Disease Comorbidity and Survival in the NHLBI Family Heart Study

Michael Province, Ph.D.

Project Overview:
Study Questions:

  1. What is the exact amount of excess familial clustering within each family for each risk factor/disease condition? 
  2. How do these familial factors co-cluster (or not) within the same families?
  3. What increased/decreased risk do combinations of family-specific clustering confer on overall/disease specific survival? 
  4. What is the independent contribution to survival risk from the familial nature of each factor?


  • To quantify the extent of familial clustering of families for each comorbidity and risk factor
  • To characterize the excess sharing of familial risk for clusters of diseases
  • To assess the independent contribution of these to overall survival

Study Methodology:
National Death Index (NDI) searches will be conducted to update current vital status. The relationship between prevalent disease and survival within individuals as well as within pedigree units will be examined using a variety of state of the art statistical models . Family Risk Score (FRS) methods (Williams et al., 2001) will allow us to quantify the extent of excess familial clustering of each family for each co-morbid condition. The degree to which these FRS scores are correlated across disease conditions within the same families will allow us to characterize the excess sharing of familial risk for clusters of diseases. We have also developed extensions of statistical frailty models for families (Siegmund et al., 1998, 1999) and bootstrapping methods (Province et al., 2000, 2001), which will allow us to assess the impact of disease and risk factors on survival in related (non-independent) samples. Finally, our general structural equations software designed for familial and genetic quantitative traits (SEGPATH: Province et al., 1995; 2001) will be extended to include survival.

Final Report: 
The NHLBI Family Heart Study (FHS) is a large, multicenter, population-based study of the familial and genetic determinants of coronary heart disease (CHD), atherosclerosis, and cerebral vascular disease (CVD) risk factors. Vital status, medical histories and lifestyle factors were collected on approximately 25,000 first degree relatives from 3,000 three generation pedigrees selected from 4 geographically diverse field centers (Winston-Salem, NC; Minneapolis, MN; Framingham MA; and Salt Lake City, UT) from 1992-1996. Conditions assessed included MI, CHD, stroke, hypertension, diabetes, high cholesterol, cancer, and asthma. Ages ranged from 0-103 years at that time. Approximately half of the families at each center were selected randomly, and the remaining were selected for high rates of early, familial CHD. A subset of over 5,000 subjects in 1,200 pedigrees (half random, half high risk) were further extensively examined clinically in all major risk domains for CHD, including demographics, diet, exercise, personal habits, concurrent medication use, anthropometry, lipid profile, biochemistries, ECG, pulmonary function, and carotid artery ultrasound measures of preclinical atherosclerosis. Washington University has been the Coordinating Center for FHS from the beginning (Province, PI) so these data are readily at hand and well understood. Read the full Final Report.