(C37) Changes in Comorbidity / Multimorbidity within 5 Years, Post-Recruitment into the UK Biobank

Publication Number: 981

Background: UK Biobank (UKB) research commonly treats comorbidity / multimorbidity status as a ‘static’ state, with particular emphasis on comorbidity status only at recruitment. However, the prevalence of multiple comorbidities increases with age; and, an individual’s comorbidity profile will change as new diagnoses accumulate over time.

Objectives: Our research calculated Elixhauser Comorbidity Scores (ECS) in UKB to identify how ECS changes over 5 years.

Methods: We used data from UKB, a prospective cohort study of over 500K participants aged 40-69 years at recruitment from 2006 to 2010, with follow-up through additional, bespoke data collection efforts as well as linkage to electronic health record data and other registries (e.g. Cancer and Mortality). General Practitioner (GP) data is only currently available for approximately half of the cohort. Consequently, we required participants in our study to have at least 1 GP registration record. We also excluded any participants with a mortality event before 5 years post-recruitment. Elixhauser et.al. defined a set of 31 comorbidity indicators, which have been translated by Quan et.al. for use in administrative databases based on ICD-9 and ICD-10 diagnostic codes; and, more recently by Metcalfe et.al for use in Read-coded (CTV3) databases. We identified events matching each of the 31 Elixhauser comorbidities in the hospital episodes and cancer registries, coded in ICD10, and in the GP data, coded in CTV3, for all events through 5 years post-recruitment. We calculated ECS at recruitment by counting the number of unique Elixhauser comorbidities in the clinical record on or before recruitment. We recalculated the ECS for each participant each time a new comorbidity was identified in the clinical record. To assess change over time, we compared the proportions of participants with ECS 0, 1, and 2 or more for each year from recruitment to Year 5.

Results: We identified n=227,600 UKB participants that met our inclusion criteria. At recruitment, 47.3% (n=107,667) participants had an ECS of 0, 28.5% (n=64,882) had an ECS of 1, and 24.2% (n=55,051) had an ECS of 2 or more. By Year 5, only 35.5% (n=80,887) of participants maintained an ECS of 0 (a decrease of 11.8%). The proportion of participants with an ECS of 1 remained relatively stable between recruitment (28.5, n=64,882) and Year 5 (27.8%, n=63,280). An additional n=28,382 (36.7%) participants had ECS 2 by Year 5 compared to recruitment (24.2%, n=55,051), an absolute increase of 12.5%.

Conclusions: Even within a relatively narrow period of follow-up (5 years), we demonstrated significant changes in comorbidity status. Relying solely on measures of comorbidity / multimorbidity status at recruitment may not accurately represent the true distribution of disease at later time points.