METHODS: This cohort study used linked administrative data from Ontario, Canada to identify adults >65 years having elective noncardiac surgery from 2010 to 2015. Frailty was identified using a validated index; comorbidities were identified with validated codes. We evaluated the presence of effect modification (also called interaction) between frailty and each comorbidity on (1) the relative (or multiplicative) scale by assessing whether the risk of mortality when both frailty and the comorbidity were present was different than the product of the risks associated with each condition; and (2) the absolute risk difference (or additive) scale by assessing whether the risk of mortality when both frailty and the comorbidity were present was greater than the sum of the risks associated with each condition.
RESULTS: 11,150 (9.7%) people with frailty died versus 7826 (2.8%) without frailty. After adjustment, frailty was associated with decreased survival (adjusted hazard ratio [HR] = 2.42; 95% confidence interval [CI], 2.31–2.54). On the relative (multiplicative) scale, only diabetes mellitus demonstrated significant effect modification (P value for interaction .03; reduced risk together). On the absolute risk difference (additive) scale, all comorbidities except for coronary disease demonstrated effect modification of the association of frailty with survival. Co-occurrence of dementia with frailty carried the greatest excess risk (Synergy Index [S; the excess risk from exposure to both risk factors compared to the sum of the risks from each factor in isolation] = 2.29; 95% CI, 1.32–10.80, the excess risk from exposure to both risk factors compared to the sum of the risks from each factor in isolation).
CONCLUSIONS: Common comorbidities modify the association of frailty with postoperative survival; however, this effect was only apparent when analyses accounted for effect modification on the absolute risk difference, as opposed to relative scale. While the relative scale is more commonly used in biomedical research, smaller effects may be easier to detect on the risk difference scale. The concurrent presence of dementia, COPD, and HF with frailty were all associated with excess mortality on the absolute risk difference scale.
Leave a Reply
You must be logged in to post a comment.