahmed N, Dyker AG, Atula S, Kalra L, Warburton EA, Baron JC, Jenkinson DF, Wahlgren NG, Walters MR; SITS-UK Group. Thrombolytic therapy for acute stroke in the united kingdom: experience from the Safe Implementation of Thrombolysis in Stroke (SITS) register. QJM. 2008; 101: 863–869.
Reeves MJ, Smith E, Fonarow G, Hernandez A, Pan W, Schwamm LH; GWTG-Stroke Steering Committee & Investigators. Off-hour admission and in-hospital stroke case fatality in the Get With The Guidelines–Stroke program. Stroke. 2009; 40: 569–576.
Rose KM, Rosamond WD, Huston SL, Murphy CV, Tegeler CH. Predictors of time from hospital arrival to initial brain-imaging among suspected stroke patients: the North Carolina Collaborative Stroke Registry. Stroke. 2008; 39: 3262–3267.
Lloyd-Jones D AR, Brown TM, Carnethon M, Dai S, De Simone G, Ferguson TB, Ford E, Furie K, Gillespie C, Go A, Greenlund K, Haase N, Hailpern S, Ho PM, Howard V, Kissela B, Kittner S, Lackland D, Lisabeth L, Marelli A, McDermott MM, Meigs J, Mozaffarian D, Mussolino M, Nichol G, Roger V, Rosamond W, Sacco R, Sorlie P, Stafford R, Thom T, Wasserthiel-Smoller S, Wong ND, Wylie-Rosett J; on behalf of the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics—2010 update. A report from the American Heart Association. Circulation. 2010; 121: e46–e4215.
Kostis WJ, Demissie K, Marcella SW, Shao YH, Wilson AC, Moreyra AE; Myocardial Infarction Data Acquisition System (MIDAS 10) Study Group. Weekend versus weekday admission and mortality from myocardial infarction. N Engl J Med. 2007; 15: 1099–1109.
Crowley RW, Yeoh HK, Stukenborg GJ, Medel R, Kassell NF, Dumont AS. Influence of weekend hospital admission on short-term mortality after intracerebral hemorrhage. Stroke. 2009; 40: 2387–2392.
Kleindorfer D, Lindsell CJ, Brass L, Koroshetz W, Broderick JP. National US estimates of recombinant tissue plasminogen activator use: ICD-9 codes substantially underestimate. Stroke. 2008; 39: 924–928.