The role of primary care data in multimorbidity estimates using administrative health data

BACKGROUND: This study aims to investigate the extent to which two frequently used multimorbidity measures, the Charlson Comorbidity Index (CCI) and the 2+ count of chronic diseases, are underestimated when primary care data are absent, and to examine if this varies by sociodemographic factors. METHODS: This register-based study included adults, aged 60 years and older, living in Stockholm, Sweden from 2017-2022 (n=497736). Multimorbidity was operationalized as 1) 2+ chronic diseases from a list of 60 and 2) CCI score of 2+. The multimorbidity prevalences were calculated using only outpatient specialist and hospital data, then after adding primary care data. The proportion of the prevalence underestimated without primary care data was estimated for both measures. Stratified analyses were conducted to assess the underestimation by age, sex, education, cohabitation, and care status. FINDINGS: Without primary care data, approximately 20% of individuals with multimorbidity were missed (2+count: 19.4%,95%CI:19.3%-19.5%; 2+CCI: 23.9%,95%CI:23.8%-24.0%). The proportion of missed cases was highest among community-dwelling individuals (2+count: 21.0%,95%CI:20.8%-21.1%; 2+CCI: 25.5%,95%CI:25.3%-25.6%) and decreased with greater social care need (home care: 2+count: 4.33%,95%CI:4.09%-4.58%; 2+CCI: 16.0%,95%CI:15.5%-16.4%; nursing homes: 2+count: 7.78%,95%CI:7.35%-8.22%, 2+CCI: 18.0%,95%CI:17.3%-18.6%). The underestimation without primary care was greater among younger age groups for the 2+count, and women and older individuals for the 2+CCI. INTERPRETATION: Multimorbidity is underestimated without primary care data, particularly among community-dwelling and independent older adults, as well as the youngest and oldest segments of the older population. Caution is warranted when comparing multimorbidity across age groups, care status, or social factors in the absence of primary care data.