Antibiotic prescribing for care-home residents: a population-based, cross-classified multilevel analysis in Scotland, UK.

Background: There is wide variation in antibiotic prescribing across care-homes for older people, with implications for resident outcomes and antimicrobial resistance.

Objective: To quantify variation in antibiotic prescribing and associations with resident, care-home and general practice characteristics.

Design: Population-based analyses using administrative data.

Setting and subjects: 148 care-homes in two Scottish regions, with 6633 residents registered with 139 general practices.

Methods: Prescriptions for any antibiotic and for broad-spectrum antibiotics between 1 April 2016 and 31 March 2017 were analysed using cross-classified multilevel negative binomial regression.

Results: For any antibiotics, the mean prescription rate was 6.61 (SD 3.06) per 1000 resident bed-days (RBD). In multivariate analysis, prescribing was associated with resident age [incidence rate ratio (IRR) 1.30 [95% confidence interval 1.19 to 1.41] for 90+ versus <80 years old] and comorbidity (1.88 [1.71 to 2.06] for Charlson Comorbidity Index 3+ versus 0), and the care-home's sampling rate for microbiological culture (1.53 [1.28 to 1.84] for >7 versus <3.5 samples per 1000 RBD), with residual unexplained variation between care-homes (median IRR 1.29 [1.23 to 1.36]) and general practices (1.11 [1.05 to 1.18]). For broad-spectrum antibiotics, the mean rate was 0.98 (0.92) per 1000 RBD. Broad-spectrum prescribing was also associated with resident age, sex, comorbidity and sampling rate, with larger residual unexplained variation between care-homes (1.56 [1.36 to 1.77]) and general practices (1.51 [1.31 to 1.72]).

Conclusion: Variation in prescribing was influenced by resident case-mix, but there is significant unexplained variation between care-homes and between general practices, indicating a need for antibiotic stewardship to target both.