Role of high-risk antibiotic use in incidence of health-care-associated Clostridioides difficile infection in Quebec, Canada: a population-level ecological study

Summary Background The incidence of health-care-associated Clostridioides difficile infections has been declining in the Canadian province of Quebec since 2015. We examined whether changes in high-risk antibiotic use could account for this decrease, as reported in other jurisdictions. Methods We did a retrospective interrupted time-series analysis of 12 hospitals in the Canadian province of Quebec, representing a quarter of all health-care-associated C difficile infections in this region between April 1, 2012, and March 31, 2017. Data for high-risk antibiotic use (eg, amoxicillin–clavulanate, cephalosporins, fluoroquinolones, and clindamycin) in defined daily doses (DDDs) were extracted from local surveillance databases, and incidences of health-care-associated C difficile infections were extracted from provincial surveillance databases. We used hierarchical segmented Poisson regression to assess whether variations in rates of health-care-associated C difficile infections followed variations in antibiotic use. Findings Overall, 4455 health-care-associated C difficile infections and 6 281 960 patient-days were reported in the 12 participating hospitals, representing around a quarter of the provincial data. A 50% decrease in the annual incidence of health-care-associated C difficile infections was recorded between 2012–13 and 2016–17 (9·4 infections per 10 000 patient-days vs 4·7 infections per 10 000 patient-days), and a 67% decrease in the proportion of these infections due to the NAP1/027 strain of C difficile was seen (64% in 2013 vs 21% in 2017). In total, 1 266 960 DDDs of high-risk antibiotics were distributed during the study period. An increasing time trend was noted in high-risk antibiotic use, reaching a total of 223 DDDs per 1000 patient-days in 2016–17. An increase of one DDD per 1000 patient-days was associated with a 0·2% increase in the rate of health-care-associated C difficile infections in the following 4-week period. A significant change in incidence of health-care-associated C difficile infections persisted despite adjustment for high-risk antibiotic use, as shown by a significant residual step change (0·825, 95% CI 0·731–0·932) and change in trend (0·987, 0·980–0·994). Interpretation Changes in use of high-risk antibiotics do not entirely account for the sudden decrease in health-care-associated C difficile infections in the Canadian province of Quebec since 2015. Further studies are needed to understand factors implicated in the change in epidemiology of health-care-associated C difficile infections. Funding Institut National de Sante Publique du Quebec.