Quality of Care and Outcomes for In-Hospital Ischemic Stroke Findings From the National Get With The Guidelines-Stroke

Transcription

1 Quality of Care and Outcomes for In-Hospital Ischemic Stroke Findings From the National Get With The Guidelines-Stroke Ethan Cumbler, MD; Heidi Wald, MD, MSPH; Deepak L. Bhatt, MD, MPH; Margueritte Cox, MS; Ying Xian, MD, PhD; Mathew Reeves, BVSc, PhD; Eric E. Smith, MD, MPH; Lee Schwamm, MD; Gregg C. Fonarow, MD Background and Purpose Analysis of quality of care for in-hospital stroke has not been previously performed at the national level. This study compares patient characteristics, process measures of quality, and outcomes for in-hospital strokes with those for community-onset strokes in a national cohort. Methods We performed a retrospective cohort study of the Get With The Guidelines-Stroke (GWTG-Stroke) database of The American Heart Association from January 2006 to April 2012, using data from 1280 sites that reported 1 in-hospital stroke. Patient characteristics, comorbid illnesses, medications, quality of care measures, and outcomes were analyzed for in-hospital ischemic strokes compared with community-onset ischemic strokes. Results Patients with in-hospital stroke had more thromboembolic risk factors, including atrial fibrillation, prosthetic heart valves, carotid stenosis, and heart failure (P), and experienced more severe strokes (median National Institutes of Health Stroke Score 9.0 versus 4.0; P). Using GWTG-Stroke achievement measures, the proportion of patients with defect-free care was lower for in-hospital strokes (60.8% versus 82.0%; P). After accounting for patient and hospital characteristics, patients with in-hospital strokes were less likely to be discharged home (adjusted odds ratio 0.37; 95% confidence intervals [ ]) or be able to ambulate independently at discharge (adjusted odds ratio 0.42; 95% confidence intervals [ ]). In-hospital mortality was higher for in-hospital stroke (adjusted odds ratio 2.72; 95% confidence intervals [ ]). Conclusions Compared with community-onset ischemic stroke, patients with in-hospital stroke experienced more severe strokes, received lower adherence to process-based quality measures, and had worse outcomes. These findings suggest there is an important opportunity for targeted quality improvement efforts for patients with in-hospital stroke. (Stroke. 2014;45: ) Key Words: outcome assessment (health care) quality indicators, health care quality of health care registries stroke thrombolytic therapy Of all patients with stroke, 4% to 17% patients have onset of symptoms while hospitalized. 1 6 There are an estimated to of these in-hospital strokes in the United States each year. 7 Relatively little is known about the characteristics, treatments, quality of care, and outcomes of these patients. Previous studies suggest that in-hospital strokes are more severe and have worse outcomes than strokes occurring in the community. 1,5,6,8,9 However, there are few previous studies of in-hospital stroke quality of care and none at the national level. 6,10 11 The purpose of this study was to analyze quality of care and clinical outcomes for in-hospital acute ischemic strokes using the national Get With The Guidelines-Stroke (GWTG-Stroke) registry of the American Heart Association (AHA)/American Stroke Association (ASA). The study aims were to compare patients with in-hospital acute ischemic strokes with patients with community-onset acute ischemic strokes with respect to (1) demographics and clinical characteristics, (2) comorbid illnesses, (3) medication use before admission, (4) quality of care measures, (5) complications of IV thrombolysis, and (6) in-hospital outcomes. Methods We performed a retrospective cohort study of the AHA GWTG-Stroke registry database. Hospitals across the United States voluntarily Received September 23, 2013; accepted October 22, From the Department of Medicine, University of Colorado School of Medicine Anschutz Medical Campus, Aurora, CO (E.C., H.W.); VA Boston Healthcare System, Brigham and Women s Hospital, and Harvard Medical School, Boston, MA (D.L.B.); Duke Clinical Research Institute, Duke University Medical Center, Durham, NC (M.C., Y.X.); Department of Epidemiology, Michigan State University, East Lansing, MI (M.R.); Department of Clinical Neurosciences, Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta, Canada (E.E.S.); Department of Neurology, Massachusetts General Hospital, Boston, MA (L.S.); and Division of Cardiology, Ronald Reagan-UCLA Medical Center, Los Angeles, CA (G.C.F.). The online-only Data Supplement is available with this article at /-/DC1. Correspondence to Ethan Cumbler, MD, E 17th Ave F782, Aurora, CO Ethan.Cumbler@ucdenver.edu 2013 American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 232 Stroke January 2014 participate in the GWTG-Stroke registry, a national program of the AHA/ASA since Description of the GWTG-Stroke program, performance measures, and validity and reliability of its data collection have been previously published Data Collection Hospital personnel performing chart abstraction used standardized definitions and codes for data entry into the Internet-based interactive Patient Management Tool (Outcome Sciences, Inc., Cambridge, MA). Case ascertainment used prospective clinical identification, retrospective identification using International Classification of Diseases (ICD)-9 at discharge, or a combination. Methods of prospective identification included surveillance of stroke alert logs, neurological consultation service census, or radiological diagnosis of new stroke. The eligibility of each acute stroke or transient ischemic attack abstraction is confirmed by chart review. In-hospital ischemic stroke is defined as acute symptomatic infarction of central nervous system tissue that occurs during hospitalization in a patient originally admitted to the hospital for another diagnosis or procedure. Recording data for in-hospital stroke cases were at the discretion of individual hospitals. Hospitals participating in the registry submitted deidentified patient-level clinical information about the medical history, hospital care, and outcomes of patients with stroke. Outcome, a Quintiles Company was the registry data collection coordination center for the GWTG programs. The Duke Clinical Research Institute served as the data analysis center. Institutional review board approval was granted to analyze aggregate deidentified data for research purposes. All participating institutions were required to comply with local regulatory and privacy guidelines and, if required, to secure local institutional review board approval. Because data were used primarily at the local site for quality improvement, sites were granted a waiver of informed consent under the common rule. Setting and Study Population The initial data set included sites participating in the GWTG-Stroke program from April 2003 to April We excluded patients admitted before January 1, 2006, when in-hospital strokes were not identifiable ( patients [8%] and 38 sites) and cases of nonischemic stroke ( [36%] patients and 29 sites). Locations of interest included in-hospital and community-onset strokes. All other locations, such as clinic or nursing home setting, were excluded ( patients [3%] and 5 sites). We excluded data from sites that did not contribute 1 in-hospital stroke ( patients [3%] and 462 sites). Patient Characteristics Patient characteristics included age, sex, race, insurance status, and body mass index. Stroke severity was measured with the National Institutes of Health Stroke Score (NIHSS) at stroke onset. Comorbid illness included atrial fibrillation/flutter, prosthetic heart valve, previous stroke/transient ischemic attack, coronary artery disease/ previous myocardial infarction, carotid stenosis, diabetes mellitus, peripheral vascular disease, hypertension, smoking, dyslipidemia, and heart failure. Classes of medication used before admission included antiplatelet, anticoagulant, antihypertensive, cholesterolreducing, and diabetic medication. Ambulatory status (recorded as unable, with assistance, or independent) was documented from status before admission. Hospital characteristics, obtained from the American Hospital Association, included hospital type (academic/ nonacademic), location (urban/rural), geographic region (west, south, midwest, northeast), and number of beds. 15 Annual number of stroke discharges (categorized as 0 100, , or >300) were obtained from GWTG-Stroke. Quality Measures The primary outcomes included performance on process measures after the stroke defined by GWTG-Stroke as either achievement or quality measures. 12 The achievement measures included IV tissue plasminogen activator (tpa) treatment within 3 hours if arrived by 2 hours, early antithrombotics within 48 hours, deep venous thrombosis prophylaxis for nonambulatory patients within 48 hours, antithrombotic therapy on discharge, anticoagulation on discharge for atrial fibrillation/flutter, statin therapy, and smoking cessation counseling if appropriate on discharge. The quality measures included cholesterol treatment if low-density lipoprotein >100 mg/dl or not documented, dysphagia screening, stroke education, rehabilitation assessment, IV tpa treatment within 4.5 hours if arrived by 3.5 hours, door to IV tpa within 60 minutes, low-density lipoprotein documented, and intensive statin therapy. 12 For time from admission-based measures, the date and time of symptom recognition was substituted for time of admission for in-hospital strokes. We calculated a composite achievement measure (opportunity composite) defined as the total number of achievement measure interventions performed among eligible patients divided by the total number of possible achievement measure interventions among eligible patients. We also defined a defect-free (all-or-none) AHA-GWTG-Stroke summary measure reflecting the proportion of patients who received all of the achievement measures for which they were eligible. 12 Outcome Measures The primary outcome measures were in-hospital mortality, discharge to home, and independent ambulation at discharge. For the subset of patients who received IV tpa, symptomatic intracranial hemorrhage was an additional primary outcome of interest. Additional complications of tpa that were recorded included serious bleeding within 36 hours of treatment, or other serious complications that required additional medical interventions or prolonged length of stay. Statistical Analysis Univariate tables were generated to compare in-hospital strokes with community-onset strokes across the variables of interest. Pearson χ 2 tests were used for categorical variables, and Wilcoxon Rank-Sum tests were used for continuous variables. Single imputation based on most likely category (or median) was used for patient characteristics that had <10% missing observations. For imputed medical history variables, missing categories were assigned to no. Body mass index was imputed to the sex-specific median. The percentage of missing data for individual variables is included in the tables. Differences between outcomes (in-hospital mortality, discharge to home for survivors, independent ambulation at discharge, and, for the subset of strokes treated with IV tpa, symptomatic intracranial hemorrhage) were further explored with multivariate logistic regression models using generalized estimating equations to account for within-hospital clustering. 16 The models were adjusted for prespecified patient and hospital characteristics including age, sex, race, history of atrial fibrillation/flutter, previous stroke/ transient ischemic attack, coronary artery disease/previous myocardial infarction, carotid stenosis, diabetes mellitus, peripheral vascular disease, hypertension, dyslipidemia, smoking, hospital region, hospital type, annual number of stroke discharges, and number of beds. Because stroke severity is a key prognostic factor but was not available in all patients, sensitivity analyses, adjusting for NIHSS, were performed on the subset of patients with complete NIHSS data available. All P values were 2-sided with P<0.05 considered significant. Analysis was performed using SAS version 9.2 (SAS Institute, Cary, NC). Results There were ischemic strokes from 1280 sites in the analyses after exclusions, of which (2.2%) strokes had onset in hospital. Table 1 describes demographic and clinical characteristics of in-hospital and community-onset groups of patients with stroke. Although the median age in both groups was similar (73 years), more in-hospital strokes occurred in women (54.3% versus 51.9%; P). Compared with patients with community-onset stroke, in-hospital patients

3 Cumbler et al Quality of Care for In-Hospital Ischemic Stroke 233 Table 1. Patient Characteristics Variables In-Hospital Stroke Community-Onset Stroke P Value N (%) (2.2%) (97.8%) Patient demographics Age- n responding, missing (%) (0%) (0%) Median (25th to 75th percentile) Mean (SD) 73 (62 82) 71.4 (14.1) 73 (60 82) 70.7 (14.7) Sex- n, missing (%) (0%) (0%) (% female) (54.3%) (51.9%) Race- n responding, missing (%) (0.2%) (0.2%) White (72.5%) (71.3%) Black or African American 2910 (13.7%) (16.1%) Hispanic 1448 (6.8%) (6.3%) Other or unable to determine 1504 (7.1%) (6.4%) Insurance- n responding, missing (%) (11.4%) (26.0%) Private/VA/other 8319 (44.0%) (43.4%) Medicare 7520 (39.7%) (39.2%) Medicaid 2110 (11.2%) (9.7%) Self-pay/uninsured 773 (4.1%) (6.7%) Not documented 200 (1.1%) 6887 (1.0%) Medical history- n responding, missing (%) (1.6%) (2.6%) Previous stroke/tia 5400 (25.7%) (31.4%) Atrial fibrillation 4824 (23.0%) (18.4%) Prosthetic heart valve 585 (2.8%) (1.4%) Heart failure 2833 (13.5%) (6.9%) Coronary artery disease/previous MI 8110 (38.6%) (26.8%) Carotid stenosis 1422 (6.8%) (4.1%) Diabetes mellitus 7546 (35.9%) (32.1%) Peripheral vascular disease 1735 (8.3%) (4.8%) Hypertension (76.0%) (77.2%) Smoker 3149 (15.0%) (19.3%) Dyslipidemia 9136 (43.5%) (41.4%) Body mass index- n responding, missing (%) (20.4%) (23.8%) Median (25th to 75th percentile) 26.8 ( ) 26.8 ( ) Medication use before admission Antiplatelet- n responding, missing (%) Anticoagulation- n responding, missing (%) Initial INR* median (IQR), mean (SD) Antihypertensive- n responding, missing (%) Cholesterol reducing- n responding, missing (%) Diabetic medication- n responding, missing (%) Functional status- independent ambulation Before admission- n responding, missing (%) n (%) Stroke severity NIHSS- n responding (% missing) Median (25th to 75th percentile) (24.1%) 7949 (49.1%) (24.2%) 2725 (16.8%) 1.4 ( ) 1.84 (1.37) (3.2%) (74.9%) (0.6%) 9494 (44.8%) (3.5%) 5979 (29.0%) (16.4%) (82.5%) (50.9%) 9 (4 17) (39.1%) (45.4%) (39.2%) (11.3%) 1.5 ( ) 1.81 (1.04) (4.6%) (69.6%) (0.3%) (40.0%) (5.2%) (25.2%) (24.0%) (90.0%) (42.9%) 4 (2 11) INR indicates international normalized ratio; IQR, interquartile range; MI, myocardial infarction; NIHSS, National Institutes of Health Stroke Scale; TIA, transient ischemic attack; and VA, Veterans Administration. *Initial INR limited to patients on anticoagulation before admission for

4 234 Stroke January 2014 with stroke had more thromboembolic risk factors including atrial fibrillation, prosthetic heart valves, carotid stenosis, and congestive heart failure (P for all comparisons). Before hospitalization, a higher proportion of patients with in-hospital stroke used antiplatelet, anticoagulant, antihypertensive, and lipid lowering agents (P for all comparisons). Functional status, as measured by ability to ambulate independently before admission, was lower for the in-hospital stroke cohort than for that of community-onset strokes (82.6% versus 90.0%; P). Patients with in-hospital stroke experienced more severe strokes (median NIHSS 9.0 versus 4.0; P). In-hospital strokes were more likely to be reported by larger urban academic medical centers. Hospitals in the northeast United States contributed a relatively larger proportion of in-hospital stroke cases, whereas those in the south contributed fewer (Table I in the onlineonly Data Supplement). When assessed using the GWTG-Stroke quality of care metrics, in-hospital stroke care had significantly lower adherence to all 7 of the achievement measures and 5 of the 8 quality measures (Table 2). There was no significant difference in adherence to stroke education between groups. The only Table 2. GWTG-Stroke Achievement and Quality Measures Achievement measures IV tpa (arrive by 2 h, treat by 3 h)- n (%) Early antithrombotics- n (%) DVT prophylaxis- n (%) Antithrombotics on discharge- n (%) Anticoagulants on discharge for afib/flutter- n (%) Statin for LDL >100 (or if LDL not determined)- n (%) Smoking cessation counseling- n (%) Quality measures LDL 100 or ND- n (%) Dysphagia screening- n (%) Stroke education- n (%) Rehabilitation assessment- n (%) IV tpa (arrive by 3.5 h treat by 4.5 h)- n (%) Door to IV tpa in 60 min- n (%) LDL documented- n (%) Intensive statin therapy- n (%) Composite achievement measure* Achievement measures performed/achievement measure opportunities Defect-free achievement measure n (%) In-Hospital Strokes N= (31.6%) (89.4%) (88.8%) (96.1%) (90.6%) (79.1%) (92.2%) (81.1%) (65.6%) (66.8%) (96.8%) (14.6%) (19.7%) (71.5%) (20.8%) % / (60.8%) Community-Onset Strokes N= P Value (73.4%) (96.4%) (92.2%) (97.7%) (93.8%) (82.9%) (95.6%) (87.2%) (76.8%) (67.1%) (91.4%) (34.3%) (29.4%) (86.3%) (15.9%) % / (82.0%) AHA indicates American Heart Association; CDC, Centers for Disease Control; DVT, deep venous thrombosis; GWTG, Get With The Guidelines; LDL, low-density lipoprotein; ND, not determined; and tpa, tissue plasminogen activator. *Composite achievement measure is defined as the total number of achievement measure interventions performed among eligible patients divided by the total number of possible achievement measure interventions among eligible patients. Defect-free achievement measure includes all achievement measures and reflects the proportion of patients who received all of the achievement measures for which they were eligible.

5 Cumbler et al Quality of Care for In-Hospital Ischemic Stroke 235 measures for which in-hospital strokes had significantly higher adherence were rehabilitation assessment and intensive statin therapy. Adherence to opportunity composite was lower for in-hospital strokes (82.6% versus 92.8%; P). Overall, the proportion of patients having defect-free care for all of the achievement measures was lower for in-hospital strokes (60.8% versus 82.0%; P). This difference remained significant even after adjustment for stroke severity, patient, and hospital characteristics (adjusted odds ratio [OR] for defect-free care 0.33 [95% confidence intervals {CI} ]; P<0.001; Table II in the online-only Data Supplement). Patients with inhospital stroke had longer total hospital length of stay (median 6 days [25th to 75th percentile 4 11] versus 4 days [25th to 75th percentile 3 6] for community-onset strokes; P). On discharge, a lower proportion of in-hospital strokes were ambulating independently (31.0% versus 50.4%; P; Table 3). Patients with in-hospital stroke were half as likely to be discharged home compared with patients with communityonset stroke (27.7% versus 49.9%; P). The in-hospital mortality rates were higher for in-hospital strokes; 13.9% of patients with in-hospital stroke died during hospitalization compared with only 5.0% of patients with community-onset strokes (P). In multivariate models adjusting for patient demographics, comorbidities, and hospital characteristics, patients with in-hospital stroke were less likely to be able to ambulate independently at discharge (adjusted OR 0.42, 95% CI [ ]; P<0.001), and less likely to be discharged home (OR 0.37, 95% CI [ ]; P<0.001; Table 4). Mortality was higher for in-hospital stroke (OR 2.72, 95% CI [ ]; P<0.001). Results were similar in models that included stroke severity (Table 4). Eleven percentage of in-hospital strokes were treated with IV thrombolysis (2352/21 349) compared with 6.6% of community-onset strokes in this database (61 447/ ) although the proportion of those receiving tpa who were within the approved time window was significantly lower for in-hospital stroke (Table 2). Median time to IV tpa of 100 Table 3. Outcomes minutes for in-hospital strokes compared with 76 minutes for community-onset strokes (P). Patients with in-hospital stroke treated with IV tpa had overall higher adherence to quality measures (81.7% defect-free AHA-GWTG-Stroke summary measure) compared with in-hospital strokes not treated with IV tpa (57.9% defect-free AHA-GWTG-Stroke summary measure). The discrepancy between in-hospital and community-onset stroke defect-free AHA-GWTG-Stroke summary measures was diminished but remained significant (81.7% versus 83.8%; P=0.007). The narrowing of the gap was driven by higher defect-free care for in-hospital strokes in the subgroup treated with IV tpa. In univariate comparisons, patients with in-hospital stroke treated with IV tpa had similar rates of intracranial hemorrhage (4.5% versus 5.3%; P=0.099) and serious bleeding (1.6% versus 1.2%; P=0.087) but experienced higher rates of other serious complications (3.2% versus 2.5%; P=0.029) when compared with patients with community-onset stroke treated with IV tpa. After multivariable adjustment in patients treated with IV thrombolysis, those with in-hospital stroke had lower rates of intracranial hemorrhage (OR 0.80, 95% CI [ ]; P=0.049) although this was no longer significant in models adjusting for stroke severity (Table 4). Discussion In this largest study to date we found that in-hospital strokes were of greater severity and experienced higher rates of mortality and subsequent disability when compared with patients with community-onset strokes. Even after adjustment for clinical characteristics, risk factors, severity, and treatments, inhospital strokes portended worse outcomes in comparison with those in the community. Our data suggest that there was lower adherence to the majority of stroke achievement and quality measures representing opportunities for quality improvement. Previous work suggests that the risk factors for in-hospital strokes differ from those of patients with community-onset stroke. Cardioembolic pathogeneses have been found to be In-Hospital Strokes (N=21 349) Functional status- independent ambulation On discharge- n responding, missing (%) n (%) (26.4%) 4862 (31.0%) Outcomes Discharged home- n responding n (%) 5029 (27.7%) In-hospital mortality- n responding n (%) 2949 (13.9%) Complications of IV tpa Symptomatic ICH-, missing (%) 2312 (1.7) n (% experiencing this complication) 105 (4.5%) Serious bleeding-, missing (%) 2312 (1.7%) n (% experiencing this complication) 38 (1.6%) Other serious complications-, missing (%) 2312 (1.7%) n (% experiencing this complication) 75 (3.2%) ICH indicates intracranial hemorrhage; and tpa, tissue plasminogen activator. Community-Onset Strokes (N= ) (18.7%) (50.4%) (49.9%) (5.0%) (2.0%) 3208 (5.3%) (2.0%) 747 (1.2%) (2.0%) 1515 (2.5%) P Value

6 236 Stroke January 2014 Table 4. Odds Ratios for Outcomes Comparing In-Hospital Strokes With Community-Onset Strokes Unadjusted OR P Value Adjusted OR P Value Sensitivity Analysis Including NIHSS* Adjusted OR Overall cohort- OR (95% CI) Independent ambulation at discharge 0.44 ( ) < ( ) < ( ) <0.001 Discharge home 0.38 ( ) < ( ) < ( ) <0.001 In-hospital mortality 3.01( ) < ( ) < ( ) <0.001 Patients receiving IV tpa - OR (95% CI) Symptomatic intracranial hemorrhage 0.84 ( ) ( ) ( ) Independent ambulation at discharge 0.84 ( ) < ( ) ( ) Discharge home 0.75 ( ) < ( ) < ( ) <0.001 In-hospital mortality 1.28 ( ) < ( ) ( ) Generalized estimating equations models adjusted for age, sex, race, history of atrial fibrillation/flutter, previous stroke/transient ischemic attack, coronary artery disease/previous myocardial infarction, carotid stenosis, diabetes mellitus, peripheral vascular disease, hypertension, dyslipidemia, smoking, hospital region, hospital type, annual number of stroke discharges, and number of beds. Interaction terms were not included in the multivariable logistic regression models. CI indicates confidence interval; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; and tpa, tissue plasminogen activator. *Sensitivity analysis adjusting for stroke severity limited to patients with NIHSS recorded (n= overall). Analysis limited to patients receiving IV tpa (n= overall). P Value more common for in-hospital strokes. 1,8,10,17 Although we could not identify embolic strokes, our findings that risk factors for embolic events including atrial fibrillation, congestive heart failure, prosthetic heart valves, and carotid stenosis were more common in patients who suffer in-hospital strokes are consistent with the earlier results. This may help explain our observation that in-hospital strokes are more severe than community-onset strokes. Our work also validates previous observations that patients with in-hospital stroke are less likely to return directly home from the hospital and are affected by greater in-hospital mortality. 5,6 Previous publications on quality of care processes for inhospital stroke suggest in-hospital stroke evaluations are less comprehensive and take more time than community-onset stroke but are regional and limited by smaller sample size. Previous research finds brain imaging is ordered less often for in-hospital strokes, and time goals are rarely met. 1,6,10,18,19 Two previous analyses of consensus-based process measures of care have been performed using statewide registries for patients with in-hospital stroke. Analysis of a Michigan registry found lower rates of evaluation for pathogenesis but otherwise relatively minor differences in process measures between in-hospital and community-onset strokes. 6 The second, from a Colorado stroke registry, found higher rates of adherence to assessment for rehabilitation and stroke education for in-hospital strokes, driving higher overall quality scores for in-hospital stroke. 11 Our analysis overcomes sample size limitations and is the first to draw conclusions about quality of care for in-hospital stroke nationally. We confirmed that a few metrics of quality including rehabilitation assessment are higher for in-hospital stroke but, contrary to previous research, we find in-hospital strokes on the whole are less likely to receive care consistent with quality and achievement targets. The magnitude in the difference observed in defect-free care is substantial. Our observation of a quality discrepancy generates the disquieting hypothesis that some of the relatively poor outcomes of in-hospital stroke compared with community-onset stroke may result from differences in delivery of quality care during hospitalization. Patients who have an in-hospital stroke are, in general, not initially cared for by neurologists with only a minority having a neurological or neurosurgical admitting diagnosis. 6,7,18,20 Nearly one quarter of patients with in-hospital stroke are admitted with cardiovascular illness, and many occur postoperatively, particularly after cardiac valve, head and neck, or vascular surgery. 6,8,17,21 23 Multiple concurrent acute illnesses may create conflicting priorities in care and interact in a complex fashion on adherence to disease-specific quality metrics. The presence of interacting acute illnesses may increase the chance that select quality metrics such as the intensity of statin therapy or assessment for rehabilitation needs will be met for patients with in-hospital stroke. Conversely, providers for these patients may be less familiar with the stroke achievement and quality process measures that represent the current standard of care. Systems of care such as admission stroke order sets that bundle elements of quality care may be lacking for patients who have stroke after admission. These factors may explain why adherence to the majority of quality and achievement measures is lower for in-hospital strokes. Patients with in-hospital stroke have more medical contraindications to thrombolysis than those with community-onset strokes. 1,11 However, smaller studies of patients with in-hospital stroke have reported use of IV tpa for 2.6% to 13.7% of inhospital strokes. 1,6,8,11 Our study suggests such IV thrombolysis, for appropriately selected in-hospital strokes, does not carry increased risk for symptomatic intracranial hemorrhage. Our data are the first to suggest that time to treatment significantly differs between in-hospital and community-onset strokes with median difference in time to IV tpa of 24 minutes between groups. Only 1 in 5 in-hospital strokes in our study met the goal of thrombolytic treatment within 60 minutes of symptom recognition by hospital staff. The explanation for this gap is complex and may include a culture of rapid diagnosis and treatment in the Emergency Department, greater physical distance between

7 Cumbler et al Quality of Care for In-Hospital Ischemic Stroke 237 hospital wards and brain imaging/pharmacy, and less experience with thrombolytic stroke treatment by responders to stroke symptoms outside of the Emergency Department. It is possible that conflicting priorities, instability, or greater medical complexity can make thrombolytic treatment decision-making for in-hospital stroke more time consuming. However, with explicit quality improvement efforts, there is evidence that response times can be reduced Our findings suggest in-hospital stroke time to thrombolytic treatment requires targeted attention. There are several limitations of this research. Hospitals participating in GWTG-Stroke were self-selected and tended to be larger, teaching institutions, and had an interest in stroke quality improvement. Findings from this study may not be generalizable to in-hospital strokes in other US hospitals. The large sample size provided the ability to detect statistically significant differences that may not be of clinical importance. In-hospital strokes were likely under-reported in the GWTGstroke registry. The fraction of strokes with onset in the hospital in the GWTG-Stroke registry was lower than that seen in other studies. 1,3 Sites not reporting in-hospital strokes were significantly more likely to be nonacademic, rural, have smaller number of hospital beds, and have lower annual stroke volumes (data not shown). More severe strokes, or those receiving tpa, may be more likely to be reported among the population of inhospital strokes, which may introduce reporting bias. Several processes of care measures, particularly time-related measures, were modified for application to the in-hospital stroke population, and thus may not be directly comparable with the community-onset stroke population. Data were not available from the registry for the admission diagnosis or previous procedures for patients who developed in-hospital stroke. Findings were based on observational data and may be limited by residual measured and unmeasured confounders. Missing data on ambulatory status before admission or on discharge limit the certainty of findings relating to functional outcome. Finally, NIHSS was not available in 43% of patients, and therefore, we did not adjust for stroke severity in our main models. Although we performed sensitivity analyses in patients with NIHSS documented and found that adjusting for NIHSS produced similar results, we cannot be completely sure that we have fully accounted for potential confounding by stroke severity. Conclusions Several hypotheses can be generated to explain the observed worsened outcomes for in-hospital strokes. Patients with inhospital stroke may have a greater burden of acute illness, more comorbid conditions, and greater stroke severity. The observed gap in adherence to processes of stroke care is an important opportunity for targeted quality improvement because this represents a modifiable factor with potential to improve outcomes. There is an imperative to assure that all patients with stroke, regardless of location at onset, receive timely evidence-based therapy and the highest quality of care possible. Sources of Funding The Get With The Guidelines-Stroke (GWTG-Stroke) program is provided by the American Heart Association/American Stroke Association. The GWTG-Stroke program is currently supported in part by a charitable contribution from Janssen Pharmaceutical Companies of Johnson & Johnson. GWTG-Stroke has been funded in the past through support from Boeringher-Ingelheim, Merck, Bristol-Myers Squib/Sanofi Pharmaceutical Partnership, and the AHA Pharmaceutical Roundtable. Role of the Sponsors None of the above sponsors of GWTG-Stroke had any participation in the design, analysis, or preparation of this article. Disclosures Dr Cumbler is the Course Director for the In-hospital Stroke Quality Improvement initiative of the National Stroke Association. Dr Bhatt discloses the following relationships: Advisory Board in Elsevier Practice Update Cardiology, Medscape Cardiology, Regado Biosciences; Board of Directors in Boston VA Research Institute, Society of Cardiovascular Patient Care; Chair in American Heart Association Get With The Guidelines Steering Committee (AHA GWTG); Honoraria from American College of Cardiology (Editor, Clinical Trials, Cardiosource), Belvoir Publications (Editor in Chief, Harvard Heart Letter), Duke Clinical Research Institute (clinical trial steering committees), Population Health Research Institute (clinical trial steering committee), Slack Publications (Chief Medical Editor, Cardiology Today s Intervention), WebMD (CME steering committees); Other as Senior Associate Editor, Journal of Invasive Cardiology; Data Monitoring Committees in Duke Clinical Research Institute, Harvard Clinical Research Institute, Mayo Clinic, and Population Health Research Institute; Research Grants from Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, Sanofi Aventis, and The Medicines Company; and unfunded research from FlowCo, PLx Pharma, Takeda. Dr Fonarow is an UCLA employee who holds a patent on stroke retriever devices. Schwamm is chair of the AHA GWTG national Steering Committee; steering committee of Efficacy and Safety Study of Desmoteplase to Treat Acute Ischemic Stroke (DIAS4) (Lundbeck), Evaluation of the Phased Radio Frequency Ablation System (VICTORY-AF) (Medtronic); DSMB Separator 3D (Penumbra); and Genentech provides alteplase free of charge to Massachusetts General Hospital for the multicenter, National Institute of Neurological Disorders and Stroke funded MR WITNESS: A Study of Intravenous Thrombolysis With Alteplase in MRI-Selected Patients trial (PI Schwamm). References 1. Dulli D, Samaniego EA. Inpatient and Community Ischemic Strokes in a Community Hospital. Neuroepidemiology. 2007;28: Iguchi Y, Kimura K, Kobayashi K, Ueno Y, Shibazaki K, Iwanaga T, et al. In-hospital onset ischemic stroke may be associated with atrial fibrillation and right-to-left shunt. J Neurol Sci. 2007;254: Kelley RE, Kovacs AG. Mechanism of in-hospital cerebral ischemia. Stroke. 1986;17: Azzimondi G, Nonino F, Fiorani L, Vignatelli L, Stracciari A, Pazzaglia P, et al. Incidence of stroke among inpatients in a large Italian hospital. Stroke. 1994;25: Kimura K, Minematsu K, Yamaguchi T. Characteristics of in-hospital onset ischemic stroke. Eur Neurol. 2006;55: Farooq MU, Reeves MJ, Gargano J, Wehner S, Hickenbottom S, Majid A; Paul Coverdell National Acute Stroke Registry Michigan Prototype Investigators. In-hospital stroke in a statewide stroke registry. Cerebrovasc Dis. 2008;25: Blacker DJ. In-hospital stroke. Lancet Neurol. 2003;2: Park HJ, Cho HJ, Kim YD, Lee DW, Choi HY, Kim SM, et al. Comparison of the characteristics for in-hospital and out-of-hospital ischaemic strokes. Eur J Neurol. 2009;16: Aly N, McDonald K, Leathley M, Sharma A, Watkins C. Retrospective case note review of acute and inpatient stroke outcomes. BMJ. 2000;320: Bhalla A, Smeeton N, Rudd AG, Heuschmann P, Wolfe CD. A comparison of characteristics and resource use between in-hospital and admitted patients with stroke. J Stroke Cerebrovasc Dis. 2010;19: Cumbler E, Murphy P, Jones WJ, Wald HL, Kutner JS, Smith DB. Quality of care for in-hospital stroke: analysis of a statewide registry. Stroke. 2011;42:

8 238 Stroke January Fonarow GC, Reeves MJ, Smith EE, Saver JL, Zhao X, Olson D, et al. Characteristics, Performance Measures, and In-hospital Outcomes for the First One Million Stroke and Transient Ischemic Attack Admissions in Get With the Guidelines Stroke. Circ Cardiovasc Qual Outcomes. 2010;3: Schwamm LH, Fonarow GC, Reeves MJ, Pan W, Frankel MR, Smith EE, et al. Get With the Guidelines-Stroke is associated with sustained improvement in care for patients hospitalized with acute stroke or transient ischemic attack. Circulation. 2009;119: Xian Y, Fonarow GC, Reeves MJ, Webb LE, Blevins J, Demyanenko VS, et al. Data quality in the American Heart Association Get With The Guidelines-Stroke (GWTG-Stroke): results from a national data validation audit. Am Heart J. 2012;163: American Hospital Association (AHA). AHA Hospital Statistics Chicago, IL: American Hospital Association; Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44: Vera R, Lago A, Fuentes B, Gállego J, Tejada J, Casado I, et al; Stroke Project of the Spanish Cerebrovascular Diseases Study Group. In-hospital stroke: a multi-centre prospective registry. Eur J Neurol. 2011;18: Alvaro LC, Timiraos J, Sádaba F. [In-hospital stroke: clinical profile and expectations for treatment]. Neurologia. 2008;23: Masjuan J, Simal P, Fuentes B, Egido JA, Díaz-Otero F, Gil-Núñez A, et al. In-hospital stroke treated with intravenous tissue plasminogen activator. Stroke. 2008;39: Alberts MJ, Brass LM, Perry A, Webb D, Dawson DV. Evaluation times for patients with in-hospital strokes. Stroke. 1993;24: Witt BJ, Ballman KV, Brown RD Jr, Meverden RA, Jacobsen SJ, Roger VL. The incidence of stroke after myocardial infarction: a meta-analysis. Am J Med. 2006;119:354.e1 354.e Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Borden WB, et al; American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics 2013 update: a report from the American Heart Association. Circulation. 2013;127:e6 e Selim M. Perioperative stroke. N Engl J Med. 2007;356: Cumbler E, Anderson T, Neumann R, Jones WJ, Brega K. Stroke alert program improves recognition and evaluation time of in-hospital ischemic stroke. J Stroke Cerebrovasc Dis. 2010;19: Cumbler E, Zaemisch R, Graves A, Brega K, Jones W. Improving stroke alert response time: applying quality improvement methodology to the inpatient neurologic emergency. J Hosp Med. 2012;7: Nolan S, Naylor G, Burns M. Code gray an organized approach to inpatient stroke. Crit Care Nurs Q. 2003;26: