[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Table 1.  
Baseline Characteristics for 8076 Patients with TIA or Nonsevere Stroke
Baseline Characteristics for 8076 Patients with TIA or Nonsevere Stroke
Table 2.  
Final Risk-Adjustment Model for 90-Day and 1-Year Mortality and Recurrent Ischemic Stroke Events
Final Risk-Adjustment Model for 90-Day and 1-Year Mortality and Recurrent Ischemic Stroke Events
Table 3.  
Risk-Adjusted Association of Passing an Individual Process of Care Measure With Recurrent Vascular Events
Risk-Adjusted Association of Passing an Individual Process of Care Measure With Recurrent Vascular Events
1.
Johnston  KC, Connors  AF  Jr, Wagner  DP, Knaus  WA, Wang  X, Haley  EC  Jr.  A predictive risk model for outcomes of ischemic stroke.  Stroke. 2000;31(2):448-455. doi:10.1161/01.STR.31.2.448PubMedGoogle ScholarCrossref
2.
Wu  CM, McLaughlin  K, Lorenzetti  DL, Hill  MD, Manns  BJ, Ghali  WA.  Early risk of stroke after transient ischemic attack: a systematic review and meta-analysis.  Arch Intern Med. 2007;167(22):2417-2422. doi:10.1001/archinte.167.22.2417PubMedGoogle ScholarCrossref
3.
Giles  MF, Rothwell  PM.  Risk of stroke early after transient ischaemic attack: a systematic review and meta-analysis.  Lancet Neurol. 2007;6(12):1063-1072. doi:10.1016/S1474-4422(07)70274-0PubMedGoogle ScholarCrossref
4.
Ranta  A, Dovey  S, Weatherall  M, O’Dea  D, Gommans  J, Tilyard  M.  Cluster randomized controlled trial of TIA electronic decision support in primary care.  Neurology. 2015;84(15):1545-1551. doi:10.1212/WNL.0000000000001472PubMedGoogle ScholarCrossref
5.
Lavallée  PC, Meseguer  E, Abboud  H,  et al.  A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects.  Lancet Neurol. 2007;6(11):953-960. doi:10.1016/S1474-4422(07)70248-XPubMedGoogle ScholarCrossref
6.
Luengo-Fernandez  R, Gray  AM, Rothwell  PM.  Effect of urgent treatment for transient ischaemic attack and minor stroke on disability and hospital costs (EXPRESS study): a prospective population-based sequential comparison.  Lancet Neurol. 2009;8(3):235-243. doi:10.1016/S1474-4422(09)70019-5PubMedGoogle ScholarCrossref
7.
Rothwell  PM, Giles  MF, Chandratheva  A,  et al; Early Use of Existing Preventive Strategies for Stroke (EXPRESS) Study.  Effect of urgent treatment of transient ischaemic attack and minor stroke on early recurrent stroke (EXPRESS study): a prospective population-based sequential comparison.  Lancet. 2007;370(9596):1432-1442. doi:10.1016/S0140-6736(07)61448-2PubMedGoogle ScholarCrossref
8.
Kernan  WN, Ovbiagele  B, Black  HR,  et al; American Heart Association Stroke Council; Council on Cardiovascular and Stroke Nursing; Council on Clinical Cardiology; Council on Peripheral Vascular Disease.  Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/American Stroke Association.  Stroke. 2014;45(7):2160-2236. doi:10.1161/STR.0000000000000024PubMedGoogle ScholarCrossref
9.
Bravata  DM, Myers  LJ, Arling  G,  et al.  Quality of care for veterans with transient ischemic attack and minor stroke.  JAMA Neurol. 2018;75(4):419-427. doi:10.1001/jamaneurol.2017.4648PubMedGoogle ScholarCrossref
10.
Bravata  DM, Myers  LJ, Cheng  E,  et al.  Development and validation of electronic quality measures to assess care for patients with transient ischemic attack and minor ischemic stroke.  Circ Cardiovasc Qual Outcomes. 2017;10(9):10. doi:10.1161/CIRCOUTCOMES.116.003157PubMedGoogle ScholarCrossref
11.
Johnston  SC.  Clinical practice: transient ischemic attack.  N Engl J Med. 2002;347(21):1687-1692. doi:10.1056/NEJMcp020891PubMedGoogle ScholarCrossref
12.
Phipps  MS, Fahner  J, Sager  D, Coffing  J, Maryfield  B, Williams  LS.  Validation of stroke Meaningful Use measures in a national electronic health record system.  J Gen Intern Med. 2016;31(suppl 1):46-52. doi:10.1007/s11606-015-3562-5PubMedGoogle ScholarCrossref
13.
The Joint Commission. Stroke (STK). https://manual.jointcommission.org/releases/TJC2016A/Stroke.html. Accessed May 22, 2019.
14.
Smith  EE, Saver  JL, Alexander  DN,  et al; AHA/ASA Stroke Performance Oversight Committee.  Clinical performance measures for adults hospitalized with acute ischemic stroke: performance measures for healthcare professionals from the American Heart Association/American Stroke Association.  Stroke. 2014;45(11):3472-3498. doi:10.1161/STR.0000000000000045PubMedGoogle ScholarCrossref
15.
Chobanian  AV, Bakris  GL, Black  HR,  et al; Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; National Heart, Lung, and Blood Institute; National High Blood Pressure Education Program Coordinating Committee.  Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.  Hypertension. 2003;42(6):1206-1252.PubMedGoogle ScholarCrossref
16.
Borzecki  AM, Wong  AT, Hickey  EC, Ash  AS, Berlowitz  DR.  Can we use automated data to assess quality of hypertension care?  Am J Manag Care. 2004;10(7 Pt 2):473-479.PubMedGoogle Scholar
17.
Johnson  AM, Goldstein  LB, Bennett  P, O’Brien  EC, Rosamond  WD; Investigators of the Registry of the North Carolina Stroke Care Collaborative.  Compliance with acute stroke care quality measures in hospitals with and without primary stroke center certification: the North Carolina Stroke Care Collaborative.  J Am Heart Assoc. 2014;3(2):e000423. doi:10.1161/JAHA.113.000423PubMedGoogle ScholarCrossref
18.
O’Brien  EC, Zhao  X, Fonarow  GC,  et al.  Quality of care and ischemic stroke risk after hospitalization for transient ischemic attack: findings from Get With The Guidelines-Stroke.  Circ Cardiovasc Qual Outcomes. 2015;8(6)(suppl 3):S117-S124. doi:10.1161/CIRCOUTCOMES.115.002048PubMedGoogle ScholarCrossref
19.
Ovbiagele  B, Schwamm  LH, Smith  EE,  et al.  Patterns of care quality and prognosis among hospitalized ischemic stroke patients with chronic kidney disease.  J Am Heart Assoc. 2014;3(3):e000905. doi:10.1161/JAHA.114.000905PubMedGoogle ScholarCrossref
20.
Glickman  ME, Rao  SR, Schultz  MR.  False discovery rate control is a recommended alternative to Bonferroni-type adjustments in health studies.  J Clin Epidemiol. 2014;67(8):850-857. doi:10.1016/j.jclinepi.2014.03.012PubMedGoogle ScholarCrossref
21.
Johnston  SC, Gress  DR, Browner  WS, Sidney  S.  Short-term prognosis after emergency department diagnosis of TIA.  JAMA. 2000;284(22):2901-2906. doi:10.1001/jama.284.22.2901PubMedGoogle ScholarCrossref
22.
Knoflach  M, Lang  W, Seyfang  L,  et al; Austrian Stroke Unit Collaborators.  Predictive value of ABCD2 and ABCD3-I scores in TIA and minor stroke in the stroke unit setting.  Neurology. 2016;87(9):861-869. doi:10.1212/WNL.0000000000003033PubMedGoogle ScholarCrossref
23.
Bray  BD, Ayis  S, Campbell  J,  et al.  Associations between the organisation of stroke services, process of care, and mortality in England: prospective cohort study.  BMJ. 2013;346:f2827. doi:10.1136/bmj.f2827PubMedGoogle ScholarCrossref
24.
Hillmann  S, Wiedmann  S, Rücker  V,  et al; German Stroke Register Study Group (ADSR).  Stroke unit care in Germany: the German Stroke Registers Study Group (ADSR).  BMC Neurol. 2017;17(1):49. doi:10.1186/s12883-017-0819-0PubMedGoogle ScholarCrossref
25.
Amarenco  P, Lavallée  PC, Labreuche  J,  et al; TIAregistry.org Investigators.  One-year risk of stroke after transient ischemic attack or minor stroke.  N Engl J Med. 2016;374(16):1533-1542. doi:10.1056/NEJMoa1412981PubMedGoogle ScholarCrossref
26.
Johnson  ES, Lanes  SF, Wentworth  CE  III, Satterfield  MH, Abebe  BL, Dicker  LW.  A metaregression analysis of the dose-response effect of aspirin on stroke.  Arch Intern Med. 1999;159(11):1248-1253.PubMedGoogle ScholarCrossref
27.
Sheehan  OC, Merwick  A, Kelly  LA,  et al.  Diagnostic usefulness of the ABCD2 score to distinguish transient ischemic attack and minor ischemic stroke from noncerebrovascular events: the North Dublin TIA Study.  Stroke. 2009;40(11):3449-3454. doi:10.1161/STROKEAHA.109.557074PubMedGoogle ScholarCrossref
28.
Kernan  WN, Viscoli  CM, Brass  LM,  et al.  The Stroke Prognosis Instrument II (SPI-II): a clinical prediction instrument for patients with transient ischemia and nondisabling ischemic stroke.  Stroke. 2000;31(2):456-462. doi:10.1161/01.STR.31.2.456PubMedGoogle ScholarCrossref
29.
Perry  JJ, Sharma  M, Sivilotti  ML,  et al.  A prospective cohort study of patients with transient ischemic attack to identify high-risk clinical characteristics.  Stroke. 2014;45(1):92-100. doi:10.1161/STROKEAHA.113.003085PubMedGoogle ScholarCrossref
30.
Arling  G, Reeves  M, Ross  J,  et al.  Estimating and reporting on the quality of inpatient stroke care by Veterans Health Administration Medical Centers.  Circ Cardiovasc Qual Outcomes. 2012;5(1):44-51. doi:10.1161/CIRCOUTCOMES.111.961474PubMedGoogle ScholarCrossref
31.
Haukoos  JS, Lewis  RJ.  The propensity score.  JAMA. 2015;314(15):1637-1638. doi:10.1001/jama.2015.13480PubMedGoogle ScholarCrossref
32.
Tabak  YP, Johannes  RS, Silber  JH.  Using automated clinical data for risk adjustment: development and validation of six disease-specific mortality predictive models for pay-for-performance.  Med Care. 2007;45(8):789-805. doi:10.1097/MLR.0b013e31803d3b41PubMedGoogle ScholarCrossref
33.
Wasserman  JK, Perry  JJ, Sivilotti  ML,  et al.  Computed tomography identifies patients at high risk for stroke after transient ischemic attack/nondisabling stroke: prospective, multicenter cohort study.  Stroke. 2015;46(1):114-119. doi:10.1161/STROKEAHA.114.006768PubMedGoogle ScholarCrossref
34.
Wong  ES, Wang  V, Liu  CF, Hebert  PL, Maciejewski  ML.  Do Veterans Health Administration enrollees generalize to other populations?  Med Care Res Rev. 2016;73(4):493-507. doi:10.1177/1077558715617382PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 1,528
    Original Investigation
    Health Policy
    July 3, 2019

    Processes of Care Associated With Risk of Mortality and Recurrent Stroke Among Patients With Transient Ischemic Attack and Nonsevere Ischemic Stroke

    Author Affiliations
    • 1Veterans Affairs Health Services Research and Development, Precision Monitoring to Transform Care, Quality Enhancement Research Initiative, Department of Veterans Affairs, Indianapolis, Indiana
    • 2Veterans Affairs Health Services Research and Development, Center for Health Information and Communication, Richard L. Roudebush VA Medical Center, Indianapolis, Indiana
    • 3Department of Internal Medicine, Indiana University School of Medicine, Indianapolis
    • 4Regenstrief Institute, Indianapolis, Indiana
    • 5Department of Epidemiology, Michigan State University, East Lansing
    • 6Department of Neurology, VA Greater Los Angeles Healthcare System, Los Angeles, California
    • 7Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles
    • 8Department of Biostatistics, Indiana University School of Medicine, Indianapolis
    • 9Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
    • 10Department of Neurology, Yale University School of Medicine, New Haven, Connecticut
    • 11Clinical Epidemiology Research Center, VA Connecticut Healthcare System, West Haven
    • 12Department of Pharmacy Practice, Purdue University College of Pharmacy, West Lafayette, Indiana
    • 13Department of Neurology, University of Maryland School of Medicine, Baltimore
    • 14Department of Neurology, Indiana University School of Medicine, Indianapolis
    • 15Omaha Division, VA Nebraska-Western Iowa Health Care System, Omaha
    • 16Department of Surgery, University of Nebraska, Lincoln
    • 17Purdue University School of Nursing, Lafayette, Indiana
    JAMA Netw Open. 2019;2(7):e196716. doi:10.1001/jamanetworkopen.2019.6716
    Key Points español 中文 (chinese)

    Question  Which processes of care are associated with reduced risk of mortality or recurrent stroke after transient ischemic attack or nonsevere ischemic stroke?

    Findings  In this cohort study of 8076 patients with transient ischemic attack or nonsevere ischemic stroke, only 1216 (15.3%) received without-fail care, defined as receiving all guideline-concordant processes of care for which they were eligible (ie, brain imaging, carotid artery imaging, antihypertensive intensification, high- or moderate-potency statin therapy, antithrombotics, and anticoagulation for atrial fibrillation). Receiving all 6 processes was associated with lower risk of death (31.2% reduction at 1 year) but not lower risk of recurrent stroke.

    Meaning  Clinicians should ensure that patients with transient ischemic attack and nonsevere ischemic stroke receive all guideline-concordant processes of care for which they are eligible.

    Abstract

    Importance  Early evaluation and management of patients with transient ischemic attack (TIA) and nonsevere ischemic stroke improves outcomes.

    Objective  To identify processes of care associated with reduced risk of death or recurrent stroke among patients with TIA or nonsevere ischemic stroke.

    Design, Setting, and Participants  This cohort study included all patients with TIA or nonsevere ischemic stroke at Department of Veterans Affairs emergency department or inpatient settings from October 2010 to September 2011. Multivariable logistic regression was used to model associations of processes of care and without-fail care, defined as receiving all guideline-concordant processes of care for which patients are eligible, with risk of death and recurrent stroke. Data were analyzed from March 2018 to April 2019.

    Main Outcomes and Measures  Risk of all-cause mortality and recurrent ischemic stroke at 90 days and 1 year was calculated. Overall, 28 processes of care were examined. Without-fail care was assessed for 6 processes: brain imaging, carotid artery imaging, hypertension medication intensification, high- or moderate-potency statin therapy, antithrombotics, and anticoagulation for atrial fibrillation.

    Results  Among 8076 patients, the mean (SD) age was 67.8 (11.6) years, 7752 patients (96.0%) were men, 5929 (73.4%) were white, 474 (6.1%) had a recurrent ischemic stroke within 90 days, 793 (10.7%) had a recurrent ischemic stroke within 1 year, 320 (4.0%) died within 90 days, and 814 (10.1%) died within 1 year. Overall, 9 processes were independently associated with lower odds of both 90-day and 1-year mortality after adjustment for multiple comparisons: carotid artery imaging (90-day adjusted odds ratio [aOR], 0.49; 95% CI, 0.38-0.63; 1-year aOR, 0.61; 95% CI, 0.52-0.72), antihypertensive medication class (90-day aOR, 0.58; 95% CI, 0.45-0.74; 1-year aOR, 0.70; 95% CI, 0.60-0.83), lipid measurement (90-day aOR, 0.68; 95% CI, 0.51-0.90; 1-year aOR, 0.64; 95% CI, 0.53-0.78), lipid management (90-day aOR, 0.46; 95% CI, 0.33-0.65; 1-year aOR, 0.67; 95% CI, 0.53-0.85), discharged receiving statin medication (90-day aOR, 0.51; 95% CI, 0.36-0.73; 1-year aOR, 0.70; 95% CI, 0.55-0.88), cholesterol-lowering medication intensification (90-day aOR, 0.47; 95% CI, 0.26-0.83; 1-year aOR, 0.56; 95% CI, 0.41-0.77), antithrombotics by day 2 (90-day aOR, 0.56; 95% CI, 0.40-0.79; 1-year aOR, 0.69; 95% CI, 0.55-0.87) or at discharge (90-day aOR, 0.59; 95% CI, 0.41-0.86; 1-year aOR, 0.69; 95% CI, 0.54-0.88), and neurology consultation (90-day aOR, 0.67; 95% CI, 0.52-0.87; 1-year aOR, 0.74; 95% CI, 0.63-0.87). Anticoagulation for atrial fibrillation was associated with lower odds of 1-year mortality only (aOR, 0.59; 95% CI, 0.40-0.85). No processes were associated with reduced risk of recurrent stroke after adjustment for multiple comparisons. The rate of without-fail care was 15.3%; 1216 patients received all guideline-concordant processes of care for which they were eligible. Without-fail care was associated with a 31.2% lower odds of 1-year mortality (aOR, 0.69; 95% CI, 0.55-0.87) but was not independently associated with stroke risk.

    Conclusions and Relevance  Patients who received 6 readily available processes of care had lower adjusted mortality 1 year after TIA or nonsevere ischemic stroke. Clinicians caring for patients with TIA and nonsevere ischemic stroke should seek to ensure that patients receive all guideline-concordant processes of care for which they are eligible.

    Introduction

    Patients with transient ischemic attack (TIA) and nonsevere ischemic stroke are at high risk of recurrent vascular events.1-3 However, studies have demonstrated that timely delivery of guideline-concordant care can dramatically reduce this risk.4-7 Studies reporting risk reductions of at least 70% for recurrent events among patients with TIA or nonsevere ischemic stroke have emphasized early evaluation and management; however, these studies differed in terms of the processes of care that were provided.4-8 The American Heart Association/American Stroke Association (AHA/ASA) stroke prevention guidelines recommend a broad range of processes, including diagnostic processes (eg, brain imaging) and secondary prevention interventions (eg, hypertension management).8 Our objective was to identify the guideline-concordant processes of care that were associated with a reduction in the risk of recurrent ischemic stroke or death among patients with TIA or nonsevere stroke, adjusting for baseline patient characteristics.

    Methods

    This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Human subjects approvals were obtained from the Indiana University School of Medicine Institutional Review Board and the Richard L. Roudebush VA Medical Center Research and Development Committee. Given that data were obtained from administrative sources, we obtained a waiver of written informed consent.

    Cohort Construction

    We identified patients with TIA or ischemic stroke (minor or major) cared for in Department of Veterans Affairs (VA) emergency departments or inpatient settings in fiscal year 2011 (October 2010 to September 2011) using primary discharge codes for TIA (International Classification of Diseases, Ninth Revision [ICD-9] codes: 435.0, 435.1, 435.3, 435.8, and 435.9) or ischemic stroke (ICD-9 codes: 433.X1, 434.00, 434.X1, and 436).9,10 Electronic health record data did not include a measure of stroke severity; therefore, we used a validated approach to identify and exclude patients with major stroke (ie, length of stay >6 days, ventilator use, feeding tube use, coma, intensive care unit stay, inpatient rehabilitation stay, or thrombolysis).10 Patients transferred to non–Veterans Health Administration (VHA) facilities were excluded because we could not assess quality of care.

    Outcomes

    The 4 primary outcomes were 90-day and 1-year risk of all-cause mortality and recurrent ischemic stroke.11 We chose a 90-day end point because this timeframe has been used in TIA and nonsevere ischemic stroke studies4,5,7 and a 1 year end point because some processes (eg, hypertension control) are thought to improve outcomes during longer time horizons. Ischemic stroke outcome events included emergency department or inpatient visits with a diagnosis code used for cohort construction. Ischemic stroke events occurring the day after discharge from the index event were not considered recurrent because medical record review indicated that the second visit was nearly always a continuation of care for the index event. All patients were included in the mortality outcomes. Patients transferred to non-VHA facilities or who died during the index event were excluded from recurrent stroke outcomes.

    Processes of Care

    Overall, 28 processes of care were evaluated using electronic quality measures (eQMs) previously validated against medical record review.10,12 The processes were AHA/ASA recommended elements of care8; some were similar to The Joint Commission stroke core measure set processes13 and some were in the AHA/ASA Performance Measures Set.14 To our knowledge, neither organization has a TIA-specific measure set. Numerator, denominator, and exclusion definitions are in eTable 1 in the Supplement. The processes included diagnostic and therapeutic components of care: carotid artery imaging; carotid stenosis procedure (ie, endarterectomy or stent); antihypertensive medication intensification (ie, new medication or increased dosage of existing medication); hypertension control (ie, mean blood pressure <140/90 mm Hg in the 90 days after discharge); antihypertensive medication class15 (ie, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker for patients with chronic kidney disease; thiazide for black patients; thiazide, angiotensin-converting enzyme inhibitor, or angiotensin receptor blocker for others); lipid measurement; lipid management (ie, cholesterol-reducing medication if low-density lipoprotein cholesterol [LDL-C] levels ≥100 mg/dL [to convert millimoles per liter, multiply by 0.0259] or no measurement or if LDL-C <100 mg/dL or missing but patient is using existing lipid-lowering medication); discharged receiving statin medication; cholesterol-lowering medication intensification (ie, if LDL-C ≥100 mg/dL or no measurement among patients on submaximal lipid-lowering regimen); high- or moderate-potency statin prescription (ie, high-potency statin if patient is aged <75 years; moderate- or high-potency statin if patient is aged ≥75 years); oral hypoglycemic medication intensification; hemoglobin A1c measurement; electrocardiography; telemetry; Holter monitor use; antithrombotics at day 2; antithrombotics at discharge; anticoagulation for atrial fibrillation or flutter; international normalized ratio (INR) measurement; anticoagulation quality (ie, INR 2-3 within 30 days of discharge); brain imaging; deep vein thrombosis prophylaxis; rehabilitation needs assessment; speech-language pathology consultation; substance-use treatment referral for patients with alcohol use disorder; nicotine replacement therapy for smokers; polysomnography; and neurology consultation.

    For each process, patients were classified as ineligible, eligible and passed, or eligible and failed. The carotid stenosis measure was assessed in the 14 days after presentation; patients with an outcome event in the 14 days after presentation were excluded. For measures assessed during 30 days after discharge, patients with outcomes prior to the 30-day mark were excluded. For the 2 measures assessed during 90 days after discharge (ie, hypertension control and polysomnography), we reported outcomes at 1 year and excluded patients with events prior to 90 days.

    A variety of data sources were used to construct the eQMs.10 Veterans Health Administration Central Data Warehouse inpatient and outpatient data files in the 5 years before the event were used to identify medical history, health care utilization, and procedures (using Current Procedural Terminology, Healthcare Common Procedures Coding System, and ICD-9 procedure codes).16 Pharmacy Benefits Management data were used to identify medications. Data from the VHA Central Data Warehouse were also used for vital signs, laboratory data, allergies, orders, and consultations. Linked VA/US Centers for Medicare & Medicaid Services data were used to identify hospitalizations in non-VHA facilities. Veterans Affairs Fee Basis data were used to identify inpatient and outpatient use and medical history. The date of death was obtained from VA Vital Status files.

    Without-Fail Care

    Our goal was to identify the processes of care that were associated with improvements in vascular risk and could be implemented across health care systems. It might not be possible to identify individual processes that confer benefit because patients may receive several processes at the same time or through a shared structure of care (eg, an admission order set). Although the VA system does not use a specific TIA or stroke admission order set or care pathway, we hypothesized that a bundle of processes might be routinely ordered, and therefore, we examined the 6 processes that were found to be effective in acute TIA management studies (ie, brain imaging, carotid artery imaging, hypertension medication intensification, high- or moderate-potency statin therapy, antithrombotics, and anticoagulation for atrial fibrillation).6 The 6 without-fail processes should be routinely available because they do not require advanced structures of care. A without-fail care rate17-19 was calculated for patients who were eligible for at least 1 of the 6 processes. Patients who received all without-fail care processes for which they were eligible were classified as passing the without-fail care rate.

    Statistical Analysis

    Patient characteristics, medical history, concomitant comorbidities, the final index event diagnosis (TIA or nonsevere ischemic stroke), pre–index event inpatient and outpatient use, and index event symptoms were compared among patients with outcomes and patients without outcomes. Continuous variables were summarized using means and SDs or medians and ranges. The Wilcoxon rank sum test was used to test differences in continuous variables, and Fisher exact or χ2 tests were used to compare categorical variables between groups.

    Multivariable modeling was conducted in 2 phases for each of the 4 outcomes. First, we constructed risk-adjustment models and then used the risk-adjustment models to assess associations of processes with outcomes. First, 4 logistic regression models with random facility effects (to account for clustering of patients within facilities) were used to model the outcomes. Independent variables based on patient and facility factors (but not eQMs) were divided into 2 types: those which clinical judgment deemed a priori confounders were forced into the model and those eligible for potential inclusion in the final model based on backward variable selection (P < .25 to enter, P < .05 to stay). All tests were 2-tailed.

    Second, using the models built in the first phase, the eQMs were included individually as covariates by creating 2 indicator variables for each process: (1) passing among eligible patients and (2) not eligible. The reference category was being eligible and failing; therefore, the results can be interpreted as passing vs failing. Including patients who were not eligible allowed all cohort members to be included in each model and facilitated our identification of processes that were most closely associated with outcomes. Each eQM was evaluated in a separate risk-adjusted model that did not include other eQMs. To enhance the clarity of tables, we present odds ratios (ORs), 95% CIs, and P values for passing the process with failing as the reference (data for ineligible patients are in eTables 2-5 in the Supplement). Statistical significance was set at P < .05, and tests were 2-tailed. We adjusted for multiple comparisons using the false discovery rate within each domain of interest (each process and the without-fail rate).20

    We conducted 3 sensitivity analyses. First, we repeated the modeling with patients 65 years or older because Centers for Medicare & Medicaid Services data for this population augments outcome data available from VA sources (ie, hospitalizations for ischemic stroke at non-VHA facilities that were not paid for by the VA). Second, we examined the outcomes of quality of care for patients with TIA vs nonsevere ischemic stroke. Third, we examined 3 medication measures to evaluate whether prior medication use (eg, being prescribed a statin before the index TIA or nonsevere ischemic stroke event) accounted for the association of processes with outcomes. Patients who passed each medication measure were classified into 2 groups: passed and taking medication before the index event vs passed and taking new medication after the index event. These 2 variables were entered in the final logistic regression models (the reference group included patients who failed the measure). Data analyses were performed using SAS software version 9.2 (SAS Institute).

    Results

    Among 8076 patients, 3863 (47.8%) had TIA and 4213 (52.2%) had nonsevere ischemic stroke. A total of 474 of 7802 at risk (6.1%) had a recurrent stroke within 90 days, and 793 of 7387 at risk (10.7%) had a recurrent stroke within 1 year; 320 of 8076 (4.0%) died within 90 days, and 814 of 8076 (10.1%) died within 1 year. The mean (SD) age was 67.8 (11.6) years, 7752 (96.0%) were men, and 5929 (73.4%) were white (Table 1). The multivariable risk-adjustment models are provided in Table 2. The final risk-adjustment model included 25 patient-level characteristics (included in Table 1) and no facility-level characteristics. The C statistics for the risk-adjustment models were 0.815 for 90-day mortality, 0.789 for 1-year mortality, 0.657 for 90-day recurrent stroke, and 0.659 for 1-year recurrent stroke.

    The associations of individual processes with outcomes are shown in Table 3. Overall, 9 processes were independently associated with lower odds of 90-day mortality after adjustment for multiple comparisons: carotid artery imaging (adjusted OR [aOR], 0.49; 95% CI, 0.38-0.63), antihypertensive medication class (aOR, 0.58; 95% CI, 0.45-0.74), lipid measurement (aOR, 0.68; 95% CI, 0.51-0.90), lipid management (aOR, 0.46; 95% CI, 0.33-0.65), discharged receiving statin medication (aOR, 0.51; 95% CI, 0.36-0.73), cholesterol-lowering medication intensification (aOR, 0.47; 95% CI, 0.26-0.83), antithrombotics by day 2 (aOR, 0.56; 95% CI, 0.40-0.79), antithrombotics at discharge (aOR, 0.59; 95% CI, 0.41-0.86), and neurology consultation (aOR, 0.67; 95% CI, 0.52-0.87). For 1-year mortality, 10 processes were independently associated with lower risk after adjustment for multiple comparisons: carotid artery imaging (aOR, 0.61; 95% CI, 0.52-0.72), antihypertensive medication class (aOR, 0.70; 95% CI, 0.60-0.83), lipid measurement (aOR, 0.64; 95% CI, 0.53-0.78), lipid management (aOR, 0.67; 95% CI, 0.53-0.85), discharged receiving statin medication (aOR, 0.70; 95% CI, 0.55-0.88), cholesterol-lowering medication intensification (aOR, 0.56; 95% CI, 0.41-0.77), antithrombotics by day 2 (aOR, 0.69; 95% CI, 0.55-0.87), antithrombotics at discharge (aOR, 0.69; 95% CI, 0.54-0.88), anticoagulation for atrial fibrillation (aOR, 0.59; 95% CI, 0.40-0.85), and neurology consultation (aOR, 0.74; 95% CI, 0.63-0.87). No process was independently associated with recurrent stroke outcomes after adjustment for multiple comparisons. The pass rates for the processes and complete model results are provided in eTables 2-5 in the Supplement; raw P values and P values adjusted for multiple comparisons are provided in eTable 8 in the Supplement.

    Without-Fail Care

    Among 8076 patients, 143 were not eligible for at least 1 of the 6 without-fail processes, leaving 7933 eligible. The without-fail rate was 15.3%, meaning that only 1216 patients (15.3%) received all the processes of care for which they were eligible among the 6 without-fail processes. Receiving without-fail care was associated with a 31.2% reduction in the odds of 1-year mortality (aOR, 0.69; 95% CI, 0.55-0.87) but was not independently associated with 90-day mortality or stroke outcomes after adjustment for multiple comparisons (Table 3) (eTable 8 in the Supplement).

    Sensitivity Analysis

    In analyses restricted to patients 65 years and older, the results were similar to the main results (eTable 6 in the Supplement). The magnitude of the association of the without-fail care rate with outcomes was similar to the main analysis, but the aORs were modestly higher when restricted to the population 65 years and older (eTable 6 in the Supplement). The magnitude of the association of the without-fail care rate with mortality was nearly identical when comparing patients with TIA with patients with nonsevere ischemic stroke; however, the aORs for the without-fail care rate and recurrent stroke risk were somewhat lower among patients with nonsevere ischemic stroke compared with patients with TIA (eTable 6 in the Supplement). For medication-based processes, the aORs were very similar for patients who were taking medication before the index event and patients who were given a new prescription after the index event (eTable 7 in the Supplement).

    Discussion

    These results support the association of guideline-concordant processes with improved outcomes for patients with TIA and nonsevere ischemic stroke. The 6 without-fail care processes (ie, brain imaging, carotid artery imaging, antihypertensive intensification, high- or moderate-potency statin therapy, antithrombotics, and anticoagulation for atrial fibrillation) can be provided routinely at diverse medical centers because they do not require specialized structures of care. Given the strength of the prospective trial evidence as well as the current findings supporting the association of these processes with improved outcomes, health care systems should prioritize providing patients with TIA or nonsevere ischemic stroke with the guideline-concordant processes of care for which they are eligible.

    The individual processes that were associated with improved outcomes included carotid stenosis management, hyperlipidemia, anticoagulation for atrial fibrillation, antithrombotics, and neurology consultation. These data support the AHA/ASA secondary prevention recommendations.8 Although carotid artery screening and intervention for symptomatic carotid stenosis are endorsed by guidelines, these processes are not currently the focus of existing quality measurement programs.8,13

    The observed recurrent event rates are similar to those from other settings.4,6,7,21 For example, 90-day recurrent TIA and stroke rates have been reported in the 8.5%4 to 9.9%7 range. Johnston et al21 reported a 90-day stroke rate of 10.5% for patients with TIA, but those data were from a cohort in 1997 to 1998. However, data from Australia indicated that only 4.2% of patients with TIA or nonsevere ischemic stroke who were treated in stroke units had recurrent stroke in 90 days;22 it may be that high-quality care provided in specialized stroke units conferred a lower stroke recurrence rate.23,24 The TIAregistry.org project (2009-2011) reported a 90-day recurrent stroke rate of 3.7%, a 1-year recurrent stroke rate of 5.1%, and a 1-year mortality rate of 1.8%; most patients (78%) received early stroke specialist care.25 We similarly observed that early neurology consultation was associated with a reduction in mortality.23 Therefore, the difference in risk of mortality between the TIAregistry.org project and the current cohort may be owing to differences in neurology consultation rates. We relied on emergency department and inpatient use to identify recurrent stroke events; therefore, the reported recurrent stroke rates underestimate actual rates for patients seeking care at non-VHA facilities. It was for this reason that we conducted sensitivity analyses focused on patients 65 years and older for whom we had additional health care utilization data; the results of this analysis supported the overall findings.

    We were surprised that not more individual processes of care were independently associated with improved outcomes. Our analyses may have been limited by relatively high pass rates on several processes. For example, although antithrombotic medications were associated with lower odds of mortality, they were not associated with recurrent stroke risk. A meta-regression of trials from the 1970s to 1990s26 indicated that approximately 14% of patients with stroke have a recurrent stroke during 32 months of follow-up, and the recurrent stroke risk can be reduced by 15% (95% CI, 6%-23%) with aspirin. It may be that our cohort of patients with TIA and nonsevere ischemic stroke, who had a 1-year recurrent stroke rate of 10.7% and a pass rate of antithrombotics at discharge of 87%, may have included too few eligible patients who did not receive antithrombotics to detect differences in recurrent stroke risk. However, among 67 892 patients in the Get With the Guidelines–Stroke cohort, aspirin use was also not statistically associated with a reduction in 1-year recurrent stroke risk.18 Therefore, it may be that the effect of antithrombotics on stroke risk is less robust now than was observed in the clinical trials, many of which were conducted before the advent of high-potency statins.

    Although several risk prediction scores have been validated for patients with stroke and TIA, we did not rely on any single risk adjustment score.1,3,22,27-29 Because the clinical features of cerebrovascular events are not routinely captured in electronic health record data, we could not use the age, blood pressure, clinical features, duration of symptoms, and diabetes (ABCD2) score, but we did include 3 of its components (age, blood pressure, diabetes). We evaluated a comprehensive set of risk-adjustment variables taken from the literature about post-TIA risk prediction. The final model included variables with robust associations with postevent outcomes (Table 2).

    Given that these data were derived from an observational cohort, we cannot exclude the possibility of confounding. We must consider that unmeasured confounders bias the assessment of the associations of patient characteristics with outcomes as well as of processes with outcomes. The observation that some clinical characteristics were associated with lower odds of death might be because of confounding. For example, patients with prior carotid endarterectomy or carotid stent had lower odds of 90-day mortality than patients without carotid interventions (Table 2). It may be that surgeons favor patients with strong performance status, and thereby, the carotid intervention variable is a surrogate for good health. Although we included a broad range of patient characteristics in the risk adjustment, we recognize that the risk of confounding cannot be eliminated in observational cohort studies.

    A related methodological issue is confounding by indication for the processes. We observed that patients who were not eligible for a given process generally had higher odds of poor outcomes than patients who were eligible for that process. For example, patients who were eligible for carotid imaging and received the process had lower odds of 90-day mortality compared with patients who were eligible but did not receive carotid imaging (aOR, 0.49; 95% CI, 0.38-0.63). However, patients who were not eligible for carotid imaging had higher odds of 90-day mortality compared with patients who were eligible but did not receive the process (eTable 2 in the Supplement). This observation is clinically expected because patients with very high disease severity (eg, hospice) were generally not eligible for processes (eTable 1 in the Supplement). Similar to findings from other cohort studies, we also observed that some processes were associated with an increased risk of adverse events.30 For example, the observation that rehabilitation needs assessment was associated with a higher odds of recurrent stroke (albeit not statistically significant) may be owing to confounding by stroke severity (ie, patients with severe deficits were more likely to have rehabilitation consultations); we did not have a measure of disease severity in the risk-adjustment model.

    We could have used 2 alternative methodological strategies: (1) the cohort could have been restricted to patients who were eligible for all processes of care, or (2) propensity analyses could have been conducted for each process. We favored including all patients in the study because some key processes (eg, anticoagulation for atrial fibrillation) are relevant to small subpopulations (ie, patients with atrial fibrillation), limiting the sample size and the generalizability. Propensity score adjustment is used to reduce bias when estimating the effect of an intervention on an outcome and is generally applied when investigating associations of single interventions with outcomes, especially when the characteristics of patients who receive an intervention differ from patients who do not receive an intervention.31 Given that we were interested in evaluating a broad range of processes, propensity score adjustment would have necessitated the development and presentation of more than 2 dozen models. Our approach involved defining eligibility for processes based on clinical guidelines. The risk-adjusted results (Table 3) describe the association of processes of care with outcomes among patients who were eligible for those processes.

    Strengths and Limitations

    The strengths of this study were its national scope, relatively large sample size, and inclusion of a comprehensive set of processes as well as a bundle of care. This study had limitations. First, as described earlier, although the modeling included a broad range of potential confounders,32 the possibility for residual confounding exists, especially because we did not have brain imaging results33 or data on stroke severity, stroke type, or stroke location. Second, without-fail processes were identified a priori; therefore, it is possible that the without-fail measure did not include processes with the strongest associations with outcomes. Future studies might consider identifying all-or-none quality measures empirically. Third, this study focused on veterans seeking care within VHA facilities, so the results may not generalize to other populations or health care settings.34

    Conclusions

    In this study, individual processes of care and a 6-item without-fail care measure were associated with clinically meaningful reductions in risk of stroke and death. Widespread implementation of these processes should be strongly considered for patients with TIA and nonsevere ischemic stroke. In addition, health care systems should consider routinely measuring key processes of care for patients with TIA in addition to the quality measurement that exists for patients with stroke.

    Back to top
    Article Information

    Accepted for Publication: May 16, 2019.

    Published: July 3, 2019. doi:10.1001/jamanetworkopen.2019.6716

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Bravata DM et al. JAMA Network Open.

    Corresponding Author: Dawn M. Bravata, MD, Richard L. Roudebush VA Medical Center, 1481 W 10th St, SR&D Mail Code 11H, Indianapolis, IN 46202 (dawn.bravata2@va.gov).

    Author Contributions: Dr Bravata and Mr Perkins had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Bravata, Myers, Sico, Williams, Johanning, Yu, Arling.

    Acquisition, analysis, or interpretation of data: All authors.

    Drafting of the manuscript: Bravata, Ofner.

    Critical revision of the manuscript for important intellectual content: Bravata, Myers, Reeves, Cheng, Baye, Miech, Damush, Sico, Zillich, Phipps, Williams, Chaturvedi, Johanning, Yu, Perkins, Zhang, Arling.

    Statistical analysis: Bravata, Cheng, Baye, Ofner, Miech, Yu, Perkins, Zhang, Arling.

    Obtained funding: Bravata, Zillich, Yu.

    Administrative, technical, or material support: Bravata, Myers, Miech, Damush, Sico, Zillich, Williams, Chaturvedi, Johanning.

    Conflict of Interest Disclosures: Dr Johanning reported being a managing member of FUTUREASSURE, LLC, and holding licensing and patents on systems to assess frailty. No other disclosures were reported.

    Funding/Support: This study was supported by the Department of Veterans Affairs (VA) Health Services Research and Development Service and Stroke Quality Enhancement Research Initiative (Service Directed Project 12-178). Dr Sico is supported by Career Development Award HX001388-01A1 from the VA Health Services Research and Development Service. Support for VA/Centers for Medicare & Medicaid Service data is provided by SDR 02-237 and 98-004 from the VA Information Resource Center.

    Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Additional Information: The quality-of-care data reported in this article must be maintained behind the VA firewall. Investigators interested in working with these data should contact the corresponding author.

    References
    1.
    Johnston  KC, Connors  AF  Jr, Wagner  DP, Knaus  WA, Wang  X, Haley  EC  Jr.  A predictive risk model for outcomes of ischemic stroke.  Stroke. 2000;31(2):448-455. doi:10.1161/01.STR.31.2.448PubMedGoogle ScholarCrossref
    2.
    Wu  CM, McLaughlin  K, Lorenzetti  DL, Hill  MD, Manns  BJ, Ghali  WA.  Early risk of stroke after transient ischemic attack: a systematic review and meta-analysis.  Arch Intern Med. 2007;167(22):2417-2422. doi:10.1001/archinte.167.22.2417PubMedGoogle ScholarCrossref
    3.
    Giles  MF, Rothwell  PM.  Risk of stroke early after transient ischaemic attack: a systematic review and meta-analysis.  Lancet Neurol. 2007;6(12):1063-1072. doi:10.1016/S1474-4422(07)70274-0PubMedGoogle ScholarCrossref
    4.
    Ranta  A, Dovey  S, Weatherall  M, O’Dea  D, Gommans  J, Tilyard  M.  Cluster randomized controlled trial of TIA electronic decision support in primary care.  Neurology. 2015;84(15):1545-1551. doi:10.1212/WNL.0000000000001472PubMedGoogle ScholarCrossref
    5.
    Lavallée  PC, Meseguer  E, Abboud  H,  et al.  A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects.  Lancet Neurol. 2007;6(11):953-960. doi:10.1016/S1474-4422(07)70248-XPubMedGoogle ScholarCrossref
    6.
    Luengo-Fernandez  R, Gray  AM, Rothwell  PM.  Effect of urgent treatment for transient ischaemic attack and minor stroke on disability and hospital costs (EXPRESS study): a prospective population-based sequential comparison.  Lancet Neurol. 2009;8(3):235-243. doi:10.1016/S1474-4422(09)70019-5PubMedGoogle ScholarCrossref
    7.
    Rothwell  PM, Giles  MF, Chandratheva  A,  et al; Early Use of Existing Preventive Strategies for Stroke (EXPRESS) Study.  Effect of urgent treatment of transient ischaemic attack and minor stroke on early recurrent stroke (EXPRESS study): a prospective population-based sequential comparison.  Lancet. 2007;370(9596):1432-1442. doi:10.1016/S0140-6736(07)61448-2PubMedGoogle ScholarCrossref
    8.
    Kernan  WN, Ovbiagele  B, Black  HR,  et al; American Heart Association Stroke Council; Council on Cardiovascular and Stroke Nursing; Council on Clinical Cardiology; Council on Peripheral Vascular Disease.  Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/American Stroke Association.  Stroke. 2014;45(7):2160-2236. doi:10.1161/STR.0000000000000024PubMedGoogle ScholarCrossref
    9.
    Bravata  DM, Myers  LJ, Arling  G,  et al.  Quality of care for veterans with transient ischemic attack and minor stroke.  JAMA Neurol. 2018;75(4):419-427. doi:10.1001/jamaneurol.2017.4648PubMedGoogle ScholarCrossref
    10.
    Bravata  DM, Myers  LJ, Cheng  E,  et al.  Development and validation of electronic quality measures to assess care for patients with transient ischemic attack and minor ischemic stroke.  Circ Cardiovasc Qual Outcomes. 2017;10(9):10. doi:10.1161/CIRCOUTCOMES.116.003157PubMedGoogle ScholarCrossref
    11.
    Johnston  SC.  Clinical practice: transient ischemic attack.  N Engl J Med. 2002;347(21):1687-1692. doi:10.1056/NEJMcp020891PubMedGoogle ScholarCrossref
    12.
    Phipps  MS, Fahner  J, Sager  D, Coffing  J, Maryfield  B, Williams  LS.  Validation of stroke Meaningful Use measures in a national electronic health record system.  J Gen Intern Med. 2016;31(suppl 1):46-52. doi:10.1007/s11606-015-3562-5PubMedGoogle ScholarCrossref
    13.
    The Joint Commission. Stroke (STK). https://manual.jointcommission.org/releases/TJC2016A/Stroke.html. Accessed May 22, 2019.
    14.
    Smith  EE, Saver  JL, Alexander  DN,  et al; AHA/ASA Stroke Performance Oversight Committee.  Clinical performance measures for adults hospitalized with acute ischemic stroke: performance measures for healthcare professionals from the American Heart Association/American Stroke Association.  Stroke. 2014;45(11):3472-3498. doi:10.1161/STR.0000000000000045PubMedGoogle ScholarCrossref
    15.
    Chobanian  AV, Bakris  GL, Black  HR,  et al; Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure; National Heart, Lung, and Blood Institute; National High Blood Pressure Education Program Coordinating Committee.  Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.  Hypertension. 2003;42(6):1206-1252.PubMedGoogle ScholarCrossref
    16.
    Borzecki  AM, Wong  AT, Hickey  EC, Ash  AS, Berlowitz  DR.  Can we use automated data to assess quality of hypertension care?  Am J Manag Care. 2004;10(7 Pt 2):473-479.PubMedGoogle Scholar
    17.
    Johnson  AM, Goldstein  LB, Bennett  P, O’Brien  EC, Rosamond  WD; Investigators of the Registry of the North Carolina Stroke Care Collaborative.  Compliance with acute stroke care quality measures in hospitals with and without primary stroke center certification: the North Carolina Stroke Care Collaborative.  J Am Heart Assoc. 2014;3(2):e000423. doi:10.1161/JAHA.113.000423PubMedGoogle ScholarCrossref
    18.
    O’Brien  EC, Zhao  X, Fonarow  GC,  et al.  Quality of care and ischemic stroke risk after hospitalization for transient ischemic attack: findings from Get With The Guidelines-Stroke.  Circ Cardiovasc Qual Outcomes. 2015;8(6)(suppl 3):S117-S124. doi:10.1161/CIRCOUTCOMES.115.002048PubMedGoogle ScholarCrossref
    19.
    Ovbiagele  B, Schwamm  LH, Smith  EE,  et al.  Patterns of care quality and prognosis among hospitalized ischemic stroke patients with chronic kidney disease.  J Am Heart Assoc. 2014;3(3):e000905. doi:10.1161/JAHA.114.000905PubMedGoogle ScholarCrossref
    20.
    Glickman  ME, Rao  SR, Schultz  MR.  False discovery rate control is a recommended alternative to Bonferroni-type adjustments in health studies.  J Clin Epidemiol. 2014;67(8):850-857. doi:10.1016/j.jclinepi.2014.03.012PubMedGoogle ScholarCrossref
    21.
    Johnston  SC, Gress  DR, Browner  WS, Sidney  S.  Short-term prognosis after emergency department diagnosis of TIA.  JAMA. 2000;284(22):2901-2906. doi:10.1001/jama.284.22.2901PubMedGoogle ScholarCrossref
    22.
    Knoflach  M, Lang  W, Seyfang  L,  et al; Austrian Stroke Unit Collaborators.  Predictive value of ABCD2 and ABCD3-I scores in TIA and minor stroke in the stroke unit setting.  Neurology. 2016;87(9):861-869. doi:10.1212/WNL.0000000000003033PubMedGoogle ScholarCrossref
    23.
    Bray  BD, Ayis  S, Campbell  J,  et al.  Associations between the organisation of stroke services, process of care, and mortality in England: prospective cohort study.  BMJ. 2013;346:f2827. doi:10.1136/bmj.f2827PubMedGoogle ScholarCrossref
    24.
    Hillmann  S, Wiedmann  S, Rücker  V,  et al; German Stroke Register Study Group (ADSR).  Stroke unit care in Germany: the German Stroke Registers Study Group (ADSR).  BMC Neurol. 2017;17(1):49. doi:10.1186/s12883-017-0819-0PubMedGoogle ScholarCrossref
    25.
    Amarenco  P, Lavallée  PC, Labreuche  J,  et al; TIAregistry.org Investigators.  One-year risk of stroke after transient ischemic attack or minor stroke.  N Engl J Med. 2016;374(16):1533-1542. doi:10.1056/NEJMoa1412981PubMedGoogle ScholarCrossref
    26.
    Johnson  ES, Lanes  SF, Wentworth  CE  III, Satterfield  MH, Abebe  BL, Dicker  LW.  A metaregression analysis of the dose-response effect of aspirin on stroke.  Arch Intern Med. 1999;159(11):1248-1253.PubMedGoogle ScholarCrossref
    27.
    Sheehan  OC, Merwick  A, Kelly  LA,  et al.  Diagnostic usefulness of the ABCD2 score to distinguish transient ischemic attack and minor ischemic stroke from noncerebrovascular events: the North Dublin TIA Study.  Stroke. 2009;40(11):3449-3454. doi:10.1161/STROKEAHA.109.557074PubMedGoogle ScholarCrossref
    28.
    Kernan  WN, Viscoli  CM, Brass  LM,  et al.  The Stroke Prognosis Instrument II (SPI-II): a clinical prediction instrument for patients with transient ischemia and nondisabling ischemic stroke.  Stroke. 2000;31(2):456-462. doi:10.1161/01.STR.31.2.456PubMedGoogle ScholarCrossref
    29.
    Perry  JJ, Sharma  M, Sivilotti  ML,  et al.  A prospective cohort study of patients with transient ischemic attack to identify high-risk clinical characteristics.  Stroke. 2014;45(1):92-100. doi:10.1161/STROKEAHA.113.003085PubMedGoogle ScholarCrossref
    30.
    Arling  G, Reeves  M, Ross  J,  et al.  Estimating and reporting on the quality of inpatient stroke care by Veterans Health Administration Medical Centers.  Circ Cardiovasc Qual Outcomes. 2012;5(1):44-51. doi:10.1161/CIRCOUTCOMES.111.961474PubMedGoogle ScholarCrossref
    31.
    Haukoos  JS, Lewis  RJ.  The propensity score.  JAMA. 2015;314(15):1637-1638. doi:10.1001/jama.2015.13480PubMedGoogle ScholarCrossref
    32.
    Tabak  YP, Johannes  RS, Silber  JH.  Using automated clinical data for risk adjustment: development and validation of six disease-specific mortality predictive models for pay-for-performance.  Med Care. 2007;45(8):789-805. doi:10.1097/MLR.0b013e31803d3b41PubMedGoogle ScholarCrossref
    33.
    Wasserman  JK, Perry  JJ, Sivilotti  ML,  et al.  Computed tomography identifies patients at high risk for stroke after transient ischemic attack/nondisabling stroke: prospective, multicenter cohort study.  Stroke. 2015;46(1):114-119. doi:10.1161/STROKEAHA.114.006768PubMedGoogle ScholarCrossref
    34.
    Wong  ES, Wang  V, Liu  CF, Hebert  PL, Maciejewski  ML.  Do Veterans Health Administration enrollees generalize to other populations?  Med Care Res Rev. 2016;73(4):493-507. doi:10.1177/1077558715617382PubMedGoogle ScholarCrossref
    ×