[Skip to Content]
[Skip to Content Landing]
Figure.
Time to First Stroke or Myocardial Infarction (MI) by Treatment Group and Baseline Risk Strata
Time to First Stroke or Myocardial Infarction (MI) by Treatment Group and Baseline Risk Strata

Time to first stroke or MI by treatment group and baseline risk strata as assigned using risk prediction model A. The thin lines (low-risk strata) indicate baseline predicted risk for stroke or MI below the median; the thick lines (high-risk strata) indicate baseline predicted risk for stroke or MI at or above the median. Risk strata were demarcated by the median value of linear predictor, which is 3.12.

Table 1.  
Baseline Features of Study Participants by Treatment Group
Baseline Features of Study Participants by Treatment Group
Table 2.  
Treatment Effects Overall and Within Model A Risk Strataa
Treatment Effects Overall and Within Model A Risk Strataa
Table 3.  
Internally Derived Model B
Internally Derived Model B
Table 4.  
Treatment Effects Within Model B Risk Strataa
Treatment Effects Within Model B Risk Strataa
1.
Kernan  WN, Ovbiagele  B, Black  HR,  et al; American Heart Association Stroke Council, Council on Cardiovascular and Stroke Nursing, Council on Clinical Cardiology, and 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.PubMedGoogle ScholarCrossref
2.
Kernan  WN, Viscoli  CM, Furie  KL,  et al; IRIS Trial Investigators.  Pioglitazone after ischemic stroke or transient ischemic attack.  N Engl J Med. 2016;374(14):1321-1331.PubMedGoogle ScholarCrossref
3.
Semenkovich  CF.  Insulin resistance and a long, strange trip.  N Engl J Med. 2016;374(14):1378-1379.PubMedGoogle ScholarCrossref
4.
Davidoff  F.  Can knowledge about heterogeneity in treatment effects help us choose wisely?  Ann Intern Med. 2017;166(2):141-142.PubMedGoogle ScholarCrossref
5.
Kent  DM, Rothwell  PM, Ioannidis  JPA, Altman  DG, Hayward  RA.  Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal.  Trials. 2010;11:85.PubMedGoogle ScholarCrossref
6.
Kovalchik  SA, Tammemagi  M, Berg  CD,  et al.  Targeting of low-dose CT screening according to the risk of lung-cancer death.  N Engl J Med. 2013;369(3):245-254.PubMedGoogle ScholarCrossref
7.
Sussman  JB, Kent  DM, Nelson  JP, Hayward  RA.  Improving diabetes prevention with benefit based tailored treatment: risk based reanalysis of Diabetes Prevention Program.  BMJ. 2015;350:h454.PubMedGoogle ScholarCrossref
8.
Yeh  RW, Secemsky  EA, Kereiakes  DJ,  et al; DAPT Study Investigators.  Development and validation of a prediction rule for benefit and harm of dual antiplatelet therapy beyond 1 year after percutaneous coronary intervention.  JAMA. 2016;315(16):1735-1749.PubMedGoogle ScholarCrossref
9.
Burke  JF, Hayward  RA, Nelson  JP, Kent  DM.  Using internally developed risk models to assess heterogeneity in treatment effects in clinical trials.  Circ Cardiovasc Qual Outcomes. 2014;7(1):163-169.PubMedGoogle ScholarCrossref
10.
Viscoli  CM, Brass  LM, Carolei  A,  et al; IRIS Trial Investigators.  Pioglitazone for secondary prevention after ischemic stroke and transient ischemic attack: rationale and design of the Insulin Resistance Intervention after Stroke trial.  Am Heart J. 2014;168(6):823-9.e6.PubMedGoogle ScholarCrossref
11.
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.PubMedGoogle ScholarCrossref
12.
Lovett  JK, Coull  AJ, Rothwell  PM.  Early risk of recurrence by subtype of ischemic stroke in population-based incidence studies.  Neurology. 2004;62(4):569-573.PubMedGoogle ScholarCrossref
13.
Johnston  SC, Gress  DR, Browner  WS, Sidney  S.  Short-term prognosis after emergency department diagnosis of TIA.  JAMA. 2000;284(22):2901-2906.PubMedGoogle ScholarCrossref
14.
Wolf  PA, D’Agostino  RB, Belanger  AJ, Kannel  WB.  Probability of stroke: a risk profile from the Framingham Study.  Stroke. 1991;22(3):312-318.PubMedGoogle ScholarCrossref
15.
van Wijk  I, Kappelle  LJ, van Gijn  J,  et al; LiLAC Study Group.  Long-term survival and vascular event risk after transient ischaemic attack or minor ischaemic stroke: a cohort study.  Lancet. 2005;365(9477):2098-2104.PubMedGoogle ScholarCrossref
16.
Kent  DM, Nelson  J, Dahabreh  IJ, Rothwell  PM, Altman  DG, Hayward  RA.  Risk and treatment effect heterogeneity: re-analysis of individual participant data from 32 large clinical trials.  Int J Epidemiol. 2016;45(6):2075-2088.PubMedGoogle Scholar
17.
Viscoli  CM, Brass  LM, Kernan  WN, Sarrel  PM, Suissa  S, Horwitz  RI.  A clinical trial of estrogen-replacement therapy after ischemic stroke.  N Engl J Med. 2001;345:1243-1249.PubMedGoogle ScholarCrossref
18.
Stahrenberg  R, Niehaus  CF, Edelmann  F,  et al.  High-sensitivity troponin assay improves prediction of cardiovascular risk in patients with cerebral ischaemia.  J Neurol Neurosurg Psychiatry. 2013;84(5):479-487.PubMedGoogle ScholarCrossref
19.
Weimar  C, Benemann  J, Michalski  D,  et al; German Stroke Study Collaboration.  Prediction of recurrent stroke and vascular death in patients with transient ischemic attack or nondisabling stroke: a prospective comparison of validated prognostic scores.  Stroke. 2010;41(3):487-493.PubMedGoogle ScholarCrossref
20.
Meng  X, Wang  Y, Zhao  X,  et al.  Validation of the Essen Stroke Risk Score and the Stroke Prognosis Instrument II in Chinese patients.  Stroke. 2011;42(12):3619-3620.PubMedGoogle ScholarCrossref
21.
Thompson  DD, Murray  GD, Dennis  M, Sudlow  CLM, Whiteley  WN.  Formal and informal prediction of recurrent stroke and myocardial infarction after stroke: a systematic review and evaluation of clinical prediction models in a new cohort.  BMC Med. 2014;12(58):58.PubMedGoogle ScholarCrossref
22.
Zipkin  DA, Umscheid  CA, Keating  NL,  et al.  Evidence-based risk communication: a systematic review.  Ann Intern Med. 2014;161(4):270-280.PubMedGoogle ScholarCrossref
23.
Bibbins-Domingo  K, Grossman  DC, Curry  SJ,  et al; US Preventive Services Task Force.  Statin use for the primary prevention of cardiovascular disease in adults: US Preventive Services Task Force Recommendation Statement.  JAMA. 2016;316(19):1997-2007.PubMedGoogle ScholarCrossref
24.
Stone  NJ, Robinson  JG, Lichtenstein  AH,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  Circulation. 2014;129(25)(suppl 2):S1-S45.PubMedGoogle ScholarCrossref
25.
January  CT, Wann  S, Alpert  JS,  et al; ACC/AHA Task Force Members.  2014 AHA/ACC/HRS Guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society.  Circulation. 2014;130:2071-2104.PubMedGoogle ScholarCrossref
26.
Bibbins-Domingo  K; US Preventive Services Task Force.  Aspirin use for the primary prevention of cardiovascular disease and colorectal cancer: US Preventive Services Task Force recommendation statement.  Ann Intern Med. 2016;164(12):836-845.PubMedGoogle ScholarCrossref
27.
Rothwell  PM, Warlow  CP; European Carotid Surgery Trialists’ Collaborative Group.  Prediction of benefit from carotid endarterectomy in individual patients: a risk-modelling study: European Carotid Surgery Trialists’ Collaborative Group.  Lancet. 1999;353(9170):2105-2110.PubMedGoogle ScholarCrossref
28.
Diener  H-C, Ringleb  PA, Savi  P.  Clopidogrel for the secondary prevention of stroke.  Expert Opin Pharmacother. 2005;6(5):755-764.PubMedGoogle ScholarCrossref
29.
Viscoli  CM, Inzucchi  SE, Young  LH,  et al; IRIS Trial Investigators.  Pioglitazone and risk for bone fracture: safety data from a randomized clinical trial.  J Clin Endocrinol Metab. 2017;102(3):914-922.PubMedGoogle Scholar
Original Investigation
November 2017

Targeting Pioglitazone Hydrochloride Therapy After Stroke or Transient Ischemic Attack According to Pretreatment Risk for Stroke or Myocardial Infarction

Author Affiliations
  • 1Department of Internal Medicine, Yale School of Medicine, New Haven, Connecticut
  • 2Department of Neurology, Yale School of Medicine, New Haven, Connecticut
  • 3Predictive Analytics and Comparative Effectiveness Center, Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, Boston, Massachusetts
  • 4National Institute of Neurological Disorders and Stroke, Bethesda, Maryland
  • 5Department of Neurological Sciences, University of Nebraska Medical School, Omaha
  • 6Department of Neurology, Alpert Medical School of Brown University, Providence, Rhode Island
  • 7Maine Medical Center, Portland
  • 8Statistical Center for HIV/AIDS Research Prevention, Fred Hutchinson Cancer Research Center, Seattle, Washington
  • 9University of Connecticut School of Medicine, Farmington
JAMA Neurol. 2017;74(11):1319-1327. doi:10.1001/jamaneurol.2017.2136
Key Points

Question  After an ischemic stroke or transient ischemic attack, do patients at higher risk for recurrent stroke or myocardial infarction derive more benefit from pioglitazone hydrochloride compared with patients at lower risk?

Findings  In this double-blind, placebo-controlled trial, the absolute risk reduction for recurrent stroke or myocardial infarction was larger for patients at higher risk of recurrent stroke or myocardial infarction compared with those at lower risk (4.9% vs 1.9%). The relative risk reduction, however, was similar, and neither difference reached statistical significance.

Meaning  Patients at higher risk for stroke or myocardial infarction may derive a greater absolute benefit from pioglitazone.

Abstract

Importance  There is growing recognition that patients may respond differently to therapy and that the average treatment effect from a clinical trial may not apply equally to all candidates for a therapy.

Objective  To determine whether, among patients with an ischemic stroke or transient ischemic attack and insulin resistance, those at higher risk for future stroke or myocardial infarction (MI) derive more benefit from the insulin-sensitizing drug pioglitazone hydrochloride compared with patients at lower risk.

Design, Setting, and Participants  A secondary analysis was conducted of the Insulin Resistance Intervention After Stroke trial, a double-blind, placebo-controlled trial of pioglitazone for secondary prevention. Patients were enrolled from 179 research sites in 7 countries from February 7, 2005, to January 15, 2013, and were followed up for a mean of 4.1 years through the study’s end on July 28, 2015. Eligible participants had a qualifying ischemic stroke or transient ischemic attack within 180 days of entry and insulin resistance without type 1 or type 2 diabetes.

Interventions  Pioglitazone or matching placebo.

Main Outcomes and Measures  A Cox proportional hazards regression model was created using baseline features to stratify patients above or below the median risk for stroke or MI within 5 years. Within each stratum, the efficacy of pioglitazone for preventing stroke or MI was calculated. Safety outcomes were death, heart failure, weight gain, and bone fracture.

Results  Among 3876 participants (1338 women and 2538 men; mean [SD] age, 63 [11] years), the 5-year risk for stroke or MI was 6.0% in the pioglitazone group among patients at lower baseline risk compared with 7.9% in the placebo group (absolute risk difference, –1.9% [95% CI, –4.4% to 0.6%]). Among patients at higher risk, the risk was 14.7% in the pioglitazone group vs 19.6% for placebo (absolute risk difference, –4.9% [95% CI, –8.6% to 1.2%]). Hazard ratios were similar for patients below or above the median risk (0.77 vs 0.75; P = .92). Pioglitazone increased weight less among patients at higher risk but increased the risk for fracture more.

Conclusions and Relevance  After an ischemic stroke or transient ischemic attack, patients at higher risk for stroke or MI derive a greater absolute benefit from pioglitazone compared with patients at lower risk. However, the risk for fracture is also higher.

Trial Registration  clinicaltrials.gov Identifier: NCT00091949

Introduction

Patients who survive an ischemic stroke or transient ischemic attack (TIA) are at high risk for recurrent cerebrovascular and cardiovascular events.1 The mainstays of therapy to prevent recurrence include antiplatelet agents, carotid revascularization, anticoagulation, blood pressure control, and lowering of lipids. In 2016, a study2 reported that the insulin sensitizer pioglitazone hydrochloride, when added to standard secondary prevention therapies, reduces the risk for stroke or myocardial infarction (MI) by an additional 24%. Although it is generally well tolerated, pioglitazone is associated with weight gain, peripheral edema, and increased risk of bone fracture. Before prescribing pioglitazone, clinicians are advised to carefully weigh its benefits and risks for individual patients.3

The challenge of estimating the benefits and risks of medical therapy is not new, and the field of clinical therapeutics has long recognized that treatments may work differently in different groups of patients, a concept known as heterogeneity of treatment effect.4,5 Investigators often discover heterogeneity by examining treatment effect for subgroups defined by single variables in completed clinical trials. However, multivariable risk stratification systems more effectively stratify patients by their pretreatment risk and may be more effective in identifying groups of patients for whom an intervention offers particular benefit.5 Recent examples include research suggesting that lung cancer screening has a greater net benefit for patients with a higher baseline risk for cancer,6 that lifestyle modification is more effective in preventing diabetes in patients at highest risk for diabetes,7 and that dual antiplatelet therapy beyond 1 year for patients with a coronary stent is more effective for patients with a higher risk for thrombosis.8 Heterogeneity may be observed on relative scales (eg, the hazard ratio [HR] may differ between subgroups) or absolute scales (eg, outcome rates may differ). As in the examples already given, however, heterogeneity is far more commonly found for absolute scales than for relative scales.

Analyses for heterogeneity of treatment effect on relative and absolute scales are usually performed in clinical trial cohorts. Pretreatment risk of disease may be stratified by applying published systems or, if these are not available, by applying a system that is internally derived from the trial population. Evidence from theory and simulations show that such internal risk stratification systems do not introduce bias in treatment effect estimates across risk strata when compared with external models.9

The purpose of this study was to estimate the relative and absolute effectiveness of pioglitazone after ischemic stroke or TIA in subgroups of patients defined by pretreatment risk for stroke or MI. This analysis was not prespecified in the Insulin Resistance Intervention After Stroke (IRIS) trial protocol.10 However, the methods were developed and approved by the authors before conducting the analysis.

Methods
Participants

The design of the IRIS trial has been previously published.10 In brief, IRIS was a randomized, double-blind, placebo-controlled clinical trial that tested pioglitazone for prevention of stroke and MI among patients with insulin resistance but without type 1 or type 2 diabetes who had an ischemic stroke or TIA within 180 days of trial entry. Major exclusion criteria were being younger than 40 years of age and a history of heart failure or bladder cancer. Between February 7, 2005, and January 15, 2013, a total of 3876 participants were enrolled from 179 clinical sites in 7 countries. Institutional review boards approved the protocol at each site prior to data collection. Written informed consent was obtained from the participants.

Follow-up Procedures

Participants were randomized to receive pioglitazone hydrochloride at an initial daily dose of 15 mg or matching placebo. Follow-up contacts to inquire about health events took place every 2 weeks for 3 months while the dose was increased to 45 mg of pioglitazone hydrochloride or matching placebo. Thereafter, participants were contacted every 4 months up to a maximum follow-up of 5 years or at the last scheduled contact before study end on July 28, 2015.2 Mean follow-up for participants was 4.1 years.

Internally Derived Risk Models

We used the IRIS internal whole-trial data (ie, pioglitazone and placebo groups combined) to identify subgroups with a distinct risk for stroke or MI.9 This approach avoids differential fitting on treatment and control groups, which can induce bias in estimation of treatment effect across risk strata.9 Predictive features included pretreatment variables known to be associated with increased risk for stroke or MI after an index stroke and that were available in the IRIS data set. These features included age, atrial fibrillation, stroke mechanism, stroke vs TIA for the index event, history of stroke before index event, motor weakness, aphasia, coronary artery disease, history of hypertension, systolic blood pressure, peripheral arterial disease, and current smoking status.11-15 Because patients with diabetes were excluded from the IRIS trial, hemoglobin A1c level was examined as a continuous variable. In addition, we included 2 features that we hypothesized might be associated with risk for stroke or MI: sex and modified Rankin Scale score.

Two risk stratification systems were created from these variables: model A (the main system) and model B (the supplemental system). For model A, risk of stroke or MI was estimated for each participant in a Cox proportional hazards regression model that included all the predictive variables listed. This estimated risk was used to assign participants to risk quantiles. We examined halves, tertiles, quartiles, and quintiles but reported selectively based on risk distribution and the number of outcomes per quantile, but not on treatment effect.

Performance metrics for model A included the C statistic, the extreme quantile risk ratio, and the median to mean risk ratio.16 After stratifying participants into quantiles of predicted risk from model A, we calculated the ratio of highest-risk to lowest-risk quantile predicted risk and the ratio of highest-risk to lowest-risk quantile observed risk. The extreme quantile risk ratios increase with greater risk heterogeneity in the patient population. The ratio of median to mean predicted risk is a measure of skewness in risk, with larger values reflecting greater divergence between the mean result and the risk for a typical patient.

Model B was developed contingent on finding heterogeneity of treatment effect on either relative or absolute scales from model A. To create model B, the predictive variables included in model A were entered into a Cox proportional hazards regression model, and a backward selection procedure was used to remove features not significantly associated with risk (P < .05). The model coefficients were used to assign scores to the selected features, and risk groups were defined by summing scores. No imputations were made for missing data. Performance metrics for model B included the C statistic and extreme quantile risk ratios. As a check on external validity, we also examined the discrimination of model B using data from the Women’s Estrogen for Stroke Trial (WEST).17 (The WEST trial lacked several variables needed to test model A.)

Externally Derived Risk Model

We identified an external risk stratification system in a literature search for articles that developed or evaluated prediction models meeting the following criteria: (1) they included adult patients with ischemic stroke or TIA, (2) they combined multiple predictor variables, (3) they included recurrent stroke as an outcome, (4) they predicted risk over at least 1 year, and (5) they maintained predictive value in an external cohort. Among the 3 eligible models identified in our search,11,18-20 the Stroke Prognosis Instrument II (SPI-II) was chosen for its marginally superior discriminative performance and its usability as a simple point score.19,21 The C statistic was again used to quantify the discrimination of SPI-II in the IRIS cohort.

Statistical Analysis

All analyses of heterogeneity of treatment effect were conducted using the “as randomized” principle (intention to treat). Within each risk stratum for models A and B and SPI-II, treatment effects on risk for stroke or MI were estimated on relative and absolute scales. Relative effects were estimated by HRs from the Cox proportional hazards regression. The absolute risk difference (ARD) was calculated as the difference in Kaplan-Meier 5-year survival probabilities between the pioglitazone and placebo groups. The number needed to be treated was derived from these same probabilities. The null hypothesis of no heterogeneity of treatment effect across risk strata was tested in Cox proportional hazards regression models that included the indicator variable for risk group, treatment, and a product term (“interaction”) between risk group and treatment. An interaction term significant at P < .05 would indicate the presence of heterogeneity of treatment effect on a relative scale (ie, in HRs).

Secondary outcomes examined by treatment group within risk strata were stroke alone, acute coronary syndrome, all-cause mortality, and selected adverse events that have been associated with pioglitazone (ie, heart failure, bone fracture, and weight gain). We used SAS software, version 9.3, for all analyses (SAS Institute Inc).

Results
Study Population

There were 3876 total participants (1939 randomly assigned to pioglitazone and 1937 randomly assigned to placebo) enrolled in this study (1338 women and 2538 men; mean [SD] age, 63 [11] years; range, 40-95 years). Stroke was the index event for 3375 patients (87.1%), with subtype classified as lacunar for 1146 patients (29.6%) and large vessel atherosclerosis for 1007 patients (26.0%) (Table 1). Residual motor weakness was present in 609 participants (15.7%), and aphasia was present in 242 (6.2%). Treatment groups were comparable for the specified risk features.

Internally Derived Risk Model A

The following 6 features were significantly associated with increased risk for stroke or MI: aphasia, coronary artery disease, hypertension history, current smoking, peripheral arterial disease, and older age (eTable 1 in the Supplement). The model containing all 15 risk features had a C statistic of 0.66. Predicted risk estimates for the cohort were slightly skewed (median, 10.8%; mean, 12.3%; median to mean ratio, 0.88). Stratification of patients into 2 groups (above and below the median) and tertiles resulted in a 2- to 3-fold increase in both observed and predicted risk from lowest to highest strata (eTable 2 in the Supplement). (Greater quantiles were not reported owing to limitations on events per strata.) We present the main results below in groups defined by median risk. Similar results by risk tertiles are included in eTable 3 in the Supplement.

Among patients with lower risk, the 5-year risk for stroke or MI was 6.0% in the pioglitazone group vs 7.9% in the placebo group (ARD, –1.9% [95% CI, –4.4% to 0.6%]; HR, 0.77 [95% CI, 0.53 to 1.11]). For patients with higher risk, the 5-year risk for stroke or MI was 14.7% in the pioglitazone group vs 19.6% in the placebo group (ARD, –4.9% [95% CI, –8.6% to –1.2%]; HR, 0.75 [95% CI, 0.60 to 0.95]) (Figure, Table 2; eTable 3 in the Supplement). The HRs for the 2 risk groups were not significantly different (P = .92).

For the secondary efficacy outcomes of stroke alone and acute coronary syndrome, similar findings were observed. The ARDs were higher for the higher-risk groups, but the HRs were similar. The ARDs were identical for all-cause mortality.

For adverse events, the risk increments for heart failure and weight gain were slightly lower in the high-risk group than in the low-risk group, but the risk increment for bone fracture was higher. The risk for fracture was 10.6% in the pioglitazone group vs 7.4% in the placebo group (ARD, 3.2% [95% CI, 0.4%-6.0%]; HR, 1.40 [95% CI, 1.01-1.94]) among patients at lower pretreatment risk for stroke or MI. The risk for fracture was 16.9% in the pioglitazone group vs 10.1% in the placebo group (ARD, 6.8% [95% CI, 3.3%-10.2%]; HR, 1.65 [95% CI, 1.25-2.17]) among patients at higher risk for stroke or MI.

To understand the finding for bone fracture, we examined the risk for fracture for the features in Table 1. As expected, age was strongly associated with both increased risk for stroke and MI and risk for fracture. In addition, 2 features associated with risk for stroke and MI, prior stroke and aphasia, displayed an interactive effect between treatment and fracture risk: the presence of both was associated with greater fracture risk in the pioglitazone group but not in the placebo group (eTable 4 in the Supplement).

Internally Derived Risk Model B

Predictive features selected for model B are shown in Table 3. The C statistic for the model was 0.65. The observed risk for stroke or MI increased 2- to 3-fold from the lowest to the highest quantile. When model B was tested using the WEST data, outcome rates were 17.1% for 181 patients with a score of 0 to 1 compared with 23.5% for 476 patients with a score of 2 or more (HR, 1.42 [95% CI, 0.96-2.12]) (eTable 5 in the Supplement). The C statistic in WEST was 0.59.

The risk-stratified findings in model B were similar to those in model A for the primary outcome (Table 4; eFigure and eTable 6 in the Supplement). That is, the ARD for stroke or MI was more than 2-fold higher among patients at high pretreatment risk compared with patients at lower pretreatment risk (–4.8% [95% CI, –8.7% to –0.9%] vs –2.0% [95% CI, –4.3% to 0.3%]), but the HRs were similar. Some differences between models A and B were observed for secondary efficacy outcomes. In particular, the ARD for stroke alone was identical for the 2 risk groups in model B (–1.7%). As in model A, the ARD for fracture was higher for patients with a high pretreatment risk for stroke or MI.

External Risk Model

When we repeated the analysis using the SPI-II instrument (eTable 7 in the Supplement), results were similar to those from the internal models (eTable 8 in the Supplement). However, the increment in fracture risk was less between the high- and low-risk groups for SPI-II compared with model A (1.7% vs 3.6%).

Discussion

For this analysis, we created a system to divide the IRIS trial population into 2 subpopulations of equal size with a high (17.2%) and low (6.9%) 5-year risk for stroke or MI. The relative risk reduction for stroke or MI with pioglitazone was similar in the group at high risk and the group at low risk (ie, HR, 0.75 vs 0.77). Because of effective risk stratification and stable relative risk reduction across risk groups, the absolute risk reduction was substantially larger (4.9%) for patients at high pretreatment risk compared with those at low pretreatment risk (1.9%). In terms of the number needed to be treated, 21 high-risk patients would need to be treated for approximately 5 years to prevent 1 stroke or MI compared with 53 low-risk patients.

Our finding of no substantial differences in relative risk reduction between the high- and low-risk groups (ie, homogeneity in relative treatment effect) is common and represents the rule rather than the exception in treatment research. In a recent reanalysis of participant data from 18 large trials with statistically significant treatment effects, statistically significant heterogeneity of relative risk reduction was observed in only 1 trial, but clinically meaningful distinctions in absolute risk reduction (ie, heterogeneity in absolute treatment effect) were common when patients were risk stratified.16 This pattern illustrates the important added meaning of data on absolute risk reduction. Patients who experience the same relative risk reduction with a therapy may experience a very different absolute benefit depending on underlying risk. For this reason, broader use of data on absolute risk reduction (in the form of natural frequency information) has been advocated to enhance communication with patients about the benefits and risks of specific therapy.22

The use of pretreatment risk to identify patients most likely to benefit from preventive therapy is a form of personalized medicine that is gaining increasing attention in patient care.4,5,8 For example, current guidelines by the US Preventive Services Task Force23 and the American College of Cardiology and American Heart Association24 for statin therapy to prevent cardiovascular disease advocate the use of this therapy only for patients whose 10-year risk for major cardiovascular disease exceeds a specific threshold as determined by the pooled cohort equation and other considerations. Guidelines for aspirin therapy and anticoagulation for atrial fibrillation take similar approaches.25,26 To our knowledge, our analysis is one of a very few27,28 to apply this concept of risk-based treatment selection to the field of stroke prevention and is the first to examine it in the context of the cardiovascular benefit of pioglitazone. Our results may be helpful to future committees that make recommendations for the use of this novel therapy.

In the IRIS trial, patients at high baseline risk experienced greater benefit from pioglitazone in terms of prevention of stroke and MI but also experienced a higher absolute risk increment for bone fracture compared with patients at low baseline risk (6.8% vs 3.2%). The reason for this increment in bone fracture absolute risk was identified in an exploratory analysis (eTable 4 in the Supplement). Three factors used to model risk for stroke and MI were also associated with quantitatively important increases in risk for fracture: older age, prior stroke, and aphasia (although the test for interaction was only significant for prior stroke). The association with age is well documented for both conditions. The finding that patients with prior stroke or possibly aphasia experienced a higher fracture risk in the pioglitazone group but a lower risk in the placebo group needs exploration in other populations to determine if this result was a chance finding in our data or a true and heretofore unidentified interaction effect.

As a summary metric of benefit compared with harm, we used the data in Table 2 and Table 4 to calculate the number of bone fractures per stroke or MI prevented. This ratio was greater than 1 in each stratum but decreased from 1.7 in the low-risk stratum to 1.4 in the high-risk stratum for model A and from 1.6 to 1.4 for the same strata in model B. This simple analysis ignores the relative importance that patients may assign to fractures compared with myocardial and cerebral events.

Strengths and Limitations

The strengths of our study include the high quality of the IRIS trial data, including pretreatment risk data and ascertainment of outcome events.2,29 The limitations of our analysis include the fact that it was not prespecified before the trial was completed. We attempted to mitigate this weakness by writing the analytic protocol before conducting the analysis. Second, the number of trial participants was not large enough to support every subgroup analysis or to avoid all possible error in the selection of risk factors. In particular, the sample size limited the power to detect statistically significant differences between HRs and risk differences. Finally, to our knowledge, the IRIS trial is the only randomized clinical trial of pioglitazone after stroke or TIA; therefore, our findings cannot yet be independently verified.

Conclusions

Our research was not designed to yield recommendations for prescribing pioglitazone. However, our findings may help clinicians talk with patients who are considering pioglitazone therapy after ischemic stroke or TIA. In particular, clinicians can use these findings to more precisely estimate the likelihood for specific absolute benefits and harms, allowing patients to make informed decisions based on the personal values they assign to prevention of vascular events vs risk for adverse events.

Back to top
Article Information

Corresponding Author: Walter N. Kernan, MD, Department of Internal Medicine, Yale School of Medicine, Two Church St S, Ste 515, New Haven, CT 06519 (walter.kernan@yale.edu).

Accepted for Publication: June 18, 2017.

Published Online: September 18, 2017. doi:10.1001/jamaneurol.2017.2136

Author Contributions: Dr Kernan had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Kernan, Viscoli, Kent, Furie, Guarino, Inzucchi.

Acquisition, analysis, or interpretation of data: Kernan, Viscoli, Dearborn, Kent, Conwit, Fayad, Gorman, Guarino, Inzucchi, Stuart, Young.

Drafting of the manuscript: Kernan, Viscoli, Stuart.

Critical revision of the manuscript for important intellectual content: Kernan, Viscoli, Dearborn, Kent, Conwit, Fayad, Furie, Gorman, Guarino, Inzucchi, Young.

Statistical analysis: Viscoli, Kent.

Obtained funding: Kernan, Kent, Conwit, Young.

Administrative, technical, or material support: Kernan, Viscoli, Furie, Guarino, Stuart.

Study supervision: Kernan, Viscoli, Inzucchi.

Conflict of Interest Disclosures: Dr Viscoli reported receiving a consulting fee from Takeda Pharmaceuticals International for analyzing prostate cancer data in the Insulin Resistance Intervention after Stroke trial. Dr Inzucchi reported serving as a consultant to or serving on research steering committees for AstraZeneca, Boehringer Ingelheim, Daichii Sankyo, Lexicson, Janssen, Merck, Poxel, Sanofi, and vTv Pharmaceuticals and serving on data monitoring committees for Novo Nordisk and Intarcia. No other disclosures were reported.

Funding/Support: This work was supported by award U01NS044876 from the National Institute of Neurological Disorders and Stroke/National Institutes of Health and by award RR-1705-0001 from the Patient-Centered Outcomes Research Institute.

Role of the Funder/Sponsor: The funding sources 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.

Group Information: The Insulin Resistance Intervention After Stroke (IRIS) Trial Investigators are listed as follows:

Australia: Christopher Bladin, MD (Monash University–Box Hill Hospital, Box Hill, Victoria); Stephen Davis, MD (Royal Melbourne Hospital, Parkville, Victoria); Tissa Wijeratne, BMed, FRACP, PhD (Western Hospital [University of Melbourne], Footscray, Victoria); Christopher Levi, MD, Mark Parsons, MD (John Hunter Hospital [University of Newcastle], New Lambton Heights, Newcastle, New South Wales); Amy Brodtmann, MD (Austin Health [National Stroke Research Institute], Heidelberg Heights, Victoria); Steven Ng, MD, John Archer, MD (The Northern Hospital, Epping, Victoria); Candice Delcourt, MD (George Institute for International Health–Royal Prince Alfred Hospital, Camperdown, Sydney, New South Wales).

Canada: Toni R. Winder, MD (Center for Neurologic Research, Lethbridge, Alberta); Leo Berger, MD, Jean-Martin Boulanger, MD (Hopital Charles LeMoyne, Greenfield Park, Quebec); Richard K. Chan, MD, J. David Spence, MD (Robarts Research Institute, London, Ontario); Andre Durocher, MD (Centre hospitalier de l’Université de Montréal–Centre de recherche, Hôpital Notre-Dame, Montreal, Quebec); Ariane Mackey, MD, Steve Verreault, MD (CHU de Québec–Hôpital de l’Enfant-Jésus, Quebec, Quebec); Jeffrey Minuk, MD (McGill University–Jewish General, Montreal, Quebec); Andrew M. Penn, MD (Vancouver Island Health Research Centre, Victoria, British Columbia); Ashfaq Shuaib, MD (University of Alberta, Edmonton, Alberta); Robert Cote, MD (McGill University–Montreal General, Montreal, Quebec); Daniel Selchen, MD, FRCP, Neville Bayer, MD (St Michael’s Hospital, University of Toronto, Toronto, Ontario); Margaret Sweet, MD, Salim Malik, MD (Intermountain Research Consultants, Thunder Bay, Ontario); Grant Stotts, MD (Ottawa Hospital–General Campus, Ottawa, Ontario).

Germany: Bernd Griewing, Prof Dr Med, Hassan Soda, Dr Med, Renate Weinhardt, Dr Med (Neurologische Klinik Bad Neustadt, Bad Neustadt); Jörg Berrouschot, Prof Dr Med, Anett Stoll, Dr Med (Klinikum Altenburger Land, Altenburg); Otto W. Witte, Dr Med, PhD, Albrecht Günther, Dr Med (Friedrich Schiller–University Jena, Jena); Ulf Bodechtel, Dr Med (University Hospital Dresden, Dresden); Ulf Schminke, Prof Dr Med (Ernst-Moritz-Arndt-University Greifswald, Greifswald); Carsten Hobohm, Dr Med (Leipzig University, Leipzig); Andreas Hetzel, Prof Dr Med, Johann Lambeck, Dr Med (Freiburg University, Freiburg); Katja E. Wartenberg, Dr Med, PhD (Martin-Luther-Universitaet Halle-Wittenberg, Halle); Hagen Huttner, Dr Med (University of Erlangen, Erlangen); Ralf Dittrich, Dr Med (University Hospital Münster, Münster); Darius G. Nabavi, Prof Dr Med (Neukolln Hospital, Berlin); Klaus Gröschel, Dr Med (University Hospital Mainz, Mainz); Gotz Thomalla, Prof Dr Med, M. Rosenkranz, Dr Med (University Medical Center Hamburg–Eppendorf, Hamburg); Sebastian Jander, Prof Dr Med (University Düsseldorf/Heinrich-Heine University, Düsseldorf); Andreas Meisel, Prof Dr Med (Charite–Universitätsmedizin Berlin, Berlin); Albert Ludolph, Prof Dr Med, Katharina Althaus, Dr Med, R. Huber, Dr Med (University of Ulm, Ulm); Matthias Lorenz, Prof Dr Med (University Hospital Frankfurt, Frankfurt am Main).

Israel: David Tanne, MD, Oleg Merzlyak, MD (Sheba Medical Center, Tel Hashomer); Natan M. Bornstein, MD (Tel Aviv Medical Center, Tel Aviv); Gregory Telman, MD (Rambam Health Corporation, Haifa); Yair Lampl, MD (Wolfson Medical Center, Holon); Jonathan Streifler, MD (Rabin Medical Center–Golda Campus, Petach Tikva); Boaz Weller, MD (Bnai-Zion Medical Center, Haifa); Gal Ifergane, MD, Y. Wirgin, MD (Ben Gurion Medical Center, Beer-Sheva).

Italy: Antonio Carolei, MD (University of Laquila, L’Aquila); Danilo Toni, MD, PhD (University of Rome La Sapienza, Rome); Paolo Stanzione, MD (University of Rome Tor Vergata, Roma); Giuseppe Micieli, MD (Istituto di ricovero e cura a carattere scientifico Fondazione Istituto Neurologico C. Mondino, Pavia); Giancarlo Agnelli, MD, Valeria Caso, MD (University of Perugia, Perugia); Carlo Gandolfo, MD (Genoa University Hospital, Genoa); Giancarlo Comi, MD (Hospital San Raffaele Srl, Milan); Domenico Consoli, MD (Jazzolino Hospital, Vibo Valentia); Maurizia Rasura, MD (University of Rome [S. Andrea Hospital], Roma); Vincenzo Di Lazzaro, MD (Sacred Heart Catholic University, Rome).

United Kingdom: Anand Dixit, MBBS, MD, MRCP, DGM (Newcastle upon Tyne, Newcastle upon Tyne, Tyne and Wear, England); Becky Jupp, MD (Royal Bournemouth & Christchurch Hospitals, Bournemouth, England); Louise Shaw, MB, ChB, FRCP (Royal United Hospital, Avon, England); Isam Salih, MD (Torbay Hospital [South Devon Healthcare NHS Foundation Trust], Devon, England); Bernard Esisi, MD (Queen Elizabeth Hospital Gateshead, Sheriff Hill, Gateshead, England); Michael Power, MD (Ulster Hospital, Dundonald, Belfast, Northern Ireland); William D. Strain, BSc, MB ChB, MD, Salim Elyas, MD (Royal Devon and Exeter, Exeter, England); Dulka Manawadu, MBBS, MD, MRCP, PhD, FRCP, Lalit Kalra, MD (Kings College London, London, England); Eoin O’Brien, MB, MRCPI, FRCP, MICGP, Elizabeth Warburton, MD (Addenbrookes Foundation NHS Trust [Cambridge], Cambridge, England); Kausik Chatterjee, MBBS, MRCP, MD (Countess of Chester Foundation Trust, Chester, England); David R. Hargroves, BSc, FRCP, MD (William Harvey Hospital, Ashford, Kent, England); Adrian Blight, MD (Royal Surrey County Hospital, Guildford, England); Barry Moynihan, MB, BCH, BAO, MD, MRCPI, Hugh S. Markus, MD (St George’s University of London, Tooting, England); Mary Joan Macleod, MB CHB, PHD, FRCP (Aberdeen Royal Infirmary, Foresterhill, Aberdeen, Scotland); David Lance Broughton, MBBS, MRCP, MD (James Cook University Hospital, Middlesbrough, Cleveland, England); Helen Rodgers, MB ChB, MRCP, FRCH (North Tyneside General Hospital, Newcastle upon Tyne, Tyne and Wear, England); Thant Hlaing, MD (University Hospital Aintree, Liverpool, England); Scott Muir, MD (Western Infirmary, Glasgow, Scotland); Mahmud Sajid, MD (Chesterfield Hospital, Chesterfield, Derbyshire); Philip M. W. Bath, MD, MB, FRCP (University of Nottingham, Nottingham, England); Christopher Price, MB ChB, MD, FRCP, MclinEd (Wansbeck General Hospital, Ashington, Northumberland, England); Lakshmanan Sekaran, MB BS, MD, FRCP (Luton and Dunstable Hospital, Luton, England); Djamil Vahidassr, MD (Northern Trust, Co Antrim, Northern Ireland); Keith W. Muir, MB ChB, MSc, MD, MRCP, CCST, FRCP (Southern General Hospital, Glasgow, Scotland); James McIlmoyle, MB, BCh, MRCP, FRCP (Blackpool Victoria Hospital, Blackpool, England); Prabal K. Datta, MD, FRCP, Richard Davey, MD (Dewsbury District Hospital, Dewsbury, West Yorkshire, England); Peter Langhorne, BSc, PhD, FRCP, David Stott, MD (Glasgow Royal Infirmary, Glasgow, Scotland); Prabal K. Datta, MD, FRCP (Pinderfields Hospital, Wakefield, England); Timothy John England, MD, K. Muhidden, MD (Royal Derby Hospital, Mickleover, Derby, England); Janice Elizabeth O’Connell, BSc, MBChB, MRCP, FRCP, Nikhil Majmudar, MD (Sunderland Royal Hospital, Sunderland, Tyne and Wear, England).

United States: Joseph Schindler, MD (Yale University, New Haven, CT); Wayne M. Clark, MD (Oregon Health & Science University, Portland, OR); Pramodkumar Sethi, MD (Guilford Neurologic Associates, Greensboro, NC); Guy Rordorf, MD, MPH (Massachusetts General Hospital/General Hospital Corp, Boston, MA); Dawn O. Kleindorfer, MD (University of Cincinnati, Cincinnati, OH); Scott L. Silliman, MD (University of Florida, Jacksonville); Mark Gorman, MD (University of Vermont, Burlington); Michael A. Kelly, MD, Lafayette Singleton, MD (Hektoen Institute for Medical Research, LLC, Chicago, IL); Brett C. Meyer, MD, Christy Jackson, MD (University of California, San Diego, San Diego); James Walker, MD, As’ad Ehtisham, MD, Hewitt C. Goodpasture, MD (Via Christi Regional Medical Center, Wichita, KS); David Wang, DO (OSF St Francis Medical Center, Peoria, IL); Pierre Fayad, MD (University of Nebraska, Omaha); Steve Cordina, MD, Dean Naritoku, MD (University of South Alabama, Mobile); David Chiu, MD (Methodist Hospital Research Institute, Houston, TX); Timothy Lukovits, MD, Richard Goddeau, DO, Robin Clark-Arbogast, APRN (Dartmouth, Lebanon, NH); Richard Leigh, MD, Robert J. Wityk (The Johns Hopkins University, Baltimore, MD); L. Creed Pettigrew, MD (University of Kentucky Research Foundation, Lexington); Ashis H. Tayal, MD, Judy Jarouse, NP (Allegheny Singer Research Institute, Pittsburgh, PA); Gary H. Friday, MD (Lankenau Institute for Medical Research, Wynnewood, PA); Souvik Sen, MD, MS, MPH, FAHA (University of South Carolina, Columbia); Anthony S. Kim, MD, S. Claiborne Johnston, MD, PhD, Jacob S. Elkins, MD (University of California, San Francisco, San Francisco); Anna M. Barrett, MD (Kessler Foundation, West Orange, NJ); Enrique C. Leira, MD (University of Iowa, Iowa City); Adam Kelly, MD, S. Burgin, MD, David A. Rempe, MD (University of Rochester, Rochester, NY); Michael R. K. Jacoby, MD, Bruce Hughes, MD (Ruan Neuroscience Center/Mercy Medical Center, Des Moines, IA); Jennifer Majersik, MD, Elaine J. Skalabrin, MD (University of Utah, Salt Lake City); Jin-Moo Lee, PhD, MD, Chung Hsu, MD (Washington University, St Louis, MO); Sophia Sundararajan, MD (Case Western Reserve University, Cleveland, OH); Andrew Slivka, MD (Ohio State University, Columbus); Alireza Minagar, MD (Louisiana State University Health Sciences Center, Shreveport); Radica Alicic, MD, Madeleine Geraghty, MD (Providence Medical Research Center, Spokane, WA); Carlos S. Kase, MD (Boston Medical Center Corp, Boston, MA); Maartan Lansberg, MD, Greg Albers, MD (Stanford University, Stanford, CA); Dennis W Dietrich, MD (Advanced Neurology Specialists, Great Falls, MT); Joseph P. Hanna, MD (Metrohealth Medical Center, Cleveland, OH); Nina T. Gentile, MD (Temple University, Philadelphia, PA); Fernando Santiago, MD (University of Puerto Rico, San Juan); Irene Katzan, MD (Cleveland Clinic Foundation, Cleveland, OH); Marilou Ching, MD, MPH, Robert N. Sawyer Jr, MD (Research Foundation State University of New York, University of Buffalo, Buffalo); Tanya Warwick, MD (University of California, San Francisco [Fresno], Fresno); Engin Yilmaz, MD (Ingalls Memorial Hospital, Harvey, IL); Laura Pedelty, MD, PhD (University of Illinois, Chicago, Chicago); Michael J. Schneck, MD (Loyola University Chicago, Maywood, IL); Bruce M. Coull, MD (University of Arizona, Tucson); Nina J. Solenski, MD, Karen Johnston, MD (University of Virginia, Charlottesville); Vivien Lee, MD, Shyam Prabhakaran, MD (Rush University, Chicago, IL); Mark D. Johnson, MD (University of Texas, Southwestern, Dallas); Isaac E. Silverman, MD (Hartford Hospital, Hartford, CT); Miran W. Salgado, MD, Robert Birkhahn, MD (New York Methodist Hospital, Brooklyn); Richard Strawsburg, MD (Associates in Neurology, PC, Valparaiso, IN); Irfan Altafullah, MD (Minneapolis Clinic of Neurology, Golden Valley, MN); Daniel Aaron Cohen, MD, Richard Zweifler, MD (Sentara Neurology Specialists, Norfolk, VA); Peterkin Lee Kwen, MD (Southtowns Neurology of WNY, PC, West Seneca, NY); Maxim D. Hammer, MD, Nirav Vora, MD (University of Pittsburgh, Pittsburgh, PA); Gretchen E. Tietjen, MD (University of Toledo, Toledo, OH); Erfan Albakri, MD (Florida Neurovascular Institute, Tampa); Bhuvaneswari (Bo) K. Dandapani, MD (Health First Physicians, Inc, Melbourne, FL); Glen Jickling, MD, Piero Verro, MD (University of California–Davis Medical Center, Sacramento); Matthew J. Roller, MD (Altru Health System, Grand Forks, ND); Richard L. Hughes, MD, Jennifer Simpson, MD (Denver Health and Hospital Authority, Denver, CO); Thomas R. Vidic, MD, FAAN (Indiana Medical Research, Elkhart); Stephanie Lash, MD, Bruce Sigsbee, MD (Penobscot Bay Neurology, Rockport, ME); Daniel Rosenbaum, MD (State University of New York Downstate, Brooklyn); Pasquale Fonzetti, MD, PhD (Burke Medical Research Institute, White Plains, NY); James D. Fleck, MD (Indiana University, Indianapolis); Adrian J. Goldszmidt, MD (Sinai Hospital of Baltimore, Baltimore, MD); Andrei V. Alexandrov, MD, James H. Halsey, MD (University of Alabama, Birmingham); Robert Hart, MD (University of Texas, San Antonio, San Antonio); Justin A. Sattin, MD (University of Wisconsin, Madison); Sandeep Kumar, MD (Beth Israel Deaconess, Boston, MA); Diane Book, MD, Michel Torbey, MD (Medical College of Wisconsin, Milwaukee, WI); James J. Poock, MD (Northeast Iowa Medical Education Foundation, Waterloo); Molly K. King, MD, Glenn D. Graham, MD, PhD (University of New Mexico, Albuquerque); Gene Yong Sung, MD, MPH (University of Southern California, Los Angeles); Thomas Mirsen, MD (Cooper University Hospital, Camden, NJ); Alexander W. Dromerick, MD (National Rehabilitation Hospital, Washington, DC); Andreas D. Runheim, MD (Salem Neurological Center, Winston-Salem, NC); Christy M. Jackson, MD (Scripps Clinic, La Jolla, CA); Eliahu Feen, MD (St Louis University, St Louis, MO); Raymond K. Reichwein, MD (Penn State–Hershey Medical Center, Hershey, PA); Michael F. Waters, MD (University of Florida, Gainesville); Colum Amory, MD, Gary L. Bernardini, MD (Albany Medical Center, Albany, NY); Rodney D. Bell, MD (Thomas Jefferson University, Philadelphia, PA); B. Franklin Diamond, MD (Abington Memorial Hospital, Abington, PA); Daniel M. Rosenbaum, MD (Albert Einstein, Bronx, NY); David Palestrant, MD (Cedars-Sinai Medical Center, Los Angeles, CA); Alan Z. Segal, MD (Cornell University, New York, NY); Kathleen Burger, DO (George Washington University, Washington, DC); Ronald L. Schwartz, MD (Hattiesburg Clinic, Hattiesburg, MS); Panayiotis Mitsias, MD (Henry Ford Health Sciences Center, Detroit, MI); Jeffrey Kramer, MD (Jeffrey Kramer, MDSC [formerly Kramer/Mercy], Chicago, IL); David Robbins, MD (Pines Neurological Associates, Pembroke Pines, FL); Brian Silver, MD, J. Donald Easton, MD, Edward Feldmann, MD (Rhode Island Hospital, Providence); Marilyn M. Rymer, MD, Joyce Dorssom, MD (St Luke’s Brain and Stroke Institute, Kansas City, MO); Latisha Ali, MD, Bruce Ovbiagele, MD (University of California, Los Angeles, Los Angeles); Howard S. Kirshner, MD (Vanderbilt University, Nashville, TN).

Disclaimer: The views, statements, and opinions presented in this article are solely the responsibility of the authors and do not necessarily represent the views of the National Institute of Neurological Disorders and Stroke, Takeda Pharmaceuticals International, or the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

References
1.
Kernan  WN, Ovbiagele  B, Black  HR,  et al; American Heart Association Stroke Council, Council on Cardiovascular and Stroke Nursing, Council on Clinical Cardiology, and 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.PubMedGoogle ScholarCrossref
2.
Kernan  WN, Viscoli  CM, Furie  KL,  et al; IRIS Trial Investigators.  Pioglitazone after ischemic stroke or transient ischemic attack.  N Engl J Med. 2016;374(14):1321-1331.PubMedGoogle ScholarCrossref
3.
Semenkovich  CF.  Insulin resistance and a long, strange trip.  N Engl J Med. 2016;374(14):1378-1379.PubMedGoogle ScholarCrossref
4.
Davidoff  F.  Can knowledge about heterogeneity in treatment effects help us choose wisely?  Ann Intern Med. 2017;166(2):141-142.PubMedGoogle ScholarCrossref
5.
Kent  DM, Rothwell  PM, Ioannidis  JPA, Altman  DG, Hayward  RA.  Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal.  Trials. 2010;11:85.PubMedGoogle ScholarCrossref
6.
Kovalchik  SA, Tammemagi  M, Berg  CD,  et al.  Targeting of low-dose CT screening according to the risk of lung-cancer death.  N Engl J Med. 2013;369(3):245-254.PubMedGoogle ScholarCrossref
7.
Sussman  JB, Kent  DM, Nelson  JP, Hayward  RA.  Improving diabetes prevention with benefit based tailored treatment: risk based reanalysis of Diabetes Prevention Program.  BMJ. 2015;350:h454.PubMedGoogle ScholarCrossref
8.
Yeh  RW, Secemsky  EA, Kereiakes  DJ,  et al; DAPT Study Investigators.  Development and validation of a prediction rule for benefit and harm of dual antiplatelet therapy beyond 1 year after percutaneous coronary intervention.  JAMA. 2016;315(16):1735-1749.PubMedGoogle ScholarCrossref
9.
Burke  JF, Hayward  RA, Nelson  JP, Kent  DM.  Using internally developed risk models to assess heterogeneity in treatment effects in clinical trials.  Circ Cardiovasc Qual Outcomes. 2014;7(1):163-169.PubMedGoogle ScholarCrossref
10.
Viscoli  CM, Brass  LM, Carolei  A,  et al; IRIS Trial Investigators.  Pioglitazone for secondary prevention after ischemic stroke and transient ischemic attack: rationale and design of the Insulin Resistance Intervention after Stroke trial.  Am Heart J. 2014;168(6):823-9.e6.PubMedGoogle ScholarCrossref
11.
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.PubMedGoogle ScholarCrossref
12.
Lovett  JK, Coull  AJ, Rothwell  PM.  Early risk of recurrence by subtype of ischemic stroke in population-based incidence studies.  Neurology. 2004;62(4):569-573.PubMedGoogle ScholarCrossref
13.
Johnston  SC, Gress  DR, Browner  WS, Sidney  S.  Short-term prognosis after emergency department diagnosis of TIA.  JAMA. 2000;284(22):2901-2906.PubMedGoogle ScholarCrossref
14.
Wolf  PA, D’Agostino  RB, Belanger  AJ, Kannel  WB.  Probability of stroke: a risk profile from the Framingham Study.  Stroke. 1991;22(3):312-318.PubMedGoogle ScholarCrossref
15.
van Wijk  I, Kappelle  LJ, van Gijn  J,  et al; LiLAC Study Group.  Long-term survival and vascular event risk after transient ischaemic attack or minor ischaemic stroke: a cohort study.  Lancet. 2005;365(9477):2098-2104.PubMedGoogle ScholarCrossref
16.
Kent  DM, Nelson  J, Dahabreh  IJ, Rothwell  PM, Altman  DG, Hayward  RA.  Risk and treatment effect heterogeneity: re-analysis of individual participant data from 32 large clinical trials.  Int J Epidemiol. 2016;45(6):2075-2088.PubMedGoogle Scholar
17.
Viscoli  CM, Brass  LM, Kernan  WN, Sarrel  PM, Suissa  S, Horwitz  RI.  A clinical trial of estrogen-replacement therapy after ischemic stroke.  N Engl J Med. 2001;345:1243-1249.PubMedGoogle ScholarCrossref
18.
Stahrenberg  R, Niehaus  CF, Edelmann  F,  et al.  High-sensitivity troponin assay improves prediction of cardiovascular risk in patients with cerebral ischaemia.  J Neurol Neurosurg Psychiatry. 2013;84(5):479-487.PubMedGoogle ScholarCrossref
19.
Weimar  C, Benemann  J, Michalski  D,  et al; German Stroke Study Collaboration.  Prediction of recurrent stroke and vascular death in patients with transient ischemic attack or nondisabling stroke: a prospective comparison of validated prognostic scores.  Stroke. 2010;41(3):487-493.PubMedGoogle ScholarCrossref
20.
Meng  X, Wang  Y, Zhao  X,  et al.  Validation of the Essen Stroke Risk Score and the Stroke Prognosis Instrument II in Chinese patients.  Stroke. 2011;42(12):3619-3620.PubMedGoogle ScholarCrossref
21.
Thompson  DD, Murray  GD, Dennis  M, Sudlow  CLM, Whiteley  WN.  Formal and informal prediction of recurrent stroke and myocardial infarction after stroke: a systematic review and evaluation of clinical prediction models in a new cohort.  BMC Med. 2014;12(58):58.PubMedGoogle ScholarCrossref
22.
Zipkin  DA, Umscheid  CA, Keating  NL,  et al.  Evidence-based risk communication: a systematic review.  Ann Intern Med. 2014;161(4):270-280.PubMedGoogle ScholarCrossref
23.
Bibbins-Domingo  K, Grossman  DC, Curry  SJ,  et al; US Preventive Services Task Force.  Statin use for the primary prevention of cardiovascular disease in adults: US Preventive Services Task Force Recommendation Statement.  JAMA. 2016;316(19):1997-2007.PubMedGoogle ScholarCrossref
24.
Stone  NJ, Robinson  JG, Lichtenstein  AH,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  Circulation. 2014;129(25)(suppl 2):S1-S45.PubMedGoogle ScholarCrossref
25.
January  CT, Wann  S, Alpert  JS,  et al; ACC/AHA Task Force Members.  2014 AHA/ACC/HRS Guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society.  Circulation. 2014;130:2071-2104.PubMedGoogle ScholarCrossref
26.
Bibbins-Domingo  K; US Preventive Services Task Force.  Aspirin use for the primary prevention of cardiovascular disease and colorectal cancer: US Preventive Services Task Force recommendation statement.  Ann Intern Med. 2016;164(12):836-845.PubMedGoogle ScholarCrossref
27.
Rothwell  PM, Warlow  CP; European Carotid Surgery Trialists’ Collaborative Group.  Prediction of benefit from carotid endarterectomy in individual patients: a risk-modelling study: European Carotid Surgery Trialists’ Collaborative Group.  Lancet. 1999;353(9170):2105-2110.PubMedGoogle ScholarCrossref
28.
Diener  H-C, Ringleb  PA, Savi  P.  Clopidogrel for the secondary prevention of stroke.  Expert Opin Pharmacother. 2005;6(5):755-764.PubMedGoogle ScholarCrossref
29.
Viscoli  CM, Inzucchi  SE, Young  LH,  et al; IRIS Trial Investigators.  Pioglitazone and risk for bone fracture: safety data from a randomized clinical trial.  J Clin Endocrinol Metab. 2017;102(3):914-922.PubMedGoogle Scholar
×