Operationalizing multimorbidity in relation to mortality and physical function: findings from a prospective longitudinal cohort study

BACKGROUND AND OBJECTIVESThere is little consensus on how multimorbidity should be operationalized in life course research. We set out to derive better empirical definitions for multimorbidity by examining its different operationalizations and their associations with mortality, functional independence and physical capability in a longitudinal population-representative cohort. RESEARCH DESIGN AND METHODSWe used data from 2653 (51.6% female) study members in the MRC National Survey of Health and Development (NSHD), a British cohort born in 1946. We examined predictive utility of five multimorbidity operationalizations by age 63 (binary multimorbidity, unweighted disease count, weighted disease count, clustered multimorbidity, cumulative multimorbidity) using 16 chronic disease, to predict 12-year mortality and age 69 functional independence and physical capability (grip strength, chair rise speed, balance). RESULTSThe multimorbidity operationalizations of unweighted disease count (16.3%) and weighted disease count (16.7%) explained the highest variation for mortality compared with other operationalizations. Similarly, explained variation for unweighted disease count (17.0%) and weighted disease count (16.4%) showed the highest prediction for functional independence and physical capability at age 69 years, with minor variations in physical performance measures where clustered multimorbidity also explained high variance (e.g. 8.6% for changes in chair rise speed). Binary multimorbidity, although frequently used in research was the least predictive of all outcomes. DISCUSSION AND IMPLICATIONSThe associations between various multimorbidity measures and mortality and physical functioning lend support to important influence of multimorbidity on later-life health and functioning while highlighting the differences in variance prediction and point estimates when different approaches to operationalizing multimorbidity are used. An unweighted disease count approach might be suitable for many epidemiological research questions as it is simple to estimate while being as predictive as other more complex approaches such as weighted disease counts. Translational SignificanceMultimorbidity research urgently needs a standardized approach to its measurement and operationalization. This study simultaneously compares different strengths of association between multimorbidity definitions and mortality, functional independence and physical capability. This adds possibility by using the predictive utility for mortality, functional independence and physical capability as criteria for determining the usefulness of a given operationalization. Quantifying how these vary offers practical choices between methods of operationalizing multimorbidity for various purposes in research and clinical settings, for example, a targeted strategy aimed at understanding mortality, physical function and health care utilization.