Author + information
- Received November 10, 2016
- Accepted December 14, 2016
- Published online March 13, 2017.
- Maria-Corina Serban, MD, PhDa,b,
- Lisandro D. Colantonio, MD, MSca,
- Angelika D. Manthripragada, PhDc,
- Keri L. Monda, PhDc,
- Vera A. Bittner, MDd,
- Maciej Banach, MD, PhDe,
- Ligong Chen, PhD, MSa,
- Lei Huang, MSa,
- Ricardo Dent, MDf,
- Shia T. Kent, PhDa,
- Paul Muntner, PhDa,∗ ( and )
- Robert S. Rosenson, MDg
- aDepartment of Epidemiology, University of Alabama at Birmingham, Birmingham, Alabama
- bDepartment of Functional Sciences, University of Medicine and Pharmacy Victor Babes Timisoara, Romania
- cCenter for Observational Research, Amgen, Inc., Thousand Oaks, California
- dDepartment of Medicine, Division of Cardiovascular Disease, University of Alabama at Birmingham, Birmingham, Alabama
- eDepartment of Hypertension, Medical University of Lodz, Lodz, Poland
- fEsperion Therapeutics, Ann Arbor, Michigan
- gMount Sinai Heart, Icahn School of Medicine at Mount Sinai, New York, New York
- ↵∗Address for correspondence:
Dr. Paul Muntner, Department of Epidemiology, University of Alabama at Birmingham, 1665 University Boulevard, Suite 230J, Birmingham, Alabama 35294.
Background Many patients report adverse reactions to, and may not tolerate, statin therapy. These patients may be at increased risk for coronary heart disease (CHD) events and mortality.
Objectives This study evaluated the risk for recurrent myocardial infarction (MI), CHD events, and all-cause mortality in Medicare beneficiaries with statin intolerance and in those with high adherence to statin therapy.
Methods We studied 105,329 Medicare beneficiaries who began a moderate- or high-intensity statin dosage after hospitalization for MI between 2007 and 2013. Statin intolerance was defined as down-titrating statins and initiating ezetimibe therapy, switching from statins to ezetimibe monotherapy, having International Classification of Diseases, 9th revision, diagnostic codes for rhabdomyolysis or an antihyperlipidemic adverse event, followed by statin down-titration or discontinuation, or switching between ≥3 types of statins within 1 year after initiation. High statin adherence over the year following hospital discharge was defined as proportion of days covered ≥80%. Recurrent MI, CHD events (recurrent MI or a coronary revascularization procedure), and mortality were identified from 1 year after hospital discharge through December 2014.
Results Overall, 1,741 patients (1.65%) had statin intolerance, and 55,567 patients (52.8%) had high statin adherence. Over a median of 1.9 to 2.3 years of follow-up, there were 4,450 recurrent MIs, 6,250 CHD events, and 14,311 deaths. Compared to beneficiaries with high statin adherence, statin intolerance was associated with a 36% higher rate of recurrent MI (41.1 vs. 30.1 per 1,000 person-years, respectively), a 43% higher rate of CHD events (62.5 vs. 43.8 per 1,000 person-years, respectively), and a 15% lower rate of all-cause mortality (79.9 vs. 94.2 per 1,000 person-years, respectively). The multivariate-adjusted hazard ratios (HR) comparing beneficiaries with statin intolerance versus those with high statin adherence were 1.50 (95% confidence interval [CI]: 1.30 to 1.73) for recurrent MI, 1.51 (95% CI: 1.34 to 1.70) for CHD events, and 0.96 (95% CI: 0.87 to 1.06) for all-cause mortality.
Conclusions Statin intolerance was associated with an increased risk for recurrent MI and CHD events but not all-cause mortality.
The efficacy of statins for the prevention of recurrent myocardial infarction (MI) has been demonstrated in several randomized controlled trials (1). Based on this evidence, guidelines recommend high-intensity statins following acute coronary syndromes (2). Statins are generally well tolerated, but as many as 20% of patients report adverse muscle side effects (3,4).
Statin intolerance may lead to discontinuation of therapy, which is associated with an increased risk of cardiovascular outcomes (5). However, statin intolerance does not always lead to patients discontinuing therapy (3). Patients who are statin intolerant may be down-titrated or switched among statin types. Few data are available for the risk of MI, recurrent coronary heart disease (CHD) events, and mortality among patients with statin intolerance. We determined rates of recurrent MI, CHD events, and all-cause mortality among older U.S. adults with statin intolerance following MI. For comparison, we calculated rates of these outcomes among a control population with high adherence to statin therapy following MI. We hypothesized that individuals with intolerance to statin therapy would have a higher risk for recurrent MI, CHD events, and all-cause mortality than those with high adherence to statins.
We conducted a retrospective cohort study using data from all Medicare beneficiaries hospitalized for MI between January 1, 2007, and December 31, 2013. Medicare is a program funded by the U.S. government that provides health insurance to residents ≥65 years of age and to those who are unable to work due to disability or who have end-stage renal disease. MI hospitalizations were identified by a claim, using International Classification of Disease-9th revision-Clinical Modification (ICD-9-CM) code 410.XX (excluding 410.X2, which represents a subsequent episode of care) in any discharge diagnosis position in the inpatient files. We restricted the analysis to beneficiaries with an overnight hospitalization for MI with a stay ≤30 days (n = 1,082,627) (Figure 1). All beneficiaries were required to be ≥66 and ≤110 years of age on the day of their hospital admission. In addition, we required all beneficiaries to have continuous full Medicare fee-for-service coverage for 365 days before their hospital admission date (i.e., the “look-back period”) and be alive with full Medicare fee-for-service coverage for the 365 days after hospital discharge (i.e., the “assessment period”). The look-back period was used to ensure patients were not taking lipid-lowering medication before their index MI and to define the presence of comorbid conditions (e.g., diabetes, chronic kidney disease). We defined full Medicare fee-for-service coverage as having Medicare Parts A (inpatient), B (outpatient), and D (pharmacy) coverage. We excluded Medicare beneficiaries with Part C coverage, as claims are incomplete for these individuals. Beneficiaries who did not live in the United States or were in hospice care at any point during the look-back or assessment periods were also excluded. We also restricted the analysis to beneficiaries who had an index statin prescription filled (i.e., first fill following hospital discharge for their index MI) for a high- or a moderate-intensity statin dosage within 30 days of discharge. To study patients who were beginning statin therapy, we excluded beneficiaries who had a statin prescription filled of any intensity during the look-back period. We restricted the analysis to patients who were beginning statin therapy, because beneficiaries who took statins before their index MI might have represented a select population, potentially with a lower incidence of statin intolerance. We also excluded beneficiaries who had a fill for a nonstatin lipid-lowering medication during the look-back period, as they might have been more likely to have a history of statin intolerance prior to their index MI.
Data for Medicare claims were obtained from the U.S. Centers for Medicare and Medicaid Services (CMS) Chronic Condition Data Warehouse. The Institutional Review Board at the University of Alabama at Birmingham and U.S. CMS approved the study.
Statin intolerance and statin use
A timeline for the identification of covariates, assessment of statin intolerance, high adherence to statins, and outcomes is provided in Figure 2. Statin prescription fills were identified using generic names from Medicare Part D claims and included atorvastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, and simvastatin. We defined the intensity for each statin type in accordance with American College of Cardiology (ACC)/American Heart Association (AHA) Cholesterol Guidelines (2) as shown in Online Table 1.
There is no consensus about how to define statin intolerance in claim-based studies. Therefore, we used 2 previously published definitions of statin intolerance (6). These definitions were determined by factors recognized as adverse events suggesting statin intolerance and by factors linked to the clinical management of statin intolerance consistent with current clinical practice guidelines (2,7–10).
The primary definition of statin intolerance included the following components, assessed within 365 days following hospital discharge after the index MI (i.e., the assessment period):
1. Statin discontinuation with the initiation of ezetimibe therapy;
2. Initiation of ezetimibe therapy within 7 days before or any time after down-titrating statin dose;
3. An inpatient or outpatient claim for rhabdomyolysis (defined by ICD-9-CM code 728.88 in any position), followed by statin down-titration or discontinuation;
4. An inpatient or outpatient claim for “adverse effect of an antihyperlipidemic agent” (defined by ICD-9-CM diagnostic code E942.2 in any position), followed by statin down-titration or discontinuation; and
5. Fills for ≥3 types of statins.
Many patients with statin intolerance may have their statin intensity down-titrated without beginning ezetimibe therapy or have a claim with a diagnostic code for rhabdomyolysis or an adverse effect of an antihyperlipidemic agent. Therefore, we used a secondary definition of statin intolerance, which included all components of the primary definition plus down-titrating statin intensity. More details for the primary and secondary definitions of statin intolerance used in the current analysis are provided elsewhere (6).
Among beneficiaries without statin intolerance, high statin adherence over the year following hospital discharge was defined by using a proportion of days covered (PDC) of ≥80% (11). Proportion of days covered was calculated as the number of days for which a patient had a statin available to take, based on their initial statin fill and pharmacy refill data in Medicare Part D, divided by the number of days in the year following hospital discharge. Days spent in the hospital did not contribute to the calculation of the PDC.
We used Medicare beneficiary summary files to obtain data for age at the time of admission for the index MI, sex, and race/ethnicity. We used all available claims prior to the index MI to determine comorbid conditions, including history of diabetes, chronic kidney disease, stroke, and heart failure (Online Table 2). For each beneficiary, we calculated the Charlson comorbidity index by using the method of Romano et al. (12). We used prescription claims from Medicare part D during the look-back period to determine a history of use of nonstatin lipid-lowering medication, including niacin, fibrates, bile acid sequestrants and ezetimibe, and antihypertensive medication. Antihypertensive medication classes analyzed as part of this study included angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta-blockers, calcium channel blockers, and diuretics (13). We categorized antihypertensive medication adherence during the look-back period as PDC <50%, 50% to <80%, and ≥80% (14).
We identified the occurrence of recurrent MI, CHD events, and all-cause mortality from 365 days after each beneficiary’s hospital discharge for their index MI through December 31, 2014. Patients were censored upon loss of Medicare Part A or B coverage or initiation of Part C coverage. Recurrent MI was defined by using an inpatient claim with an ICD-9-CM diagnosis code of 410.XX, except for 410.X2, in any position. Coronary heart disease events were defined by using a recurrent MI or an inpatient or outpatient revascularization procedure, including percutaneous coronary intervention, coronary artery bypass grafting, and other revascularization procedures. Percutaneous coronary intervention was defined by using the current procedure terminology code 92980-92982, 92984, 92995, or 92996 or the ICD-9-CM procedure code 00.66, 36.0, 36.01, 36.02, 36.05, or 36.07. Coronary artery bypass grafting was defined by the current procedure terminology code 33510-33519, 33521-33523, 33530, or 33533-33536 or the ICD-9-CM procedure code 36.10-36.19, or 36.2. Other revascularization procedures were defined by using ICD-9-CM procedure code 36.03, 36.04, or 36.09. All-cause mortality was identified by using Social Security Administration-validated death dates from the Medicare beneficiary summary file.
We calculated characteristics of Medicare beneficiaries with statin intolerance and those with high adherence to statins. The statistical significance of differences across these 2 groups was calculated using chi-square tests or Student t tests, as appropriate. We calculated the cumulative incidence for recurrent MI, CHD events, and all-cause mortality among beneficiaries with statin intolerance and among those with high adherence to statin therapy. The cumulative incidence of recurrent MI and CHD events was adjusted for the competing risk of all-cause mortality. We also calculated rates for recurrent MI, CHD events, and all-cause mortality for Medicare beneficiaries with statin intolerance and those with high adherence to statins. We used proportional hazard regression models to calculate the hazard ratios for recurrent MI, CHD events, and all-cause mortality associated with statin intolerance versus high adherence to statins. In addition to unadjusted models, 2 models with progressive adjustment for potential confounders were conducted. The initial multivariate-adjusted model included age, sex, and race/ethnicity. A second model included multivariate adjustment for age, sex, race/ethnicity, history of diabetes, chronic kidney disease, stroke and heart failure, antihypertensive medication use, adherence to antihypertensive medication, and Charlson comorbidity index during the look-back period. Hazard ratios (HRs) for recurrent MI and CHD events were estimated, with mortality as a competing risk, using the approach described by Fine and Gray (15).
We conducted subgroup analyses to investigate the consistency of the associations between statin intolerance and outcomes across age group, race/ethnicity, sex, history of comorbid conditions, use of antihypertensive medication, adherence to antihypertensive medication, and Charlson comorbidity index. We assessed differences in HRs across categories by modeling multiplicative interaction terms between statin intolerance and variables used to define subgroups (e.g., statin intolerance × sex).
We compared characteristics of Medicare beneficiaries who met the secondary definition of statin intolerance and those with high adherence to statins. We also calculated rates and HRs for recurrent MI, CHD event, and all-cause mortality among Medicare beneficiaries who met the secondary definition of statin intolerance versus those with high adherence to statins, as described above. All analyses were conducted using SAS version 9.3 software (SAS Institute, Cary, North Carolina), using a level of significance of <0.05.
Among 105,329 Medicare beneficiaries who began moderate- or high-intensity statins within 30 days of hospital discharge for their index MI, 1,741 subjects (1.65%) met the primary definition of statin intolerance, and 55,567 subjects (52.8%) had high adherence to statins. Compared to beneficiaries with high adherence, those with statin intolerance were less likely to be ≥80 years of age, male, have a history of chronic kidney disease, heart failure, a PDC of ≥80% for antihypertensive medication during the look-back period, and a Charlson comorbidity index ≥4 (Table 1).
Statin intolerance and recurrent MI, and CHD events and all-cause mortality
Over a maximum follow-up of 7 years (median: 1.9 to 2.3 years, depending on the outcome), there were 4,450 recurrent MIs, 6,250 CHD events, and 14,311 deaths. The cumulative incidence was higher for recurrent MI and CHD events and lower for all-cause mortality among beneficiaries with statin intolerance versus those with high adherence to statins (Figure 3). After multivariate adjustment, the HR comparing beneficiaries with statin intolerance with those with high statin adherence was 1.50 (95% confidence interval [CI]: 1.30 to 1.73) for recurrent MI, 1.51 (95% CI: 1.34 to 1.70) for CHD events, and 0.96 (95% CI: 0.87 to 1.06) for all-cause mortality (Table 2). The associations of statin intolerance with recurrent MI and CHD events were consistent across most subgroups (Figure 4). Statistically significant differences were present across age groups and participants with and without a history of stroke for recurrent MI; across age groups for CHD events; and across age groups, race/ethnicity, and sex for all-cause mortality. No other tests for interaction were statistically significant (all p values were >0.05).
Overall, 11,295 of the 105,329 Medicare beneficiaries (10.7%) met the secondary definition of statin intolerance. Beneficiaries with statin intolerance using the secondary definition were less likely to be ≥80 years of age, white, and male and to have a PDC of ≥80% for antihypertensive medication compared with those with high adherence to statins and were more likely to have a history of diabetes (Online Table 3). Beneficiaries with statin intolerance using the secondary definition had higher incidence rates for recurrent MI and CHD events than their counterparts with high adherence to statins but no differences were present in all-cause mortality (Online Table 4). After full multivariate adjustment, the HR comparing beneficiaries with statin intolerance using the secondary definition and high statin adherence was 1.26 (95% CI: 1.18 to 1.35) for recurrent MI, 1.25 (95% CI: 1.18 to 1.32) for CHD events, and 1.03 (95% CI: 0.99 to 1.08) for all-cause mortality.
In the current study, the incidence rates of recurrent MI and CHD events were higher among Medicare beneficiaries with statin intolerance than among their counterparts with high statin adherence (Central Illustration). Statin intolerance was not associated with an increased risk for all-cause mortality. Results were consistent in sensitivity analysis that included down-titration of statin intensity as an indicator of intolerance.
Based on evidence from randomized controlled trials (16–18), clinical practice guidelines recommend high-intensity statin therapy in patients with CHD (2,19). In a meta-analysis of randomized trials of patients with a history of MI, statins reduced the incidence of major coronary events by 22% (relative risk: 0.78; 95% CI: 0.74 to 0.84) per 1 mmol/l reduction in low-density lipoprotein cholesterol (20). Although statins are recommended to patients with CHD, several studies have reported suboptimal statin use in this high-risk population (21,22). For example, in the population-based REGARDS (REasons for Geographic And Racial Differences in Stroke) study, only 58% of participants with established CHD were taking statins (23). Furthermore, only 68% of eligible CHD patients in the PINNACLE (National Cardiovascular Data Practice Innovation and Clinical Excellence) registry were prescribed statins (24).
Side effects and intolerance have been reported frequently among patients following the initiation of statins. In a systematic review of 26 clinical trials, 12.7% of participants treated with statin developed myalgia (25). In the STOMP (Effect of Statins on Muscle Performance) study, 9.4% of participants randomized to 80 mg of atorvastatin developed myalgia (26). Side effects were reported by 60% of former statin users and 25% of current statin-users in an internet-based survey of 10,138 adults enrolled in the USAGE (Understanding Statin Use in America and Gaps in Education) study (27). However, not all adverse events among people taking statins are attributable to this therapy (25). In GAUSS-3 (Goal Achievement After Utilizing an Anti-PCSK9 Antibody in Statin-Intolerant Subjects 3), a randomized controlled trial that enrolled adults with a history of statin intolerance to 2 or more statins, 42.6% of participants reported intolerable muscle-related symptoms while taking atorvastatin but not placebo, whereas 26.5% of participants reported these symptoms while taking placebo but not atorvastatin (28). In the current study, the incidence of statin intolerance was lower (1.65% and 10.7% for the primary and secondary definitions of statin intolerance, respectively) compared with prior observational studies. The lower incidence of statin intolerance may be attributable to the algorithm used. The algorithms we used detected only cases that resulted in changes in patterns of statin use other than discontinuation of therapy. Among patients with a statin-related event analyzed by Zhang et al. (3), 59.2% of participants discontinued statins at least temporarily. The algorithm we used did not include statin discontinuation as people discontinue statins for several reasons beyond intolerance. Alternatively, the low incidence of intolerance could be the result of clinicians not submitting claims for some patients who reported side effects.
In the current study, no association was present between statin intolerance and all-cause mortality. Randomized controlled trials have reported that statins reduce the risk for all-cause mortality in patients with established cardiovascular disease (29,30). Many patients with statin intolerance continue to take statins or receive additional lipid-lowering therapy (e.g., ezetimibe). Additionally, the lack of an association in the current study may reflect the high mortality rate among older adults, including those with high adherence to statins. Future studies are needed to evaluate the mortality risk among individuals with statin intolerance.
The current study has several clinical implications. It reinforces the importance of patients with CHD to maintain high statin adherence. Statin intolerance may contribute to the increased risk for MI and CHD events through the subsequent discontinuation of treatment, low adherence, or use of low-intensity dosages (31). Another reason for the higher incidence of recurrent MI and CHD events in statin-intolerant versus high-adherent Medicare beneficiaries may be the presence of comorbid conditions. The excess risk for MI and CHD events indicates the need for cardiovascular risk reduction strategies among individuals with statin intolerance (22). Patients with statin intolerance may benefit from being rechallenged. In a retrospective cohort study, 59.1% of patients who discontinued treatment after a documented statin-related event were rechallenged, and more than 90% of these patients were still taking statins 1 year later (3). In a separate study, most patients referred to a specialty clinic for statin intolerance were able to be rechallenged and remained on a statin regimen (32). Reducing the residual CHD risk among patients with statin intolerance should be a high priority.
There is no standardized definition available to identify patients with statin intolerance (33,34). In the current analysis, we used an algorithm developed to identify statin intolerance using Medicare claims (6). This algorithm may be practical for investigating trends and factors associated with statin intolerance in settings where patient-reported outcomes including muscle pain and other potential statin side effects are not available. However, a potential weakness of the algorithm is that it may not detect mild cases of statin intolerance, which do not result in changes in statin use. Many patients report side effects from statins but do not change their treatment regimen (3).
There are strengths to the current study. Medicare data represent real-world clinical care of a large population of older adults across the United States. More than 90% of U.S. adults ≥65 years receive health insurance through Medicare, with most of them having fee-for-service coverage (35). Therefore, the results of this analysis have high generalizability to older U.S. adults (36). In addition to reliance on claims to define statin intolerance mentioned above, there are other potential and known limitations to the current study. The current study does not include a global definition for statin intolerance and positive and negative predictive values of the claims-based algorithm remain unknown. This study highlights the need to work toward a standard and validated definition of statin intolerance to better recognize and manage this high-risk population (33,34). High adherence to statins was defined based on administrative claims data. Pharmacy claims only identify beneficiaries who filled a drug prescription but not whether they actually took the medication (37). Also, lifestyle modifications, including physical activity, are recommended to people following MI, which may trigger muscle pain and subsequent changes in statin use (38). However, data for therapeutic lifestyle changes are not available in Medicare claims. Patients with statin intolerance were more likely to have low adherence to antihypertensive medication. Future studies are needed to explore the extent to which beliefs about medication in general are associated with statin intolerance. Some of the patients in the current study might have discontinued statins due to intolerance. We did not include statin discontinuation as part of the definition of intolerance because people discontinue statins for a variety of reasons (e.g., cost, beliefs about medication). A prior analysis suggested that including all patients who discontinue statins as intolerant would overestimate the incidence of statin intolerance (6). Laboratory results and over-the-counter medications (e.g., red yeast rice) are not available in Medicare claims. In addition, as Medicare beneficiaries <65 years of age are either disabled or have end-stage renal disease, we restricted our analyses to older adults, and results may not be generalizable to younger populations.
In a national cohort of Medicare beneficiaries hospitalized for MI, statin intolerance was common in the year after initiation of therapy. Medicare beneficiaries with statin intolerance had an increased risk for recurrent MI and CHD events. Strategies are needed to reduce this risk.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: After myocardial infarction, statin-intolerant patients face a high risk of recurrent coronary events.
TRANSLATIONAL OUTLOOK: Future studies are needed to discern the mechanisms responsible for recurrent coronary events among patients with statin intolerance and to develop alternative interventions to reduce this risk.
For supplemental tables, please see the online version of this article.
This study was supported by a research grant from Amgen, Inc. Dr. Rosenson has received research support from Amgen Inc., Eli Lilly, The Medicines Company, Regeneron, and Sanofi; serves on the advisory boards for Amgen Inc., Eli Lilly, Regeneron, and Sanofi. Drs. Manthripragada and Monda are employees of and hold stock in Amgen Inc. Dr. Dent is an employee of Esperion Therapeutics. At the time this study was conducted, Dr. Dent was an employee of Amgen Inc. Dr. Banach is member of the speakers bureau at Abbott/Mylan, Abbott Vascular, Actavis, Akcea, Amgen, Biofarm, KRKA, MSD, and Sanofi; is a consultant for Abbott Vascular, Amgen, Daichii-Sankyo, Esperion, MSD, Resverlogix, and Sanofi; and has received grants from Sanofi and Valeant. Dr. Bittner has participated in advisory panels for Amgen, Eli Lilly, and Pfizer; and has received research funding through her institution from Amgen, AstraZeneca, Dalcor, Bayer, and Sanofi. Dr. Kent has received salary support from Amgen. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary heart disease
- International Classification of Diseases-9th revision-Clinical Modification
- myocardial infarction
- proportion of days covered
- Received November 10, 2016.
- Accepted December 14, 2016.
- 2017 American College of Cardiology Foundation
- Stone N.J.,
- Robinson J.G.,
- Lichtenstein A.H.,
- et al.
- Mancini G.J.,
- Baker S.,
- Bergeron J.,
- et al.
- Tong L.S.,
- Hu H.T.,
- Zhang S.,
- Yan S.Q.,
- Lou M.
- Colantonio L.D.,
- Kent S.T.,
- Huang L.,
- et al.
- Bitzur R.,
- Cohen H.,
- Kamari Y.,
- Harats D.
- Fitchett D.H.,
- Hegele R.A.,
- Verma S.
- Gu Q.,
- Burt V.L.,
- Dillon C.F.,
- Yoon S.
- ↵Nau DP. Proportion of days covered (PDC) as a preferred method of measuring medication adherence. PQ alliance (Ed)Pharmacy Quality Alliance, Springfield, VA (2012). Available at: http://www.pqaalliance.org/images/uploads/files/PQA%20PDC%20vs%20%20MPR.pdf. Accessed February 6, 2017.
- Smith S.C.,
- Benjamin E.J.,
- Bonow R.O.,
- et al.
- Rosenson R.S.,
- Kent S.T.,
- Brown T.M.,
- et al.
- Hirsh B.J.,
- Smilowitz N.R.,
- Rosenson R.S.,
- Fuster V.,
- Sperling L.S.
- Maddox T.M.,
- Chan P.S.,
- Spertus J.A.,
- et al.
- Parker B.A.,
- Capizzi J.A.,
- Grimaldi A.S.,
- et al.
- ↵The Administration on Aging, Administration for Community Living. A profile of older Americans: 2013. Washington DC: U.S. Department of Health and Human Services, 2014. Available at: http://www.aoa.acl.gov/aging_statistics/profile/index.aspx. Accessed March 23, 2015.
- Mirel L.B.,
- Wheatcroft G.,
- Parker J.D.,
- Makuc D.M.
- Colantonio L.D.,
- Kent S.T.,
- Kilgore M.L.,
- et al.