Effects of health system limitations on the use of blood culture and sensitivity testing in Kenyan county hospitals: an interview-based qualitative study using causal loop diagrams.

BACKGROUND

There is evidence of rapidly growing resistance to antibiotics across Africa. We aimed to establish whether blood culture and sensitivity (BCS) testing is a feasible component of the response to antimicrobial resistance (AMR) in large Kenyan hospitals.

METHODS

We used a qualitative study design and conducted key informant interviews (KIIs) using iteratively developed, semi-structured interviews with purposively sampled health-care workers (HCWs) within a network of facilities in Kenya called the Clinical Information Network. Only hospital laboratories that either reported fewer than 50 BCS tests in the previous 6 months or had not recorded use of BCS tests for the past 6 months were included in this study. This selection was further limited by considerations of timing, logistics, and data saturation. Our purposive selection of interviewees was guided by the level of expertise, profession, the number of key informants per hospital, and existing relations with the hospital staff. Each interview took an average of 45 min. Our thematic analysis used inductive coding to identify key themes, and we used causal loop diagrams (CLDs) to explain interactions between themes. The CLDs illustrate how health system issues relate to each other and influence the use of BCS testing in these study settings.

FINDINGS

We conducted 72 KIIs across eight tertiary-level Kenyan hospitals between Oct 27 and Dec 2, 2021. Of the 72 HCWs interviewed, 33 (46%) were women and 39 (54%) were men. The participants consisted of 24 clinicians, 26 laboratory staff, and 22 pharmacists with a median age of 36 years (IQR 31-40). We found that the main issues that led to insufficient use of BCS testing in these hospitals related to demand and supply. A host of issues influence clinician demand for BCS testing, including: the use and uptake of BCS, normalisation of clinical diagnosis, unaffordability of the BCS test, turnaround time of the BCS test, preferential use of alternative biochemistry and haematology tests by clinicians, diagnosis by malaria confirmation, and negative results discouraging clinicians from ordering this test. Similarly, health system logistics or supply issues, including laboratory capacity, support and goodwill from hospital management, and scarcity of refresher training opportunities, hinder the availability and sustenance of BCS testing. The CLDs show that these multiple factors can create mutually reinforcing feedback loops that undermine efforts to provide BCS testing in hospitals.

INTERPRETATION

The findings suggest that sustainable and routine provision of BCS testing would require many issues to be targeted simultaneously and continuously at the health system level, which is unlikely to be feasible in the short term for Kenyan hospitals. Therefore, in these settings, alternatives to routine BCS testing-such as the adoption of a targeted or vertical approach and the use of survey-informed antimicrobial stewardship to inform local treatment guidelines-should be considered for the control of AMR.

FUNDING

Wellcome Trust.