Although the urinary tract has multiple mechanisms in place to keep it sterile, urinary tract infections (UTIs) are one of the most common bacterial infections affecting nearly 150 million people worldwide and amounting to more than $6 billion United States dollars in healthcare costs, with approximately 95% of all UTIs occurring because of periurethral contamination by enteric uropathogens.^1,2 International studies indicate that due to the high incidence of UTIs and the widespread practice of empirical treatment, the rate of antibiotic prescriptions is also increasing.³ As antibiotic use is known to be the main driver of the evolution of resistance, the increasing use of antibiotics in the treatment of UTIs has contributed to the alarming increase in antimicrobial resistance among frequently isolated urinary pathogens, thus further limiting treatment options, particularly with orally administered drugs.^2,4,5,6

Escherichia coli, Klebsiella spp., and Enterococcus spp., which are all members of the normal gut flora, are the most typically isolated organisms from urine, with E. coli being the most frequently isolated.^7,8 Empiric treatment options for UTIs are well established globally, and these include ciprofloxacin, amoxicillin/clavulanic acid, nitrofurantoin, fosfomycin, and cephalosporins.^9,10 However, studies conducted locally and internationally demonstrate that an increase in the use of most of these drugs has caused a surge in antimicrobial resistance and a concomitant increase in treatment failures.^2,11 This exerts greater pressure on an already overburdened healthcare system. Antimicrobial resistance surveillance studies thus need to be conducted regularly to identify bacterial uropathogens and their antimicrobial susceptibility patterns to guide the empiric treatment of UTIs, which is a crucial part of antimicrobial stewardship. Therefore, this study aimed to determine the prevalence of bacterial uropathogens and their antimicrobial susceptibility patterns to assist antimicrobial stewardship in a regional hospital in KwaZulu-Natal, South Africa.

From January 2018 to December 2020, the National Health Laboratory Services RK Khan Laboratory analysed 3804 urine samples from patients with suspected UTIs. After removing duplicate data and data on yeasts, data on a total of 3048 bacterial isolates were included in this study.

Escherichia coli (n = 1603; 53%) was the most frequent urinary pathogen isolated, followed by Klebsiella spp. (n = 437; 14%), including 398 (13%) Klebsiella pneumoniae and 39 (1%) Klebsiella oxytoca (Figure 1). Enterobacter spp. (n = 97; 3.2%) (including 82 [2.7%] Enterobacter cloacae and 15 [0.5%] Enterobacter aerogenes), Pseudomonas aeruginosa (n = 61; 2%), Acinetobacter baumannii (n = 84; 3%), and other Gram-negative bacilli (n = 189; 6%) made up a further 14%. Enterococcus faecalis (n = 372; 12%) was the most frequent Gram-positive organism isolated, followed by E. faecium (n = 109; 4%), Group B Streptococcus (n = 55; 2%), and Staphylococcus spp. (n = 41; 1%).

As E. coli and Klebsiella spp. made up the majority of the Gram-negative bacilli isolated, it was important to determine their antimicrobial resistance rates. The total number of cephalothin susceptibility results differed from the other antibiotics because cephalothin susceptibility was determined manually and the result was not always entered into the laboratory information system. Similarly, the total number of results was different for each antimicrobial tested due to missing results linked to random terminations that can sometimes occur during automated antimicrobial susceptibility testing with the VITEK^® 2 instrument.

E. coli demonstrated low levels of resistance to meropenem (0.56%; n = 9), amikacin (2.9%; n = 47), nitrofurantoin (6.2%; n = 99) and piperacillin/tazobactam (5.3%; n = 85) (Figure 2). A very high level of resistance was noted in recommended oral treatment options, namely trimethoprim/sulfamethoxazole (61%; n = 988), ciprofloxacin (38%; n = 605) and amoxicillin/clavulanic acid (30%; n = 478). Similarly, Klebsiella spp. strains were highly resistant to nitrofurantoin (61%; n = 268), trimethoprim/sulfamethoxazole (48%; n = 208), amoxicillin/clavulanic acid (42%; n = 185), and ciprofloxacin (30%; n = 133). Twenty-six percent of E. coli (n = 424) isolates were extended-spectrum beta-lactamase producers and 0.6% (n = 9) were carbapenem-resistant. Among the Klebsiella spp. isolates, 41.8% (n = 184) were extended-spectrum beta-lactamase producers and 5% (n = 22) were carbapenem-resistant.

Overall, Klebsiella spp. was significantly more resistant than E. coli to all the tested antimicrobials except cephalothin (E. coli – 56% vs Klebsiella spp. – 50%), ciprofloxacin (38% vs 30%), and trimethoprim/sulfamethoxazole (61% vs 48%). For all comparisons except cephalothin (p = 0.03) and ciprofloxacin (p = 0.005), the p-values were less than 0.001.

Among the Gram-positive bacteria, E. faecium isolates showed high levels of resistance to both ampicillin (83.5%; n = 91) and penicillin (83.5%; n = 91) (Table 1). Conversely, E. faecalis isolates had low rates of resistance to ampicillin (0.8%; n = 3) and penicillin (2.4%; n = 9). In addition, the Staphylococcus spp. isolates had high resistance rates to ampicillin (n = 38; 92.7%) and penicillin (n = 38; 92.7%). Vancomycin was the most effective antibiotic against E. faecium (97.2% susceptibility) and Staphylococcus spp. (100% susceptibility), while ampicillin and penicillin were most effective against Group B Streptococci (100% susceptibility).

For analysis of inpatient and outpatient data, only E. coli was studied as it constituted more than half of the bacterial pathogens. Five hundred and eleven E. coli isolates (32%) were isolated from inpatients, while 1092 (68%) were from outpatients (Table 2). Low level of resistance was observed in both inpatient and outpatient wards for meropenem (0.8%; n = 4 and 0.1%; n = 1), amikacin (3.1%; n = 16 and 2.8%; n = 31), piperacillin/tazobactam (5.5%; n = 28 and 4.7%; n = 51) and nitrofurantoin (6.3%; n = 32 and 5.5%; n = 60). However, there were high rates of resistance to trimethoprim/sulfamethoxazole (inpatients: 69.7%; n = 355, outpatients: 57.6%; n = 625), ciprofloxacin (inpatients: 34.7%; n = 177, outpatients: 39%; n = 426), and amoxicillin/clavulanic acid (inpatients: 32.9%; n = 168, outpatients: 28.4%; n = 310). The prevalence of extended-spectrum beta-lactamases and carbapenem resistance in E. coli was higher among inpatients (31.0%; n = 158 and 0.8%; n = 4) compared to outpatients (24%; n = 261 and 0.1%; n = 1).

There were no statistically significant differences in resistance rates to recommended oral antibiotics, namely amoxicillin/clavulanic acid (p = 0.07), ciprofloxacin (p = 0.10) and nitrofurantoin (p = 0.54) between the E. coli isolates from inpatients and outpatients. However, for trimethoprim/sulfamethoxazole, E. coli isolates from inpatient wards were significantly more resistant than those from outpatient wards (p < 0.001).

Of the 1603 E. coli strains isolated between 2018 and 2020, 625 (39.1%) strains were isolated in 2018, 539 (33.6%) were isolated in 2019, and 439 (27.4%) were isolated in 2020, demonstrating slight decreases in prevalence over the study period (Figure 3). Amikacin and meropenem resistance rates, while low, increased slightly between 2018 (2.4% and 0.5%) and 2020 (3.6% and 0.9%). Conversely, the E. coli isolates had higher resistance rates to the more commonly prescribed oral antibiotics, but these rates decreased steadily between 2018 and 2020 (cephalothin – 63.9% vs 49.8%; ciprofloxacin – 42.9% vs 30.4%; amoxicillin/clavulanic acid – 36.6% vs 22.4%).

Over the three years (2018–2020), there were no statistically significant differences in resistance rates to amikacin (p = 0.50), meropenem (p = 0.75), nitrofurantoin (p = 0.16), piperacillin/tazobactam (p = 0.34), and trimethoprim/sulfamethoxazole (p = 0.2). In contrast, resistance rates to amoxicillin/clavulanic acid, ceftazidime, cephalothin and ciprofloxacin decreased significantly between 2018 and 2020 (p < 0.001). The E. coli isolates also demonstrated a significant decrease in resistance to cefotaxime or ceftriaxone (p = 0.04) and gentamicin (p = 0.02).

In this study, we found that E. coli was the most frequently isolated uropathogen at the RK Khan Hospital, Durban, KwaZulu-Natal, South Africa, between 2018 and 2020, followed by Klebsiella spp. and E. faecalis. We also found that the commonly identified urinary pathogens had relatively high resistance rates to the widely administered antibiotics for UTI treatment.

Our findings are consistent with those of previous local and international studies.^13,14 A study conducted among patients with community-acquired UTIs in India showed that E. coli (55.1%), E. faecalis (15.8%), and K. pneumoniae (13.7%) were the most prevalent uropathogens isolated.¹³ Similarly, a six-year (2011–2016) study conducted in the obstetric departments of six public-sector hospitals in Durban, KwaZulu-Natal, South Africa, showed that E. coli (54.2%) was the most common uropathogen, followed by K. pneumoniae (12.9%).¹⁴

In 2010, the Infectious Diseases Society of America and the European Society of Microbiology and Infectious Disease published a clinical practice guideline outlining their recommendations for the treatment of uncomplicated UTIs, with nitrofurantoin, fosfomycin, trimethoprim/sulfamethoxazole, and ciprofloxacin recommended as the drugs of choice.⁶ Similarly, in the South African Standard Treatment Guideline, published in 2019, fosfomycin, ciprofloxacin, amoxicillin/clavulanic acid, and nitrofurantoin are also considered as empiric oral treatment options for UTIs.⁹ In their guidelines, both the Infectious Diseases Society of America and the European Society of Microbiology and Infectious Disease recommend that ciprofloxacin should be used as an empiric antibiotic therapy provided that resistance rates remain below 10%.⁶ It was therefore disconcerting to discover that, in our study, both E. coli and Klebsiella spp. showed high resistance rates against ciprofloxacin and amoxicillin/clavulanic acid. Over 30% of the E. coli and Klebsiella spp. isolates were resistant to ciprofloxacin and amoxicillin/clavulanic acid. In contrast, while only 6.2% of E. coli were resistant to nitrofurantoin, Klebsiella spp. showed a high rate (61.3%) of nitrofurantoin resistance. Although nitrofurantoin is still an option for treating UTIs caused by E. coli, our study suggests that it is ineffective against the majority of Klebsiella spp. infections, leaving physicians with limited empiric oral therapy alternatives.

Similar to our findings, in a 9-year (2011–2019) study conducted in Europe,¹⁵ resistance rates to ciprofloxacin and amoxicillin/clavulanic acid exceeded the 10% threshold for both E. coli (42.1% and 14.3%) and Klebsiella spp. (68.7% and 38.8%). Although the E. coli nitrofurantoin resistance rate (4.8%) in our study was below this 10% threshold, the resistance rate for Klebsiella spp. (46.0%) far exceeded the threshold. These high rates of resistance to recommended oral empiric therapy are of major concern as this could lead to increased treatment failure rates.

As our study was laboratory-based and the relevant information was not available on the laboratory information system, we were unable to distinguish between community-acquired and hospital-acquired UTIs. We, therefore, examined the antimicrobial patterns of E. coli isolates from the inpatient wards versus the outpatient wards. Although resistance rates were generally higher (between 1% and 12% difference) among inpatients than outpatients, these differences were not statistically significant, except for trimethoprim/sulfamethoxazole. A Japanese study done in 2020 showed that the differences in E. coli resistance rates between hospital-acquired and community-acquired UTIs were between 5% and 10%.¹⁶

According to the South African public healthcare system, new patients do not directly present to hospitals. Instead, they are initially seen at local clinics and only referred to hospitals if required.⁹ As a result, patients who are seen at hospitals are either known hospital patients with chronic diseases or patients who need a higher level of care than the primary healthcare offered in the clinics. This may explain the high resistance rates in frequently isolated urinary bacteria in our study. It is however interesting to note that in a local study conducted in KwaZulu-Natal, South Africa, between 2011 and 2016, lower levels of resistance to recommended empiric oral antimicrobials were reported among E. coli isolates from pregnant, but otherwise healthy, individuals who were generally not on antimicrobial therapy.¹⁴ Despite these rates being lower than those of our study, they are still above the empirical antibiotic resistance threshold of 10% recommended by the Infectious Diseases Society of America and the European Society of Microbiology and Infectious Disease.