A methodical qualitative and quantitative description of the steps undertaken between August 2021 and March 2023 to establish an AMR surveillance system in the water and environment sector is provided. Several surveillance documents, guidelines, and reports were reviewed to collect data appropriately aligned to AMR surveillance in a low-resource setting [21]. The description below provides the steps (processes and methods) undertaken to set up the program.

The Government of Uganda, through the MWE, with support from the Infectious Diseases Institute at Makerere University through the Fleming Fund Country Grant 2 project, instituted a stepwise approach with incremental targets and sequential phases from August 2021. This involved establishing a foundation, consolidating and refining gains, scaling up, and further expanding the surveillance system.

A 10-member sector-specific AMR technical working group (TWG) with focal persons was established to coordinate the AMR containment efforts, including surveillance activities in the sector. The TWG developed AMR surveillance documents (plan, protocol, and standard operating procedures) focused on the monitoring of priority environmental bacteria (Escherichia coli, Klebsiella species, Enterococcus species, and Salmonella species) in water samples. These documents were aligned to the different international and national guidelines [22-25]. The National Water Quality Reference Laboratory (NWQRL) was identified as the sentinel site for AMR surveillance in the sector using conventional culture-based techniques. One sample type (water samples) was also designated for the initial AMR surveillance efforts. The TWG also reviewed the generated AMR data to inform AMR programming and policy formulation in the sector.

The NWQRL microbiology section was equipped and its infrastructure was enhanced. The human resource capacity was enhanced through in-service microbiology, biosecurity and biosafety, and laboratory quality management system trainings and mentorships. This capacity enhancement was done through a One Health approach, leveraging the established AMR surveillance capacity in the human and animal sectors. The NWQRL was also enrolled in the national laboratory external quality assessment and proficiency testing scheme of the Uganda National Health Laboratory and Diagnostic Services to enhance the microbiology testing quality.

The developed sector-specific AMR surveillance documents were pretested through an active survey. This involved the collection of samples from the Kampala-Wakiso region and analyzing them at the NWQRL. A total of 9 strategic surface water (nonpoint sources) and wastewater (point sources) sampling sites were identified in Kampala and Wakiso, and 15 grab samples were collected using the standard procedures as stipulated in the different surveillance documents. The samples were transported to the NWQRL under appropriate conditions and analyzed using standard, conventional, culture-based procedures. The lessons learned during the pretest were used to refine the surveillance documents.

The documents were then rolled out to generate AMR surveillance data through a hybrid of passive and active surveillance. The initial efforts focused on nonpoint sources or surface water sources (rivers, streams, and other open channels), drinking water (national water grid and other potable water), and point sources or wastewater (sewer, wastewater treatment plants and ponds, and septic tanks) samples. Passive surveillance used remnants of samples routinely referred from across the country to the NWQRL for water quality monitoring. Active surveillance used samples collected through quarterly targeted surveys from strategic sites (shown in the map provided in Multimedia Appendix 1) in the Kampala-Wakiso region as stipulated in the surveillance protocol. This region was chosen because it has vast human economic activities that involve the intense use of antibiotics and thus provided an ideal setting for integrated AMR surveillance in the sector.

Conventional culture-based bacteriology techniques were used for the enumeration, isolation, and identification of priority bacteria. Water sample enrichment was done by inoculating 2-5 ml of the sample into 10 ml of brain heart infusion and incubating for 16-18 hours at 37 °C. Following the enrichment, culture media and biochemical tests were used to isolate and identify the bacteria.

For E coli and Klebsiella species, MacConkey agar with crystal violet was used for the primary culture and purity plating. The distinct colonies (E coli: pink/red lactose-fermenting colonies; Klebsiella species: pink-yellow mucoid lactose-fermenting colonies) that grew after 18-24 hours of incubation at 37 °C were subjected to Gram stain (microscopy) and biochemical tests including oxidase, urea, citrate, triple sugar iron, and sulfide indole motility tests using standard procedures.

For Salmonella species, MacConkey agar with crystal violet was used for the primary culture. The distinct colonies (colorless colonies) that grew after 18-24 hours of incubation at 37 °C were purity plated on xylose lysine deoxycholate media. The distinct colonies (red colonies with black centers) that grew after 18-24 hours of incubation at 37 °C were subjected to Gram stain and biochemical tests including oxidase, urea, citrate, triple sugar iron, and sulfide indole motility tests using standard procedures. For Enterococcus species, Slanetz and Bartley agar was used for the primary culture. The distinct colonies (red or purple colonies) that grew after 18-24 hours of incubation at 37 °C were purity plated on 5% sheep blood agar media. The distinct colonies (white or gray colonies) that grew after incubation were subjected to Gram stain (microscopy) and biochemical tests including catalase and bile esculin tests.

Other than the priority bacterial isolates, we also recovered other bacteria species including Citrobacter freundii, Enterobacter, Pseudomonas, Proteus, Providencia, and Raoultella species.

Antimicrobial susceptibility testing was done using the Kirby-Bauer disc diffusion method with the appropriate isolate-antibiotic combinations and techniques as stipulated in the 31st edition of the Clinical and Laboratory Standards Institute guidelines [26]. Penicillins, fluoroquinolones, cephalosporins, aminoglycosides, carbapenems, glycopeptides, macrolides, oxazolidinones, folates (sulphonamide-trimethoprim), tetracycline, and phenicols were the antibiotic classes considered. Isolates were cryopreserved in a 20% glycerol and brain heart infusion and archived at the Uganda National Biorepository at the Central Public Health Laboratories and National Animal Diseases Diagnostics and Epidemiology Center laboratory at –80 °C.

Excel 2016 (Microsoft Corporation) and Stata 16 (StataCorp) were used for data entry, cleaning, and analysis. The percentage resistance of the isolates to each antibiotic was generated and visuals (charts and graphs) were developed. The chi-square test and binary logistic regression were used to test whether the resistance of the priority pathogens (E coli, Klebsiella, and Enterococcus species) to the different antibiotics was significantly different across the point and nonpoint sources. A P value <.05 indicated a significant statistical difference.

The work reported here is part of the Uganda National AMR surveillance program that was approved by the National AMR Sub-Committee of the National One Health Platform.

The sector-specific TWG was constituted and comprised of 10 members from the three directorates of the MWE, including the directorates of water resources management, water development, and environmental affairs. Focal persons for AMR surveillance in the sector were identified. The TWG finalized the sector surveillance documents that stipulate the target priority pathogens and sample type, designated the NWQRL as the sentinel site, designed the routine and targeted surveys, and reviewed and reported the AMR data generated to the MWE. The TWG also published a report on the AMR burden in the sector in the annual Natural Resources, Environment, Climate Change, Land and Water Management Program Performance Report 2022 [27]. Further, the TWG functioned as a linkage for the sector surveillance program to the national AMR surveillance program under the National AMR Sub-Committee of the National One Health Platform, the human and animal sector-specific AMR surveillance programs, the academia, and implementing partners.

Following the infrastructural and equipment enhancement, the NWQRL now houses a fully-fledged biosafety level 2 microbiology laboratory, the minimum level required for AMR surveillance. The laboratory has staff trained in microbiology, including antimicrobial susceptibility testing, biosafety biosecurity, AMR data management, and laboratory quality management systems. These staff have supported the laboratory analysis of environmental samples to generate AMR data.

Up to 460 samples were collected and processed at the NWQRL between August 2021 and March 2023, of which 363 (78.9%) were from passive surveillance while 97 (21.1%) were from active surveillance. Of the 460 samples, 158 (34.3%) were from point sources (wastewater samples), while 284 (61.7%) were from nonpoint sources: 108 (23.5%) ground, 95 (20.7%) surface, and 81 (17.6%) drinking water samples. Up to 328 (71.3%) samples had significant growth and yielded 602 bacterial isolates, of which, 399 (66.3%) and 203 (33.7%) were priority and nonpriority pathogens, respectively. Of the 602 isolates, 223 (37%), 166 (27.6%), 132 (21.9%), and 67 (11.1%) were from waste, surface, ground, and drinking water sources, respectively. Of the 602 isolates, 399 (66.3%) were priority pathogens: 156 (39.1%) E coli, 140 (35.1%) Klebsiella species, 96 (24.1%) Enterococcus species, and 7 (1.8%) Salmonella species. The 203 nonpriority isolates were Citrobacter species (n=57, 28.1%), Enterobacter species (n=52, 25.6%), Proteus species (n=23, 11.3%), Pseudomonas species (n=19, 9.4%), Acinetobacter species (n=8, 3.9%), and other bacterial species (n=44, 21.7%).

E coli (n=156) had a resistance of 79% to ampicillin, 55% to trimethoprim/sulfamethoxazole, 29% to ceftriaxone and ciprofloxacin, 18% to cefepime and chloramphenicol, 11% to imipenem, and 0% to amikacin and meropenem. Other antibiotic resistance is shown in Table 1.

Klebsiella species isolates (n=140) had a resistance of 100% to ampicillin, 33% to trimethoprim/sulfamethoxazole, 28% to amoxicillin-clavulanate, 27% to cefuroxime, 17% to ciprofloxacin, 2% to imipenem, and 0% to amikacin. Salmonella species isolates (n=7) had a resistance of over 50% to ampicillin, ciprofloxacin, ceftriaxone, trimethoprim/sulfamethoxazole, and tetracycline. Enterococcus species (n=96) had a resistance of 45% to vancomycin, 56% to erythromycin, 54% to tetracycline, and 52% to ciprofloxacin (Table 1).

Overall, there was no significant difference between the resistance observed in E coli and Klebsiella species isolates recovered from point and nonpoint sources. Among the Enterococcus species isolates, a significant difference (odds ratio 5.318182, 95% CI 1.793498-15.76977; P=.003) was observed in the resistance to chloramphenicol between the isolates recovered from point and nonpoint sources. The Enterococcus isolates recovered from point sources were 5 times more likely to be resistant to chloramphenicol than those recovered from nonpoint sources.

Several isolates recovered exhibited nonsusceptibility to more than one antibiotic and to two or more antibiotic classes, a phenomenon referred to as multidrug resistance (MDR) and extensive drug resistance, respectively. Up to 254 (63.7%) of the 399 priority pathogens recovered exhibited MDR or extensive drug resistance, of which 99 (39%), 85 (33.5%), 65 (25.6%), and 5 (2%) were E coli, Enterococcus species, Klebsiella species, and Salmonella species, respectively. Of the MDR E coli isolates (n=99), 40 (40%), 25 (25%), 18 (18%), 12 (12%), 1 (1%), and 3 (3%) showed resistance to 2, 3, 4, 5, 6, and 7 antibiotics from different classes, respectively. Of the MDR Enterococcus species isolates (n=85), 27 (32%), 34 (40%), 16 (19%), and 8 (9%) showed resistance to 2, 3, 4, and 5 antibiotics from different classes, respectively. Of the MDR Klebsiella species isolates (n=65), 36 (55%), 20 (32%), 2 (3%), 6 (9%), and 1 (2%) showed resistance to 2, 3, 4, 5, and 7 antibiotics from different classes, respectively (Table 2).