Hospital Care for Patients Experiencing Weekend vs Weekday Stroke: A Comparison of Quality and Aggressiveness of Care

Background  We examined the quality and aggressiveness of care for the treatment of acute ischemic stroke (AIS) on weekends vs weekdays. Acute ischemic stroke is a leading cause of death and disability in the United States, and aggressive treatment must be provided within 3 hours for optimal patient outcomes. Because of this short treatment window for the administration of tissue plasminogen activator, patients need around-the-clock access to high-quality and aggressive care.

Objective  To determine whether there is a difference in the quality or aggressiveness of care for patients experiencing AIS on weekends vs weekdays.

Design  Retrospective study.

Setting  Academic research.

Patients  We conducted a retrospective study of patients with AIS in Virginia. Two logistic regression analyses assessed the relationship between weekend admission and quality and aggressiveness of care, while controlling for appropriate patient-level and hospital-level control variables. A propensity score stratification approach controlled for selection bias.

Main Outcome Measures  Treatment with tissue plasminogen activator and in-hospital mortality.

Results  Patients with AIS admitted on weekends are more likely to receive tissue plasminogen activator than those admitted on weekdays (P < .05). No statistically significant difference was noted in patient mortality based on day of admission (P ≥ .05). We detected no difference in the likelihood to seek hospital care on weekends between patients with AIS vs patients with hemorrhagic stroke.

Conclusions  Patients experiencing AIS are more likely to receive tissue plasminogen activator on weekends than on weekdays. Patients experiencing AIS who are admitted on weekends are no more likely to die than those who are admitted on weekdays. Further research is necessary to understand differences in weekend vs weekday care.

An intense focus has been placed on the quality of health care in the United States. One challenge is the notion that quality is difficult to measure, yet it has been identified as an aim of the processes and outcomes of care.¹ While all patients are presumably interested in the outcomes of their care, many are likely interested in the aggressiveness of the treatment they receive as they face serious medical conditions. Despite not being universally used, aggressive care is appropriate and cutting edge and represents a desired standard of care. Variability and disparities in health care have been shown to be related to many factors, including sex, race/ethnicity, residential proximity to a stroke center, and time of hospital admission.^2-4 One area of concern (and a focus for research) has been the temporal effect of seeking care and the quality of care received. Although hospitals operate around the clock every day of the year, there are inevitable staffing differences during shifts, and there may be differences in the availability of diagnostic modalities or treatment options for care. According to Bell and Redelmeier,⁵ hospital staffing levels are lower on weekends than on weekdays. Other researchers have speculated that the best-trained, brightest, and most competent staff members do not work on weekends.^6,7 Some investigators claim that it is a combination of shortages in resources, expertise, and providers (ie, physicians and allied health professionals) on weekends that leads to differences in the processes of care.⁸ Regardless of the potential reasons for the difference in care, a body of literature has evolved that examines variations in the processes and outcomes of patient care based on day of admission.

Stroke is a leading cause of death in the United States.⁹ Besides being an unexpected and sudden event requiring medical care, it is a leading cause of long-term disability in the United States, leading to significant expense in treatment and lost quality-of-life years for patients.^9,10

In 1996, the use of tissue plasminogen activator for the treatment of acute ischemic stroke (AIS) began following approval by the Food and Drug Administration.¹¹ Tissue plasminogen activator has been shown to significantly decrease disability for patients with AIS.¹² Despite these facts, tissue plasminogen activator use remains infrequent and must be administered to patients with AIS within 3 hours from the onset of symptoms. Alberts et al¹² note that tissue plasminogen activator is appropriately given in hospitals that have the following features: an acute stroke team, a stroke unit, written care protocols, an integrated emergency response system, 24-hour computed tomography availability, and rapid laboratory testing. Many hospitals that have these key elements attain stroke center status through state guidelines, the Joint Commission on Accreditation of Healthcare Organizations, or the American Stroke Association–American Heart Association.^13,14 To accommodate the significant resources necessary to treat patients with AIS using tissue plasminogen activator, Medicare increased reimbursement substantially for this treatment in October 2005.¹⁵

The objectives of this study were to further explore the notion that weekend care in the United States is inferior to weekday care in quality and that patients treated on weekends are less likely to receive aggressive care and are more likely to die. Because stroke is unexpected, is potentially serious, and requires emergency care, it is appropriate to examine the rates of aggressive treatment and mortality and the relationship to day of hospital admission. We hypothesized that there would be no difference in the quality or aggressiveness of stroke care based on whether a patient was admitted on a weekend or a weekday.

Using an archival billing data set, we assessed the relationship between weekend and weekday stroke care. A control group of patients with hemorrhagic stroke was used to examine any unobservable confounding patient factors that influenced patterns of seeking care. Regression techniques were used to analyze the relationship between the day of hospital admission (weekend or weekday) and the quality and aggressiveness of care.

We used inpatient data for all patients with AIS discharged from all Virginia hospitals from January 1, 1998, to June 30, 2006. The control group was captured from the same data set and included patients with hemorrhagic stroke for the same period. These data are available from the Virginia Patient Data System provided by Thomson Reuters (http://www.ThomsonReuters.com/products_services/healthcare/). Patients were identified using International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes for AIS. Data included patients from 112 hospitals, 8 of which became designated as stroke centers during the study period. The hospitals are geographically dispersed throughout Virginia, including urban and rural areas.

The sample was identified using ICD-9 diagnosis codes within diagnosis related groups 14 and 559 in a primary or secondary position. Tissue plasminogen activator use was identified using ICD-9 procedure code 99.10.

Aggressive treatment was determined by the use of tissue plasminogen activator. Treatment outcome was assessed by mortality during the hospital stay for patients with AIS and for patients with hemorrhagic stroke. For the regression analysis, the models were adjusted for age, sex, race/ethnicity, and residential proximity to a stroke center, as well as atrial fibrillation treatment for patients with AIS. Age, sex, and race/ethnicity were included to control for any potential treatment disparities of care. Because tissue plasminogen activator must be administered within 3 hours from the onset of symptoms, we included the patient's residential proximity to a stroke center, calculated based on estimated driving time from a patient's zip code. Atrial fibrillation treatment was included in the model because patients with AIS are clinically indicated to receive tissue plasminogen activator unless they are receiving anticoagulant therapy. These patients accounted for almost 20% of our study population. Hospital variables included size, stroke case volume, and stroke center status as defined by the Joint Commission on Accreditation of Healthcare Organizations. Hospital size was defined as small (<200 beds), medium (200-499 beds), or large (≥500 beds). The stroke case volume seen in each hospital is defined as low (<100 cases), medium (100-249 cases), or high (≥250 cases). We also included a variable to control for the time of treatment (before or after the change in tissue plasminogen activator treatment reimbursement) because we recognize that this may have potentially influenced the type of care a patient received.

This study used 2 logistic regression analyses. The first logistic analysis had tissue plasminogen activator treatment as the dependent variable. The second logistic regression analysis had patient mortality as the dependent variable.

A propensity score stratification approach was used to identify any selection bias in the 2 groups and specified in the means as described by Rubin.¹⁶ Scores were calculated using logistic regression analysis of organizational and patient variables to predict the likelihood of being admitted on weekends vs weekdays. The scores were then divided into 5 equal strata to represent the levels of likelihood of weekend admission in the form of dummy variables. Included in the logistic regression analyses, these dummy variables adjust for the potential confounding patient factors that may increase the likelihood of weekend admission for any particular stratum of patients, allowing the model to provide unbiased estimates of the relationships for this patient population.¹⁶

Because there is also a chance that patients with AIS who are admitted on weekends differ from other patients because of lifestyle factors and patient or family reactions to symptoms, we used a comparison group of patients with hemorrhagic stroke to test for unobserved factors influencing the care-seeking behavior of patients such as marital status. This group of patients has symptoms that are often similar to those of AIS, but their treatment is not limited by a 3-hour response period. We reasoned that observed differences in weekend outcomes for the AIS and hemorrhagic stroke groups were likely differences in seeking care and not owing to variations in the timeliness of receiving tissue plasminogen activator. We tested for differences by using a logistic regression model to assess the likelihood of weekend admission, while controlling for other patient and organizational factors.

Expedited institutional review board approval was obtained. Informed consent was unnecessary.

There were 78 657 patients with AIS and 20 101 patients with hemorrhagic stroke identified in our sample. The baseline characteristics of these 2 populations are given in Table 1. There were 43 486 women with AIS and 35 165 men with AIS. Women accounted for 10 476 hemorrhagic strokes and men for 9624 hemorrhagic strokes. Six of the observations were excluded because of missing sex. The mean (71.11 years) and median (73.00 years) ages of patients with AIS were slightly older than the mean (66.07 years) and median (69.00 years) ages of patients with hemorrhagic stroke. In the AIS mortality analysis, 2978 patients (3.8%) were excluded because survival data were unavailable. Of the patients included in our analyses, 772 patients with AIS (1.0%) were treated with tissue plasminogen activator, and 5413 patients with AIS (6.9%) died in the hospital during the study period (January 1, 1998, to June 30, 2006).

The likelihood of weekend admission was not different for the AIS and hemorrhagic stroke groups. The model was significant, with a −2 log likelihood of 113 020.44, and the hemorrhagic stroke variable was not significant (P ≥ .05) in predicting weekend admission.

Tissue plasminogen activator was used in 229 patients with AIS admitted on weekends and in 543 patients with AIS admitted on weekdays. According to the model, patients with AIS admitted to the hospital on weekends are 20% more likely to be treated aggressively using tissue plasminogen activator (P < .05). Other significant predictors of tissue plasminogen activator use include patient age, sex, race/ethnicity, and atrial fibrillation treatment. Hospital factors that significantly relate to tissue plasminogen activator use include low stroke case volume, stroke center status, and period relating to reimbursement change. Patients at hospitals with low stroke case volume are less likely than those at hospitals with higher stroke case volume to receive tissue plasminogen activator. These results are summarized in Table 2.

The –2 log likelihood of the model is 8463.71 (Table 3), and inclusion of the propensity strata does not change the relationship between day of admission and aggressiveness of care. Patients admitted on weekends are still more likely to be treated with tissue plasminogen activator.

The mortality model demonstrated no difference in patient death related to day of hospital admission. A total of 3993 patients with AIS were admitted on weekdays compared with 1420 patients with AIS admitted on weekends. Significant predictors of mortality in our model include patient age, hospital size, hospital stroke case volume, patient residential proximity to a stroke center, tissue plasminogen activator treatment, and atrial fibrillation treatment. Patients who seek treatment at stroke centers are slightly less likely to die than those who do not seek treatment at stroke centers. Patients who live farther than a 30-minute drive from a stroke center are more likely to die. These results are given in Table 4.

Adjustment for selection bias using the propensity score did not substantially change the results. These findings are summarized in Table 5.

Because tissue plasminogen activator must be administered to a patient within 3 hours from the onset of stroke symptoms, aggressive identification and treatment regardless of the day of the week are essential for successful outcomes. For this reason, patients with AIS must quickly receive appropriate and standardized medical care to prevent the debilitating effects of this condition. If a patient experiences a stroke on Saturday or Sunday, waiting until Monday for aggressive treatment is not an option.

Recently, Kostis et al¹⁷ published the results of a study in which they found that patients admitted on weekends were less likely to receive aggressive care and experienced higher mortality. Using a sample of patients who sought care for myocardial infarction, their research shows that mortality is more common and care is less aggressive in cardiac patients admitted to the hospital on weekends. This study examined a single clinical condition in a single state using panel data. In a similar study, Becker¹⁸ found that cardiac patients admitted on weekends were significantly less likely to receive immediate intensive procedures and were more likely to experience adverse outcomes such as readmission and mortality.

Conversely, Hixson et al¹⁹ found no difference in mortality associated with weekend vs evening admission in pediatric intensive care units or the pediatric emergency department. This study examined a large patient population from 1996 to 2003 using standardized mortality ratios in a multivariable logistic regression analysis. When examining the crude mortality rates for weekend vs weekday admission, weekend mortality rates were higher. However, when the researchers adjusted for clinical factors, no statistically significant difference was found in patient mortality based on day of admission.

In a similar study, Rudd et al⁴ examined access to stroke care and the effect of patient age, sex, and weekend admission in England, Wales, and Northern Ireland. Using data collected from an audit of 8718 patients in 246 hospitals, they found that patients experiencing stroke on weekends had to wait longer for admission to a stroke unit. Patients admitted on weekends for stroke were also less likely to undergo a brain scan, with the researchers concluding that “the health service is ill-equipped to provide high-quality emergency stroke care at weekends.”^4(p253) In another study examining patient stroke outcomes in Canada, Saposnik et al⁸ found that patients with stroke admitted on weekends were at greater risk for mortality than those admitted on weekdays.

Our results suggest that hospital care on weekends vs weekdays may differ in aggressiveness, but no differences in quality based on day of admission were detected. Patients with AIS included in this sample and admitted on weekends were more likely to receive tissue plasminogen activator than those admitted on weekdays. Our finding that tissue plasminogen activator use is positively and significantly related to increased reimbursement for ICD-9 procedure code 99.10 suggests that the higher payment was successful in increasing tissue plasminogen activator use.

This study shows no difference in the likelihood of death related to patient day of admission. This may indicate that the centralization of resources in these stroke centers is a beneficial practice for patient outcomes. These results indicate that patients who receive tissue plasminogen activator seem to be more likely to experience hospital mortality. This may be because of the tendency for physicians to treat more severe strokes with tissue plasminogen activator, as 2 of the exclusion criteria for its use are minor symptoms and rapidly recovering symptoms. Another possibility has been demonstrated by other studies^20,21 that suggest factors such as patient sex and race/ethnicity or physician training are likely to influence patient mortality more than tissue plasminogen activator treatment. In the overall sample, 7.1% of patients with AIS who did not receive tissue plasminogen activator died in the hospital compared with 12.5% of patients with AIS who received tissue plasminogen activator. It is possible that patients with AIS arrive earlier for hospital care on weekends, when they do not face possible traffic or job obligations that may slow their arrival to the emergency department. This may explain the increased likelihood of receiving tissue plasminogen activator treatment on weekends and should be explored in future research. There was also likely an increase in the acceptance of tissue plasminogen activator treatment during the study period, with higher use in 2006 than in 1998. Other aggressive treatments such as intra-arterial therapies exist for patients with AIS. In this article, we only examine treatment with tissue plasminogen activator.

The results of our study differ from the findings by Kostis et al.¹⁷ We hypothesize that part of the reason why mortality and aggressiveness of treatment for our population did not differ on weekends vs weekdays is that most hospitals that provide tissue plasminogen activator treatment maintain Joint Commission on Accreditation of Healthcare Organizations Stroke Center Certification, which requires access to computed tomography systems and neurologists 24 hours a day. If this is the case, these hospitals may not provide lower-quality care on weekends; rather, they may be required to operate differently to maintain their stroke center status and provide uniform care during “off” hours. Similar requirements and treatment standards may not be as common or enforced for other conditions such as myocardial infarction.

The operational and policy implications of this study are important. Although hospital care provided on weekends may not be identical to that provided on weekdays, it does not necessarily mean that it is associated with poorer quality. Patient treatment and outcomes may vary based on geography and clinical conditions. These findings may be the result of certain hospital standards, and regulations required for certification of particular treatments may be beneficial. Joint Commission on Accreditation of Healthcare Organizations Stroke Center Certification may help ensure less care variation and more consistent patient outcomes regardless of the day of admission to the hospital.

It is possible that the findings of this study indicating that hospital care on weekends is more aggressive may be the result of better access to equipment and diagnostic modalities on weekends. Elective surgical procedures on weekends are rare, and this may contribute to decreased traffic and waiting time for diagnostic equipment, and result in quicker and more efficient diagnosis and determination of treatment. Reduced road traffic and job obligations on weekends may contribute to the possibility that patients with AIS arrive sooner at the hospital. An alternative explanation may be that physicians are busy with other patient care responsibilities on weekdays and are more available to quickly treat patients with AIS on weekends because of the lack of scheduled appointments on Saturdays and Sundays. The creation of the ICD-9 code and increased reimbursement for administering tissue plasminogen activator may provide incentive for hospitals to make arrangements with physicians to be more available to treat patients with AIS at all times through alternative employment or reimbursement arrangements.

Our study is limited by possible underidentification of patients who received tissue plasminogen activator before Medicare developed the diagnosis related group code for this type of treatment in August 2005. Before then, hospitals may not have properly identified these patients because reimbursement incentives were not yet in place. ICD-9 procedure code 99.10 (injection or infusion of thrombolytic agent) was initially issued in 1998, but it was not accompanied by an increase in reimbursement. Therefore, hospitals may not have been motivated to begin using this code until the increase in reimbursement for administering tissue plasminogen activator was established in October 2005. Likewise, it is possible that the number of patients who received tissue plasminogen activator was higher than reported before this time. We have not directly found any evidence of this, and every effort has been made to accurately identify applicable cases.

Another limitation is the possible selection bias of patients who arrived for care on weekends. We have controlled for this through propensity scores and through comparison with patients having hemorrhagic stroke. Previous health services research has used the propensity score adjustment method with claims data to control for patient-level covariates that may confound analyses by influencing care and outcomes.²² This approach removes more than 90% of potentially confounding bias from the estimates.²² The patients with hemorrhagic stroke served as a nonequivalent control group that may experience symptoms and circumstances similar to those of patients with AIS. Finally, although we believe that Virginia is a representative state in its population and in rural and urban areas, the results of this study may not be generalizable to other populations. Overall, our study findings suggest that hospital stroke care on weekends is not necessarily inferior to care provided on weekdays. In fact, certain aspects of care such as aggressiveness may be improved. Further study on care variations that may improve patient outcomes is needed.

Correspondence: Abby S. Kazley, PhD, Department of Health Professions, Medical University of South Carolina, 151 Rutledge Ave, Charleston, SC 29425 (swansoaj@musc.edu).

Accepted for Publication: August 27, 2009.

Author Contributions: Drs Kazley and Gartner Hillman 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: Kazley, Gartner Hillman, Johnston, and Simpson. Acquisition of data: Gartner Hillman and Simpson. Analysis and interpretation of data: Kazley, Gartner Hillman, Johnston, and Simpson. Drafting of the manuscript: Kazley. Critical revision of the manuscript for important intellectual content: Kazley, Gartner Hillman, Johnston, and Simpson. Statistical analysis: Kazley and Simpson. Administrative, technical, and material support: Gartner Hillman. Study supervision: Johnston.

Financial Disclosure: None reported.