Background: Manual blood culture bottles (BCBs) are frequently used in low-resource settings. There are few BCB performance evaluations, especially evaluations comparing them with automated systems. We evaluated two manual BCBs (Bi-State BCB and BacT/ALERT BCB) and compared their yield and time to growth detection with those of automated BacT/ALERT system. Methods: BCBs were spiked in triplicate with 177 clinical isolates representing pathogens common in low-resource settings (19 bacterial and one yeast species) in adult and paediatric volumes, resulting in 1056 spiked BCBs per BCB system. Growth in manual BCBs was evaluated daily by visually inspecting the broth, agar slant, and, for BacT/ALERT BCB, colour change of the growth indicator. The primary outcomes were BCB yield (proportion of spiked BCB showing growth) and time to detection (proportion of positive BCB with growth detected on day 1 of incubation). 95% CI for yield and growth on day 1 were calculated using bootstrap method for clustered data using. Secondary outcomes were time to colony for all BCBs (defined as number of days between incubation and colony growth sufficient to use for further testing) and difference between time to detection in broth and on agar slant for the Bi-State BCBs. Findings: Overall yield was 95·9% (95% CI 93·9–98·0) for Bi-State BCB and 95·5% (93·3–97·8) for manual BacT/ALERT, versus 96·1% (94·0–98·1) for the automated BacT/ALERT system (p=0·61). Day 1 growth was present in 920 (90·8%) of 1013 positive Bi-State BCB and 757 (75·0%) of 1009 positive manual BacT/ALERT BCB, versus 1008 (99·3%) of 1015 automated bottles. On day 2, detection rates were 100% for BI-State BCB, 97·7% for manual BacT/ALERT BCB, and 100% for automated bottles. For Bi-State BCB, growth mostly occurred simultaneously in broth and slant (81·7%). Sufficient colony growth on the slant to perform further tests was present in only 44·1% of biphasic bottles on day 2 and 59·0% on day 3. Interpretation: The yield of manual BCB was comparable with the automated system, suggesting that manual blood culture systems are an acceptable alternative to automated systems in low-resource settings. Bi-State BCB outperformed manual BacT/ALERT bottles, but the agar slant did not allow earlier detection nor earlier colony growth. Time to detection for manual blood culture systems still lags that of automated systems, and research into innovative and affordable methods of growth detection in manual BCBs is encouraged. Funding: Médecins Sans Frontières and Department of Economy, Science and Innovation of the Flemish Government.
The Lancet Microbe
e124 - e132