THE CORRELATION BETWEEN THREE MORTALITY INDICES TO DETERMINE THE LIKELIHOOD OF BENEFIT FROM COLORECTAL CANCER SCREENING IN COMMUNITY-DWELLING OLDER ADULTS

Purpose: To target medical decision making appropriately among older adults regarding CRC screening, one must distinguish among those most and least likely to receive a net benefit. We sought to examine three strategies one might use for determining the likelihood of benefit from colorectal cancer screening in community-dwelling older adults, including a modification of the widely used Charlson Comorbidity Index (CCI) to determine how they compare within the same population. 

Methods: As part of a larger trial of a decision support tool, English-speaking primary care patients aged ≥70 years who were not currently up-to-date with CRC screening were evaluated using three mortality indices: a novel age-modified Charlson Comorbidity Index (CCI), a 4-year mortality index (the Lee index), and a 5-year mortality index (the Schonberg index).  Older adults were then categorized into best, intermediate, and worst health groups according to their modified-CCI and compared to the Lee and Schonberg indices. 

Results: Of 349 older adults, 150 were categorized in the best health according to the adjusted-CCI, 107 in intermediate health, and 92 in the worst health. The mean un-modified CCI score for the full sample was 1.8 (S.D. ± 2.1)(min, max= 0, 19); the mean Lee index score was 7.5 (S.D. ± 2.5) (min, max= 3, 16); and the mean Schonberg index score was 9.2 (S.D.±4.2) (min, max= 1, 23). The Lee and Schonberg indices correlated fairly closely with the modified-CCI, with a Pearson correlation coefficient of .51 and .46, respectively.  The Lee and Schonberg indices’ correlation with each other was 0.90 (p<0.001).  Five patients who were placed in the best health group according to the CCI were in the worst health group according to the Lee and Schonberg indices.  Seven were placed in the worst health group according to the CCI but the best health group according to the Lee and Schonberg indices.

Conclusions: Among a cohort of adults 70 and older, the age-modified CCI, Lee and Schonberg mortality indices correlate fairly closely. While the modified-CCI index was a useful tool, it did disagree occasionally with the others likely because these incorporate domains beyond comorbidity, such as degree and type of functional impairment.  When choosing a mortality index to assess likelihood of benefit, it is important to understand the strengths and limitations of the index.