Healthy Volunteer Effect and Cardiovascular Risk

Leening, Maarten J. G.; Heeringa, Jan; Deckers, Jaap W.; Franco, Oscar H.; Hofman, Albert; Witteman, Jacqueline C. M.; Stricker, Bruno H. Ch.

doi: 10.1097/EDE.0000000000000091

Department of Epidemiology, Erasmus MC–University Medical Center Rotterdam, Rotterdam, The Netherlands

Department of Cardiology, Erasmus MC–University Medical Center Rotterdam, Rotterdam, The Netherlands

Department of Epidemiology, Erasmus MC–University Medical Center Rotterdam, Rotterdam, The Netherlands,

Supplemental digital content is available through direct URL citations in the HTML and PDF versions of this article ( This content is not peer-reviewed or copy-edited; it is the sole responsibility of the author.

Article Outline
Back to Top | Article Outline

To the Editor:

Most cardiovascular risk prediction functions are developed using data from cohort studies. Invariably, a proportion of invitees will not participate in such studies because relatively good health status is required for a person to agree to undergo the examinations. This implies that persons enrolled in a study requiring active participation are healthier than those who declined to participate. It is thus unclear whether the cardiovascular risk distributions among study participants adequately reflect the risk distribution of the source population. We aimed to quantify the consequences of this “healthy volunteer effect.”

Within the Rotterdam Study,1 a prospective population-based cohort, we investigated the association between participation in the third examination (1997–1999), all-cause mortality,2 and coronary risk (Framingham point score; assessed at enrollment 1990–1993)3 (see eAppendix, for details). Of 5423 eligible invitees (mean age 73.5 years; 39% men), 87% participated, of whom 76% visited the research center (eTable 1, Nonparticipants had lost interest (50%), had physical complaints (34%), or considered themselves too old to participate (12%; mean age 86.9 years). Persons who were elderly, women, less educated, and with higher levels of specific cardiovascular risk factors were less likely to participate (eTable 1).

Nonparticipation was strongly associated with mortality (hazard ratio [HR] = 1.71 [95% confidence interval (CI) = 1.56–1.88]). This was most pronounced shortly after invitation (0–3 months, HR = 4.85 [2.43–9.71]), with a diminishing healthy volunteer effect during follow-up (test for trend, P< 0.001) (Table). Every percentage-point increase in coronary risk yielded an approximately 3% lower probability of participating (eTable 2, Those categorized as “high risk” were least likely to participate (odds ratio = 0.56 [95% CI = 0.45–0.71]; eTable 2). There was a slightly lower proportion of high-risk persons among the examined participants compared with all invitees (23% vs. 24%) (eTable 3,

More than 5 decades ago, investigators from the Framingham Study observed higher mortality rates in those who refused to participate.4 They speculated that mortality in participants and nonparticipants might converge later during follow-up. We indeed noticed declining differences in mortality rates at long-term follow-up. The large difference shortly after invitation is presumably attributable to clinical or subclinical disease that makes invitees less likely to volunteer.5 Residual differences at long-term follow-up could reflect a health-care-aversive attitude or a lower awareness of health in general.6 However, long-term benefits of participation in population-based research cannot be ruled out because most studies (including the Rotterdam Study) disclose results from measurements (eg, blood pressure and cholesterol levels) and incidental findings (eg, aneurysms and indolent cancers) to the participants.7

Our focus on distributions of absolute cardiovascular risk is new information. This is important in light of the recent shift in cardiovascular epidemiologic research toward evaluating new risk markers on the basis of their clinical utility in risk prediction (ie, whether the predicted risks with addition of a new marker change sufficiently to alter recommended therapy) rather than the marker’s association with cardiovascular disease expressed as an odds ratio or HR. In contrast to the study of associations, the risk distribution in the study population affects the magnitude of measures of risk reclassification, such as net reclassification improvement.8 Given that most people have a low predicted cardiovascular risk (eTable 3), underrepresentation of persons at higher cardiovascular risk will usually mean that smaller proportions of persons are reclassified, resulting in lower estimates of net reclassification improvement8; therefore, the contribution of emerging risk markers could be underestimated. We observed a small underrepresentation of persons at high cardiovascular risk (1.2% less at high risk) (eTable 3); thus, results from analyses on absolute cardiovascular risk are not likely to be severely affected. However, future simulation studies could quantify the degree of bias on measures of risk reclassification.

Maarten J. G. Leening

Jan Heeringa

Department of Epidemiology

Erasmus MC–University Medical

Center Rotterdam

Rotterdam, The Netherlands

Jaap W. Deckers

Department of Cardiology

Erasmus MC–University Medical

Center Rotterdam

Rotterdam, The Netherlands

Oscar H. Franco

Albert Hofman

Jacqueline C. M. Witteman

Bruno H. Ch. Stricker

Department of Epidemiology

Erasmus MC–University Medical

Center Rotterdam

Rotterdam, The Netherlands

Back to Top | Article Outline


1. Hofman A, Darwish Murad S, van Duijn CM, et al. The Rotterdam Study: 2014 objectives and design update. Eur J Epidemiol. 2013;28:889–926
2. Leening MJG, Kavousi M, Heeringa J, et al. Methods of data collection and definitions of cardiac outcomes in the Rotterdam Study. Eur J Epidemiol. 2012;27:173–185
3. Koller MT, Leening MJG, Wolbers M, et al. Development and validation of a coronary risk prediction model for older U.S. and European persons in the Cardiovascular Health Study and the Rotterdam Study. Ann Intern Med. 2012;157:389–397
4. Gordon T, Moore FE, Shurtleff D, Dawber TR. Some methodological problems in the long-term study of cardiovascular disease: observations on the Framingham Study. J Chronic Dis. 1959;10:186–206
5. Lindsted KD, Fraser GE, Steinkohl M, Beeson WL. Healthy volunteer effect in a cohort study: temporal resolution in the Adventist Health Study. J Clin Epidemiol. 1996;49:783–790
6. Shahar E, Folsom AR, Jackson R. The effect of nonresponse on prevalence estimates for a referent population: insights from a population-based cohort study. Atherosclerosis Risk in Communities (ARIC) Study Investi gators. Ann Epidemiol. 1996;6:498–506
7. Benfante R, Reed D, MacLean C, Kagan A. Response bias in the Honolulu Heart Program. Am J Epidemiol. 1989;130:1088–1100
8. Leening MJG, Vedder MM, Witteman JCM, Pencina MJ, Steyerberg EW. Net reclassification improvement: computation, interpretation, and controversies. A literature review and clinician’s guide. Ann Intern Med. 2014;160:122–131

Supplemental Digital Content

Back to Top | Article Outline
© 2014 by Lippincott Williams & Wilkins, Inc