Study Design and Setting
This retrospective study was conducted at the Edward Francis Small Teaching hospital (EFSTH) in Banjul, The Gambia. EFSTH is the only tertiary hospital in the country offering specialist care, dialysis and intensive care, in addition to around the clock services for laboratory requests. We extracted baseline characteristics (age, gender, and ward), bacterial findings and antimicrobial susceptibility testing (AST) in patients who had blood cultures drawn over a 12-month period from September 2022 to August 2023. Data was retrieved from paper-based records at the Department of Microbiology at EFSTH, pseudonymized and then analysed using Microsoft Excel (Version 16.83). Graphs were designed using datawrapper software (https://www.datawrapper.de/).
Study Participants
Case eligibility was any patient with a blood culture (one pair) drawn from September 2022 to August 2023 at the EFSTH. No patients were excluded. Samples were drawn in all departments of emergency and acute medicine, surgery, internal medicine, paediatrics, paediatric surgery, outpatient clinics, obstetrics and gynaecology. In general, patients 14 years or younger are defined and treated as children (< 28 days as neonates, > 28 day to 14 years as paediatric patients) while patients aged 15 years or older are treated in adult departments at EFSTH. We defined patients from neonatal intensive care unit (NICU) as neonates if information on age was not documented. Patient results from intensive care unit (ICU) were labelled as adults, if information on age was missing.
Microbiological methods
Specimen collection procedures followed established protocols for the collection of blood cultures. Typically, they were drawn prior to the administration of antibiotics or upon deterioration in a patient despite ongoing antibiotic therapy. Samples were drawn via venepunctures from new sites and preceded by sterilization of the access sites. Ideally, eight to ten millilitres of blood were collected and inoculated into a set of BD BACTEC™ PLUS culture bottles each (one aerobic and one anaerobic) for adults and for children a minimum of 0.5 ml blood was drawn to a maximum of 5 ml in one BD BACTEC™ PLUS paediatric culture bottle.
Blood culture bottles were incubated in a BD BACTEC™ FX40 incubator for 5 days. A gram stain was performed from positive bottles for preliminary identification followed by sub-culturing onto blood, Chocolate and MacConkey agar. Blood agar was prepared using 5% human blood obtained from expired blood donated for transfusions from the hospital blood bank. Sheep blood is not routinely available. Capacity for anaerobic culture were limited due to the lack of an anaerobic chamber or gas packs for producing an anaerobic atmosphere in jars.
Blood and MacConkey plates were incubated aerobically whereas Chocolate plates were incubated in 5% CO2. All were kept at 35℃ and reviewed after 24 hours when growth was compared across the plates.
Species identification
Species identification was based on the following procedures. Most gram-positive organisms will grow on blood and chocolate media, while gram-negative cocci may grow on either or both of blood and chocolate agar and gram-negative bacilli on all three plates. In case of polymicrobial growth across the various plates or growth patterns that do not conform to the above, contamination was suspected and the plates were discarded if confirmed. Phenotypic species identification was performed using manual biochemical methods. In gram-positive cocci, identified through gram stain and morphology appearance, a catalase test was performed using 1% hydrogen peroxide. If positive, the isolate was further speciated using Staphaurex™ Plus Latex Agglutination Test. This classified isolates as either Staphylococcus aureus or coagulase negative Staphylococcus. Should clinical information suggest further need, a novobiocin disc test was performed to differentiate S. epidermidis from S. saprophyticus. For catalase negative gram-positive cocci, the haemolytic pattern was considered. β-haemolytic colonies on blood agar were subjected to a bacitracin disk susceptibility test with Group A streptococci, specifically S. pyogenes, exhibiting sensitivity while Group B (S. agalactiae), Group D (including Enterococci spp.) were expected to demonstrate resistance. For resistant β-haemolytic cocci a CAMP test was done to confirm Group B streptococci. For α-haemolytic organisms an optochin disc test was performed to differentiate Streptococcus pneumoniae from other α-haemolytic Streptococci. NaCl 6.5% growth media were used to differentiate Enterococcus from non-enterococcal group D Streptococci.
Gram-negative bacilli identified as lactose fermenters from MacConkey agar were phenotypically identified using triple sugar iron (TSI), 2% urea and citrate media. In these instances identification was generally done to genera level. Should the identification process give ambiguous results or subsequent antimicrobial testing showed a multidrug resistant organism BIOMÉRIEUX API® ID 20E was used for further species identification if available. Lactose negative bacilli were screened using the oxidase test. Oxidase negative isolates would be further tested as for the lactose fermenters.
Pseudomonas spp. isolates (lactose negative) were identified through colonial morphology, gram stain and oxidase reaction (oxidase positive). Finally gram-negative coccobacilli, if isolates grew only on chocolate agar, would be screened using factor X and V on nutrient agar to speciate Haemophilus spp. Suspected Neisseria species on the other hand would be identified based on gram stain, colonial morphology, and a positive oxidase test.
Antimicrobial susceptibility testing
AST was done per Clinical and Laboratory Standards Institute (CLSI) guidelines with Kirby-Bauer disk diffusion (16). Inoculum for plating were prepared using MacFarland saline suspension of target organism. For the identification of methicillin-resistant Staphylococcus aureus (MRSA), a cefoxitin test was performed. Extended-spectrum beta-lactamase production was defined as resistance to third generation cephalosporins in pathogens of the Enterobacterales group using double disc synergy testing. Acquired antimicrobial resistance in Acinetobacter spp. or Pseudomonas spp. was defined as any additional acquired resistance beyond wild-type susceptibility (16). Antibiotic discs available were often impacted by stock outs and availability.
Isolates identified as coagulase negative Staphylococci, Bacillus spp., Micrococcus spp., Streptococcus viridans and coryneform bacteria were regarded as contaminants unless they were isolated in two or more independent blood culture specimens from the same patient within a 48-hour period, in which case they were included as a pathogen.
If bacteraemia was polymicrobial, each separate finding was noted regarding species distribution of positive cultures, with the exception of if polymicrobial growth was due to contaminants. Then they were specified as one contaminant.
Outcomes and Variables
The primary analysis focused on blood culture positivity rates and distribution of pathogens identified grouped in neonatal, paediatric or adult patients, highlighting the five most common pathogens identified. The secondary analyses highlighted resistance patterns in the five most commonly identified pathogens and investigated differences in distribution of pathogens and AMR in patients cultured in ICU vs non-ICU.
Bias
The aim of the study was to assess patterns and frequency of AMR at the EFSTH. Given its retrospective design, selection bias is expected with an overrepresentation of severely ill as well as children due to restrictive testing. Hence, to highlight the extent of bias, secondary analyses compared distribution of pathogens and AMR in ICU vs non-ICU patients as well as between age groups.
Additionally, constraints in funding and supply shortages at the hospital, the availability of kits for species identification and antibiotic discs for AST were often limited. Consequently, our analysis focus on the most commonly encountered pathogens and antibiotics. If data on AST were unavailable due to not having been performed, complete case analysis was performed. To minimize seasonal bias, we designed the study to cover a 12-months period to include seasonal variations.
Data analysis
Descriptive statistics were mainly reported. For categorical variables, counts and percentages were presented with Fischer’s exact test used for statistical comparisons. For continuous variables, median and interquartile range (IQR) were presented. The study sample size was determined by the total amount of blood cultures performed during the study period. No formal power analysis was performed a priori.
Solely complete case analysis is presented due to the lack of reliable predictors for multiple imputation to be valid. If data was missing on age or where the blood culture was drawn, age group could not be defined, why potential blood culture findings were presented in total numbers, yet not in sub-groups.