INTRODUCTION
Microbial resistance poses a growing challenge to South Africa's health system that is compounded by overuse and inappropriate prescription of antibiotics in the paediatric population.(1) This problem is illustrated in paediatric urinary tract infection (UTI) where there is an expansion of resistant organisms arising from the community. Presently, there are no published data on paediatric UTI organism's antibiotic profiles in South Africa. Data trends on uropathogens have shown the emergence of new organisms; however, antibiotic prescriptions remain unchanged.
The overall incidence of UTIs in infants presenting with fever is 7%, and they account for 5%–14% of visits to emergency departments annually in high-income countries.(2,3) UTI is the second most common infection following otitis media, peaking during infancy and toilet training.(4) Uropathogenic Escherichia coli accounts for 60%–70% of community-acquired infections.(5) A review by Alberici et al. of data from the ESCAPE study showed that the bacterial profile is evolving in both hospital and community settings. E. coli caused <50% of UTIs with Klebsiella species and Enterococci emerging as prominent community- acquired uropathogens.(6)
An Italian study showed that E. coli UTIs had increasing rates of resistance to co-amoxicillin and ciprofloxacin (23.6% and 6.5%, respectively, in 2007 compared to 35.6% and 9.4% in 2011–2014; P < 0.0001). Similar trends were seen amongst resistance of Enterobacteriaceae species to ciprofloxacin, with resistance rates of 1.2% in 2007 and 6.1% in 2011–2014 (P = 0.0007).(7) Studies from Taiwan demonstrated that E. coli extended-spectrum beta- lactamase (ESBL) activity was 2% in 2003 increasing to 11% in 2012.(8) Cephalosporin prophylaxis increases the rate of ESBL activity and decreases antimicrobial susceptibility to almost all classes of antibiotics. Resistance is further exacerbated by the use of repeated courses of different antibiotics.(9)
Antibiotic resistance in South Africa is a known obstacle. A retrospective study amongst adult females with UTI conducted at a tertiary hospital in the Western Cape, South Africa found the overall antibiotic resistance in E. coli isolates to amoxicillin, co-amoxicillin and trimethoprim/sulfamethoxazole (TMP/SMX) to be 65.1%, 18.7% and 47.3%, respectively.(10) Furthermore, a study that reviewed adult female patients in Gauteng, South Africa in 2011 showed uropathogen antibiotic resistance to co-amoxicillin to be 82.8%.(11) Antibiotic prescription at primary level health facilities inherently predisposes patients to developing resistance for the subsequent 6 months.(12)
This study evaluated the uropathogen population in paediatric (birth to 14 years) bacterial UTIs and the respective antibiotic susceptibility profile at a government tertiary hospital in the Western Cape, South Africa and compared this to international data and trends. Identified resistance profiles were compared to current hospital antibiotic policies and UTIs classified into community, hospital-associated and hospital-acquired infections.
MATERIALS AND METHODS
Setting
This study was conducted at Tygerberg Hospital, a tertiary level hospital in the Western Cape, South Africa, serving the Northern and Eastern rural districts. The paediatric service comprises 309 beds providing neonatal, general and sub-specialist care. All urine cultures were performed on-site at the medical microbiology laboratory of the National Health Laboratory Service (NHLS).
Study design
This was a retrospective descriptive analysis, using data from the NHLS database, of urine cultures from patients aged birth–14 years with suspected UTIs admitted to Tygerberg Hospital, between 1 January 2012 and 31 December 2013. UTI was defined as the growth of a single organism (>100,000 cfu/ml) and leukocyte levels >1000 cells/ml. The American Academy of Paediatrics defines significant bacteriuria as >50,000 cfu/ml; however, this definition was amended for this study as there was no access to clinical information or sample collection methods.(13) There were no interventions or diagnostic methods implemented in this study.
Investigation of urinary tract infection
Urine samples were collected and sent to the laboratory based either on clinical suspicion of a UTI, or as part of the diagnostic work-up for pyrexia in paediatric patients.
The NHLS reported on cell counts and the bacterial culture results of all urine specimens sent for analysis. Microbial identification and susceptibility was performed using the automated Vitek II platform and interpreted using annually published Clinical and Laboratory Standards Institute antibiotic breakpoints. Full antimicrobial sensitivity analyses of the uropathogens were not released if the organism was sensitive to penicillin and cephalosporin antibiotics. Organisms were classified into multi- or extreme drug resistance as per definitions published by the European Centre of Disease Prevention and Control (ECDC) and the Centre of Disease Prevention and Control (CDC).(14) Multi-drug resistance was defined as acquired non-susceptibility to at least one agent in three or more antimicrobial classes. Extreme-drug resistance was defined as non-susceptibility to at least one agent in all but two or fewer antimicrobial classes.(14)
The data provided by the NHLS was cross-referenced with data on the electronic Clinicom Data Management System to stratify the specimens into community, hospital-associated and hospital-acquired infections; and the date of admission was compared to the date of urine sample.
Samples were further linked with concurrent blood culture and human immunodeficiency virus (HIV) results. Blood cultures were only drawn if the treating physician deemed them to be clinically indicated. Voluntary counselling and testing for HIV was routinely offered to patients and parents. If the HIV status of patient was known at the time of initial assessment, then formal HIV testing was not performed. HIV exposure history was not available during this investigation.
Statistical analysis
All data received from the laboratory was captured on electronic spread sheets. Data variables were allocated numerical codes and exported to statistical software (STATA v12; StataCorp) for further analysis. Data analysis was done in partnership with the Department of Biostatistics at Stellenbosch University. Factors associated with positive culture were assessed bivariately using Pearson's chi-square test and Fisher's exact test for categorical variables. Statistical significance was defined as P <0.05.
RESULTS
From 1 January 2012 until 31 December 2013, a total of 3616 urine samples were sent to the NHLS for analysis. These samples were reviewed for eligibility and 3334 samples were excluded (Figure 1). Two hundred and eighty-two samples met the study definition and were included in the study. Of those samples, 170/282 (60.3%) were community-derived UTIs, 52/282 (18.4%) were hospital-associated and 60/282 (21.3%) were hospital- acquired UTIs.
Table 1 reports the descriptive data for the culture-confirmed bacterial uropathogens included in the study. The median age for the group was 12.1 months (IQR: 2.9–42.3). The majority of samples, 142/282 (50.4%), were from patients younger than 1 year old; of whom, 90/142 (63.4%) were male. The data showed a slight overall male predominance which was predictive for developing a UTI (160/282; 56.74%; P = 0.004). Of the culture-confirmed UTIs, 235/282 (83.3%) were sourced from the general and specialist paediatric wards. Figure 2 shows that the most common organism cultured isolated was E. coli (143/282; 50.7%). This was followed by Klebsiella pneumoniae (64/282; 22.7%), Proteus mirabilis (13/282; 4.6%) and Enterococcus species (13/282; 4.6%).
Descriptive data for culture-confirmed bacterial uropathogens2
| Total (n = 282) | E. coli (n = 143) | K. pneumoniae (n = 64) | P. mirabilis (n = 13) | Enterococus spp (n = 13)* | P value | |
|---|---|---|---|---|---|---|
| Sex ( n/% ) | ||||||
| Male | 160 (56.7%) | 69 (43.1%) | 34 (21.3%) | 9 (5.6%) | 13 (8.1%) | 0.04 (x 2) |
| Female | 122 (43.3%) | 74 (60.7%) | 30 (24.6%) | 4 (3.3%) | 0 (0%) | |
| Ward ( n/% ) | ||||||
| General | 235 (83.3%) | 136 (95.1%) | 46 (71.9%) | 11 (84.6%) | 11 (84.6%) | |
| Neonatal | 27 (9.6%) | 3 (2.1%) | 12 (30.3%) | 2 (15.4%) | 2 (15.4%) | |
| ICU | 20 (7.1%) | 4 (2.8%) | 5 (7.8%) | 0 (0%) | 0 (0%) | |
| HIV status ( n/% ) | ||||||
| Positive | 16 (6.7%) | 3 (18.8%) | 9 (56.3%) | 0 (0%) | 2 (12.5%) | |
| Negative | 80 (28.4%) | 30 (37.5%) | 24 (30%) | 3 (3.75%) | 2 (2.5%) | |
| Not tested | 186 (66%) | 110 (59.1%) | 31 (16.7%) | 10 (5.4%) | 10 (5.4%) | |
| Blood culture ( n/% ) | ||||||
| Positive | 9 (3.2%) | 7 (77.8%) | 0 (0%) | 0 (0%) | 0 (0%) | |
| Negative | 91 (32.3%) | 32 (35.2%) | 32 (35.2%) | 3 (3.3%) | 7 (7.7%) | |
| Not done | 167 (59.2%) | 97 (58.1%) | 28 (16.8%) | 10 (6%) | 7 (4.2%) | |
| Contaminated | 10 (3.6%) | 5 (50%) | 3 (30%) | 0 (0%) | 0 (0%) | |
| Positive; alternative organism to UTI | 5 (1.8%) | 2 (40%) | 1 (20%) | 0 (0%) | 0 (0%) | |
| Type of infection ( n/% )** | ||||||
| Community acquired | 170 (60.3%) | 114(67.1%) | 17 (10%) | 9 (5.3%) | 8 (4.7%) | |
| Hospital associated | 52 (18.4%) | 20 (38.5%) | 22 (42.3%) | 0 (0%) | I (1.9%) | |
| Nosocomial | 60 (21.3%) | 9 (15%) | 25 (41.7%) | 4 (6.7%) | 5 (8.3%) | |
* Enterococcus spp. includes Enterococcus faecalis (1/0.35%), Enterococcus faecium (1/0.35%) and Enterococcus species (13/4.61%).
** Community-acquired infection: Infection identified on presentation to the paediatric emergency unit in the absence of any prior admission in the preceding 30 days.
Hospital-associated Infection: Infection identified on presentation to the paediatric emergency unit with prior hospital admission in the preceding 30 days.
Hospital-acquired Infection: Infection after 72 h post admission.
HIV serology based testing results were not available for 186/282 (65.9%) patients. Only 115/282 (40.8%) of the samples had a concurrent blood culture performed. Of those, 9/115 (7.83%) samples had a positive blood culture of which 7/9 (77.8%) isolated E. coli.
A total of 75/282 (26.6%) cultured organisms were positive for ESBL with the distribution shown in Figure 3. Of those, 19/75 (25.3%) were community-acquired infections vs 32/75 (42.7%) which were hospital-acquired (P < 0.05). E. coli displayed high-level resistance to amoxicillin and ampicillin (129/142; 90.8%) and TMP/SMX (102/142; 71.8%). K. pneumoniae was highly resistant to penicillins, penicillins with beta-lactam antagonists, all cephalosporin categories and folate pathway inhibitors (Table 2). K. pneumoniae also showed marked resistance to gentamicin (49/64; 76.6%) but showed greater susceptibility to amikacin (38/51; 74.5%). Similar levels of resistance were observed in the Enterobacter species group.
Antibiotic resistance profile of identified uropathogens
| Antibiotic Category | E. coli (n = 143) | K. pneumoniae (n = 64) | P. mirabilis (n = 13) | |||
|---|---|---|---|---|---|---|
| Tested | Resistance (n/%) | Tested | Resistance (n/%) | Tested | Resistance (n/%) | |
| Penicillin | ||||||
| Amoxicillin/Ampicillin | 142 | 129 (90.8) | 64 | 64 (100) | 13 | 11 (84.6) |
| Penicillin (+B lactam inhibitors) | ||||||
| Co-amoxicillin | 129 | 49 (38) | 62 | 55 (88.7) | 7 | 3 (42.9) |
| Piperacillin-tazobactam | 15 | 9 (60) | 38 | 28 (73.7) | 2 | 2 (100) |
| First/Second generation cephalosporin | ||||||
| Cefuroltime (PO) | 138 | 26 (18.8) | 64 | 55 (85.9) | 13 | 0 (0) |
| Third/Fourth generation cephalosporin | ||||||
| Cefotaltime | 52 | 15 (28.8) | 55 | 54 (98.2) | 3 | 0 (0) |
| Ceftriaxone | 52 | 15 (28.8) | 55 | 54 (98.2) | 3 | 0 (0) |
| Ceftazidime | 19 | 15 (78.9) | 55 | 54 (98.2) | ||
| Cefepime | 18 | 15 (83.3) | 51 | 50 (98) | ||
| Aminoglycoside | ||||||
| Gentamicin | 141 | 18 (12.8) | 64 | 49 (76.6) | 13 | 3 (23. 1) |
| Amikacin | 22 | 4 (18.2) | 51 | 13 (25.5) | 3 | 0 (0) |
| Fluoroquinolone | ||||||
| Ciprofloxacin | 52 | 11 (21.2) | 48 | 23 (47.9) | 7 | 1 (14.3) |
| Folate pathway inhibitors | ||||||
| Co-trimoxazole | 142 | 102 (71.8) | 64 | 49 (76.6) | 11 | 6 (54.5) |
Resistance amongst E. coli, K. pneumoniae and P. mirabilis infections were further described according to definitions by the ECDC and CDC.(14) A positive HIV status was predictive of more resistant E. coli strains of infection (P = 0.037). Multi- and extremely drug-resistant K. pneumoniae species were more frequent with hospital-associated and nosocomial infections (Figure 4) (P = 0.006). Multi- and extremely drug-resistant isolates of P. mirabilis (n = 13) showed higher rates of resistance to penicillins and penicillins with beta-lactams in male patients and in community infections but did not reach statistical significance due to the small numbers.
DISCUSSION
The majority of urine samples taken from the patients included in the study with suspected UTI were sterile and did not culture a uropathogen (n = 2350). The median age for the group was 12.1 months (IQR: 2.9–42.3) that was congruent with the published literature describing patients younger than 5-year old were at a higher risk of UTI.(15) Gram-negative enteric bacteria predominated in this study, in-line with available data. E. coli remains the dominant organism (143/282; 50.7%) found in this study, followed by K. pneumoniae (64/282; 22.7%), P. mirabilis and Enterococcus spp. Rates of E. coli infection causing UTI have been reported to be between 60%–80.3% and 7%–26% for K. pneumoniae.(5,7,16–18) The lower rates of E. coli infection in the current study demonstrate an emergence of newer organisms as a consequence of drug selection pressure. Further reasons for this are not immediately apparent.
E. coli and K. pneumoniae accounted for 69/75 (92%) of total ESBL infections; 19/75 (25.3%) of which were attributed to community-acquired infections. Hospital-acquired ESBL UTIs represented 42.7% of total ESBL infections in the current study; E. coli represented 20% and K. pneumoniae 72% (P < 0.05). At a tertiary hospital in Pretoria, South Africa, the rates of hospital-acquired ESBL infection for E. coli and K. pneumoniae were 11.9% and 40.6%, respectively.(19) Dramowski et al. described bacterial infections at the same site as the current study and reported that Klebsiella spp. accounted for 16% of community- acquired and 51% of hospital-acquired bacteraemias.(20) This reflects an increase in Klebsiella spp. blood and urine infections in our environment. There is a global increase in both community- and hospital-acquired ESBL UTI infections. The prevalence of community-acquired ESBL UTI in Thailand was reported as 19.2% (95% CI: 13.8–25.7) in 2015.(21) Factors implicated in exacerbating community ESBL infections were previous hospital admissions, previous use of penicillins or fluoroquinolones, recurrent UTI and genitourinary tract anomalies.(22)
A meta-analysis reported that the pooled prevalence of ESBL Enterobacteriaceae was 14% (95% CI: 8–21), with vesico-ureteric reflux (OR 2.79), history of UTI (OR 2.89) and recent antibiotic use (OR 2.92) identified as risk factors.(23) However, in the current study, rates of 3/75 (4%) were described. Of ESBL E. coli UTI found in this study, 1/15 (6.7%) and 5/15 (33.3%) were sensitive to piperacillin–tazobactam and amikacin, respectively. ESBL K. pneumoniae showed 18.5% (10/54) and 68.5% (37/54) sensitivity to piperacillin–tazobactam and amikacin, respectively.
Table 3 displays E. coli resistance rates from this study compared to data from other published sources.(6,8,24–26) The antibiotics listed are those commonly used at primary and secondary level health services and thus E. coli may be inherently resistant to agents available at these centres. The pooled global prevalence of E. coli UTI resistance to ampicillin and co-amoxicillin is 79.8% (95% CI: 73%–87.7%) and 60.3% (95% CI: 40.9%–79%), respectively.(12)
E. coli resistance rates from index study compared to international data
| Study | Site | E. Coli drug resistance | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Amoxicillin (%) | Co-amoxicillin (%) | TMP/SMX (%) | Cefotaxime (%) | Ceftriaxone (%) | Ciprofloxacin (%) | Amikacin (%) | Piptaz (%) | ||
| Index study | RSA | 90.8 | 38 | 71.8 | 28.8 | 28.8 | 21.2 | 18.2 | 60 |
| Alberici et al | European Multicentre | >50 | 20–50 | >50 | 0–50 | ||||
| Chen et al | Taiwan | 80 | 39 | 53 | 55.4 | 15 | |||
| Rajiv et al | India | 98 | 88 | 73.5 | 73.5 | 63.3 | 21.4 | ||
| Mishra et al | India | 39 | 36 | 31 | 36 | 31 | 38 | ||
| Gunduz et al | Turkey | 16 | 31.5 | 2.7 | 7.9 | 0.2 | |||
Cefazolin resistance has been progressively increasing in China as sensitivity rates for K pneumoniae and Proteus spp. were 72% and 60%, in 2008, and 44% and 10% in 2012, respectively.(8) The authors found that K. pneumoniae exhibited 85.9% resistance to first and second generation cephalosporins and >98% resistance to third and fourth generation cephalosporins. In the current study, P. mirabilis showed 100% sensitivity to all cephalosporins tested by the NHLS.
Male gender, age <3 months, higher creatinine and underlying urological abnormalities are risk factors for the development of bacteraemia associated with urosepsis.(27) However, the current study showed only 9 (3.2%) concurrent positive blood cultures although the majority of patients did not have a culture sample drawn. A previous retrospective study conducted at the same study site looking at blood stream infections in paediatric patients reported a 5.5% pathogen yield from all blood cultures drawn.(28)
Further investigation is warranted to look at the true influence of HIV infection and risk of UTI as there is a high burden of HIV-associated disease in sub-Saharan Africa. Asharam et al. showed no significant impact of HIV on antibiotic sensitivity, response to therapy or duration of hospitalisation.(29) The high TMP/SMX resistance in this study may be related to prophylaxis given to patients with HIV infection. Due to small numbers of HIV positive patients however, this could not be further analysed.
Antibiotic prophylaxis is indicated for children with vesico-ureteric reflux to reduce the risk of recurrent infection but does not reduce the risk of renal scarring.(30,31) Prophylactic antibiotics need to be prescribed judiciously, as the odds of multi-drug resistant infection increases 6.4 fold. Prophylaxis is not recommended following the first or second episode of febrile UTI in otherwise well children and should be reserved for children at higher risk.(13,31,32)
The following limitations of this study should be noted. Laboratory and study definitions were used to diagnose UTI and thus some UTI cases may have been excluded. Clinical data to distinguish febrile vs non-febrile UTI, urine dipstick results, method of urine sample collection and documented HIV status were not available if sample collection was performed at a referral hospital. Urine samples were not repeated after antibiotic treatment to assess in vivo response. Risk factors for community-acquired ESBL infection were not reported in this retrospective study.
CONCLUSION
The results of this study show that the organism profile of paediatric UTI in this unit is changing in-line with international data trends. K. pneumoniae attributed to 10% of culture positive infections. ESBL producing organisms were noted in 75/282 (26.6%) UTIs, of which K. pneumoniae accounted for 72% of infections. Furthermore, 25.3% of ESBL producers arose from the community setting. Extremely drug-resistant K. pneumoniae species were significantly more likely to originate from hospital-acquired infections. These antibiograms support the current hospital policy to treat hospital-associated and hospital-acquired infections with piperacillin–tazobactam and amikacin until urine culture and sensitivity are available, thereby limiting carbapenem drug pressure.
Health-care practitioners should demonstrate measured antibiotic practices to curb growing resistance through inappropriate prescriptions. Therapy should be rationalised based on urine microscopy culture and sensitivity results. Should patients not clinically respond to chosen antibiotics, urine samples should be retested to ascertain therapeutic response or resistance. Future studies are required to further explore resistance profiles so to better guide clinician's antibiotic prescription practices.