|Home | About | Journals | Submit | Contact Us | Français|
Little is known about in-hospital care for hemorrhagic stroke. We examined quality of care in intracerebral hemorrhage (ICH) and subarachnoid hemorrhage (SAH) admissions in the national Get With The Guidelines–Stroke (GWTG-Stroke) database, and compared them to ischemic stroke (IS) or TIA admissions.
Between April 1, 2003, and December 30, 2007, 905 hospitals contributed 479,284 consecutive stroke and TIA admissions. The proportions receiving each quality of care measure were calculated by dividing the total number of patients receiving the intervention by the total number of patients eligible for the intervention, excluding ineligible patients or those with contraindications to treatment. Logistic regression models were used to determine associations between measure compliance and stroke subtype, controlling for patient and hospital characteristics.
Stroke subtypes were 61.7% IS, 23.8% TIA, 11.1% ICH, and 3.5% SAH. Performance on care measures was generally lower in ICH and SAH compared to IS/TIA, including guideline-recommended measures for deep venous thrombosis (DVT) prevention (for ICH) and smoking cessation (for SAH) (multivariable-adjusted p < 0.001 for all comparisons). Exceptions were that ICH patients were more likely than IS/TIA to have door-to-CT times <25 minutes (multivariable-adjusted p < 0.001) and to undergo dysphagia screening (multivariable-adjusted p < 0.001). Time spent in the GWTG-Stroke program was associated with improvements in many measures of care for ICH and SAH patients, including DVT prevention and smoking cessation therapy (multivariable-adjusted p < 0.001).
Many hospital-based acute care and prevention measures are underutilized in intracerebral hemorrhage and subarachnoid hemorrhage compared to ischemic stroke /TIA. Duration of Get With The Guidelines–Stroke participation is associated with improving quality of care for hemorrhagic stroke.
Hemorrhagic stroke, comprising intracerebral hemorrhage (ICH) and subarachnoid hemorrhage (SAH), accounts for 10%–15% of all stroke and is the most severe form of stroke. Case fatality rates for ICH and SAH are 30%–40%,1 and survivors are frequently left with significant disabilities requiring rehabilitation.2 Those who survive are at risk for in-hospital medical complications also seen in ischemic stroke, including aspiration pneumonia and deep venous thrombosis (DVT). Hospital clinical practices can affect outcomes such as mortality in both ICH3 and SAH.4 In addition, the hospital stay represents an opportunity to initiate management of risk factors for hemorrhagic stroke, many of which are also risk factors for ischemic stroke and cardiovascular disease.
Get With The Guidelines–Stroke (GWTG–Stroke) is an ongoing voluntary, continuous quality-improvement initiative that collects patient level data on characteristics, treatments, in-hospital outcomes, and adherence to quality measures in stroke, including ischemic stroke, ICH, and SAH.5 Duration of participation in GWTG-Stroke has previously been associated with improved processes of care in ischemic stroke.6 In contrast, little has been reported regarding provision of care for hemorrhagic stroke. We hypothesized that adherence with GWTG performance measures would be lower in hemorrhagic stroke compared to IS or TIA. We also hypothesized that participation in GWTG-Stroke would be associated with improved provision of care over time for patients with SAH and ICH.
Characteristics of the GWTG-Stroke program have been previously described.5 Participating hospitals use an Internet-based Patient Management Tool (PMT) (Outcome Sciences Inc., Cambridge, MA) to enter data, receive decision support, and obtain feedback via on-demand reports of performance on quality measures. Hospitals were instructed to record data from consecutive stroke and TIA admissions; hospitals could choose whether to record data from consecutive hemorrhagic stroke admissions in addition to consecutive IS and TIA admissions. Case ascertainment was through clinical identification during hospitalization, retrospective identification by DRG codes, or both. The eligibility of each case was confirmed at chart review prior to abstraction.
Trained hospital personnel abstracted data using the Internet-based PMT with standardized data definitions and detailed coding instructions. The Internet-based system performs checks to ensure that the reported data are complete and internally consistent. In addition, data quality is monitored for completeness and accuracy.
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. Outcome Sciences, Inc. serves as the data collection and coordination center for GWTG. The Duke Clinical Research Institute serves as the data analysis center and has Institutional Review Board approval to analyze the aggregate de-identified data for research purposes.
The prespecified stroke quality measures have been published previously.5 Only quality measures that were applicable to the care of both hemorrhagic and ischemic stroke were analyzed in this study; these consisted of 3 process measures and 4 secondary prevention measures. The process measures were door-to-CT time <25 minutes in patients with symptoms <3 hours duration, dysphagia screen prior to oral intake, and DVT prophylaxis started by the second hospital day for nonambulatory patients. Acceptable options for DVT prophylaxis included pneumatic compression devices as well as anticoagulants; therefore, nonambulatory patients with hemorrhagic stroke were eligible for this measure. The included discharge prevention measures were counseling for smoking cessation, counseling for weight management, discharge lipid-lowering therapy according to National Cholesterol Education Panel Adult Treatment Panel III guidelines,7 and diabetic treatment at discharge. Patients with documented reasons for nontreatment on individual measures were recorded by the local sites, and were excluded from analysis for that measure. For example, nonsmokers are not eligible for the smoking cessation measure. Patients who died, were discharged to hospice care, or were treated only with comfort measures were not considered eligible for the DVT prophylaxis or discharge prevention measures.
Only some of the analyzed care measures are recommended by American Heart Association (AHA) guidelines for ICH and SAH. The 1999 AHA ICH guidelines recommend measures to prevent DVT8 while the 2007 ICH guideline,9 published late in the study period, also recommends cessation of smoking to prevent recurrent ICH and acknowledges the high prevalence of oral dysphagia without specific recommendations for screening. The most recent SAH guideline recommends measures to prevent DVT and that smoking cessation may decrease the incidence of SAH.10 Rapid CT scanning for discrimination of hemorrhagic stroke from ischemic stroke has been recommended for all stroke patients.11
The proportion of eligible patients who received each care measure was calculated by dividing the total number of patients receiving the intervention by the total number of patients eligible for the intervention, excluding those who were ineligible or who had contraindications to treatment. As a result, the denominator, which indicates the total number of eligible patients, varied according to the measure. Using χ2 tests for categorical variables and Kruskal-Wallis test for continuous variables, the patient characteristics, hospital characteristics, and compliance with the specified measures were compared among the ICH, SAH, and combined IS and TIA (IS/TIA) groups. Separate multivariable logistic regression models, one for each measure, were performed to determine the independent effect of stroke type (ICH, SAH, or IS/TIA) on measure compliance, with IS/TIA as the reference group. The generalized estimating equations (GEE) approach was used in the analysis to account for within-hospital clustering,12 using an exchangeable working covariance structure except when comparing DVT prophylaxis in SAH or ICH vs IS/TIA where an independence working covariance structure was required to achieve stable estimates. Covariates included in all the models were age, sex, nonwhite race, comorbid medical conditions (atrial fibrillation, prosthetic heart valve, previous stroke/TIA, coronary artery disease or prior myocardial infarction, carotid stenosis, diabetes, peripheral vascular disease, hypertension, dyslipidemia, and smoking), and hospital characteristics (bed size, annual number of stroke discharges, academic teaching status, and geographic region). Hospital bed size was entered as a continuous variable, while annual number of stroke discharges was categorized as 0–100, 101–300, or >300. Hospital teaching status and hospital region (defined as Northeast, Midwest, South, or West) were determined using statistics published by the American Hospital Association.13
Univariate changes in measure performance over time were tested using the Cochran-Mantel-Haenszel row-mean score test for categorical variables, and the Cochran-Mantel-Haenszel non-zero correlation test for continuous variables. Separate multivariable logistic regression analyses, using the GEE method to account for within-hospital clustering, were performed to evaluate whether measure performance changed over time for each measure. A “time in GWTG-Stroke” variable, indicating the duration of a hospital's participation in GWTG-Stroke program, was included in these models in addition to the patient and hospital characteristics variables described above. Time in GWTG-Stroke was categorized as follows: for each hospital the first 30 admissions were used as a baseline, and each subsequent 91 days were considered a quarter of time in GWTG-Stroke. The linear trend across these categories was tested. Also included in the model was a variable for calendar time, indicating the actual calendar date of patient discharge, to take into account secular trends in stroke care. Because hospitals entered the GWTG-Stroke program in a staggered fashion, duration of program participation is not highly correlated with calendar time, allowing estimates of the effects of calendar time and time in GWTG-Stroke in the same model.6 Thus the variable time in GWTG-Stroke estimates the independent effect of GWTG-Stroke participation on improving quality of care after adjusting for patient case-mix, hospital characteristics, and general trends toward better medical care over time.
To compare the stroke subtype-specific improvements in quality of care associated with time in GWTG-Stroke, the abovementioned multivariable analysis included model terms for the stroke type (ICH and SAH, with IS/TIA as the reference group) and their interaction with time in GWTG-Stroke. The effect of time in GWTG-stroke, that is, OR of measure compliance per 1 additional year of participation in GWTG-stroke, was determined for each stroke type—ICH, SAH, and IS/TIA. An interaction term with p < 0.05 indicates a significant difference between IS/TIA and the relevant comparator group, ICH or SAH, in the multivariable-adjusted OR of measure compliance per additional year in GWTG-Stroke.
All p values are 2-sided. Analyses were performed using SAS version 9.13 (SAS Institute, Cary, NC).
Between April 1, 2003, and December 30, 2007, there were 1,050 hospitals that contributed data on 515,414 stroke and TIA admissions. There were 102 hospitals (12,263 admissions) that were excluded because they recorded data on IS/TIA patients only, without recording data on hemorrhagic stroke patients. Among the remaining 948 hospitals, 13,966 admissions were excluded because stroke type was not recorded or recorded as “unknown,” 4,898 admissions were excluded because hospital characteristics were missing, and 5,003 admissions were excluded because information on patient demographics or past medical history was missing. Therefore, in total there were 479,284 admissions (93.0% of the total) from 905 hospitals that were included in the study. Most hospitals reported both ICH and SAH as well as IS/TIA, but some reported only consecutive ICH and IS/TIA admissions (n = 195) and a few reported only SAH and IS/TIA admissions (n = 9). Among the study eligible cohort, there were 52,989 ICH admissions, 16,723 SAH admissions, and 409,572 IS/TIA admissions (295,682 ischemic stroke and 113,890 TIA admissions). For analyses of measure compliance over time, 46 hospitals with 849 admissions were additionally excluded because they had not yet contributed complete data for a baseline and follow-up quarter. Median duration of hospital participation was 9 complete quarters (that is, 2.25 years; interquartile range 6 quarters to 13 quarters).
Characteristics of patients with stroke and TIA are shown in table 1. Overall, 61.7% of admissions were for ischemic stroke, 23.8% for TIA, 11.1% for ICH, and 3.5% for SAH. Significant differences were seen among the groups for all patient and hospital characteristics. SAH patients were younger, more likely to be women, more likely to be smokers, and less likely to have medical comorbidities than either ICH or IS/TIA patients. ICH and SAH patients were more likely than IS/TIA patients to be nonwhite and to be cared for in larger academic hospitals, and had worse outcomes with higher mortality. ICH patients had a relatively high prevalence of vascular risk factors, but not as high as for IS/TIA patients.
Compared to IS/TIA, ICH patients were less likely to receive care for 5/7 measures, and SAH patients were less likely to receive care for 6/7 measures (table 2). Table 3 shows unadjusted and adjusted odds ratios for performance of each measure by stroke subtype (with IS/TIA used as the reference group). For most measures, performance was less frequent for ICH or SAH compared to IS/TIA. Exceptions included the odds of door-to-CT in <25 minutes and oral dysphagia screen, which were higher in ICH compared to IS/TIA. The largest differences were observed for cholesterol therapy and counseling for smoking cessation, which was provided much less frequently for ICH or SAH than for IS/TIA.
Performance of most measures increased with increasing year of participation in the GWTG program for both ICH and SAH admissions (figure). After adjustment for patient and hospital characteristics, and calendar time, significant improvements associated with time in the GWTG-Stroke program were seen in many of the measures (table 4). Improvements associated with calendar time were also seen for all measures (p ≤ 0.01, data not shown) and generally of similar or slightly greater magnitude than the effect of time in the GWTG program, with odds ratios for improvements per calendar year ranging from 1.07 for door-to-CT time <25 minutes to 1.57 for smoking cessation counseling. Significant model interaction terms between stroke subtype and time in GWTG-Stroke, indicating that rate of improvement over time was statistically different between ICH vs IS/TIA or SAH vs IS/TIA, were observed for the following subtypes and measures: ICH patients had more rapid improvement per year in door-to-CT time (p = 0.004) and DVT prevention (p = 0.01) but less rapid improvement in guideline-recommended lipid-lowering therapy (p = 0.008), while SAH patients had more rapid improvement in DVT prevention (p = 0.003) but less rapid improvement in dysphagia screening (p = 0.02).
GWTG-Stroke is among the largest registry and quality improvement programs in hospitalized stroke, with data from more than 500,000 admissions, including more than 69,000 hemorrhagic stroke admissions. The prevalence and characteristics of hemorrhagic stroke patients in this study are similar to those previously reported in epidemiologic studies.14,15 Mortality rates for ICH and SAH reported here are lower than some other estimates of 30-day mortality,1,16,17 at least partly because out-of-hospital deaths and emergency room deaths were, by design, not included in GWTG-Stroke. Stroke admissions were recorded from a large variety of hospitals from all regions of the United States, and included a mix of academic and nonacademic, and small and large hospitals.6
Many ICH and SAH patients did not receive the measured interventions, including those specifically recommended by AHA guidelines for ICH and SAH. Adjusted and unadjusted odds ratios for measure adherence were similar (table 3), suggesting that the differences in measure adherence among stroke subtypes were not predominantly caused by differences in patient and hospital characteristics (table 1).
AHA and Brain Attack Coalition guidelines, in place during the entire study period, recommended rapid CT scan for stroke,11 prevention of DVT following ICH8 and SAH,10 and cessation of smoking after SAH.10 Smoking cessation following ICH is recommended in the 2007 guideline9 but not the 1999 guideline.8 Provision of this evidence-based care was less frequent in ICH and SAH compared to IS/TIA, with the exception that ICH patients were more likely to have rapid CT scanning and oral dysphagia screening, and SAH patients were similarly likely to receive DVT prophylaxis (table 3). The relatively better, but still suboptimal, performance on rapid CT scanning and oral dysphagia screening in ICH could be related to the known increased severity of ICH compared to IS/TIA. We are unable to directly test this hypothesis in GWTG-Stroke because the NIH Stroke Scale is not routinely collected in clinical practice and is not a mandatory data element in the registry. Provision of smoking cessation therapy showed the greatest variation across stroke types, despite the guideline recommendations, and was disappointingly low in SAH, for which it is a strong risk factor.
AHA guidelines for hemorrhagic stroke do not address management of dyslipidemia, obesity, or diabetes, because these conditions are relatively weakly associated with ICH or SAH. However, there is some evidence that ICH incidence is associated with diabetes18 and obesity,19 suggesting that management of these conditions could reduce the risk of recurrent ICH. Regardless of whether treating these risk factors will reduce the risk of recurrent ICH or SAH, their management is recommended by guidelines for primary prevention of coronary heart disease and ischemic stroke,20,21 for which ICH survivors are at substantial risk.22 The lack of specific disease-based guideline recommendations for management of these risk factors may explain why measure performance was lower in ICH and SAH than IS/TIA.
ICH patients were less likely than IS/TIA patients to receive lipid-lowering therapy. In contrast to many risk factors, hypercholesterolemia increases the risk of ischemic stroke but may decrease the risk of SAH23 and ICH.24–28 A randomized trial of high-dose atorvastatin showed more ICH in the treatment arm29; however, a meta-analysis of published trials30 and a prospective cohort study of ICH survivors31 found no elevated risk across a range of statin types and dosages. Therefore, the role of cholesterol-lowering therapy, and particularly statins, is controversial in ICH and may partly explain the lower rate of provision of cholesterol-lowering therapy in ICH compared to IS/TIA.
For many measures, increased time spent in the GWTG-Stroke program was associated with better measure performance. These increases in measure performance were detected on a background of increasing measure performance independent of time in the GWTG-Stroke program, probably reflecting national secular trends toward improving stroke care. Sustained increases in measure compliance over time has previously been reported for IS/TIA patients in the GWTG-Stroke database.6 The current study shows that hemorrhagic stroke patients have also experienced substantial increases in measure performance over time. Despite the evidence of increased performance over time for most measures, quality of care for hemorrhagic stroke patients is still far from optimal and further improvements are needed.
There are some limitations to this study. Our results may not be generalizable to non-GWTG hospitals because hospital participation in GWTG-Stroke is voluntary and therefore limited to hospitals with an interest in stroke quality improvement. Study data were collected based on the medical record and depend on the accuracy and completeness of clinical documentation. For most interventions, there is limited information on reasons why care was not provided, other than there was documentation of a contraindication. Our data suggest that GWTG-Stroke participation improves quality of care, by showing that duration of program participation is associated with increasing quality of care after adjusting for hospital characteristics, case mix, and secular time trends; however, the study does not include a control group for comparison. Although a link between process measures during hospitalization and postdischarge stroke outcomes may be valid, we were not able to directly explore the process-outcome relationships for hemorrhagic stroke because GWTG-Stroke does not collect postdischarge mortality. Thus, the long-term implications of suboptimal processes of care among patients with SAH and ICH remain to be determined.
Previous reports from GWTG-Stroke have shown that the program is associated with sustained improvement in quality of care for ischemic stroke.6 These data show that a Web-based data repository with real-time feedback and decision support can also lead to sustained improvements in quality of care for hemorrhagic stroke. Further research is needed to determine the durability of GWTG-Stroke related improvements, reasons for noncompliance with quality measures, and to evaluate local quality improvement initiatives launched in partnership with the GWTG-Stroke program.
Quality improvement in hemorrhagic stroke would be facilitated by updated guidelines for evidence-based ICH and SAH care. Specifically, we suggest that oral dysphagia screening be explicitly recommended by ICH and SAH guidelines. Our data suggest that guideline recommendations alone will not be sufficient to homogenize care across stroke subtypes, however. Quality improvement initiatives targeted to hemorrhagic stroke patients, aimed at improving processes of care as well as prevention of recurrence, are warranted.
The Get With The Guidelines Program (GWTG) is funded by the American Heart Association (AHA) and the American Stroke Association. The program is also supported in part by unrestricted educational grants to the American Heart Association by Pfizer, Inc. (New York, NY) and the Merck-Schering Plough Partnership (North Wales, PA), who did not participate in the design, analysis, or manuscript preparation or approval. Dr. Smith serves as a member of the Get With the Guidelines Science Subcommittee and receives research support from the NIH (NINDS R01 NS062028) and the Canadian Stroke Network, and salary support from the Heart and Stroke Foundation of Canada and Canadian Institute of Health Research. Dr. Liang is a member of the Duke Clinical Research Institute, which serves as the AHA GWTG data coordinating center. Dr. Hernandez is a member of the Duke Clinical Research Institute, which serves as the AHA GWTG data coordinating center; reports receiving research grants from GlaxoSmithKline, Johnson & Johnson (Scios, Inc.), Medtronic, Merck, and Roche Diagnostics; and has served on the speaker's bureau or received honoraria within the past 5 years from AstraZeneca, Medtronic, Novartis, Sanofi-Aventis, and Thoratec Corporation. Dr. Reeves has received salary support from the Michigan Stroke GWTG Registry and serves as a member of the American Heart Association's Get With the Guidelines Quality Improvement Subcommittee. Dr. Cannon serves as a member of the American Heart Association's Get With the Guidelines Steering Committee; reports research funding from Accumetrics, AstraZeneca, Bristol-Myers Squibb/Sanofi Partnership, Glaxo Smith Kline, Merck and Merck/Schering Plough Partnership; and serves as a clinical advisor and holds equity in Automedics Medical Systems. Dr. Fonarow chairs the AHA GWTG Steering Committee; serves as a consultant to Pfizer, Merck, Schering Plough, Bristol Myers Squibb, and Sanofi-Aventis; receives speaker honoraria from Pfizer, Merck, Schering Plough, Bristol Myers Squibb, and Sanofi-Aventis; and receives research support from Pfizer and National Institutes of Health. Dr. Schwamm serves as vice chair of the AHA GWTG Steering Committee; serves as a consultant to the Research Triangle Institute, CryoCath, and to the Massachusetts Department of Public Health; and has provided expert medical opinions in malpractice lawsuits regarding stroke treatment and prevention.
Address correspondence and reprint requests to Dr. Eric E. Smith, Calgery Stroke Program, Hotchkiss Brain Institute, University of Calgary, Department of Clinical Neurosciences, Foothills Medical Centre, 1403 29th Street NW, Calgary, AB, Canada T2N 2T9 ac.yraglacu@htimsee
Disclosure: Author disclosures are provided at the end of the article.
Received November 16, 2008. Accepted in final form May 6, 2009.