Fig. 1
Variation in adjusted 36-month ASH rates according to GP and practice characteristics, entire samplea, b. a GP- and practice-level variables separately added to the full base model for the entire sample that includes group assignment in the BRIGHT trial, number of ASH events in the 18-month period prior to baseline, age, ethnicity, marital status, living arrangement, deprivation decile of participant’s home address, education, number of health problems, number of medications, Abbreviated Mental Test Score (cognition), Geriatric Depression Scale (GDS-15) score (depression), social support score, physical activity, alcohol consumption, frequency of alcohol intake, and nutritional risk score. b Estimates < 0.6 and > 1.8 are marked as ▼and ▲, respectively