The study was conducted in the Bacteriology-Virology laboratory of the Department of Medical Biology at Dalal Jamm National Teaching Hospital, located at Ancien Parcours du Golf, Guédiawaye (Dakar region, Senegal). This teaching hospital comprises a general intensive care unit, clinical haematology, nephrology and haemodialysis, cardiology and infectious diseases departments, as well as several paediatric units (neonatology, general paediatrics, paediatric oncology and paediatric surgery). The laboratory provides diagnostic bacteriology services for all of these departments.

This was a retrospective descriptive study covering the period from 1 January to 31 December 2024. All patients admitted to Dalal Jamm National Teaching Hospital, irrespective of age or sex, from whom at least one blood culture was drawn during the study period and in whom a KES-group organism was isolated, were included. Specimens other than blood cultures were excluded. Blood cultures whose isolate was judged to be a contaminant according to routine laboratory criteria (commensal skin flora recovered from a single bottle in the absence of supporting clinical evidence) were not counted in the denominator of significant isolates.

Units of analysis were defined a priori and are used consistently throughout. A blood culture bottle is a single vial; a blood culture set is the group of bottles filled at one venepuncture; a blood culture request (the unit recorded by the laboratory information system) covers the set or sets received under one accession number, and is what the laboratory reports as a single positive or negative culture. An isolate is one organism recovered from one request. A blood culture episode was defined as all requests from the same patient within a 14-day window, the first positive request marking the start of the episode; a new positive request beyond that window was counted as a new episode.

Because the laboratory information system in use over the study period records an accession number rather than a unique patient identifier, de-duplication could be performed reliably at the level of the request but not at the level of the patient. Requests recorded twice under the same accession number were merged before analysis. Patient and episode characteristics are therefore reported per episode, and microbiological results per isolate; the two denominators differ because a single episode may yield more than one KES organism. The number of distinct patients is reported as a range, bounded below by grouping episodes that shared an identical requesting ward, age and sex within 14 days.

Blood was drawn at the bedside under aseptic conditions, distributed into adult or paediatric blood culture bottles, and incubated in an automated system (BACTEC or BACT/ALERT) at 35 - 37˚C under aerobic conditions. Any detected growth prompted microscopic examination after Gram staining and subculture onto chocolate agar, MacConkey agar and EMB agar.

Bacterial identification was performed using API 20E biochemical strips or the automated Vitek 2 system (GN cards). Antimicrobial susceptibility testing was carried out on Vitek 2 (AST-N cards), supplemented where appropriate by disc diffusion. Results were categorised as susceptible (S), intermediate (I) or resistant (R) according to the prevailing recommendations of the Antibiogram Committee of the French Society for Microbiology (CA-SFM/EUCAST) [15]. The agents actually reported over the study period were: aminopenicillins (amoxicillin, ampicillin); amoxicillin-clavulanate; ticarcillin and ticarcillin-clavulanate; piperacillin and piperacillin-tazobactam; the cephamycin cefoxitin; third-generation cephalosporins (cefotaxime, ceftriaxone, ceftazidime, cefixime); the fourth-generation cephalosporin cefepime; the monobactam aztreonam; carbapenems (imipenem, meropenem, ertapenem); aminoglycosides (gentamicin, tobramycin, amikacin, kanamycin); fluoroquinolones (ciprofloxacin, levofloxacin, ofloxacin, norfloxacin, pefloxacin, moxifloxacin); co-trimoxazole and trimethoprim; fosfomycin; nitrofurantoin; tetracyclines (doxycycline, tetracycline); and chloramphenicol. Panels were not uniform across the study period, so the set of agents tested varies between isolates. Time to positivity was recorded by the automated system as the number of days elapsed between bottle loading and detection of bacterial growth; a value of 0 or 1 day was taken as positivity within 24 hours and a value of 2 days as positivity between 25 and 48 hours.

Data were extracted from the laboratory information system (FileMaker Pro version 21) and exported to a spreadsheet for analysis. The variables collected were: anonymised specimen identifier, age, sex, requesting department, clinical indication for the request, sampling date, time to positivity, isolated species and detailed susceptibility results.

Results were analysed by individual antimicrobial agent, by antimicrobial class and by species. In line with standard practice in resistance epidemiology, the intermediate and resistant categories were combined under the term non-susceptibility (non-S), and percentages were calculated using the number of isolates actually tested for each agent or class as the denominator. For class-level results an isolate was considered non-susceptible as soon as at least one agent of that class was categorised as I or R. This convention is deliberately conservative and, where the agents of a class do not behave uniformly, it overstates resistance to the class as a whole; class-level results are therefore accompanied by the corresponding agent-level results, and the proportion of isolates showing discordant categorisations within a class is reported. Categorical variables are expressed as counts and percentages, and continuous variables as median and interquartile range (IQR). Denominators are stated for every proportion, and isolates for which a given agent or variable was not documented are excluded from that denominator and reported separately rather than counted as susceptible.

β-lactam resistance phenotypes were inferred from antibiogram profiles using the following hierarchical algorithm, applied in the order shown:

carbapenem non-susceptibility: at least one carbapenem (imipenem, meropenem or ertapenem) categorised as I or R; this category takes precedence over the others;ESBL-type phenotype: non-susceptibility to at least one third-generation cephalosporin, with a cefoxitin result of S and no carbapenem non-susceptibility;AmpC-type phenotype (± associated ESBL): non-susceptibility to at least one third-generation cephalosporin, with a cefoxitin result of I or R and no carbapenem non-susceptibility;wild-type phenotype: susceptibility preserved to all third-generation cephalosporins tested;not classifiable: non-susceptibility to at least one third-generation cephalosporin without a documented cefoxitin result, or no third-generation cephalosporin tested. Isolates in this category were reported separately and were not redistributed among the other phenotypes.

These phenotypes are reported as probable, given the absence of systematic synergy testing and of molecular confirmation.

Because Enterobacter spp. and Serratia spp. possess an inducible chromosomal AmpC cephalosporinase, cefoxitin non-susceptibility in these genera does not discriminate between a derepressed cephalosporinase and an associated ESBL. The ESBL-type and AmpC-type labels are therefore mechanistically informative only for K. pneumoniae, which lacks a chromosomal AmpC, and are reported for the other two genera as a descriptive summary of the antibiogram.

Multidrug resistance (MDR) was defined according to Magiorakos et al. as non-susceptibility to at least one agent in at least three antimicrobial categories, after exclusion of species-specific intrinsic resistance (aminopenicillins for K. pneumoniae; aminopenicillins, penicillin-inhibitor combinations and cephamycins for Enterobacter spp.; with the addition of polymyxins, nitrofurantoin and tetracyclines for Serratia spp.) [16][17].

The study relied exclusively on the retrospective use of laboratory data generated during routine care, with no additional intervention on patients. Data were anonymised before analysis and handled confidentially. The protocol was submitted to and authorised by the management of Dalal Jamm National Teaching Hospital. Given the retrospective and anonymised nature of the study, individual informed consent was not required.

During 2024, the Bacteriology-Virology laboratory of Dalal Jamm National Teaching Hospital processed 1844 blood cultures, of which 694 (37.6%) were positive and 1150 (62.4%) negatives. After exclusion of 105 isolates considered to be contaminants (15.1% of positive blood cultures), 589 isolates were retained as significant: 386 Gram-negative bacilli (65.5%), 201 Gram-positive cocci (34.1%) and 2 yeasts (0.3%).

KES-group Enterobacterales accounted for 117 isolates, that is 30.3% of Gram-negative bacilli and 19.9% of all significant isolates. Four of these isolates had no analysable antibiogram and were excluded, leaving 113 isolates that constituted the study population. These 113 isolates were recovered from 110 blood culture episodes: three episodes yielded two distinct KES organisms each, and a further three episodes yielded a non-KES co-isolate (Acinetobacter baumannii, Escherichia coli and Sphingomonaspaucimobilis, one each) that was not analysed here. Because the laboratory information system does not carry a unique patient identifier, episodes were de-duplicated at the level of the blood culture request; six groups of episodes sharing an identical ward, age and sex fell within 14 days of one another, so the 110 episodes correspond to at least 103 and at most 110 distinct patients (Figure 1).

Figure 1. Flow diagram of blood cultures processed in the Bacteriology-Virology laboratory of Dalal Jamm National Teaching Hospital, January-December 2024. KES: Klebsiella-Enterobacter-Serratia group; GNB: Gram-negative bacilli. Percentages of Gram-negative bacilli and of significant isolates refer to the 117 KES isolates identified.

Characteristics of the 110 blood culture episodes and of the 113 isolates they yielded are presented in Table 1. A slight female predominance was observed (59 episodes in female patients; 53.6%), giving a male-to-female ratio of 0.86. Median age was 20 years (IQR 0.4 - 50; range 1 day - 96 years). Children aged 10 years or younger accounted for 40.9% of episodes (n = 45), including 28 infants under one year of age (25.5% of the total); patients older than 60 years accounted for 12.7% of episodes.

Paediatric units taken together (neonatology, general paediatrics and paediatric oncology) accounted for 43 episodes, that is 39.1% of cases, with neonatology alone contributing 23 episodes (20.9%). Next came the intensive care unit (n = 16; 14.5%), clinical haematology (n = 14; 12.7%), general paediatrics (n = 12; 10.9%), cardiology (n = 10; 9.1%) and nephrology (n = 9; 8.2%). Febrile infectious syndrome was the leading reason for requesting a blood culture, followed by neonatal infections and febrile episodes occurring in haematology patients.

Time to positivity was documented for 106 of the 110 episodes (96.4%). Among these, the median was 1 day (IQR 1 - 2; range 0 - 7), and 89 of 106 bottles (84.0%) flagged positive within the first 48 hours of incubation.

Table 1. Characteristics of patients and of KES-group isolates recovered from blood cultures (n = 113) in 2024.

a. Dermatology, gynaecology and orthopaedics (1 case each). b. Percentages calculated on the 106 episodes for which time to positivity was documented; the 4 episodes without a documented value are shown separately and are not included in the denominator. Sex, age and requesting department are reported per episode (n = 110); the species distribution is reported per isolate (n = 113), since three episodes yielded two KES isolates each.

Enterobacter spp. was the most frequently isolated group (n = 53; 46.9%), very closely followed by Klebsiella pneumoniae (n = 51; 45.1%) and then by Serratia spp. (n = 9; 8.0%). Together, the first two accounted for 92% of isolates in the group. Identification was resolved to species level for only 10 of the 53 Enterobacter isolates (18.9%), all reported as Enterobacter cloacae; the remaining 43 (81.1%) were reported at genus level as Enterobacter spp., and no isolate was assigned to a species complex. Among the Serratia isolates, 8 were Serratia marcescens and 1 was Serratiaplymuthica.

Non-susceptibility rates by antimicrobial class and species are detailed in Table 2 and illustrated in Figure 2.

Non-susceptibility to third-generation cephalosporins was very high in K. pneumoniae (48/51; 94.1%) and frequent in Enterobacter spp. (35/53; 66.0%), whereas it remained lower in Serratia spp. (3/9; 33.3%). These class-level figures were robust: among the 100 isolates for which at least two third-generation cephalosporins were tested, only 4 (4.0%) gave discordant categorisations, and non-susceptibility rested on a resistant rather than an intermediate result in 84 of the 86 non-susceptible isolates. Cefepime offered no decisive advantage in K. pneumoniae (20 of 22 isolates non-susceptible; 90.9%). Non-susceptibility to cefoxitin clearly separated the two main groups: 85.7% in Enterobacter spp. (30/35) versus 20.7% in K. pneumoniae (6/29), consistent with expression of a chromosomal cephalosporinase in the former. Piperacillin-tazobactam was inactive against 70.3% of K. pneumoniae and 60.0% of Enterobacter spp.

Carbapenems remained the most active β-lactams, although non-susceptibility was found in 21 of 108 tested isolates (19.4%), distributed comparably across K. pneumoniae (10/49; 20.4%), Enterobacter spp. (9/50; 18.0%) and Serratia spp. (2/9; 22.2%). This figure should be read with caution: 14 of the 21 non-susceptible isolates were classified on the basis of a single carbapenem, in 6 cases ertapenem alone, and 4 were categorised intermediate rather than resistant. Only 2 of the 44 isolates tested against at least two carbapenems gave discordant results.

Among non-β-lactam agents, class-level non-susceptibility to fluoroquinolones reached 85.7% in K. pneumoniae (42/49) versus 41.5% in Enterobacter spp. (22/53), and class-level non-susceptibility to aminoglycosides 69.4% (34/49) and 55.1% (27/49) respectively. The aminoglycoside figure is, however, the clearest illustration of the limits of a class-level rule: 38 of the 77 isolates tested against at least two aminoglycosides (49.4%) gave discordant results within the class. Agent by agent, amikacin was non-susceptible in none of the 28 K. pneumoniae isolates tested (0%) and in only 4 of 31 Enterobacter isolates (12.9%), whereas gentamicin was non-susceptible in 25/39 K. pneumoniae (64.1%) and 21/36 Enterobacter (58.3%), and tobramycin in 21/28 (75.0%) and 15/25 (60.0%) respectively. Co-trimoxazole was inactive against 85.7% of K. pneumoniae (24/28). Oral alternatives offered little fallback: non-susceptibility to fosfomycin exceeded 77% in both main groups, and non-susceptibility to tetracyclines exceeded 63%. Nitrofurantoin retained activity against a majority of K. pneumoniae isolates (10 of 17 susceptible), without clinical relevance in the context of systemic infection.

Agent-level results for the classes whose members did not behave uniformly are given in the lower part of Table 2.

Table 2. Non-susceptibility (intermediate and resistant categories combined) of KES-group isolates by antimicrobial class and species.

Results are expressed as number of non-susceptible isolates/number of isolates tested (percentage). An isolate is considered non-susceptible to a class as soon as at least one agent of that class is categorised as I or R; agent-level results are given for the classes in which the agents did not behave uniformly. Denominators differ between rows because susceptibility panels were not uniform over the study period. -: no isolate tested.

Figure 2.Non-susceptibility rates to the main antimicrobial classes by species, KES-group blood culture isolates (n = 113). 3GC: third-generation cephalosporins. Denominators vary according to the agents tested (see Table 2); combinations tested on fewer than five isolates are shown as a fraction rather than as a percentage.

The distribution of β-lactam resistance phenotypes is presented in Table 3 and Figure 3. In K. pneumoniae, 18 isolates (35.3%) met the criteria for an ESBL-type phenotype and 2 (3.9%) those for an AmpC-type phenotype ± ESBL; 10 isolates (19.6%) were carbapenem non-susceptible and 3 (5.9%) were wild type. The remaining 18 isolates (35.3%) could not be classified because cefoxitin had not been tested, even though all 18 were non-susceptible to third-generation cephalosporins. Restricted to the 20 K. pneumoniae isolates for which the ESBL/AmpC distinction could actually be made, an ESBL-type phenotype accounted for 18 (90.0%). In Enterobacter spp., the distribution was more balanced: 17 wild-type isolates (32.1%), 12 AmpC-type phenotypes ± ESBL (22.6%), 11 not classifiable for want of a cefoxitin result (20.8%), 9 carbapenem non-susceptible isolates (17.0%) and 4 ESBL-type phenotypes (7.5%). Among Serratia spp., 5 of 9 isolates (55.6%) displayed a wild-type phenotype, 2 were carbapenem non-susceptible and 2 were not classifiable. Overall, cefoxitin was not tested in 20 of the 48 third-generation-cephalosporin non-susceptible K. pneumoniae isolates (41.7%), in 13 of 35 such Enterobacter isolates (37.1%) and in all 3 such Serratia isolates.

Table 3. Distribution of probable β-lactam resistance phenotypes by species, n (%).

ESBL: extended-spectrum β-lactamase. Phenotypes were assigned by the hierarchical algorithm defined in the Methods, in which carbapenem non-susceptibility takes precedence and an ESBL-type or AmpC-type assignment requires a documented cefoxitin result. a. Isolates non-susceptible to at least one third-generation cephalosporin, not carbapenem non-susceptible, for which cefoxitin was not tested. In Enterobacter spp. and Serratia spp., which possess an inducible chromosomal AmpC cephalosporinase, cefoxitin non-susceptibility does not distinguish a derepressed cephalosporinase from an associated ESBL; the phenotypes shown for these genera should be read as descriptive only.

Figure 3.Distribution of probable β-lactam resistance phenotypes by species. Carbapenem non-S: non-susceptibility to at least one carbapenem. Not classifiable: cefoxitin not tested.

The proportion of multidrug-resistant isolates as defined by Magiorakos et al. was 94.1% in K. pneumoniae (48/51), 69.8% in Enterobacter spp. (37/53) and 22.2% in Serratia spp. (2/9), amounting to 77.0% of all KES-group isolates (87/113) (Table 4).

Table 4.Frequency of multidrug resistance among KES-group blood culture isolates.