Use of Angiotensin‐Converting Enzyme Inhibitors and Angiotensin Receptor Blockers for Geriatric Ischemic Stroke Patients: Are the Rates Right?
Background Our objective is to estimate the effects associated with higher rates of renin‐angiotensin system antagonists, angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers (ACEI/ARBs), in secondary prevention for geriatric (aged >65 years) patients with new ischemic strokes by chronic kidney disease (CKD) status.
Methods and Results The effects of ACEI/ARBs on survival and renal risk were estimated by CKD status using an instrumental variable (IV) estimator. Instruments were based on local area variation in ACEI/ARB use. Data abstracted from charts were used to assess the assumptions underlying the instrumental estimator. ACEI/ARBs were used after stroke by 45.9% and 45.2% of CKD and non‐CKD patients, respectively. ACEI/ARB rate differences across local areas grouped by practice styles were nearly identical for CKD and non‐CKD patients. Higher ACEI/ARB use rates for non‐CKD patients were associated with higher 2‐year survival rates, whereas higher ACEI/ARB use rates for patients with CKD were associated with lower 2‐year survival rates. While the negative survival estimates for patients with CKD were not statistically different from zero, they were statistically lower than the estimates for non‐CKD patients. Confounders abstracted from charts were not associated with the instrumental variable used.
Conclusions Higher ACEI/ARB use rates had different survival implications for older ischemic stroke patients with and without CKD. ACEI/ARBs appear underused in ischemic stroke patients without CKD as higher use rates were associated with higher 2‐year survival rates. This conclusion is not generalizable to the ischemic stroke patients with CKD, as higher ACEI/ARBS use rates were associated with lower 2‐year survival rates that were statistically lower than the estimates for non‐CKD patients.
- angiotensin receptor
- chronic kidney disease
- instrumental variables
- ischemic stroke
- renin angiotensin system
- secondary prevention
- treatment effectiveness
What Is New?
Although secondary stroke prevention guidelines recommend use of angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers (ACEI/ARB) therapy post‐stroke and kidney guidelines recommend ACEI/ARB use in CKD, our study suggests that increasing rates of ACEI/ARB prescribing in a population with both indications for ACEI/ARB use (secondary stroke prevention and CKD) would not improve clinical outcomes and, in fact, may worsen outcomes.
However, If ACEI/ARB treatment effects are heterogeneous across patients with both stroke and CKD and care is currently individualized across patients, extrapolating our estimates to ACEI/ARB use rates far outside our study ranges may not be appropriate.
What Are the Clinical Implications?
Our estimates suggest, at a minimum, that a subset of patients exists post‐stroke with CKD for whom ACEI/ARB use may worsen survival outcomes.
Further research is needed to provide more specific guidance on which cohorts of stroke patients and which cohorts of CKD patients would benefit from more versus less ACEI prescribing, as these populations are diverse and current guidelines do not address this diversity.
Until then, it appears that ACEI/ARB use in practice should be “individualized” and “judicious” as is stressed in the current guideline.
It has been suggested that renin‐angiotensin system antagonists, angiotensin‐converting enzyme inhibitors and angiotensin receptor blockers (ACEI/ARBs), are underutilized in secondary stroke prevention.1 The current guideline recommends blood pressure reduction for all stroke patients to prevent recurrent stroke and other vascular events.2 The guideline also adds that treatment choice including the use of ACEI/ARBs should be “individualized” in light of patient conditions such as renal impairment,2 and “judicious” use of ACEI/ARBs has been stressed.3 Although ACEI/ARBs are thought protective of renal function,4 higher ACEI/ARB use rates have been associated with higher prevalence of acute kidney injury and end‐stage kidney disease.5, 6, 7 ACEI/ARBs may accelerate the progression of chronic kidney disease (CKD) to end‐stage kidney disease through intrarenal hemodynamic effects.8, 9, 10, 11, 12, 13 ACEI/ARBs may also disrupt the capacity for auto‐regulation in other vital organ systems which could be harmful in high‐risk geriatric patients with cardiovascular comorbidities.8, 9, 10, 11
The evidence guiding ACEI/ARB use for geriatric patients with comorbidities is sparse. Guidelines are based largely on studies of younger patients with well‐preserved renal function,13, 14, 15, 16, 17, 18 that did not focus on long‐term renal, cardiovascular, or survival outcomes.19, 20 Observational studies assessing ACEI/ARBs effects using risk‐adjustment approaches have shown mixed results,6, 21, 22, 23, 24, 25 likely stemming from varying ability to control for confounders26, 27 and heterogeneity of ACEI/ARB effects across patients.6, 10, 17, 24, 26, 27, 28, 29
If ACEI/ARB effects are heterogeneous across patients, the relevant question is not whether all geriatric ischemic stroke patients should be treated with an ACEI/ARB or not, but rather whether existing ACEI/ARB use rates for ischemic stroke patients are “right”?30 The ACEI/ARB use rate for ischemic stroke patients may reflect the proper tradeoffs of ACEI/ARB benefits and detriments across individual patients. Estimates of ACEI/ARB effects for ischemic stroke patients on the “extensive margin”31, 32, 33 are needed to address this question. Patients on the extensive margin are those who would be next to receive an ACEI/ARB if use rates increased, or the first not to receive an ACEI/ARB if rates fell.
Our objective is to estimate the effects of ACEI/ARBs on geriatric patients with ischemic stroke on the extensive margin. Instrumental variable (IV) estimators are applied to data from Medicare insurance system in the United States using measures of local area ACEI/ARB prescribing practice styles as instruments. This approach aligns our estimates with what might be expected from a policy intervention meant to change ACEI/ARB use rates for ischemic stroke patients.33, 34, 35 Because of the postulated heterogeneity of ACEI/ARB effects, we contrasted IV estimates across patient subpopulations based on prior CKD. In addition, data from medical charts for a sample of patients with CKD in our study population were used to evaluate the assumptions underlying our IV estimator.
Data from 2009 to 2012 for all Medicare fee‐for‐service enrollees diagnosed with stroke during 2010 were obtained from Chronic Conditions Data Warehouse (CCW). For these patients all Medicare claims and enrollment files were obtained. Also used were ZIP code level socioeconomic data from the US Census Bureau and a file developed by the project team containing driving distance between any 2 ZIP code centroids throughout the United States. Chart abstraction was performed for a stratified random sample of the ischemic stroke patients with CKD (see Data S1, Figures S1 and S2, Tables S1 and S2).2, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46 The data, analytical methods, and study materials will be made available to other researchers for the purpose of reproducing the results or replicating the procedures, but because of the sensitive nature of the data collected for this study, requests to access the data set from qualified researchers trained in human subject confidentiality protocols may be sent to the corresponding author. This study design was accepted by the University of South Carolina Institutional Review Board.
Medicare fee‐for‐service enrollees with incident ischemic stroke during 2010 with sufficient observation windows to measure all study variables were included.47 For each patient we found the first inpatient stay with a primary stroke diagnosis,48 during 2010 and designated this as each patient's index stroke stay. The observation window was built around the admission date for the index stroke—referred to as the index date. Table 1 shows the impact of inclusion criteria on our study population: 35 769 patients were included, 9092 with a diagnosis of CKD (International Classification of Diseases, Ninth Edition [ICD‐9‐CM] diagnosis codes: 585.1, 585.2, 585.3 585.4 585.5, 585.9) in the period 12 months before index through the index stay, and 26 677 without a diagnosis of CKD.
If a patient filled an ACEI/ARB prescription in the 30 days after discharge or had sufficient ACEI/ARBs at home before discharge to cover the 30 days after discharge, the ACEI/ARB treatment variable was set to one, zero otherwise. We used “days supplied” on prescriptions filled before discharge to estimate the days of supply remaining at discharge. Patients were excluded who died or had an inpatient stay during the 30 days after discharge to ensure a consistent treatment observation period.47
Two‐year survival: 1 if the patient survived 2 years after index discharge, 0 otherwise;
Two‐year secondary stroke‐free survival: 1 if the patient survived 2 years after index discharge without a recurrent stroke,48 0 otherwise;
Two‐year renal events: 1 if the patient had inpatient or outpatient claims with ICD‐9‐CM codes with acute kidney injury (584.xx and 580.xx) or end‐stage renal disease (585.6) within 2 years of index discharge, 0 otherwise.
Covariates measured at baseline included patient demographics, financial and insurance variables, comorbidities, prior adverse events related to ACEI/ARB use, complications during the index stay, therapy during the index stay, lengths of stay by unit (eg, intensive care) and facility type (skilled nursing facility), medication use before index stroke, and other medications used after discharge. Definitions and data sources for the covariates are in Data S2.48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64 For a stratified random sample of patients with CKD we measured confounders which are unmeasurable using Medicare data through chart abstraction (Data S1).
We measured ACEI/ARB local area practice style measures around each patient residence ZIP code using a driving time approach refined in previous studies based on driving times (Data S3).32, 65, 66, 67, 68 For each ZIP code, an area treatment ratio (ATR) was estimated as the ratio of the number of patients in the local area who used ACEI/ARBs after stroke over the sum of the predicted probabilities of these same patients receiving ACEI/ARBs after stroke. Larger ATR values indicate stronger provider preference in the local area for prescribing an ACEI/ARB after stroke. The instrument was specified in estimation models either using continuous variables (the patient's ZIP code ATR value and ATR value squared) or grouping patients into quintiles based on their ZIP code ATR values using dummy variables.
Patients were stratified into CKD and non‐CKD subpopulations. For each subpopulation we tested the association of the measured covariates with ACEI/ARB use and for trends in each covariate across patients grouped by ATR quintiles.69 Linear 2‐stage least squares (2SLS) IV estimators were used (Data S4).29, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80 In this study 2SLS yields estimates of the absolute average effect of ACEI/ARBs for the patients whose ACEI/ARB choice was sensitive to local area practice styles71, 80 or what is known as the local average treatment effect. Our large sample size ensures that our 2SLS estimates will be distributed normally via the central limit theorem.76 All models were estimated with robust standard errors using STATA software. We tested for differences in local average treatment effect estimates between the CKD and non‐CKD patients.77 To further contrast ACEI/ARB effects between CKD and non‐CKD patients, we estimated empirical distributions of each effect using bootstrap methods by CKD status.78 We created 3000 patient samples by randomly selecting from each subpopulation with replacement and applied the IV models to each of the 3000 samples for each subpopulation. To evaluate IV estimator assumptions, we grouped the patients from our abstraction sample based on local area ACEI/ARB practice styles and tested the mean differences in laboratory values (eg, blood pressure, kidney function, electrolytes) between groups.
Tables S3 and S4 summarize the relationships between the covariates measured using Medicare data and ACEI/ARB use and local area practice styles, for CKD and non‐CKD patients, respectively. ACEI/ARBs were used by 45.9% of CKD patients and 45.2% of non‐CKD patients. ACEI/ARB users were younger, less frail, more likely a minority, lived in areas with lower socioeconomic conditions, and had fewer prior conditions thought to be adverse events of ACEI/ARBs, but had higher rates of obesity, diabetes mellitus, and hypertension. ACEI/ARB users were more likely to have used an ACEI/ARB before their stroke and other medications after discharge. Higher‐staged CKD patients were less likely to use ACEI/ARB after stroke. For CKD patients in the first quintile of the instrument, 40.0% of patients had an ACEI/ARB available within 30 days of discharge compared with 53.5% of patients in the fifth quintile. For non‐CKD patients the range across quintiles was 38.5% to 51.7%. Local areas with higher ACEI/ARB use rates had higher minority percentages and poverty rates.
Table 2 summarizes the IV estimates. Column 3 contains F‐statistics testing whether the instrument had a statistically significant impact on ACEI/ARB use.81 All F‐statistics were much greater than 10 so that our instrumental variable is considered “non‐weak”.79 Columns 4 to 6 contain the IV estimates of ACEI/ARB use on each study outcome. For non‐CKD patients higher ACEI/ARB rates are associated with higher 2‐year survival rates (column 4). ACEI/ARB effects on stroke‐event‐free survival rates were in the same range, but not statistically different from zero (column 5). Higher ACEI/ARB rates for non‐CKD patients were associated with lower 2‐year renal event rates that were also not statistically different from zero (column 6). For patients with CKD, higher ACEI/ARB rates were associated with lower 2‐year survival rates, 2‐year renal event rates but neither was statistically different from zero.
IV estimates using the entire population are clearly averages of the CKD and non‐CKD IV estimates and hide the treatment effect heterogeneity across subpopulations. To investigate heterogeneity further, we tested for treatment effect differences between CKD and non‐CKD patients (Table 3). The IV estimates of ACEI/ARB effects on 2‐year survival were statistically different between subpopulations. Figure contains the empirical distributions of the treatment effect estimates across the bootstrapped samples by outcome, estimator, and subpopulation. A clear difference can be seen in the empirical distributions of ACEI/ARB survival effects by CKD status. The two‐sample Komogorov‐Smirnov goodness of fit test82 statistic (0.458) rejects the null hypothesis that the distributions of ACEI/ARB survival effect estimates were the same across CKD status at the P<0.01 level.
Table 4 shows laboratory values found in the index inpatient stay for ischemic stroke patients with CKD from our abstraction sample. Patients are grouped by ACEI/ARB use after discharge and local area ACEI/ARB practice styles. Patients using ACEI/ARBs after stroke had higher blood pressure and international normalized ratio levels, but lower serum creatinine than non‐treated patients. No statistically significant differences in these values were found between patients grouped by local area ACEI/ARB practice styles.
Controversy exists as to whether ACEI/ARBs are over‐ or underused for secondary prevention by stroke patients.1 Guidelines highlight the benefits of blood pressure control from ACEI/ARBs2 but the possibility of renal and survival risks from ACEI/ARBs for complex patients has been noted.5, 6, 7 Previous observational studies assessing ACEI/ARB effects are susceptible to confounding bias26, 27 and have not addressed the over/underuse question directly. Instrumental variable (IV) estimators were used in this study to assess the average effects of ACEI/ARBs for ischemic stroke patients who were Medicare fee‐for‐service beneficiaries whose ACEI/ARB use was sensitive to local area practice styles. We stratified our analysis by CKD status. We assessed confounding assumptions underlying IV estimators through chart abstraction and the contrast of baseline laboratory values across patients grouped by local area ACEI/ARB practice styles.
We found comparable ACEI/ARB use patterns for ischemic stroke patients with and without CKD, suggesting providers had similar ACEI/ARB effect expectations for both subpopulations. Global ACI/ARB use rates were nearly the same for both subpopulations, along with the range in use rates across local areas. Despite these parallels, the survival effects associated with ACEI/ARB use variation across local areas differed dramatically by CKD status. Higher ACEI/ARB use rates for non‐CKD stroke patients were associated with higher 2‐year survival rates. For non‐CKD patients with ischemic stroke, ACEI/ARBs appear underused, as higher rates would have improved survival rates with no increased renal risk. Perhaps providers overestimated the adverse‐event risks of ACEI/ARBs for these patients. We can also assert from our results that higher ACEI/ARB use rates for older ischemic stroke patients have different implications for CKD and non‐CKD patients. While the negative survival effects for patients with CKD were not statistically different from zero, they were statistically lower than the estimates for non‐CKD patients. Our estimates suggest it would be inappropriate to generalize the relationships found for non‐CKD ischemic stroke patients to patients with CKD.
IV estimates are consistent if unmeasured confounders related to study outcomes are unrelated to the instrument specified. We found no relationships in blood pressure levels, international normalized ratio values, creatinine levels, and glomerular filtration rates (GFR) across ischemic stroke patients with CKD grouped by local area ACEI/ARB practice styles. These results suggest that the local area ACEI/ARB practice styles provided a natural experiment in ACEI/ARB use and the bias risk in our IV estimates is minimal.
The IV estimates in this study should be interpreted locally to avoid improper generalization. These estimates apply directly to the ischemic stroke patients whose ACEI/ARB use in 2010 would have changed had they resided in areas with different ACEI/ARB practice styles. The ranges in ACEI/ARB use rates associated with our instrument, (38.5–51.7%) for non‐CKD patients and (40.0–53.5%) for CKD patients, are those around which our results should be interpreted. Extrapolating our estimates to changes in ACEI/ARB use rates far outside these ranges is problematic if ACEI/ARB effects are heterogeneous across patients and ACEI/ARB use in practice was individualized across patients. Without additional information on how ACEI/ARBs are sorted across ischemic stroke patients in practice, policy‐makers should be cautious using our estimates, or estimates from any observational study, as the basis guidelines on the uniform use ACEI/ARBs for CKD and non‐CKD ischemic stroke patients.
Our study uses an instrumental variable (IV) estimator with a commonly used instrument to address the question of whether ACEI/ARBs were over‐ or underused in secondary prevention for ischemic stroke patients with and without CKD who were Medicare fee‐for‐service enrollees. The survival effects associated with higher ACEI/ARB use rates clearly differed between CKD and non‐CKD patients. Higher ACEI/ARBS use rates for non‐CKD patients were associated with higher 2‐year survival rates. Whereas, 2‐year survival estimates for CKD patients were negative and statistically distinct from the estimates for non‐CKD patients. It would be a mistake to generalize the estimates for non‐CKD ischemic stroke patients to CKD patients or to apply the estimates from the entire ischemic stroke population to either subpopulation. Policies to increase ACEI/ARB uses rates for non‐CKD ischemic stroke patients should be considered but these policies should be limited to only the non‐CKD ischemic stroke patients.
Sources of Funding
This work was funded by the Patient‐Centered Outcomes Research Institute (PCORI) under project number (ME‐1303‐6011).
Dr Robinson reports support from Amarin, Amgen, Astra‐Zeneca, Esai, Esperion, Merck, Pfizer, Regeneron, Sanofi, Takeda, and Eli Lilly. The remaining authors have no disclosures to report.
Data S1. Medical chart procurement, abstraction, and quality measurement.
Data S2. Covariate definitions for Medicare claims data analysis of ACEI/ARB effectiveness after ischemic stroke.
Data S3. Instrument strategy background.
Data S4. 2‐stage least squares (2SLS) instrumental variable estimator background.
The authors wish to acknowledge the terrific support of Information Collection Enterprises (ICE) in the chart abstraction activity associated with this project.
This article was handled independently by Eric E. Smith, MD, MPH as a guest editor. The editors had no role in the evaluation of the manuscript or in the decision about its acceptance.
- ↵Kernan WN, Ovbiagele B, Black HR, Bravata DM, Chimowitz MI, Ezekowitz MD, Fang MC, Fisher M, Furie KL, Heck DV, Johnston SC, Kasner SE, Kittner SJ, Mitchell PH, Rich MW, Richardson D, Schwamm LH, Wilson JA; American Heart Association Stroke Council CoC, Stroke Nursing CoCC, Council on Peripheral Vascular D . 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;45:2160–2236.
- ↵Swift PA, Andrews PA. Not yet time for a paradigm shift away from angiotensin inhibitors in chronic kidney disease, but due diligence required. Int J Clin Pract. 2010;64:1332–1333.
- ↵Onuigbo MA. Analytical review of the evidence for renoprotection by renin‐angiotensin‐aldosterone system blockade in chronic kidney disease—a call for caution. Nephron Clin Pract. 2009;113:C63–C69.
- ↵Shlipak MG, Katz R, Kestenbaum B, Siscovick D, Fried L, Newman A, Rifkin D, Sarnak MJ. Rapid decline of kidney function increases cardiovascular risk in the elderly. J Am Soc Nephrol. 2009;20:2625–2630.
- ↵Perkins RM, Bucaloiu ID, Kirchner HL, Ashouian N, Hartle JE, Yahya T. GFR decline and mortality risk among patients with chronic kidney disease. Clin J Am Soc Nephrol. 2011;6:1879–1886.
- ↵Maschio G, Alberti D, Janin G, Locatelli F, Mann JF, Motolese M, Ponticelli C, Ritz E, Zucchelli P. Effect of the angiotensin‐converting‐enzyme inhibitor benazepril on the progression of chronic renal insufficiency. The Angiotensin‐Converting‐Enzyme Inhibition in Progressive Renal Insufficiency Study Group. N Engl J Med. 1996;334:939–945.
- ↵Ruggenenti P, Fassi A, Ilieva AP, Bruno S, Iliev IP, Brusegan V, Rubis N, Gherardi G, Arnoldi F, Ganeva M, Ene‐Iordache B, Gaspari F, Perna A, Bossi A, Trevisan R, Dodesini AR, Remuzzi G; Bergamo Nephrologic Diabetes Complications Trial I . Preventing microalbuminuria in type 2 diabetes. N Engl J Med. 2004;351:1941–1951.
- ↵Brenner BM, Cooper ME, de Zeeuw D, Keane WF, Mitch WE, Parving HH, Remuzzi G, Snapinn SM, Zhang Z, Shahinfar S; RENAAL Study Investigators . Effects of losartan on renal and cardiovascular outcomes in patients with type 2 diabetes and nephropathy. N Engl J Med. 2001;345:861–869.
- ↵Novak JE, Szczech LA. Feast and famine: epidemiology and clinical trials in chronic kidney disease. J Am Soc Nephrol. 2008;19:2–4.
- ↵Ahmed A, Jorna T, Bhandari S. Should we STOP angiotensin converting enzyme inhibitors/angiotensin receptor blockers in advanced kidney disease? Nephron. 2016;133:147–158.
- ↵James PA, Oparil S, Carter BL, Cushman WC, Dennison‐Himmelfarb C, Handler J, Lackland DT, LeFevre ML, MacKenzie TD, Ogedegbe O, Smith SC Jr., Svetkey LP, Taler SJ, Townsend RR, Wright JT Jr., Narva AS, Ortiz E. 2014 evidence‐based guideline for the management of high blood pressure in adults: report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA. 2014;311:507–520.
- ↵Hsu TW, Liu JS, Hung SC, Kuo KL, Chang YK, Chen YC, Hsu CC, Tarng DC. Renoprotective effect of renin‐angiotensin‐aldosterone system blockade in patients with predialysis advanced chronic kidney disease, hypertension, and anemia. JAMA Intern Med. 2014;174:347–354.
- ↵Hsing SC, Lu KC, Sun CA, Chien WC, Chung CH, Kao SY. The association of losartan and ramipril therapy with kidney and cardiovascular outcomes in patients with chronic kidney disease: a Chinese nation‐wide cohort study in Taiwan. Medicine (Baltimore). 2015;94:e1999.
- ↵Oh YJ, Kim SM, Shin BC, Kim HL, Chung JH, Kim AJ, Ro H, Chang JH, Lee HH, Chung W, Lee C, Jung JY. The impact of renin‐angiotensin system blockade on renal outcomes and mortality in pre‐dialysis patients with advanced chronic kidney disease. PLoS One. 2017;12:e0170874.
- ↵Schmidt M, Mansfield KE, Bhaskaran K, Nitsch D, Sorensen HT, Smeeth L, Tomlinson LA. Serum creatinine elevation after renin‐angiotensin system blockade and long term cardiorenal risks: cohort study. BMJ. 2017;356:j791.
- ↵Molnar MZ, Kalantar‐Zadeh K, Lott EH, Lu JL, Malakauskas SM, Ma JZ, Quarles DL, Kovesdy CP. Angiotensin‐converting enzyme inhibitor, angiotensin receptor blocker use, and mortality in patients with chronic kidney disease. J Am Coll Cardiol. 2014;63:650–658.
- ↵Park M, Hsu CY. An ACE in the hole for patients with advanced chronic kidney disease? JAMA Intern Med. 2014;174:355–356.
- ↵Brooks JM, Chapman CG, Schroeder MC. Understanding treatment effect estimates when treatment effects are heterogeneous for more than one outcome. Appl Health Econ Health Policy. 2018;6:381–393.
- ↵Phelps CE. Health Economics. Reading, Massachusetts: Addison‐Wesley; 1997.
- ↵Cook EA, Schneider KM, Robinson J, Wilwert J, Chrischilles E, Pendergast J, Brooks J. Field methods in medical record abstraction: assessing the properties of comparative effectiveness estimates. BMC Health Serv Res. 2014;14:391.
- ↵Bang D, Piccirillo JF, Littenberg B, Johnston A. The adult comorbidity evaluation‐27 (ACE‐27) test: a new comorbidity index for patients with cancer. 36th Annual Meeting of the American Society of Clinical Oncology. 2000.
- ↵Jauch EC, Saver JL, Adams HP Jr., Bruno A, Connors JJ, Demaerschalk BM, Khatri P, McMullan PW Jr., Qureshi AI, Rosenfield K, Scott PA, Summers DR, Wang DZ, Wintermark M, Yonas H; American Heart Association Stroke Council; Council on Cardiovascular Nursing; Council on Peripheral Vascular Disease; Council on Clinical Cardiology . Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2013;44:870–947.
- ↵Bonita R, Beaglehole R. Recovery of motor function after stroke. Stroke. 1988;19:1497–1500.
- ↵National Stroke Association . Stroke scales: an update. Stroke Clin Updates. 2006. Available at: https://www.yumpu.com/en/document/view/11589747/stroke-scales-an-update-national-stroke-association. Accessed February 12, 2016.
- ↵Deutsch A, Kline T, Lines LM, Coots L, Garfinkel D. Analysis of crosscutting Medicare functional status quality metrics using the continuity and assessment record and evaluation (CARE) item set: Final report. 2012.
- ↵Chrischilles E, Schneider K, Wilwert J, Lessman G, O'Donnell B, Gryzlak B, Wright K, Wallace R. Beyond comorbidity: expanding the definition and measurement of complexity among older adults using administrative claims data. Med Care. 2014;52(suppl 3):S75–S84.
- ↵Cook EA, Schneider KM, Chrischilles EA, Brooks JM. Accounting for immeasurable time bias when using Medicare prescription drug data. Medicare Medicaid Res Rev. 2013. Available at: https://www.cms.gov/mmrr/Downloads/MMRR2013_003_04_a01.pdf. Accessed May 19, 2018.
- ↵Chronic Condition Warehouse . 27 chronic condition algorithms. 2011. Available at: https://www.ccwdata.org/web/guest/condition-categories. Accessed March 15, 2016.
- ↵Carnahan RM. Mini‐Sentinel's systematic reviews of validated methods for identifying health outcomes using administrative data: summary of findings and suggestions for future research. Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):90–99.
- ↵Carnahan R, Gagne J, Nelson J, Fireman B, Shoaibi A, Reichman M, Zhang R, Levenson M, Graham D, Tiwari R, Southworth M, Archdeacon P, Chakravarty A, Goulding M, Brown J, Fuller C, Toh D, Chrischilles E. Mini‐sentinel prospective routine observational monitoring program tools (PROMPT): rivaroxaban surveillance, version 2. 2014.
- ↵Lucas FL, DeLorenzo MA, Siewers AE, Wennberg DE. Temporal trends in the utilization of diagnostic testing and treatments for cardiovascular disease in the United States, 1993–2001. Circulation. 2006;113:374–379.
- ↵Andrade SE, Harrold LR, Tjia J, Cutrona SL, Saczynski JS, Dodd KS, Goldberg RJ, Gurwitz JH. A systematic review of validated methods for identifying cerebrovascular accident or transient ischemic attack using administrative data. Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):100–128.
- ↵Rochon PA, Anderson GM, Tu JV, Clark JP, Gurwitz JH, Szalai JP, Lau P. Use of beta‐blocker therapy in older patients after acute myocardial infarction in Ontario. CMAJ. 1999;161:1403–1408.
- ↵Murray CJL, Kulkarni SC, Michaud C, Tomijima N, Bulzacchelli MT, Iandiorio TJ, Ezzati M. Eight Americas: investigating mortality disparities across races, counties, and race‐counties in the United States (vol 3, pg 4, 2005). PLoS Med. 2006;3:1513–1524.
- ↵Angrist JD, Krueger AB. Instrumental variables and the search for identification: from supply and demand to natural experiments. J Econ Perspect. 2001;15:69–85.
- ↵Stukel TA, Fisher ES, Wennberg DE, Alter DA, Gottlieb DJ, Vermeulen MJ. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA. 2007;297:278–285.
- ↵Baum CF. An Introduction to Modern Econometrics Using Stata. College Station, TX: Stata Press; 2006.
- ↵Greene WH. Econometric Analysis. Englewood Cliffs, NJ: Prentiss‐Hall; 2008.
- ↵Chapman CG, Brooks JM. Treatment effect estimation using nonlinear two‐stage instrumental variable estimators: another cautionary note. Health Serv Res. 2016;51:2375–2394.
- ↵Wooldridge JM. Introductory Econometrics: A Modern Approach. Mason, OH: Thomson South‐Western; 2003.
- ↵Chakravart IM, Laha RG, Roy J. Handbook of Methods of Applied Statistics. Hoboken: John Wiley and Sons; 1967.