Use of Medicare data to identify coronary heart disease outcomes in the Women's Health Initiative

Geographic Variations in Incremental Costs of Heart Disease Among Medicare Beneficiaries, by Type of Service, 2012

Using 2012 data on fee-for-service Medicare claims, we documented regional and county variation in incremental standardized costs of heart disease (ie, comparing costs between beneficiaries with heart disease and beneficiaries without heart disease) by type of service (eg, inpatient, outpatient, post-acute care). Absolute incremental total costs varied by region. Although the largest absolute incremental total costs of heart disease were concentrated in southern and Appalachian counties, geographic patterns of costs varied by type of service. These data can be used to inform development of policies and payment models that address the observed geographic disparities.

Linkage of a Population-Based Cohort With Primary Data Collection to Medicare Claims: The Reasons for Geographic and Racial Differences in Stroke Study

We described the linkage of primary data with administrative claims using the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study and Medicare. REGARDS study data were linked with Medicare claims by use of Social Security numbers. We compared REGARDS participants by Medicare linkage status, having fee-for-service (FFS) coverage or not, and with a 5% sample of Medicare beneficiaries who had FFS coverage in 2005, overall, by age (45-64 and ≥65 years), and by race. Among REGARDS participants who were ≥65 years of age, 80% had data linked to Medicare on their study-visit date (64% with FFS coverage). No differences except race and sex were present between REGARDS participants without Medicare linkage and those with data linked to Medicare with and without FFS coverage. After the age-sex-race adjustment, comorbid conditions and health-care utilization were similar for those with FFS coverage in the REGARDS study and the 5% sample of Medicare beneficiaries. Among REGARDS participants aged 45-64 years, 11% had FFS coverage on their study-visit date. In this age group, differences were present between participants with and without FFS coverage and the Medicare 5% sample with FFS coverage. In conclusion, REGARDS participants aged ≥65 years with FFS coverage are representative of the study cohort and the US population aged ≥65 years with FFS coverage.

Differences in Short- and Long-Term Outcomes Among Older Patients With ST-Elevation Versus Non–ST-Elevation Myocardial Infarction With Angiographically Proven Coronary Artery Disease

Background—

Among older patients with acute myocardial infarction (MI), it remains uncertain whether there is a time-dependent difference in the risk of recurrent mortality and nonfatal cardiovascular and cerebrovascular events for those with ST-segment–elevation MI (STEMI) compared with those with non–ST-segment–elevation MI.

Methods and Results—

Older patients ≥65 years with acute MI and significant coronary artery disease identified with coronary angiography from the ACTION Registry-GWTG (Get With the Guidelines) were linked to Medicare claims data from 2007 to 2010. We examined the unadjusted cumulative incidence of each outcome studied from hospital discharge through 2 years with log-rank tests and then performed a piece-wise proportional hazards modeling with 2 time periods: discharge to 90 days and 90 days to 2 years. Among the 46 199 patients linked with Medicare data, 17 287 (37.4%) presented with STEMI. Through 2 years, the unadjusted cumulative incidence of all-cause mortality (16.0% versus 19.8%; P <0.001) and the composite outcome (21.9% versus 27.9%; P <0.001) was lower for STEMI patients. Within the first 90 days, unadjusted rates of mortality (5.5% versus 5.3%) and the composite outcome (7.9% versus 8.1%) were similar but diverged from 90 days to 2 years (mortality, 11.1% versus 15.4%; P <0.001; composite outcome, 15.2% versus 21.5%; P <0.001). After multivariable adjustment, the adjusted risks of mortality and the composite outcome through 90 days were higher for STEMI patients, whereas risks of mortality and the composite outcome were attenuated from 90 days through 2 years.

Conclusions—

Among older acute MI patients with angiographically confirmed coronary artery disease discharged alive, STEMI patients (compared with non–ST-segment–elevation MI patients) were found to have a lower frequency of unadjusted postdischarge mortality and composite cardiovascular and cerebrovascular outcomes through 2 years after hospital discharge. This analysis provides unique insight into differential short- and long-term risks of ischemic cardiovascular and cerebrovascular outcomes by MI classification among older MI patients with confirmed coronary artery disease surviving to hospital discharge.

Development and Validation of Algorithms to Identify Statin Intolerance in a US Administrative Database

OBJECTIVES

To develop and validate algorithms to define statin intolerance (SI) in an administrative database using electronic medical records (EMRs) as the reference comparison.

METHODS

One thousand adults with one or more qualifying changes in statin therapy and one or more previous diagnoses of hyperlipidemia, hypercholesterolemia, or mixed dyslipidemia were identified from the Henry Ford Health System administrative database. Data regarding statin utilization, comorbidities, and adverse effects were extracted from the administrative database and corresponding EMR. Patients were stratified by cardiovascular (CV) risk. SI was classified as absolute intolerance or titration intolerance on the basis of changes in statin utilization and/or the occurrence of adverse effects and laboratory testing for creatine kinase. Measures of concordance (Cohen's kappa [κ]) and accuracy (sensitivity, specificity, positive predictive value [PPV], and negative predictive value) were calculated for the administrative database algorithms.

RESULTS

Half of the sample population was white, 52.9% were women, mean age was 60.6 years, and 35.7% were at high CV risk. SI was identified in 11.5% and 14.0%, absolute intolerance in 2.2% and 3.1%, and titration intolerance in 9.7% and 11.8% of the patients in the EMR and the administrative database, respectively. The algorithm identifying any SI had substantial concordance (κ = 0.66) and good sensitivity (78.1%), but modest PPV (64.0%). The titration intolerance algorithm performed better (κ = 0.74; sensitivity 85.4%; PPV 70.1%) than the absolute intolerance algorithm (κ = 0.40; sensitivity 50%; PPV 35.5%) and performed best in the high CV-risk group (n = 353), with robust concordance (κ = 0.73) and good sensitivity (80.9%) and PPV (75.3%).

CONCLUSIONS

Conservative but comprehensive algorithms are available to identify SI in administrative databases for application in real-world research. These are the first validated algorithms for use in administrative databases available to decision makers.

Electronic health records to facilitate clinical research

Electronic health records (EHRs) provide opportunities to enhance patient care, embed performance measures in clinical practice, and facilitate clinical research. Concerns have been raised about the increasing recruitment challenges in trials, burdensome and obtrusive data collection, and uncertain generalizability of the results. Leveraging electronic health records to counterbalance these trends is an area of intense interest. The initial applications of electronic health records, as the primary data source is envisioned for observational studies, embedded pragmatic or post-marketing registry-based randomized studies, or comparative effectiveness studies. Advancing this approach to randomized clinical trials, electronic health records may potentially be used to assess study feasibility, to facilitate patient recruitment, and streamline data collection at baseline and follow-up. Ensuring data security and privacy, overcoming the challenges associated with linking diverse systems and maintaining infrastructure for repeat use of high quality data, are some of the challenges associated with using electronic health records in clinical research. Collaboration between academia, industry, regulatory bodies, policy makers, patients, and electronic health record vendors is critical for the greater use of electronic health records in clinical research. This manuscript identifies the key steps required to advance the role of electronic health records in cardiovascular clinical research.

Use of administrative data to increase the practicality of clinical trials: Insights from the Women's Health Initiative

BACKGROUND

To reduce research costs in the context of pragmatic trials, consideration is given to using administrative data (Medicare claims) to ascertain clinical outcomes.

METHODS

In the historical context of the Women's Health Initiative, the correspondence between selected cardiovascular events derived from Medicare claims was compared to those documented and adjudicated in this large-scale prevention trial.

RESULTS

Classification performance varies somewhat by type of outcome, but hazard ratios and confidence intervals derived from the two data sources were quite comparable.

CONCLUSION

These encouraging results provided the needed support to launch a new embedded pragmatic trial of physical activity that will rely heavily on Medicare claims to ascertain cardiovascular disease incidence in the majority of those randomized.

Kidney Function and Cardiovascular Events in Postmenopausal Women: The Impact of Race and Ethnicity in the Women's Health Initiative

BACKGROUND

Kidney disease disproportionately affects minority populations, including African Americans and Hispanics; therefore, understanding the relationship of kidney function to cardiovascular (CV) outcomes within different racial/ethnic groups is of considerable interest. We investigated the relationship between kidney function and CV events and assessed effect modification by race/ethnicity in the Women's Health Initiative.

STUDY DESIGN

Prospective cohort study.

SETTING & PARTICIPANTS

Baseline serum creatinine concentrations (assay traceable to isotope-dilution mass spectrometry standard) of 19,411 postmenopausal women aged 50 to 79 years who self-identified as either non-Hispanic white (n=8,921), African American (n=7,436), or Hispanic (n=3,054) were used to calculate estimated glomerular filtration rates (eGFRs).

PREDICTORS

Categories of eGFR (exposure); race/ethnicity (effect modifier).

OUTCOMES

The primary outcome was the composite of 3 physician-adjudicated CV events: myocardial infarction, stroke, or CV-related death.

MEASUREMENTS

We evaluated the multivariable-adjusted associations between categories of eGFR and CV events using proportional hazards regression and formally tested for effect modification by race/ethnicity.

RESULTS

During a mean follow-up of 7.6 years, 1,424 CV events (653 myocardial infarctions, 627 strokes, and 297 CV-related deaths) were observed. The association between eGFR and CV events was curvilinear; however, the association of eGFR with CV outcomes differed by race (P=0.006). In stratified analyses, we observed that the U-shaped association was present in non-Hispanic whites, whereas African American participants had a rather curvilinear relationship, with lower eGFR being associated with higher CV risk, and higher eGFR, with reduced CV risk. Analyses among Hispanic women were inconclusive owing to few Hispanic women having very low or high eGFRs and very few events occurring in these categories.

LIMITATIONS

Lack of urinary albumin measurements; residual confounding by unmeasured or imprecisely measured characteristics.

CONCLUSIONS

In postmenopausal women, the patterns of association between eGFR and CV risk differed between non-Hispanic whites and African American women.

Study of Cardiovascular Health Outcomes in the Era of Claims Data: The Cardiovascular Health Study

BACKGROUND

Increasingly, the diagnostic codes from administrative claims data are being used as clinical outcomes.

METHODS AND RESULTS

Data from the Cardiovascular Health Study (CHS) were used to compare event rates and risk factor associations between adjudicated hospitalized cardiovascular events and claims-based methods of defining events. The outcomes of myocardial infarction (MI), stroke, and heart failure were defined in 3 ways: the CHS adjudicated event (CHS[adj]), selected International Classification of Diseases, Ninth Edition diagnostic codes only in the primary position for Medicare claims data from the Center for Medicare & Medicaid Services (CMS[1st]), and the same selected diagnostic codes in any position (CMS[any]). Conventional claims-based methods of defining events had high positive predictive values but low sensitivities. For instance, the positive predictive value of International Classification of Diseases, Ninth Edition code 410.x1 for a new acute MI in the first position was 90.6%, but this code identified only 53.8% of incident MIs. The observed event rates for CMS[1st] were low. For MI, the incidence was 14.9 events per 1000 person-years for CHS[adj] MI, 8.6 for CMS[1st] MI, and 12.2 for CMS[any] MI. In general, cardiovascular disease risk factor associations were similar across the 3 methods of defining events. Indeed, traditional cardiovascular disease risk factors were also associated with all first hospitalizations not resulting from an MI.

CONCLUSIONS

The use of diagnostic codes from claims data as clinical events, especially when restricted to primary diagnoses, leads to an underestimation of event rates. Additionally, claims-based events data represent a composite end point that includes the outcome of interest and selected (misclassified) nonevent hospitalizations.

Ongoing Use of Data and Specimens From National Cancer Institute-Sponsored Cancer Prevention Clinical Trials in the Community Clinical Oncology Program

Large cancer prevention trials provide opportunities to collect a wide array of data and biospecimens at study entry and longitudinally, for a healthy, aging population without cancer. This provides an opportunity to use pre-diagnostic data and specimens to evaluate hypotheses about the initial development of cancer. We report on strides made by, and future possibilities for, the use of accessible biorepositories developed from precisely annotated samples obtained through large-scale National Cancer Institute (NCI)-sponsored cancer prevention clinical trials conducted by the NCI Cooperative Groups. These large cancer prevention studies, which have enrolled more than 80,000 volunteers, continue to contribute to our understanding of cancer development more than 10 years after they were closed.

Limited impact of drug exposure misclassification from non-benefit thiazolidinedione drug use on mortality and hospitalizations from Saskatchewan, Canada: a cohort study

PURPOSE

Our purpose was to measure the effect of non-benefit drug use on observed associations between exposure and outcome, thereby documenting an empirical example of the potential magnitude of biases introduced when exposure status is misclassified from a restrictive drug coverage policy.

METHODS

New users of antidiabetic agents were identified with a 1-year washout period between January 1, 1995, and December 31, 2005, in Saskatchewan, Canada, and were followed until December 31, 2008. Within this population-based cohort, persons were classified as users of benefit or non-benefit thiazolidinediones (TZDs) according to their first prescription record between January 1, 2006, and December 31, 2006 (non-benefit prescription records were not captured before 2006). An intention-to-treat approach was used to categorize TZD exposure over time. We evaluated the potential bias introduced by drug exposure misclassification by evaluating bootstrapped differences in hazard ratio (HR) estimates of all-cause hospitalization or death between users and nonusers of TZDs obtained from analyses that contained complete drug use (non-benefit and benefit drug use) versus benefit drug use only (non-benefit drug use was misclassified as unexposed). All analyses were replicated within the same cohort of new users of antidiabetic agents for clopidogrel and β-blocker (bisoprolol or carvedilol) users versus nonusers because these agents were also subject to exposure misclassification from non-benefit drug use during the period of the study.

FINDINGS

Among 27,333 new users of antidiabetic agents, we identified 5759 TZD users (28% non-benefit) and 21,574 nonusers of TZDs. The crude HR for hospitalization or death among TZD users versus nonusers was higher in a database that contained benefit-only prescriptions than in a database that contained all prescriptions (HR = 1.11 [95% CI, 1.05-1.18] vs HR = 0.99 [95% CI, 0.94-1.04]). However, the differences in HRs after adjustment for demographic characteristics, health care utilization, comorbidities, and medications suggested minimal bias was introduced when TZD exposure was misclassified in the benefit-only database (adjusted HR [aHR] = 1.04 [95% CI. 0.98-1.10] vs aHR = 0.99 [95% CI, 0.94-1.04]; bootstrapped aHR difference = +0.05 [95% CI, 0.02-0.08]). Minimal differences in aHRs were also observed within analyses of clopidogrel (1551 users [24% non-benefit]; bootstrapped aHR difference = +0.01 [95% CI, -0.04 to 0.06]) and β-blocker users (351 users [42% non-benefit]; bootstrapped aHR difference = +0.06 [95% CI, -0.09 to 0.20]) versus nonusers.

IMPLICATIONS

Although patient characteristics and outcomes differed between users of non-benefit and benefit drugs, misclassification of drug exposure did not meaningfully bias estimates of all-cause mortality and hospitalization after covariate adjustment in our study.

Statin use following hospitalization among Medicare beneficiaries with a secondary discharge diagnosis of acute myocardial infarction

Background

Patients with coronary heart disease are recommended to use statins following hospital discharge. Acute myocardial infarction (AMI) is a common complication of hospitalization, but the use of statins following discharge among patients who were not initially hospitalized for AMI has not been assessed adequately.

Methods and Results

Using the Medicare 5% national random sample, we determined statin use among beneficiaries who were hospitalized and who had a secondary discharge diagnosis of AMI and among beneficiaries who had a primary discharge diagnosis of AMI, coronary artery bypass grafting, or percutaneous coronary intervention in 2007–2009. Statin use was defined by a pharmacy (Medicare Part D) claim within 90 days following discharge. Of 8175 Medicare beneficiaries who did not take statins prior to hospitalization, 31.2% with AMI as a secondary discharge diagnosis, 60.5% with AMI as the primary discharge diagnosis, 67.6% with coronary artery bypass grafting, and 63.9% with a percutaneous coronary intervention initiated statins. After multivariable adjustment, the risk ratio for statin initiation comparing beneficiaries with a secondary versus primary discharge diagnosis of AMI was 0.59 (95% CI 0.54 to 0.65). Among 5468 Medicare beneficiaries taking statins prior to hospitalization, statin use following discharge was lower for those with AMI as a secondary discharge diagnosis (71.8%) compared with their counterparts with AMI, coronary artery bypass grafting, and percutaneous coronary intervention (84.1%, 83.8%, and 87.3%, respectively) as the primary discharge diagnosis.

Conclusion

Medicare beneficiaries with a secondary hospital discharge diagnosis of AMI were less likely to fill statins compared with those with other coronary heart disease events.

Impact of drug-eluting stents on the comparative effectiveness of coronary artery bypass surgery and percutaneous coronary intervention

BACKGROUND

Drug-eluting stents (DES) have largely replaced bare-metal stents (BMS) for percutaneous coronary intervention (PCI). It is uncertain, however, whether introduction of DES had a significant impact on the comparative effectiveness of PCI versus coronary artery bypass graft surgery (CABG) for death and myocardial infarction (MI).

METHODS

We identified Medicare beneficiaries aged ≥66 years who underwent multivessel CABG or multivessel PCI and matched PCI and CABG patients on propensity score. We defined the BMS era as January 1999 to April 2003 and the DES era as May 2003 to December 2006. We compared 5-year outcomes of CABG and PCI using Cox proportional hazards models, adjusting for baseline characteristics and year of procedure and tested for a statistically significant interaction (P(int)) of DES era with treatment (CABG or PCI).

RESULTS

Five-year survival improved from the BMS era to the DES era by 1.2% for PCI and by 1.1% for CABG, and the CABG:PCI hazard ratio was unchanged (0.90 vs 0.90; P(int) = .96). Five-year MI-free survival improved by 1.4% for PCI and 1.1% for CABG, with no change in the CABG:PCI hazard ratio (0.81 vs 0.82; P(int) = .63). By contrast, survival-free of MI or repeat coronary revascularization improved from the BMS era to the DES era by 5.7% for PCI and 0.9% for CABG, and the CABG:PCI hazard ratio changed significantly (0.50 vs 0.57, P(int) ≤ .0001).

CONCLUSIONS

The introduction of DES did not alter the comparative effectiveness of CABG and PCI with respect to hard cardiac outcomes.

Validation of the atherosclerotic cardiovascular disease Pooled Cohort risk equations

IMPORTANCE

The American College of Cardiology/American Heart Association (ACC/AHA) Pooled Cohort risk equations were developed to estimate atherosclerotic cardiovascular disease (CVD) risk and guide statin initiation.

OBJECTIVE

To assess calibration and discrimination of the Pooled Cohort risk equations in a contemporary US population.

DESIGN, SETTING, AND PARTICIPANTS

Adults aged 45 to 79 years enrolled in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study between January 2003 and October 2007 and followed up through December 2010. We studied participants for whom atherosclerotic CVD risk may trigger a discussion of statin initiation (those without clinical atherosclerotic CVD or diabetes, low-density lipoprotein cholesterol level between 70 and 189 mg/dL, and not taking statins; n = 10,997).

MAIN OUTCOMES AND MEASURES

Predicted risk and observed adjudicated atherosclerotic CVD incidence (nonfatal myocardial infarction, coronary heart disease [CHD] death, nonfatal or fatal stroke) at 5 years because REGARDS participants have not been followed up for 10 years. Additional analyses, limited to Medicare beneficiaries (n = 3333), added atherosclerotic CVD events identified in Medicare claims data.

RESULTS

There were 338 adjudicated events (192 CHD events, 146 strokes). The observed and predicted 5-year atherosclerotic CVD incidence per 1000 person-years for participants with a 10-year predicted atherosclerotic CVD risk of less than 5% was 1.9 (95% CI, 1.3-2.7) and 1.9, respectively, risk of 5% to less than 7.5% was 4.8 (95% CI, 3.4-6.7) and 4.8, risk of 7.5% to less than 10% was 6.1 (95% CI, 4.4-8.6) and 6.9, and risk of 10% or greater was 12.0 (95% CI, 10.6-13.6) and 15.1 (Hosmer-Lemeshow χ2 = 19.9, P = .01). The C index was 0.72 (95% CI, 0.70-0.75). There were 234 atherosclerotic CVD events (120 CHD events, 114 strokes) among Medicare-linked participants and the observed and predicted 5-year atherosclerotic CVD incidence per 1000 person-years for participants with a predicted risk of less than 7.5% was 5.3 (95% CI, 2.8-10.1) and 4.0, respectively, risk of 7.5% to less than 10% was 7.9 (95% CI, 4.6-13.5) and 6.4, and risk of 10% or greater was 17.4 (95% CI, 15.3-19.8) and 16.4 (Hosmer-Lemeshow χ2 = 5.4, P = .71). The C index was 0.67 (95% CI, 0.64-0.71).

CONCLUSIONS AND RELEVANCE

In this cohort of US adults for whom statin initiation is considered based on the ACC/AHA Pooled Cohort risk equations, observed and predicted 5-year atherosclerotic CVD risks were similar, indicating that these risk equations were well calibrated in the population for which they were designed to be used, and demonstrated moderate to good discrimination.