Association between change in alcohol use reported during routine healthcare screening and change in subsequent hospitalization: A retrospective cohort study.

BACKGROUND AND AIMS: Primary care systems often screen for unhealthy alcohol use with brief self-report tools such as the 3-item Alcohol Use Disorders Identification Test for consumption (AUDIT-C). There is little research examining whether change in alcohol use measured on the AUDIT-C captures meaningful change in outcomes affected by alcohol use. This study aimed to measure the association between change in AUDIT-C and change in all-cause hospitalization risk, measured in the year after each AUDIT-C. DESIGN: Retrospective cohort study. SETTING: Health system in the state of Washington, USA, that conducts annual screening with the AUDIT-C in outpatient care. PARTICIPANTS: Adults (n = 165 101) who had completed at least two AUDIT-Cs 11-24 months apart (2016-2020). MEASUREMENTS: AUDIT-C scores were grouped into five risk categories reflecting no drinking (0), drinking without unhealthy alcohol use [1-2 (female)/1-3 (male)] and unhealthy alcohol use with moderate risk [3-6 (female)/4-6 (male)], high risk (7-8), and very high risk (9-12). Changes in AUDIT-C were based on the number of category levels that changed (0-4). Hospitalizations were binary, reflecting one or more hospitalizations in the 365 days after each AUDIT-C, identified from insurance claims. FINDINGS: Of 165 101 eligible patients, 5.7% and 6.1% were hospitalized the year after the first and second AUDIT-C, respectively. Decreases in AUDIT-C risk category of 1 or ≥2 levels were associated with statistically significant decreases in risk of hospitalization, compared with the change in hospitalization risk for those with no change in AUDIT-C [1-level decrease: ratio of adjusted risk ratios (aRR) = 0.92, 95% confidence interval (CI) = 0.86-0.99; ≥2-level decrease: ratio of aRR = 0.68, 95% CI = 0.58-0.81]. Increases in AUDIT-C risk category of 1 or ≥2 levels were not associated with statistically significant differences in risk of hospitalization, compared with those with no change in AUDIT-C. CONCLUSIONS: A decrease in AUDIT-C score risk category is associated with a decreased risk of both all-cause hospitalizations and hospitalizations with conditions directly or potentially attributable to alcohol. An increase in AUDIT-C score does not appear to be associated with a change in risk of hospitalization in the following year.