The prevalence and clinical context of antimicrobial resistance amongst medical inpatients at a referral hospital in Rwanda: a cohort study

Antimicrobial resistance (AMR) is a growing global concern. AMR surveillance is a crucial component of the international response; however, passive surveillance of laboratory data is limited without corresponding patient-level clinical data. This study sought to examine the burden of AMR amongst medical inpatients in Rwanda, in the context of their clinical presentations and prior antibiotic exposures. This cohort study was conducted over a 9-month period at a tertiary referral hospital in Kigali, Rwanda. We enrolled 122 adult medical inpatients with a history of fever and a positive microbiological culture result. Data were collected regarding the clinical and microbiological aspects of their admission. The most common diagnoses were urinary tract infection (n = 36, 30%), followed by pneumonia (n = 30, 25%) and bacteraemia (11 primary [9%] and 10 catheter-related [8%]). The most common pathogens were E. coli (n = 40, 33%) and Klebsiella pneumoniae (n = 36, 30%). The cohort were heavily antibiotic-exposed at the time of culture with 98% of patients (n = 119) having received an antibiotic prior to culture, with a median exposure of 3 days (IQR 2–4 days). Eighty patients (66%) were specifically prescribed ceftriaxone at the time of culture. Gram-negative organisms predominated (82% [100/122]) and exhibited high rates of resistance, with only 27% (21/77) being susceptible to ceftriaxone, 2.4% (2/82) susceptible to co-amoxiclav and 44% (8/18) susceptible to ciprofloxacin. Susceptibility amongst Gram-negatives was relatively preserved to amikacin (91%, 79/87) and imipenem (85%, 70/82). There were no cases of methicillin-resistant Staphylococcus aureus (0/12) or vancomycin-resistant enterococci (0/2). Discordant antibiotic therapy was significantly associated with in-hospital mortality (OR 6.87, 95%CI 1.80–45.1, p = 0.014). This cohort highlights high rates of resistance amongst Gram-negative organisms in Rwanda, including the presence of carbapenem resistance. Nonetheless, the detailed prescribing data also highlight the challenges of using routine laboratory data to infer broader AMR prevalence. The significant exposure to empiric broad-spectrum antibiotic therapy prior to culturing introduces a selection bias and risks over-estimating the burden of resistant organisms. Broadening access to microbiological services and active surveillance outside of teaching hospitals are essential to support national and international efforts to curb the growth of AMR in low-resource settings.