Charlson Comorbidity Index Varies More among Older People in Four U.S. Health Systems: Taylor’s Law and Fluctuation Scaling

Joel E. E. Cohen, Rockefeller University and Columbia University
Jonathan N. Tobin, Clinical Directors Network

The Charlson Comorbidity Index (CCI) combines a person’s concurrent medical conditions (comorbidities) in 17 categories, such as heart disease, AIDS, or cancer, to predict the person’s probability of death within 10 years. The average CCI among people of a given age increases with increasing age. We report here that the variability of CCI among people of each age increases with age and is closely related to the age group’s average CCI. Specifically, we relate here the variance of CCI as a function of the mean of CCI for single year-of-age groups of 82,512 adults in low-income communities cared for by four health systems, two in Chicago, Illinois, and two in New York City. In Chicago’s health systems, as age increases, the variance of CCI approximates a power function of the mean CCI. This quantitative relationship, not previously recognized, confirms Taylor's law. By contrast, in New York systems, the variance of CCI increases with the mean CCI more slowly than a power function. Within each age group and health system, the frequency distribution of CCI is well approximated by a negative binomial distribution. We speculate on possible sources of the differences between Chicago and New York.

Keywords: Health and Morbidity, Data and Methods, Mortality and Longevity, Older Adults and Intergenerational Relations

See extended abstract.