Implementation of Automated Blood Culture With Quality Assurance in a Resource-Limited Setting

• Published in: Frontiers in medicine 2021-05, Vol.8, p.627513-627513
• Main Authors: von Laer, Anja, N'Guessan, Micheline Ahou, Touré, Fidèle Sounan, Nowak, Kathrin, Groeschner, Karin, Ignatius, Ralf, Friesen, Johannes, Tomczyk, Sara, Leendertz, Fabian H., Eckmanns, Tim, Akoua-Koffi, Chantal
• Format: Article
• Language: English
• Subjects: bacterial infection; blood culture; laboratory automation; Medicine; quality control; sub-Saharan Africa
• ISSN: 2296-858X; 2296-858X
• DOI: 10.3389/fmed.2021.627513

Background: Blood cultures (BC) have a high clinical relevance and are a priority specimen for surveillance of antimicrobial resistance. Manual BC are still most frequently used in resource-limited settings. Data on automated BC performance in Africa are scarce. We implemented automated BC at a surveillance site of the African Network for improved Diagnostics, Epidemiology and Management of Common Infectious Agents (ANDEMIA). Methods: Between June 2017 and January 2018, pairs of automated BC (BacT/ALERT ® FA Plus) and manual BC (brain-heart infusion broth) were compared at a University hospital in Bouaké, Côte d'Ivoire. BC were inoculated each with a target blood volume of 10 ml from the same venipuncture. Automated BC were incubated for up to 5 days, manual BC for up to 10 days. Terminal subcultures were performed for manual BC only. The two systems were compared regarding yield, contamination, and turnaround time. For quality assurance, isolates were retested in a German routine microbiological laboratory. Results: BC sampling was increased from on average 24 BC to 63 BC per month. A total of 337 matched pairs of BC were included. Automated BC was positive in 36.5%, manual BC in 24.0% ( p -value < 0.01), proportion of contamination was 47.9 and 43.8%, respectively ( p -value = 1.0). Turnaround time of positive BC was shortened by 2.5 days with automated compared to manual BC ( p < 0.01). Most common detected pathogens in both systems were Klebsiella spp. (26.0%) and Staphylococcus aureus (18.2%). Most contaminants were members of the skin flora. Retesting of 162 isolates was concordant in 79.6% on family level. Conclusions: Implementing automated BC in a resource-limited setting is possible and improves microbiological diagnostic performance. Automated BC increased yield and shortened turnaround times. Regular training and mentorship of clinicians has to be intensified to increase number and quality of BC. Pre-analytical training to improve diagnostic stewardship is essential when implementing a new microbiological method. Retesting highlighted that manual identification and antimicrobial susceptibility testing can be of good quality and sustainable. The implementation of automated tools should be decided individually according to economic considerations, number of samples, stable supply chain of consumables, and technical sustainability.