You are here

Article Text

Article menu
Ischemic stroke
Original research
Comparison of outcomes of patients with inpatient or outpatient onset ischemic stroke
  1. Kimon Bekelis1,2,
  2. Symeon Missios3,
  3. Shannon Coy4,
  4. Todd A MacKenzie2,5,6
  1. 1Section of Neurosurgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA
  2. 2The Dartmouth Institute for Health Policy and Clinical Practice, Lebanon, New Hampshire, USA
  3. 3Department of Neurosurgery, Louisiana State University Health Sciences Center, Shreveport, Louisiana, USA
  4. 4Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA
  5. 5Department of Medicine, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA
  6. 6Department of Community and Family Medicine, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA
  1. Correspondence to Dr Kimon Bekelis, Dartmouth-Hitchcock Medical Center, One Medical Center Dr, Lebanon, NH 03756, USA; kbekelis{at}gmail.com

Abstract

Background Reperfusion times for ischemic stroke occurring in the outpatient setting have improved significantly in recent years. However, quality improvement efforts have largely ignored ischemic stroke occurring in patients hospitalized for unrelated indications.

Methods We performed a cohort study involving patients with ischemic stroke (with inpatient or outpatient onset) from 2009 to 2013 who were registered in the Statewide Planning and Research Cooperative System (SPARCS) database. A propensity score-adjusted regression analysis was used to assess the association of location of onset and outcomes. Mixed effects methods were employed to control for clustering at the hospital level.

Results Of the 176 571 ischemic strokes, 160 157 (90.7%) occurred outside of a hospital and 16 414 (9.3%) occurred in patients hospitalized for unrelated indications. Using a logistic regression model with propensity score adjustment, we demonstrated that inpatient stroke onset was associated with increased inpatient mortality (OR 3.09; 95% CI 2.81 to 3.38), rate of discharge to rehabilitation (OR 2.57; 95% CI 2.37 to 2.79), and length of stay (LOS) (β=11.58; 95% CI 10.73 to 12.42). In addition, it was associated with lower odds (OR 0.69; 95% CI 0.62 to 0.77) of undergoing stroke-related interventions (mechanical thrombectomy and intravenous tissue plasminogen activator) compared with outpatient stroke onset.

Conclusions Using a comprehensive all-payer cohort of patients with ischemic stroke in New York State, we identified an association of inpatient stroke onset with fewer stroke-related interventions and increased mortality, rate of discharge to rehabilitation, and LOS.

  • Embolic
  • Stroke

https://doi.org/10.1136/neurintsurg-2015-012145

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Timely revascularization is crucial for patients with acute ischemic stroke in order to improve survival and preserve function.1 Administration of intravenous (IV) tissue plasminogen activator (tPA) is central in this process.1 Several initiatives have led to the establishment of pre-hospital protocols for the expedited evaluation and treatment of this population.2–8 Telemedicine,9 mobile stroke units,10 and ‘drip-and-ship’ approaches11 are steps in this direction. In addition, national organizations such as the American Heart/Stroke Association (AHA/ASA) with their ‘Get with the Guidelines-Stroke’ (GWTG-S) program2 are monitoring the effectiveness of these strategies and are constantly setting new benchmarks. More recently, the multiple randomized trials12–14 supporting the efficacy of newer mechanical thrombectomy devices in the treatment of acute ischemic stroke are used as evidence to enhance the efficiency of the pre-hospital triage networks. However, little attention has been paid to the creation of similar initiatives to address ischemic stroke occurring in inpatients.

    Although significant literature has supported the association of the setting of myocardial infarction onset and outcomes,15–19 limited such information exists in the setting of cerebrovascular disease. Previous single-center analyses20–22 focusing on ischemic stroke of inpatient onset have described the increased comorbidity burden of this patient population. Moradiya and Levine23 have demonstrated that administration of IV tPA to inpatients is associated with increased mortality in comparison to outpatients. A study of a few centers in Michigan demonstrated that inpatient stroke was associated with inferior functional outcomes.24 However, the authors did not differentiate between ischemic and hemorrhagic stroke and did not use multivariable techniques to adjust for confounding. There has been no previous study investigating the association of the setting of ischemic stroke onset and outcomes using advanced observational techniques in a multicenter setting.

    We used the New York Statewide Planning and Research Cooperative System (SPARCS)25 to study the association of the setting of stroke onset with mortality, discharge to rehabilitation, length of stay, mechanical thrombectomy, and IV tPA administration for patients presenting with ischemic stroke. A propensity-adjusted regression model with mixed effects was used to control for confounding and account for clustering at the hospital level.

    Methods

    New York Statewide Planning and Research Cooperative System (SPARCS)

    All patients with acute ischemic stroke who were registered in the SPARCS database25 between 2009 and 2013 were included in the analysis. For these years, the database contains patient-level details for every hospital discharge, ambulatory surgery, and emergency department admission in New York State as coded from admission and billing records. More information about SPARCS is available at https://www.health.ny.gov/statistics/sparcs/.

    Cohort definition

    In order to establish the cohort of patients we used the International Classification of Disease-9-Clinical Modification (ICD-9-CM) codes to identify patients in the database who had a diagnosis code of ischemic stroke as a primary or secondary diagnosis (ICD-9-CM code 433.x1, 434.x1) between 2009 and 2013. The SPARCS database allows the determination of whether or not a diagnosis code was present on admission. Patients with a diagnosis of ischemic stroke present on admission were classified in the ‘outpatient stroke’ group, while those who had a diagnosis code of ischemic stroke which was not present on admission, developed their stroke during their inpatient stay and were thus categorized in the ‘inpatient stroke’ cohort.

    Outcome variables

    The primary outcome variables were in-hospital mortality, length of stay (LOS), and rate of discharge to a rehabilitation facility. The secondary outcome we examined was whether patients underwent a stroke-related intervention. The latter was defined as endovascular mechanical thrombectomy (ICD-9 code 39.74) or administration of IV tPA (ICD-9 code 99.10).

    Exposure variables

    The primary exposure variable was the setting of ischemic stroke onset (inpatient vs outpatient). Covariates (see online supplementary table S1) used for risk adjustment were age, gender, race (African-American, Hispanic, Asian, Caucasian, and other), and insurance coverage (private, Medicare, Medicaid, uninsured, and other). The comorbidities used for risk adjustment were diabetes mellitus, smoking, chronic lung disease, hypertension, hypercholesterolemia, peripheral vascular disease (PVD), congestive heart failure (CHF), coronary artery disease, transient ischemic attack, alcohol abuse, obesity, chronic renal failure, and coagulopathy. The Charlson Comorbidity Index (CCI)26 ,27 was calculated and used in part of the analysis. Only variables that were defined as ‘present on admission’ were considered part of the patient's preadmission comorbidity profile.

    Statistical analysis

    The association of stroke onset setting with our outcome measures was examined in a multivariable setting. A logistic regression was employed for categorical outcomes (mortality, discharge to rehabilitation, stroke-related intervention)28 and a linear regression for linear outcomes (LOS). The covariates used for risk adjustment in these models were age, gender, race, insurance, and all the comorbidities mentioned previously. Mixed effects methods were employed, with hospital facility used as a random effects variable, in order to account for clustering at the hospital level.

    In an alternative method to control for confounding, we used a propensity-adjusted (with deciles of propensity score) regression model. Mixed effects with hospital facility key as random effects variable were employed. We calculated the propensity score of inpatient stroke onset with a separate logistic regression model using all the covariates mentioned previously. Propensity adjustment compares subjects within the same decile of propensity score, balancing potential differences in covariates, while using the entire cohort without any observations being discarded. In this study we favored this approach over matching because the latter would significantly restrict the sample size to matched individuals only. We repeated our analysis using propensity score matching (with a significantly smaller sample size) and our results were robust.

    In prespecified sensitivity analysis we repeated the above analyses separately for the two subcategories of interventions (mechanical thrombectomy, IV-tPA). Additionally, logarithmic transformation of the values of LOS did not change the direction of the associations. We therefore elected to only report the untransformed LOS results to facilitate interpretation. Last, all previous analyses were repeated using the calculated CCI instead of individual comorbidities. The direction of the associations did not change and therefore these results are not reported further.

    Regression diagnostics were used for all models. All results are based on two-sided tests and the level of statistical significance was set at 0.05. This study, based on 176 571 patients, has sufficient power (90%) at a 5% type I error rate to detect differences in mortalit, as small as 2.6%. Statistical analyses were performed using Stata V.13 (StataCorp, College Station, Texas, USA), the 64-bit version of R.3.1.0 (R Foundation for Statistical Computing), and SPSS V.22 (IBM, Armonk, New York, USA).

    Results

    Patient characteristics

    In the selected study period there were 176 571 patients presenting with ischemic stroke (mean age 71.3 years, 53.0% women) who were registered in SPARCS (figure 1). Of those, 160 157 (90.7%) had an outpatient stroke onset and 16 414 (9.3%) had an inpatient stroke onset. The characteristics of the two cohorts at baseline are shown in table 1. Stroke patients with inpatient onset were older and more likely to have multiple comorbidities. We adjusted for these imbalances with further analyses.

    View this table:
    Table 1

    Patient characteristics

    Figure 1
    Figure 1

    Cohort creation. SPARCS, Statewide Planning and Research Cooperative System.

    Inpatient mortality

    Overall, 11 330 (7.1%) inpatient deaths were recorded after outpatient onset and 4050 (24.7%) after inpatient onset of ischemic stroke (table 2). Inpatient onset was associated with increased mortality (OR 4.30; 95% CI 4.13 to 4.78) in the unadjusted analysis. This association persisted (OR 3.08; 95% CI 2.82 to 3.37) after using a mixed effects logistic regression (table 3). The same was true in a propensity score-adjusted model (OR 3.09; 95% CI 2.81 to 3.38).

    View this table:
    Table 2

    Primary and secondary outcomes

    View this table:
    Table 3

    Correlation of outcome measures with the setting of ischemic stroke

    Discharge to rehabilitation

    Overall, 73 828 (49.6%) were discharged to rehabilitation after outpatient stroke onset and 9153 (74.0%) after inpatient stroke onset (table 2). Inpatient onset was associated with an increased rate of discharge to rehabilitation (OR 2.90; 95% CI 2.78 to 3.02) in the unadjusted analysis. This persisted (OR 2.59; 95% CI 2.39 to 2.80) after using a mixed effects logistic regression model (table 3). We found similar results in a propensity score-adjusted model (OR 2.57; 95% CI 2.37 to 2.79).

    Stroke-related interventions

    Overall, 9973 (6.2%) patients with outpatient stroke onset underwent a stroke-related intervention compared with 778 (4.7%) after inpatient stroke onset (table 2). Inpatient onset was associated with decreased odds of intervention in comparison with outpatient onset (OR 0.75; 95% CI 0.70 to 0.81) in the unadjusted analysis. This persisted (table 3) in a mixed effects logistic regression model (OR 0.68; 95% CI 0.62 to 0.76) and a propensity score-adjusted model (OR 0.69; 95% CI 0.62 to 0.77).

    In sensitivity analysis we demonstrated that inpatient stroke onset was associated with decreased odds of undergoing mechanical thrombectomy (OR 0.72; 95% CI 0.54 to 0.96) or receiving IV tPA (OR 0.75; 95% CI 0.70 to 0.81) compared with outpatient onset in a propensity score-adjusted logistic regression model (table 3). For the entire cohort, receiving an intervention did not demonstrate an association with mortality (OR 1.04; 95% CI 0.96 to 1.10) or LOS (β=0.10; 95% CI −0.21 to 0.40). However this observation is difficult to interpret, given the lack of adjustment for stroke severity.

    Length of stay (LOS)

    The mean (SD) LOS was 7.9 (13) days after outpatient stroke onset and 21 (30) days after inpatient stroke onset (table 2). Inpatient onset was associated with increased LOS (β=13.18; 95% CI 12.94 to 13.42) in the unadjusted analysis. This persisted (β=11.68; 95% CI 10.83 to 12.54) after using a mixed effects linear regression model (table 3). We found similar results in a propensity score-adjusted linear regression model (β=11.58; 95% CI 10.73 to 12.42).

    Discussion

    Using a comprehensive all-payer cohort of patients with ischemic stroke in New York State, we identified an association between inpatient stroke onset and increased mortality, discharge to rehabilitation, and LOS. Outpatients developing ischemic stroke were more likely to undergo mechanical thrombectomy or receive IV tPA. Our results were robust when considering several advanced observational techniques to account for confounders. Patients with inpatient stroke onset are excluded from quality improvement registries such as GWTG-Stroke. The impact of these initiatives on inpatient onset of stroke and the outcomes associated with the latter compared with outpatient onset of stroke remain an issue of debate. Similar observations have been made before for patients suffering inpatient or outpatient myocardial infarction.15–19 Despite the similarities in urgency between cardiovascular and cerebrovascular pathology, there is limited literature on the association of the setting of stroke onset and outcomes.

    The inferior outcomes observed in patients in the current study with inpatient stroke onset can be partially explained by the increased age and comorbidity burden of this population. Inpatients developing strokes are already sufficiently ill to justify inpatient admission compared with outpatients. However, these observations persisted after using several techniques to balance potential confounders between the two groups. Although causal inference in this setting can be challenging, some potential insights into this phenomenon are offered by our data. We observed limited exposure of inpatient onset patients to IV tPA and mechanical thrombectomy compared with patients with ictus onset in the outpatient setting. IV tPA is considered the gold standard for the treatment of patients with ischemic stroke presenting within the appropriate time frame, barring any contraindications.1 Likewise, mechanical thrombectomy of appropriate candidates with new generation devices has been shown to improve the outcomes of this patient population.12–14 Inpatients are less likely to be candidates for such interventions due to their increased risk of bleeding, postoperative status, or heavy comorbidity burden. Moreover, they are more likely to have advanced directives, communicate their wishes with the care team, and potentially refuse intervention for a new stroke event.

    Inefficient recognition of stroke symptoms and timely intervention can also explain the inferior outcomes of inpatient stroke onset. Many patients may be intubated, on anxiolytic or sedative medications, or are not undergoing frequent neurologic checks, which means their strokes may go unrecognized. There is a growing body of literature focusing on the effect of integrated health systems,29 ,30 process measures, and speed of intervention on the outcomes of patients with ischemic stroke of outpatient onset. Although some hospitals have implemented ‘stroke code’ initiatives for inpatients, their universal use and effectiveness remain questionable. Previous studies have shown that inpatients developing ischemic stroke are more likely to experience delays in recognition of symptoms and decision-making.20 ,21 ,31 Stroke mimickers trigger the activation of rapid response protocols more commonly in inpatients.20 ,21 This is particularly prominent in the setting of competing comorbidities, polypharmacy, and a care team without special expertise in neurologic diagnosis and treatment.20 ,21

    Our study has several limitations common to administrative databases. Residual confounding could account for some of the observed associations. We used several advanced observational techniques to control for confounding (by balancing covariates among comparison groups) and clustering at the hospital level. In addition, coding inaccuracies will undoubtedly occur and can affect our estimates. However, several reports have demonstrated that coding for stroke has shown nearly perfect association with medical record review.32 ,33 Although SPARCS includes all hospitals from the entire New York State, the generalization of this analysis to the entire US population is uncertain. SPARCS does not provide any clinical information on stroke severity (ie, National Institutes of Health Stroke Scale), which is important in order to optimize risk adjustment. However, we adjusted for several cardiovascular disease risk factors which are expected to be correlated with stroke severity.

    Additionally, we were lacking post-hospitalization and long-term data on our patients. Quality metrics (ie, modified Rankin score) are also not available through this database and therefore we cannot compare the stroke onset settings on these outcomes. Although discharge to home does not always indicate a favorable outcome, discharge status has been shown to correlate well with modified Rankin Scale score and provides an important insight into differences within and between treatment modalities.34 ,35 The time period of the study was before the availability of class IA evidence for mechanical thrombectomy, with substantial differences in regional practice. Although drawing any conclusions in this setting about the efficacy of thrombectomy is of limited value, we do not expect the lack of class IA evidence to have any differential impact on the rate of this procedure in the inpatient and outpatient settings. Finally, causality cannot be definitively established based on observational data, despite the use of advanced techniques. Future analyses are necessary to identify the reasons behind the observed associations.

    Conclusions

    Reperfusion times for ischemic stroke occurring in the outpatient setting have improved significantly in recent years. However, quality improvement efforts have largely ignored ischemic stroke occurring in patients hospitalized for unrelated indications. Using a comprehensive all-payer cohort of patients with ischemic stroke in New York State, we identified an association of inpatient stroke onset with increased mortality, discharge to rehabilitation, and LOS. Outpatients developing ischemic stroke were more likely to undergo mechanical thrombectomy or receive IV tPA. Our results were robust when considering several advanced observational techniques to account for confounders.

    References

    1. ↵
      1. Fonarow GC,
      2. Smith EE,
      3. Saver JL, et al
      . Timeliness of tissue-type plasminogen activator therapy in acute ischemic stroke: patient characteristics, hospital factors, and outcomes associated with door-to-needle times within 60 minutes. Circulation 2011;123:750–8. doi:10.1161/CIRCULATIONAHA.110.974675
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Ekundayo OJ,
      2. Saver JL,
      3. Fonarow GC, et al
      . Patterns of emergency medical services use and its association with timely stroke treatment: findings from Get With the Guidelines-Stroke. Circ Cardiovasc Qual Outcomes 2013;6:262–9. doi:10.1161/CIRCOUTCOMES.113.000089
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Fonarow GC,
      2. Smith EE,
      3. Saver JL, et al
      . Improving door-to-needle times in acute ischemic stroke: the design and rationale for the American Heart Association/American Stroke Association's Target: Stroke initiative. Stroke 2011;42:2983–9. doi:10.1161/STROKEAHA.111.621342
      OpenUrlAbstract/FREE Full Text
    4. ↵
      1. Fonarow GC,
      2. Zhao X,
      3. Smith EE, et al
      . Door-to-needle times for tissue plasminogen activator administration and clinical outcomes in acute ischemic stroke before and after a quality improvement initiative. JAMA 2014;311:1632–40. doi:10.1001/jama.2014.3203
      OpenUrlCrossRefPubMed
    5. ↵
      1. Gumbinger C,
      2. Reuter B,
      3. Stock C, et al
      . Time to treatment with recombinant tissue plasminogen activator and outcome of stroke in clinical practice: retrospective analysis of hospital quality assurance data with comparison with results from randomised clinical trials. BMJ 2014;348:g3429. doi:10.1136/bmj.g3429
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Sauser K,
      2. Levine DA,
      3. Nickles AV, et al
      . Hospital variation in thrombolysis times among patients with acute ischemic stroke: the contributions of door-to-imaging time and imaging-to-needle time. JAMA Neurol 2014;71:1155–61. doi:10.1001/jamaneurol.2014.1528
      OpenUrlCrossRefPubMed
    7. ↵
      1. Schellinger PD,
      2. Köhrmann M
      . 4.5-hour time window for intravenous thrombolysis with recombinant tissue-type plasminogen activator is established firmly. Stroke 2014;45:912–13. doi:10.1161/STROKEAHA.113.002700
      OpenUrlFREE Full Text
    8. ↵
      1. Xian Y,
      2. Smith EE,
      3. Zhao X, et al
      . Strategies used by hospitals to improve speed of tissue-type plasminogen activator treatment in acute ischemic stroke. Stroke 2014;45:1387–95. doi:10.1161/STROKEAHA.113.003898
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Müller-Barna P,
      2. Hubert GJ,
      3. Boy S, et al
      . TeleStroke units serving as a model of care in rural areas: 10-year experience of the TeleMedical project for integrative stroke care. Stroke 2014;45:2739–44. doi:10.1161/STROKEAHA.114.006141
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Parker SA,
      2. Bowry R,
      3. Wu TC, et al
      . Establishing the first mobile stroke unit in the United States. Stroke 2015;46:1384–91. doi:10.1161/STROKEAHA.114.007993
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Martin-Schild S,
      2. Morales MM,
      3. Khaja AM, et al
      . Is the drip-and-ship approach to delivering thrombolysis for acute ischemic stroke safe? J Emerg Med 2011;41:135–41. doi:10.1016/j.jemermed.2008.10.018
      OpenUrlCrossRefPubMed
    12. ↵
      1. Berkhemer OA,
      2. Fransen PS,
      3. Beumer D, et al
      . A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med 2015;372:11–20. doi:10.1056/NEJMoa1411587
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Goyal M,
      2. Demchuk AM,
      3. Menon BK, et al
      . Randomized assessment of rapid endovascular treatment of ischemic stroke. Stroke 2015;372:1019–30.
      OpenUrl
    14. ↵
      1. Campbell BC,
      2. Mitchell PJ,
      3. Kleinig TJ, et al
      . Endovascular therapy for ischemic stroke with perfusion-imaging selection. N Engl J Med 2015;372:1009–18. doi:10.1056/NEJMoa1414792
      OpenUrlCrossRefPubMed
    15. ↵
      1. Kaul P,
      2. Federspiel JJ,
      3. Dai X, et al
      . Association of inpatient vs outpatient onset of ST-elevation myocardial infarction with treatment and clinical outcomes. JAMA 2015;312:1999–2007. doi:10.1001/jama.2014.15236
      OpenUrl
    16. ↵
      1. Dai X,
      2. Bumgarner J,
      3. Spangler A, et al
      . Acute ST-elevation myocardial infarction in patients hospitalized for noncardiac conditions. J Am Heart Assoc 2013;2:e000004. doi:10.1161/JAHA.113.000004
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Zahn R,
      2. Schiele R,
      3. Seidl K, et al
      . Acute myocardial infarction occurring in versus out of the hospital: patient characteristics and clinical outcome. Maximal Individual TheRapy in Acute Myocardial Infarction (MITRA) Study Group. J Am Coll Cardiol 2000;35:1820–6. doi:10.1016/S0735-1097(00)00629-X
      OpenUrlFREE Full Text
    18. ↵
      1. Garberich RF,
      2. Traverse JH,
      3. Claussen MT, et al
      . ST-elevation myocardial infarction diagnosed after hospital admission. Circulation 2014;129:1225–32. doi:10.1161/CIRCULATIONAHA.113.005568
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Maynard C,
      2. Lowy E,
      3. Rumsfeld J, et al
      . The prevalence and outcomes of in-hospital acute myocardial infarction in the Department of Veterans Affairs Health System. Arch Intern Med 2006;166:1410–16. doi:10.1001/archinte.166.13.1410
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Hand PJ,
      2. Kwan J,
      3. Lindley RI, et al
      . Distinguishing between stroke and mimic at the bedside: the brain attack study. Stroke 2006;37:769–75. doi:10.1161/01.STR.0000204041.13466.4c
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Iguchi Y,
      2. Kimura K,
      3. Kobayashi K, et al
      . In-hospital onset ischemic stroke may be associated with atrial fibrillation and right-to-left shunt. J Neurol Sci 2007;254:39–43. doi:10.1016/j.jns.2006.12.013
      OpenUrlCrossRefPubMed
    22. ↵
      1. Park HJ,
      2. Cho HJ,
      3. Kim YD, et al
      . Comparison of the characteristics for in-hospital and out-of-hospital ischaemic strokes. Eur J Neurol 2009;16: 582–8. doi:10.1111/j.1468-1331.2009.02538.x
      OpenUrlCrossRefPubMed
    23. ↵
      1. Moradiya Y,
      2. Levine SR
      . Comparison of short-term outcomes of thrombolysis for in-hospital stroke and out-of-hospital stroke in United States. Stroke 2013;44:1903–8. doi:10.1161/STROKEAHA.113.000945
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Farooq MU,
      2. Reeves MJ,
      3. Gargano J, et al
      . In-hospital stroke in a statewide stroke registry. Cerebrovasc Dis 2008;25:12–20. doi:10.1159/000111494
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      New York State Department of Health. Statewide Planning and Research Cooperative System (SPARCS). Secondary Statewide Planning and Research Cooperative System (SPARCS) 2015. https://www.health.ny.gov/statistics/sparcs/
    26. ↵
      1. Romano PS,
      2. Roos LL,
      3. Jollis JG
      . Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol 1993;46:1075–9. doi:10.1016/0895-4356(93)90103-8
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Charlson ME,
      2. Ales KL,
      3. Simon R, et al
      . Why predictive indexes perform less well in validation studies. Is it magic or methods? Arch Intern Med 1987;147:2155–61.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Foster EM
      . Instrumental variables for logistic regression: an illustration. Soc Sci Res 1997;26:287–504. doi:10.1006/ssre.1997.0606
      OpenUrl
    29. ↵
      1. Lichtman JH,
      2. Jones SB,
      3. Wang Y, et al
      . Outcomes after ischemic stroke for hospitals with and without Joint Commission-certified primary stroke centers. Neurology 2011;76:1976–82. doi:10.1212/WNL.0b013e31821e54f3
      OpenUrlCrossRefPubMed
    30. ↵
      1. Alberts MJ,
      2. Wechsler LR,
      3. Jensen ME, et al
      . Formation and function of acute stroke-ready hospitals within a stroke system of care recommendations from the brain attack coalition. Stroke 2013;44:3382–93. doi:10.1161/STROKEAHA.113.002285
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Cumbler E,
      2. Simpson J
      . Code stroke: multicenter experience with in-hospital stroke alerts. J Hosp Med 2015;10:179–83. doi:10.1002/jhm.2311
      OpenUrlCrossRefPubMed
    32. ↵
      1. Kokotailo RA,
      2. Hill MD
      . Coding of stroke and stroke risk factors using international classification of diseases, revisions 9 and 10. Stroke 2005;36:1776–81. doi:10.1161/01.STR.0000174293.17959.a1
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Tirschwell DL,
      2. Longstreth WTJ
      . Validating administrative data in stroke research. Stroke 2002;33:2465–70. doi:10.1161/01.STR.0000032240.28636.BD
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Bonita R,
      2. Beaglehole R
      . Recovery of motor function after stroke. Stroke 1988;19:1497–500. doi:10.1161/01.STR.19.12.1497
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Qureshi AI,
      2. Chaudhry SA,
      3. Sapkota BL, et al
      . Discharge destination as a surrogate for Modified Rankin Scale defined outcomes at 3- and 12-months poststroke among stroke survivors. Arch Phys Med Rehabil 2012;93:1408–13. doi:10.1016/j.apmr.2012.02.032
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors KB: concept, design, manuscript preparation, statistical analysis, data interpretation. SM: data interpretation, critical review of manuscript. SC: data interpretation, critical review of manuscript. TAM: data interpretation, cohort creation, critical review of manuscript.

    • Funding Supported by grants from the National Institutes of Health through the National Center for Advancing Translational Sciences (UL1TR001086). The funders had no role in the design, execution, or interpretation of the study, or the manuscript preparation.

    • Competing interests None declared.

    • Ethics approval Ethics approval was obtained from the Committee for Protection of Human Subjects.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement All data are included in the study.