CVD indicates cardiovascular disease.
Murabito JM, Pencina MJ, Nam B, D’Agostino RB, Wang TJ, Lloyd-Jones D, Wilson PWF, O’Donnell CJ. Sibling Cardiovascular Disease as a Risk Factor for Cardiovascular Disease in Middle-aged Adults. JAMA. 2005;294(24):3117-3123. doi:10.1001/jama.294.24.3117
Author Affiliations: National Heart, Lung, and Blood Institute's Framingham Heart Study, Framingham, Mass (Drs Murabito, Pencina, Nam, D’Agostino, Wang, and O’Donnell); Section of General Internal Medicine, School of Medicine (Dr Murabito) and Statistics and Consulting Unit (Drs Pencina, Nam, and D’Agostino), Boston University, Boston, Mass; Cardiology Division, Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston (Dr Wang); Department of Preventive Medicine and Bluhm Cardiovascular Institute, Division of Cardiology, Feinberg School of Medicine, Northwestern University, Chicago, Ill (Dr Lloyd-Jones); Departments of Endocrinology, Diabetes, and Medical Genetics, Medical University of South Carolina, Charleston (Dr Wilson); and National Heart, Lung, and Blood Institute, Bethesda, Md (Dr O’Donnell).
Context While parental cardiovascular disease (CVD) doubles the risk for CVD in offspring, the extent of increased risk associated with sibling CVD is unclear.
Objective To determine, using validated events, whether sibling CVD predicts outcome in middle-aged adults independent of other risk factors.
Design, Setting, and Participants The Framingham Offspring Study, an inception cohort of the Framingham Heart Study, a prospective population-based cohort study initiated in 1948 with the offspring cohort initiated in 1971. Participants (n = 2475) were members of the offspring cohort aged 30 years or older, free of CVD, and with at least 1 sibling in the study; all were followed up for 8 years.
Main Outcome Measures Association of sibling CVD with 8-year personal risk for CVD using pooled logistic regression. A secondary analysis restricted to offspring with both parents in the study assessed the joint impact of parental and sibling CVD occurrence.
Results Among 973 person-examinations in the sibling CVD group (mean age, 57 years) and 4506 person-examinations in the no sibling CVD group (mean age, 47 years), 329 CVD events occurred during follow-up. Baseline risk factors were more prevalent in the sibling CVD group compared with the no sibling CVD group. Sibling CVD was associated with a significantly increased risk for incident CVD (age- and sex-adjusted odds ratio [OR], 1.55; 95% confidence interval [CI], 1.19-2.03). Adjustment for risk factors did not substantially attenuate the risk (adjusted OR, 1.45; 95% CI, 1.10-1.91). In the analysis restricted to persons with both parents in the study, in models adjusting for both sibling and parental CVD, the multivariable-adjusted OR for sibling CVD (1.99; 95% CI, 1.32-3.00) exceeded that for parental CVD (1.45; 95% CI, 1.02-2.05).
Conclusion Using validated events, sibling CVD conferred increased risk of future CVD events above and beyond established risk factors and parental CVD in middle-aged adults.
Cardiovascular disease (CVD) in a first-degree relative confers increased risk for CVD,1 but whether familial CVD is truly an independent risk factor remains controversial. Parental CVD doubles the risk of CVD in adult offspring.2 A seminal report by Marenberg et al3 established increased risk for death from coronary heart disease in twins. Risk associated with CVD in siblings in multiplex families is less certain because published estimates are largely derived from case-control studies that generally lack sibling CVD event validation.4- 6 Furthermore, estimates regarding magnitude of risk associated with a history of sibling CVD vary greatly. Some studies have reported CVD risk similar to that conferred by a history of parental CVD5,7; others have described much greater CVD risk in relation to sibling history than in relation to parental history.6
Accurate information about familial CVD will have increasing importance in prevention and treatment of CVD in the post-genome era.8 Recent national survey data document that adults believe family history information is important to their health, but few have systematically collected this information from relatives.9 Recall bias, especially for premature CVD in the family, may reduce the usefulness of reported family history information.10 It is also possible that risk estimates for sibling history in case-control studies are exaggerated due to bias related to differences in family size and age at disease onset.11 Risk estimates for sibling history may also be inflated due to confounding such as that caused by the sharing of higher risk factor levels by siblings of persons with CVD.12,13 Associations between reported sibling history and subclinical CVD are attenuated when adjusted for CVD risk factors.14,15
We sought to determine whether the occurrence of a validated sibling CVD event independently and prospectively predicted CVD events in a cohort of middle-aged adults. We further sought to examine the impact of sibling CVD over and above that of parental CVD.
In 1971, 5124 participants (offspring of the original Framingham Heart Study cohort and offspring spouses) aged 5 to 70 years were enrolled in the Framingham Offspring Study, a prospective epidemiologic cohort study.16,17 The offspring cohort included 3498 participants who were members of an identified Framingham Study family. Participants have undergone follow-up examinations approximately every 4 years since study inception. Study design and entry criteria for both the offspring cohort and the original cohort of the Framingham Heart Study have been previously reported.16,18 All participants provided written informed consent at each examination attended, and all study protocols were reviewed by the institutional review board at Boston Medical Center.
Data from 4 offspring examinations, each with 8 years of follow-up, were pooled: offspring cohort examinations 1 (1971 to 1975), 2 (1979 to 1982), 4 (1987 to 1990), and 6 (1995 to September 2, 1998). Follow-up for the final examination cohort ended in December 2004. Since the first and second examinations were about 8 years apart, to ascertain comparable lengths of follow-up after each of the 4 examinations, we chose to examine the 8-year occurrence of CVD events. All offspring participants who were members of a Framingham Study family and were aged 30 years and older at any of the 4 examinations were eligible for inclusion in our study if they had at least 1 sibling enrolled in the Framingham Offspring Study and if they were free of CVD at the time of examination (thus, the sibling contributing the positive CVD occurrence was excluded) (Figure). In addition, we randomly excluded 1 sibling from families with no sibling CVD to provide comparable structure between families with and without sibling CVD. Our final study sample included 2475 unique participants (1188 men) contributing 973 person-examinations from participants in a family with sibling CVD and 4506 person-examinations from participants in a family without sibling CVD (Figure).
Sibling CVD was defined as the occurrence of a validated sibling CVD event prior to an examination. Cardiovascular disease events included angina pectoris, coronary insufficiency, myocardial infarction, stroke or transient ischemic attack, intermittent claudication, coronary heart disease death, and CVD death. All CVD events (both sibling events and incident events occurring in participants in the study sample) were adjudicated by a panel of 3 senior investigators (or a panel of study neurologists for cerebrovascular disease events) who were unaware of sibling CVD status, using standardized criteria previously reported.19
Parental occurrence of premature CVD was available in a subsample of offspring participants with both parents enrolled in the original Framingham cohort (Figure). Parental events were adjudicated using the same protocol and standardized criteria. Parental premature CVD was defined as the occurrence of a validated parental event prior to an offspring examination and before age 55 years in fathers or age 65 years in mothers. These age cut points were derived from existing guidelines regarding family history of premature CVD.20
Risk factors were measured at each examination. Height and weight were obtained by trained technicians, and body mass index was calculated as weight in kilograms divided by the square of height in meters. Blood pressure was measured twice at rest by the examining physician, and the mean of the 2 blood pressure readings was used. Hypertension was defined as systolic blood pressure of 140 mm Hg or greater, diastolic blood pressure of 90 mm Hg or greater, or use of antihypertensive medication. Current smoking was defined as smoking 1 or more cigarettes per day in the year preceding examination. Blood was obtained in the fasting state, and the ratio of total cholesterol to high-density lipoprotein cholesterol was calculated. The presence of diabetes was defined by a fasting glucose level of 126 mg/dL (7.0 mmol/L) or greater or use of insulin or oral hypoglycemic agents.
Pooled logistic regression analyses weighted for sibship size were used to examine the risks of incident CVD events associated with the occurrence of sibling CVD. The method of pooling person-examinations allows for time-dependent covariance of risk factors and sibling CVD events and has been shown to provide estimates of effect similar to those provided by time-dependent Cox analyses.21 A weight for family size was used to diminish bias related to differences in the number of siblings across families. For all logistic regression analyses, the reference group consisted of participants with no sibling CVD prior to the examination. Odds ratios (ORs) and 8-year event rates with corresponding 95% confidence intervals (CIs) were calculated in unadjusted, age- and sex-adjusted, and multivariable-adjusted models. These analyses were repeated examining sibling premature CVD (defined as a CVD event in a brother before age 55 years or in a sister before age 65 years). Covariates in the multivariable model included age, sex, systolic blood pressure, use of antihypertensive medication, ratio of total cholesterol to high-density lipoprotein cholesterol, body mass index, diabetes mellitus, and current cigarette smoking. Significant interactions with sibling CVD were noted for age, systolic blood pressure, and hypertension treatment. Therefore, we repeated the analyses stratifying by age group above and below the median age (≤48 years, >48 years) and hypertension status (yes or no).
In secondary analyses limited to participants with both parents in the study, logistic regression analyses were repeated with the multivariable-adjusted model including parental premature CVD in addition to the other covariates. Finally, to examine whether there was a dose effect for occurrence of CVD in any first-degree relative, the multivariable analysis was repeated, this time entering predictor variables for parental premature CVD only, sibling CVD only, and both parental premature CVD and sibling CVD.
To assess the added usefulness of the sibling CVD status in predicting future CVD events for each participant, we calculated a risk score according to the risk guidelines from the National Cholesterol Education Program Adult Treatment Panel III.20 Participants were then stratified into 10-year risk categories for coronary heart disease (low = less than 10% risk; intermediate = 10% to 19% risk; and high = 20% risk or greater). These guidelines also place persons with known diabetes into the high-risk group. We used sex- and age-adjusted logistic regression analyses to compare 8-year CVD event rates between individuals with and without sibling CVD within each risk strata defined above, as well as across strata of individual CVD risk factors. All statistical analyses were performed using SAS version 8.0 (SAS Institute Inc, Cary, NC). P values were 2-sided; P<.05 was used to determine statistical significance.
Most sibling events occurred prematurely, with a mean age at onset of 48.2 (SD, 4.79), 48.7 (SD, 7.36), 50.7 (SD, 8.27), and 53.6 (SD, 9.94) years at the 4 pooled examinations, respectively. Compared with the group with no sibling CVD, those with sibling CVD were older and had higher prevalence of all traditional risk factors except current smoking (Table 1). Of the 329 incident CVD events during follow-up, there were 11 coronary deaths, 8 other CVD deaths, 99 nonfatal cases of myocardial infarction or coronary insufficiency, 106 cases of angina pectoris, 59 strokes or transient ischemic attacks, and 46 cases of intermittent claudication. There were 223 events in the group with no sibling CVD over 34 110 person-years of follow-up, yielding a crude rate of 6.54 events per 1000 person-years; there were 106 events in the sibling CVD group over 6943 person-years of follow-up, yielding a crude rate of 15.27 events per 1000 person-years.
The 8-year CVD event rates and ORs for CVD events associated with the occurrence of sibling CVD are shown in Table 2. Sibling CVD was associated with a significantly increased risk for incident CVD (age- and sex-adjusted OR, 1.55; 95% CI, 1.19-2.03); this association persisted even after adjustment for risk factors (multivariable-adjusted OR, 1.45; 95% CI, 1.10-1.91). The attributable risk percentage for sibling CVD was 27.4%; this represents the proportion of the 8-year CVD risk among those in the sibling CVD group that theoretically could be prevented if members of the group had not had a sibling with CVD. Because there were significant interactions in the multivariable model, we examined the model stratified by age (≤48 years vs >48 years) and hypertension status (yes or no). The impact of sibling CVD was stronger in the younger age group (multivariable-adjusted OR for sibling CVD vs no sibling CVD in the younger group: 2.22; 95% CI, 1.22-4.02; in the older group: 1.33; 95% CI, 0.98-1.80) and in participants free of hypertension (multivariable-adjusted OR in those without hypertension: 1.98; 95% CI, 1.31-2.99; in those with hypertension: 1.19; 95% CI, 0.82-1.72), but the differences were not statistically significant. In a secondary analysis taking into account the age at onset of sibling CVD, only premature onset of sibling CVD was significantly related to CVD incidence; multivariable ORs were 1.58 (95% CI, 1.18-2.12) for sibling premature CVD and 1.04 (95% CI, 0.61-1.77) for sibling nonpremature CVD. However, of the 973 person-examinations in the sibling CVD group, only 180 person-examinations came from families with nonpremature onset of sibling CVD.
We examined the impact of parental CVD in the secondary analysis restricted to participants with both parents in the study. The occurrence of sibling CVD remained a significant predictor of incident CVD, even after adjusting for all risk factors and for parental premature CVD (OR, 1.56; 95% CI, 1.11-2.18) (Table 2). Sibling CVD was associated with a stronger risk for incident CVD than was parental premature CVD; multivariable-adjusted ORs were 1.45 (95% CI, 1.02-2.05) for parental premature CVD alone, 1.99 (95% CI, 1.32-3.00) for sibling CVD alone, and 1.53 (95% CI, 0.93-2.51) for both parental premature CVD and sibling CVD. Of note, the lower multivariable-adjusted risks among participants with both parental and sibling CVD might be explained by the particularly high prevalence of risk factors in this group (prevalence of diabetes and treatment for hypertension was 12.2% and 31.5%, respectively, in those with both parental and sibling CVD, compared with 5.8% and 22.3%, respectively, in those with sibling CVD only and 5.8% and 14.1%, respectively, in those with parental premature CVD only).
When we stratified participants by elevated levels of individual risk factors and estimates of overall CVD risk categories, sibling CVD information added substantially to discrimination of observed 8-year event rates (Table 3). Of note, when we stratified by age, sibling CVD was associated with a significantly higher event rate in those aged 30 to 59 years but not in those aged 60 years and older. Sibling CVD was associated with increased risk in persons with adverse levels of most risk factors and in persons in the intermediate and high Adult Treatment Panel III risk categories. However, sibling CVD was not associated with significantly increased risk among persons with known diabetes or hypertension who already had substantially higher event rates.
Using a prospective design and validated sibling CVD events, we found that sibling CVD was associated with a significantly increased risk for incident CVD events in middle-aged adults. The OR remained statistically significantly elevated in age- and sex-adjusted and multivariable-adjusted models, suggesting that age and traditional risk factors explain part but not all of the increased risk associated with sibling CVD. Furthermore, in analyses restricted to participants with both parents in the Framingham Study, the presence of sibling CVD conferred an increased CVD risk independent of parental premature CVD, and sibling CVD may be more strongly associated with incident events than is parental premature CVD. Our findings provide strong evidence that sibling CVD is an important risk factor for incident CVD and represents a useful marker of familial vulnerability to CVD events.
Results from a previously reported, population-based, case-control study conducted to estimate risk of coronary heart disease associated with various definitions of a family history clearly established the value of going beyond the simple yes or no response to questions about presence of disease in a first-degree relative.4 This work demonstrated that distinguishing between an affected parent and an affected sibling was important, particularly for younger ages of disease onset. Moreover, having a sibling with coronary heart disease was associated with significantly increased risk, even in families with a parent already affected at a young age. The importance of a sibling history is consistent with twin studies in which coronary heart disease death in a young twin was especially associated with increased risk of death in the other twin.3,22 Our study adds to these important previous studies, because we examined nontwin siblings and were able to confirm that sibling CVD increased risk for incident events principally if onset of sibling CVD was premature and if the sibling at risk was a young adult.
Some case-control studies have reported that the risk conferred by a history of sibling CVD is similar to that conferred by a history of parental CVD,5,23 but others have found that sibling CVD confers a greater risk than parental CVD.6 A recent study that examined the association between family history and subclinical measures of coronary atherosclerosis found a much stronger association between the presence and extent of coronary artery calcification on electron beam tomography and sibling history compared with parental history. In the analysis considering both together, the ORs for coronary calcification associated with sibling history and parental history were 2.3 and 1.3, respectively.14 When we conducted an analysis comparing parental premature CVD with sibling CVD, we also found stronger associations for sibling than for parental CVD. The magnitude of risk for incident CVD associated with sibling CVD and parental premature CVD in our study was strikingly similar (ORs, 1.99 and 1.45, respectively) to that in the prior report. In the prior report, recall bias cannot be excluded as an explanatory factor contributing to the stronger ORs for sibling history compared with parental history; however, our data are not susceptible to this limitation, because sibling and parental events were based on medical records rather than on self-report.
Concern has been raised that risk associated with familial CVD history can be largely explained by familial aggregation of traditional risk factors. Consistent with other investigations of sibling history,12,13,24- 27 we found that participants with sibling CVD had higher prevalence of risk factors compared with participants with no sibling CVD. This finding was especially striking in participants with both sibling CVD and parental premature CVD, who had high prevalence of diabetes and use of medication for hypertension. Nevertheless, the risk associated with sibling CVD remained significant in multivariable models. This finding suggests that a significant proportion of risk is explained by factors other than traditional risk factors, in turn suggesting that other genetic risk factors may influence susceptibility to CVD. Among the possible factors for increased risks conferred by sibling CVD are shared early environmental exposures (in utero or early childhood) and a shared genetic background.
Several strengths and limitations of our study merit comment. Our prospective study design allowed the examination of incident CVD events, and all events within the family were adjudicated by a panel of senior investigators using the same standardized criteria. Thus, sibling events were validated by review of medical records rather than reliance on self-reports. Risk factors were directly measured for all participants and updated over time, independent of sibling CVD. Thus, measures of lipid levels and diabetes were ascertained directly and not obtained by self-report, which is more susceptible to misclassification. The use of reported risk factors in multivariable analyses may not fully account for shared risk factors within families and in turn results in an overestimation of the OR associated with sibling CVD. The original and offspring cohorts of the Framingham Heart Study are primarily white, potentially limiting the extent to which our findings can be generalized to other groups. Further, we did not have risk factor or event information on siblings who declined enrollment in the Framingham Study. It is possible that siblings with early-onset CVD died prior to enrollment or declined enrollment due to poor health status; this would be expected to bias our results toward an underestimation of the risk associated with sibling CVD.
Using validated events and a prospective design, our study substantially extends the knowledge base regarding the importance of sibling CVD. We observed that sibling CVD confers increased risks of CVD events above and beyond traditional risk factors and parental premature CVD. Thus, sibling CVD should be considered as important as parental premature CVD in the assessment of risk. Further investigation is needed to better understand why sibling history may be a stronger predictor for CVD than parental history, including exploration of the contribution of an early shared environment to increased sibling risk. Moreover, investigation of whether to incorporate sibling CVD as well as parental CVD into existing risk prediction and prevention algorithms is warranted.
Corresponding Author: Joanne M. Murabito, MD, ScM, Framingham Heart Study, 73 Mt Wayte Ave, Suite 2, Framingham, MA 01702 (firstname.lastname@example.org).
Author Contributions: Drs Murabito, Nam, Pencina, and O’Donnell had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Murabito, Pencina, Nam, D’Agostino, Wang, Lloyd-Jones, O’Donnell.
Acquisition of data: Murabito, D’Agostino, Wilson, O’Donnell.
Analysis and interpretation of data: Murabito, Pencina, Nam, D’Agostino, Wang, Lloyd-Jones, Wilson, O’Donnell.
Drafting of the manuscript: Murabito, O’Donnell.
Critical revision of the manuscript for important intellectual content: Murabito, Pencina, Nam, D’Agostino, Wang, Lloyd-Jones, Wilson, O’Donnell.
Statistical analysis: Pencina, Nam, D’Agostino, Lloyd-Jones, O’Donnell.
Obtained funding: D’Agostino.
Administrative, technical, or material support: Murabito, Wilson, O’Donnell.
Study supervision: Murabito, O’Donnell.
Financial Disclosures: None reported.
Funding/Support: This work was supported by the National Heart, Lung, and Blood Institute's Framingham Heart Study (contract N01-HC-25195).
Role of the Sponsor: The National Heart, Lung, and Blood Institute approved the manuscript for publication but had no direct involvement in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation of the manuscript.