The study took place at four sites in Southeast Asia: Mahosot Hospital Microbiology Laboratory affiliated with the Lao-Oxford-Mahosot Hospital-Wellcome Trust Research Unit (LOMWRU) in Vientiane, Lao PDR; the Cambodia Oxford Medical Research Unit (COMRU) in Siem Reap, Cambodia; the Shoklo Malaria Research Unit (SMRU) in Mae Sot, Thailand; and the Department of Microbiology, Chiangrai Prachanukroh Hospital (CPH), Chiangrai, Thailand. Mahosot Hospital is an ~ 400-bed government hospital providing primary to tertiary care and admitting ~ 2000 patients/month . COMRU supports the diagnostic microbiology service at Angkor Hospital for Children, an ~ 100-bed non-governmental paediatric hospital . SMRU supports several clinics located along the Thailand-Myanmar border that provide healthcare for marginalised populations . Chiangrai Prachanukroh Hospital is an ~ 800-bed government hospital providing primary to tertiary care and admitting ~ 4,000 patients/month .
Human packed red blood cells (PRBC) approaching expiry were acquired from local blood banks and spiked with a target organism (one bag per organism per site) – Escherichia coli, Haemophilus influenzae, Staphylococcus aureus, Streptococcus agalactiae and Streptococcus pneumoniae. These organisms were selected to reflect common causes of adult and paediatric BSIs. Stored clinical isolates from each site, previously identified using standard identification methods (biochemical tests [including Triple Sugar Iron, urease, oxidase, catalase, slide coagulase and Sulphide Indole Motility] and/or API and/or MALDI-TOF MS), were used to imitate real-world BSIs in preference to ATCC reference strains subjected to multiple passages that may have resulted in the selection of fitter organisms. S. pneumoniae and S. agalactiae were processed at LOMWRU, H. influenzae, S. pneumoniae and E. coli at COMRU, and S. aureus at SMRU. Due to the potential impact of the different bottles and an a priori assumption that S. pneumoniae would be the organism most susceptible to delayed sub-culture, both COMRU and LOMWRU processed this organism.
Bacterial colonies less than 24 h old from sheep / goat blood agar (chocolate agar for bottles spiked with H. influenzae) were suspended in normal saline to create a solution equivalent to a 1 McFarland standard (approximately 3 × 108 CFU/ml). This was diluted using 10-fold dilutions to approximately 300 CFU/ml. A 5-fold dilution was then made to give a concentration of ~ 60 CFU/ml. A final 1 in 10 dilution of the suspension with the PRBCs was made to obtain a final concentration of ~ 6 CFU/ml, 1 ml of which was inoculated into each of 15 BC bottles (five replicates for each of the three temperature conditions). Six CFU/ml was chosen to approximate the concentration of bacteria present in human BSIs [7, 8]. The Miles and Misra method was used to confirm the actual concentration inoculated .
Inoculated bottles were incubated in the automated detection system until they flagged positive or for at least five days, whichever occurred sooner. PEDS Plus/F bottles (Becton Dickinson, NJ, USA) were used with the automated BD BACTEC FX system at LOMWRU, and PF Plus bottles (bioMérieux, Marcy-l’Étoile, France) were used with the automated BacT/ALERT 3D system at COMRU and SMRU (Fig. 1).
When a bottle flagged positive, 2–4 drops of BC broth were sub-cultured (SC1) onto sheep / goat blood agar (chocolate agar for H. influenzae) using the streak plate technique, and 5–7 drops of broth dropped into a test tube, which was used to saturate an Amies bacterial transport swab (Copan eSwab, Brescia, Italy for LOMWRU and SMRU [and CPH for Phase Two], and ThermoFisher BactiSwab, MA, USA for COMRU). Agar plates were incubated in aerobic conditions (and with 5% CO2 for H. influenzae and S. pneumoniae) at 35 ± 2 °C for 24 h, extended to 48 h if there was no growth after 24 h. The bottles and swabs were stored at the pre-specified temperatures (2–8 °C [fridge], 22–27 °C [incubator or monitored air-conditioned room temperature, RT] and 35 ± 2 °C [incubator]) to simulate a range of feasible real-world transit temperatures. Bottles were then sub-cultured at approximately 22–26 h (SC2), 30–34 h (SC3), 46–50 h (SC4) and seven days (SC5) after removal from the machine. The Amies transport swab was sub-cultured once at 46–50 h (SC4 swab). Growth of the target organism on sub-culture was confirmed by Gram stain and colony morphology (conventional biochemicals and/or MALDI-TOF MS [VITEK MS, bioMérieux] were used in the case of uncertainty), and semi-quantified using standard techniques (No growth; Light growth = growth in quadrant 1 and < 10 colonies in quadrant 2 [± or +]; Moderate growth = growth in quadrant 2 and < 10 colonies in quadrant 3 [2+]; Heavy growth = growth in quadrant 3 ± quadrant 4 [3 + or 4+]).
Consecutive clinical BC bottles from children and adults that flagged positive on the respective automated detection systems at LOMWRU, COMRU and SMRU, and the BACTEC at CPH were included. In addition to the bottles used for Phase One, BD BACTEC Plus aerobic bottles were used with the BACTEC system at CPH and LOMWRU and BACT/ALERT FA PLUS were used with the BacT/ALERT system at SMRU. Duplicate patient samples were not excluded. Positive bottles were sub-cultured onto the appropriate agar according to local standard operating procedures (SOP) and incubated overnight as in Phase One. Approximately 5–7 drops were used to saturate an Amies bacterial transport swab as above. The bottles and swabs were stored on the bench (monitored air-conditioned room; Additional file 1: Fig. S1) and sub-cultured as in Phase One (Fig. 1). This temperature was selected balancing the results from Phase One with what might be practical in resource-constrained tropical environments. Growth on the first sub-culture was identified phenotypically using conventional biochemicals and/or MALDI-TOF MS, according to local SOPs. Sequential sub-culture growth was confirmed as the original isolate based on colonial morphology (or further testing if unclear), and growth semi-quantified as described above. Agar plates were incubated for 24 h, extended to 48 h if there was no growth after 24 h.
Ambient temperature was monitored continuously at all four sites during Phase Two using the TinyTag Transit 2 (Gemini Data Loggers, UK) at COMRU and LOMWRU, MicroLite (Fourtec, USA) at SMRU and Escort iMiniPlus (Escort Data Logging Systems, Germany) at CPH.
All sites complete media quality control following their internal quality control system. All sites participate in quality assurance programs: LOMWRU participates in the UK National External Quality Assurance Scheme; COMRU participates in the Pacific Pathology Training Centre Microbiology Quality Assurance Programme; SMRU participates in the Thailand Clinical Microbiology Proficiency Testing Scheme, Division of Proficiency Testing, Department of Medical Science, Ministry of Public Health; and CPH participates in the Thailand Medical Technology Council accreditation program. Results are reported following the MICRO Checklist .
All data were entered into Microsoft Excel (Richmond, WA). For Phase One, growth of S. pneumoniae between different sub-cultures and temperature conditions was compared using the Wilcoxon matched-pairs signed-rank test. For Phase Two, semi-quantitative growth between Streptococcus spp. and all other organisms isolated was compared using the Fisher’s exact test. P-values less than 0.05 were considered statistically significant. Analysis was performed using Stata version 16.1 (StataCorp, College Station, TX).