Multimorbidity measures from health administrative data using ICD system codes: A systematic review

Abstract

Background: We aimed to identify and characterize adult population-based multimorbidity measures using health administrative data and the International Classification of Diseases (ICD) codes for disease identification.Methods: We performed a narrative systematic review of studies using or describing development or validation of multimorbidity measures. We compared the number of diseases included in the measures, the process of data extraction (case definition) and the validation process. We assessed the methodological robustness using eight criteria, five based on general criteria for indicators (AIRE instrument) and three multimorbidity-specific criteria.Results: Twenty-two multimorbidity measures were identified. The number of diseases they included ranged from 5 to 84 (median = 20), with 19 measures including both physical and mental conditions. Diseases were identified using ICD codes extracted from inpatient and outpatient data (18/22) and sometimes including drug claims (10/22). The validation process relied mainly on the capacity of the measures to predict health outcome (5/22), or on the validation of each individual disease against a gold standard (8/22). Six multimorbidity measures met at least six of the eight robustness criteria assessed. Conclusion:There is significant heterogeneity among the measures used to assess multimorbidity in administrative databases, and about a third are of low to moderate quality. A more consensual approach to the number of diseases or groups of diseases included in multimorbidity measures may improve comparison between regions, and potentially provide better control for multimorbidity-related confounding in studies.

The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: A cross-sectional study using a large, primary care population dataset

“…Only 8 conditions (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure) were considered in at least half of the studies, and a quarter of studies did not consider any mental health condition. Simard and colleagues [ 23 ] reviewed existing literature to examine how studies used, developed, and validated methods for measuring multimorbidity. They found heterogeneity in the grouping of conditions, validation processes, number of ICD-10 code digits used to define included conditions, and use of additional data sources.…”

The impact of varying the number and selection of conditions on estimated multimorbidity prevalence: a cross-sectional study using a large, primary care population dataset

“…Only eight conditions (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure) were considered in at least half of the studies, and a quarter of studies did not consider any mental health condition. Simard et al [23] reviewed existing literature to examine how studies used, developed, and validated methods for measuring multimorbidity. They found heterogeneity in the grouping of conditions, validation processes, number of ICD-10 code digits used to define included conditions, and use of additional data sources.…”

Multimorbidity patterns across race/ethnicity as stratified by age and obesity

“…Our analysis includes multimorbidities based on prevalent codes as prevalence-based selection of ICD-10-CM codes in multimorbidity research has been shown to be robust 34 . Patients were classified with obesity if they had an average BMI of 30+ during the study period, and without obesity if their BMI was less than this cutoff.…”