Differences in Acute Ischemic Stroke Quality of Care and Outcomes by Primary Stroke Center Certification Organization

Introduction

To improve stroke care, the Brain Attack Coalition recommended that Primary Stroke Centers (PSC) capable of meeting certain standards and providing evidence-based care for stroke be established and that they preferentially provide care for stroke patients.¹ The Joint Commission (JC) began certifying PSCs in 2003. Other national organizations, including Healthcare Facilities Accreditation Program (HFAP) and Det Norske Veritas (DNV), started providing PSC certification later. Several states also passed legislation to designate hospitals as PSCs. The basic standards for providing PSC certification are similar for each organization providing PSC certification; yet, there may be important differences in how these standards are assessed and applied. For example, the JC requires extensive documentation that standards are met, including in-person site visits, whereas other organizations do not require site visits or even allow hospitals to self-designate.

Current evidence suggests that certified stroke centers provide higher quality of care and better outcomes than hospitals that are not PSC certified.^2,3 However, whether there are differences in care quality and outcomes among PSCs certified by different organizations has not been studied. The majority of PSCs participate in the Get With The Guidelines (GWTG)-Stroke program, which collects prospective data for evidence-based, guideline-directed process measures, safety measures, efficacy measures, and in-hospital outcomes, providing an opportunity to compare performance as a function of which organization provided PSCs certification.^4,5

Methods

GWTG-Stroke is an ongoing voluntary Web-based registry and performance improvement initiative that collects patient-level data on characteristics, diagnostic testing, treatments, adherence to quality measures, and in-hospital outcomes in patients hospitalized with stroke. Details of the design and conduct of the GWTG-Stroke program have been previously described.^4,6 The GWTG-Stroke program was made available in April 2003 to any hospital in the United States.^4,6 Trained hospital personnel are instructed to ascertain consecutive patients admitted with acute ischemic stroke by either prospective clinical identification, retrospective identification using International Classification of Diseases-Ninth Revision discharge codes, or a combination.^4,6 The eligibility of each acute ischemic stroke admission was confirmed through chart review by trained hospital personnel. Additional descriptions of the case ascertainment, data collection, and quality auditing methods have been previously published.^4,6,7 Each participating hospital received either human research approval to enroll cases without individual patient consent under the common rule or a waiver of authorization and exemption from subsequent review by their institutional review board. The Duke Clinical Research Institute serves as the data analysis center and has an agreement to analyze the aggregate deidentified data for research purposes.

Patient Population

The study population consisted of patients who were admitted to PSCs that participated in the GWTG-Stroke program between January 1, 2010, and December 31, 2012, with a final diagnosis of acute ischemic stroke. During the study period, there were 4791 acute care hospitals in the United States. Among them, 1356 hospitals were certified as PSCs and 977 (72.1%) were participating in the GWTG-Stroke program. Patients without ischemic stroke diagnosis (n=1 086 119) and discharge status missing or left against medical advice (n=5580) were excluded. We excluded 88 369 transfer-in cases. The final study population consisted of 477 297 ischemic stroke hospitalizations, ≈50% of all ischemic stroke hospitalization in the United States during the study period.

Hospital Classification

The hospitals that were listed as having maintained or obtained the PSC certification by JC, HFAP, and DNV during the study period were publicly available online at http://www.qualitycheck.org, http://www.hfap.org, and http://dnvglhealthcare.com. The list of PSCs certified by state agencies was obtained from the state health department websites. The policies of each organization and state for PSC certification are detailed in Table I in the online-only Data Supplement. PSCs certified by both JC and state were counted as JC PSCs in this study (158 sites). PSCs certified by both DNV and state certification were counted as DNV PSC (12 sites). PSCs certified by both JC and DNV were counted as JC PSC (2 sites). Hospitals in New Jersey that were classified as state Comprehensive Stroke Centers and not otherwise certified by JC/DNV/HFAP were included as State PSCs (10 sites). There were 3 Texas state–certified PSCs that required another certification, which we were unable to identify and, thus, could not be accurately assigned to a PSC category were excluded from this analysis. Twenty-three PSCs had <25 patients during the study period. Among them, 12 were JC PSCs, 1 were DNV PSC, and 10 were state-certified PSCs. Data on hospital characteristics (ie, bed size, academic status, and geographical region) were obtained from the American Hospital Association database.

Stroke Measure Definitions

The GWTG-Stroke program developed a set of process-based performance measures to quantify the quality of care for acute ischemic stroke patients. In 2007, the American Heart Association/American Stroke Association came to an agreement with The JC’s PSC certification program and the Centers for Disease Control and Prevention Coverdell Registry to jointly release a set of standardized stroke performance measures.^4,8 These performance measures have been endorsed by the National Quality Forum, and the detailed measure specifications were previously published.^4,8 The 7 performance measures that were selected as primary targets for stroke quality improvement efforts in GWTG-Stroke are as follows^4–6: intravenous tissue-type plasminogen activator (IV tPA) treated within 3 hours in patients who arrive within 2 hours after symptom onset, antithrombotic medication within 48 hours of admission, deep vein thrombosis prophylaxis within 48 hours of admission for nonambulatory patients, discharge use of antithrombotic medication, discharge use of anticoagulation for atrial fibrillation, treatment for low-density lipoprotein ≥100 mg/dL in patients meeting National Cholesterol Education Program Adult Treatment Panel III guidelines, and counseling or medication for smoking cessation. Two different measures were used to summarize the overall conformity with acute and discharge performance measures for each hospital. An all-or-none measure of care (also termed defect-free care) was used, which is defined as the proportion of patients who received all of the 7 performance measure interventions for which they were eligible. A composite measure of care, defined as the total number of 7 performance measure interventions performed among eligible patients divided by the total number of possible performance measure interventions among eligible patients, was also calculated.^4–6

An additional 8 measures, referred to as GWTG-Stroke quality or reporting measures by the program, were also used to compare the processes of care provided to patients, as previously described.^4,5 These additional measures were intravenous tPA treated within 4.5 hours in patients who arrive within 3.5 hours after symptom onset, door-to-brain imaging time ≤25 minutes, door-to-needle (DTN) time ≤60 minutes, documented National Institutes of Health Stroke Score, door-to-arrival time, DTN time, annual volume of IV tPA, and the rate of ischemic stroke patients treated with IV tPA. Summary and defect-free composite measures were also analyzed.^4,5 The in-hospital outcome measures used in this study were mortality, length of stay, independent ambulatory status at discharge, and discharge home rate.

Statistical Analysis

The patient characteristics, hospital characteristics, and measures of care for acute ischemic stroke were compared among the 4 PSC certification mechanisms: JC, DNV, HFAP, and state-based. Differences in these characteristics were compared using Pearson χ² tests for categorical row variables and Wilcoxon Rank-Sum tests for continuous row variables. Multivariable logistic regression models were performed to examine the relationship between PSC certification mechanism and quality measures or outcomes of interest using JC PSC as the reference group. Unadjusted and adjusted models using generalized estimating equations to account for in-hospital clustering were created for each binary outcome. The adjusted model controlled for potential confounders: age, sex, race, medical history of atrial fibrillation/flutter, previous stroke/transient ischemic attack, coronary artery disease/prior myocardial infarction, carotid stenosis, diabetes mellitus, peripheral vascular disease, hypertension, dyslipidemia, smoking, heart failure, arrival by emergency medical service, arrival during off hours (versus regular hours), National Institutes of Health Stroke Scale score, hospital type, geographic region, number of beds, annual ischemic stroke volume, and rural location. Multiple imputation was used to handle missing data in the models. Hospital characteristics were not imputed. The collinearity between all covariates in the models were assessed using variance inflation factors; all variance inflation factors were <5, indicating collinearity was not a problem in the models. A sensitivity analysis was performed where the PSCs certified by >1 organizations were excluded. An additional sensitivity analysis was performed where hospitals with <25 patients during the study period were excluded. We further analyzed process measures and outcomes for those states with ≥5 state-certified PSCs. All statistical analyses were performed using SAS Version 9.4 software (SAS Institute). All hypothesis tests were 2-sided, with P<0.05 considered statistically significant.

Results

This study included 477 297 acute ischemic stroke admissions from 977 certified PSCs. Among them, 78.3% were treated at JC PSCs (n=721), 16.7% were treated at State PSCs (n=208), 4.0% were treated at DNV PSCs (n=36), and 1.0% were treated at HFAP PSCs (n=12). The 977 PSCs included in this study represented 72.1% of all US PSC–certified hospitals in 2012 (n=1356). The 977 PSC-certified hospitals included in this study represented 69.5% of all JC PSCs (n=1038), 79.7% of all State PSCs (n=261), 83.7% of all DNV PSCs (n=43), and 85.7% of all HFAP PSCs (n=14).

Patient characteristics are provided in Table 1. The patients who were treated in the 4 groups of PSCs showed little differences in their demographic characteristics or comorbidities. There were moderate differences observed in race/ethnicity distribution. Hospital characteristics did show some marked differences among the 4 PSC groups (Table 2). State-based PSCs were smaller (fewer beds) than PSCs certified by the other 3 organizations. The annual stroke volume was higher in DNV and JC PSCs and lower in HFAP and State PSCs (Median 276, 233, 175, and 173, respectively).

We examined the performance, quality, and outcomes measures among the 4 PSC groups (Table 3). Overall, all the hospitals had high conformity with the 7 performance measure. HFAP programs had modestly lower defect-free care performance conformity than the other 3 mechanisms (88.5% versus 91.5%, 91.1%, 90.9%, respectively, P<0.0001). There were significant differences in several process measures among groups, especially in IV tPA use and timing (Figure). DNV PSCs had much higher annual IV tPA volume, followed by JC PSC, State PSC, and HFAP PSC (median 22, 18, 12, and 8, respectively; P<0.0001). The rate of IV tPA utilization among all ischemic stroke admissions was highest at DNV PSCs, followed by JC PSCs, State PSCs, and HFAP PSCs (9.8%, 9.0%, 7.1%, and 5.9% respectively, P<0.0001; Table 3).

• Download figure
• Download PowerPoint

For patients who arrived by 2 hours of stroke onset without IV tPA contraindications, more patients at DNV PSCs received IV tPA within the 3-hour time window, followed by JC PSC, State PSCs, and HFAP PSCs (90.1%, 85.6%, 82.2%, and 76.5%, respectively). Similarly, a greater percentage of patients who arrived by 3.5 hours of stroke onset received IV tPA by 4.5 hours at DNV PSCs, followed by JC PSCs, State PSCs, and HFAP PSCs (75.8%, 68.9%, 62.4%, and 53.1% respectively). The percentages of patients received IV tPA with DTN times within 60 minutes were higher in DNV, JC, and State PSCs than in HFAP PSCs (36.6%, 35.9% 35.5%, and 16.8%, respectively; Table 3). Analyses of stroke care process measures for those states with ≥5 state-certified PSCs reveal important differences in performance between states (Table II in the online-only Data Supplement).

Multivariable logistic regression analysis was performed to examine the association between the type of PSC certification and conformity with specific quality measures and patient-centered outcomes, using JC PSC as the reference group (Table 4). After adjustment of patient and other hospital characteristics, conformity to defect-free performance measure was comparable. However, the type of PSC certifying organization was independently associated with some individual quality measures. Patients with ischemic stroke were less likely to receive IV tPA at State PSCs (adjusted odds ratio [OR] 0.76, 95% confidence interval [CI] 0.68–0.86). Patients who arrived by 3.5 hours were less likely to receive IV tPA within 4.5 hours at State PSCs in both unadjusted and adjusted model (Table 4). Compared with JC PSCs, HFAP PSCs had a lower IV tPA rate in the unadjusted model but not in the adjusted model. HFAP PSCs were less likely to achieve DTN time of ≤60 minutes in both unadjusted (OR 0.47, 95% CI 0.29–0.74) and adjusted model (OR 0.49, 95% CI 0.31–0.77). DNV PSCs did not differ from JC PSCs in most measures in the regression analysis. Patients at DNV PSCs were more likely to start antithrombotics early in both unadjusted and adjusted models.

The in-hospital outcome measures also showed some differences (Table 3). Prior to risk adjustment, HFAP PSCs had lower unadjusted in-hospital mortality, followed by JC PSCs, DNV PSCs, and State PSCs (P<0.0001). After adjustment, HFAP PSCs had lower in-hospital mortality (OR 0.66, 95% CI 0.47–0.92), whereas State PSCs had higher in-hospital mortality (OR 1.23, 95% CI 1.07–1.41) compared with JC PSCs (Table 4). More patients at HFAP PSCs gained independence at discharge, followed by DNV PSCs, JC PSCs, and State PSCs (Table 3). Analysis among state-certified PSCs also reveal differences in outcomes between states (Table II in the online-only Data Supplement).

The sensitivity analysis where the PSCs certified by >1 organizations were excluded generated similar finding (Table III in the online-only Data Supplement) as did a sensitivity analysis where PSCs with <25 patients were excluded (data not shown).

Discussion

PSC certification was established to provide the best evidence-based care to stroke patients from acute stroke care to secondary prevention, rehabilitation, and return to community. Our study suggests that PSC certified by different certification bodies differed in performance and outcome. PSCs certified by JC and DNV, in general, showed comparable adjusted performance and outcome measures. But PSCs certified by state-based agencies as a group underperformed JC sites in some key measures. State-based PSCs had significantly lower rate of IV tPA utilization and less likely to reach DTN time of ≤60 minutes. Patients who arrived at State PSCs within the eligible time window from symptom onset had a lower probability of receiving tPA than those at JC PSCs. State PSCs had higher in-hospital mortality and lower chance of independence at discharge. While state-based certified PSCs overall showed lower performance on many measures, when analyzed by individual states, certain State PSCs did have performance that matched or exceeded that of other certifying bodies. These findings may have important implications for stroke or other systems of care continue to evolve, with potential opportunities to create more uniform and rigorous standards for PSC certification.

Intravenous tPA has been proven to improve the outcome of acute ischemic stroke.^9,10 Earlier thrombolytic treatment was associated with reduced mortality and symptomatic intracranial hemorrhage and higher rates of independent ambulation at discharge and discharge to home after acute ischemic stroke.¹¹ This study suggests that there is opportunity for State PSCs and certification bodies to improve the identification of hospitals that are more effectively delivering acute stroke diagnosis and thrombolytic therapy. For example, HFAP PSCs had lower IV tPA utilization and less likely to achieve DTN time within 60 minutes. Further analysis of PSCs certified by different states are needed given the disparities in several key measures among different states. This information may be useful because those organizations that certify PSCs examine ways to improve stroke care processes and standards being assessed.

The present study finds that PSCs certified by state agencies had higher in-hospital mortality, while HPAP PSCs had lower in-hospital mortality. The reasons for these differences in mortality are not clear. The hospital quality of care and preventing and managing complications may impact in-hospital mortality. Differences in use of comfort care, transfer patterns, discharge policy, and geographic location may be confounding these findings.^12,13 It is unlikely that lower IV tPA use or longer DTN times in HFAP PSCs is contributing to these observations given that State PSCs also had lower tPA use and longer DTN times, but higher mortality. It is also possible that preexisting quality of care may lead hospitals to seek certification from certain organizations. Further study is needed to identify the potential contributing factors along with devising effective approaches to improve stroke mortality.

There are important differences in how stroke care is organized in different countries and healthcare delivery systems. We note that the average stroke volume in US PSCs is lower than that reported by some Canadian and European regional stroke centers.¹⁴ Although the population density in Europe is much higher than that in the United States, there may be important opportunities to further regionalize acute ischemic stroke care among higher performing PSCs. Some European stroke centers have also reported higher overall IV tPA rates than that reported in the present study of PSCs in the United States.¹⁵ This may reflect differences in systems of care and geographic differences that contributes to variances in transportation times and percentages of patients arriving within IV tPA eligible time window.

There are important limitations to this study. Data on patient characteristics and quality measures were self-reported by participating hospitals without external validation. However, prior quality audits of GWTG-Stroke data show high concordance rates with source documentation.⁷ PSCs that did not participate in GWTG-Stroke program were not included in this study (estimated to be 30% of PSCs in 2012). There may be unmeasured confounders that influence which organization a hospital chose for its type of PSC certification program, and these may contribute to the measured stroke care quality. Because GWTG-Stroke participation may directly impact stroke care quality and outcomes, there may have been more limited ability to identify differences among the 4 groups of PSC-certified hospitals. The sample might be underpowered to show differences in individual measures because of the small number of DNV and HFAP PSCs. We did not analyze Comprehensive Stroke Centers or whether longer duration of PSC certification was associated with differences in quality measure conformity or outcomes. Residual measured and unmeasured confounding may influence the results of the multivariable analyses. It was not possible to account for stroke severity in all patients because the National Institutes of Health Stroke Score is not documented for all patients in the database, and so National Institutes of Health Stroke Score inclusion in the sensitivity analyses may have introduced selection bias. Although in-hospital mortality is an important outcome measure, and we adjusted for comorbid conditions and other risk factors, it can be influenced by stroke severity, transfer out policies, length of stay, and other unmeasured confounding factors. Other measures to assess care quality were not collected in this study, including other process measures, functional outcomes, procedure complications, health status, patient satisfaction, and post discharge outcomes, such as mortality and preventable readmissions. Additional studies should be performed to access the relationship between which organizations provide PSC certification and these additional quality measures and outcomes.

Conclusions

In summary, using GTWG-Stroke data, this study showed some differences in quality measures and in-hospital outcomes among hospitals that received PSC certifications depending on which organization provided certification. PSCs certified by the state and HFAP had lower IV tPA utilization rate, and HFAP PSCs were less likely to achieve DTN times within 60 minutes. State agency–certified PSC hospitals had higher risk-adjusted in-hospital mortality rates and HFAP lower mortality rates. These findings have important implications for the PSC certification process and optimizing stroke, and potentially other, systems of care.

Footnotes