Prevalence, antimicrobial susceptibility patterns of bacterial uropathogens and evaluation of urine dipstick parameter accuracy among individuals with diabetes in Ayder Comprehensive Specialized Hospital and Mekelle General Hospital, Tigray, Ethiopia

Abstract

Background

Urinary tract infections represent the primary cause of morbidity in individuals with diabetes due to various factors, including high glycosuria, low bladder emptying and weak immunity. However, there is a paucity of scientific studies in the region dealing with bacterial Urinary tract infections, antimicrobial resistance and assessment of urine dipstick parameters. Therefore, the primary aim of this research endeavour was to determine the prevalence, antimicrobial susceptibility profile of bacterial uropathogens, and the reliability of urine dipstick values among patients with diabetes undergoing treatment in public health facilities in the study area.

Methods

A hospital-based cross-sectional study involving 422 individuals with diabetes was conducted at Mekelle General Hospital and Ayder Comprehensive Specialized Hospital between June 1, 2020 and September 20, 2020, using a consecutive sampling method. The data was collected using a structured questionnaire for socio-demographic and clinical details, and mid-stream urine samples were collected for laboratory investigation (dipstick tests, culture and antimicrobial susceptibility testing). The urine dipstick tests, including leukocyte esterase, blood, and nitrite tests, were evaluated as predictive parameters for urinary tract infections using a Laboquick urine reagent strip 10-test parameter following the manufacturer’s directions. In culture, bacterial counts exceeding 10⁵ Colony-forming units (CFU)/ml in midstream urine for asymptomatic UTI and 10³ CFU/ml for symptomatic UTI were regarded as indicative of significant bacteriuria. Additionally, after a successful characterization, antimicrobial susceptibility testing of pure isolates was performed aseptically on a Muller-Hinton agar plate following the Kirby-Bauer disc-diffusion method. To explore the association of independent variables and an outcome variable (bacterial uropathogen), the bivariate and multivariate logistic regression analysis were performed using SPSS version 26. Statistically significant relationships were considered at p < 0.05 and respective 95% confidence interval, and the direction of the association was interpreted based on the adjusted odds ratios generated from the multivariate analysis. Performance metrics for each dipstick parameter were calculated, including specificity, sensitivity, likelihood ratios, and predictive values.

Results

The overall prevalence of bacterial uropathogens in this study was 13.3% (56/422). The most prevalent bacterial uropathogen was Escherichia coli (24, 42.8), followed by Coagulase-negative Staphylococci (21.9%), and Klebsiella pneumoniae (17%). The odds of having positive bacterial uropathogens were higher among females (Adjusted odds ratio = 3.21 (95% CI: 1.43–7.17) and individuals with diabetes mellitus for five or more years (Adjusted odds ratio = 2.27 (95% CI: 1.06–4.63). Additionally, those having current symptoms of urinary tract infection (Adjusted odds ratio = 17.80 (95% CI: 7.38–42.93), a history of antibiotic use before two weeks (Adjusted odds ratio = 4.45 (95%CI: 2.11–9.39), and those 56 years of age or older (Adjusted odds ratio = 4.3 (95% CI: 1.01–18.48) were significantly more likely to have positive culture for bacterial uropathogens. Most of the gram-negative bacteria (52, 94.5%) were sensitive to amikacin, followed by ciprofloxacin (42, 76.3%). A significant prevalence of resistance to Tetracycline and Trimethoprim-sulfamethoxazole was noted in Escherichia coli (79.1% and 66%, respectively) and Klebsiella pneumoniae (80% for both) among the tested gram-negative organisms. Overall, 24 of 56 bacterial isolates (42.8%) were multidrug-resistant. Specifically, 50% of Escherichia coli (12/24), Klebsiella pneumoniae (5/10), and Enterobacter species (1/2), 25% of Pseudomonas aeruginosa (1/4) and Coagulase-negative Staphylococci (3/12), and 33.3% of Staphylococcus aureus isolates were found to be multidrug-resistant. Leukocyte Esterase was the most sensitive of the dipstick parameters (94.4%), whereas the nitrite test was the most specific (99.7%). The test with the least sensitivity (51.8%) and specificity (85.8%) was the blood test.

Conclusion

In conclusion, Escherichia coli was the primary bacterial uropathogen found in individuals with diabetes, with a notable prevalence of multidrug resistance among uropathogens. Among diagnostic tests, the nitrite test demonstrated superior potential compared to the leukocyte esterase and blood tests. Both the leukocyte esterase and nitrite tests are significant for confirming bacterial urinary tract infections. Nonetheless, in diabetic patients, the diagnosis and management of bacterial urinary tract infections must advance to include the identification of multidrug-resistant pathogens, tailored treatment strategies, and effective glycemic control.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-13289-4.

Background

Urinary tract infections (UTI) rank third in terms of frequency of infection worldwide, following respiratory and gastrointestinal tract infections. The impact of UTI is substantial in both hospital-acquired and community-acquired infections [1, 2]. Particularly complicated UTI, often leading to the hospitalization, imposes a tremendous burden on healthcare systems. Hospital-acquired UTI range from 1.4% to 5.1%, with catheter-related UTI accounting for the bulk of cases [2].

Over 404.6 million persons worldwide had UTIs, resulting in nearly 236,786 UTI-related deaths and 5.2 million disability-adjusted life years [3]. Sub-Saharan Africa exhibits a high morbidity rate of UTI at 32.2%, with Escherichia coli(E. coli) being the most frequently isolated bacterium, responsible for 86.4% of bacterial UTI [4]. In Ethiopia, UTIs rank as the second most prevalent infectious disease, significantly contributing to morbidity in the general population [5].

Annually, UTI impacts over 150 million people globally, resulting in a cost exceeding six billion US dollars to the world’s economy [2]. The distribution of UTI burden is expected to be uneven, with a heavier toll on middle-class and low-income nations due to scarce resources [6]. Urinary tract infection is a health issue that affects diabetic people of all ages in the majority of Sub-Saharan Africa and other underdeveloped nations [4, 7]. In 2019, drug-resistant bacteria in UTI were linked to 0.26 million fatalities and 64.89 thousand deaths overall. Particularly, sub-Saharan Africa, tropical and southern Latin America and Europe have the highest mortality rates across all age groups [3].

In resource-constrained settings, the treatment of UTI is primarily empirical. Consequently, a significant health concern in the treatment of UTI is the rise in antibiotic resistance among the bacteria that cause UTI [8]. Physicians are primarily concerned about the rise in drug-resistant and multidrug-resistant bacterial strains and the fall in the quantity of novel antibiotics available for the treatment of UTI [9]. In Ethiopia, evidence shows that treatment of bacterial UTI has become challenging due to the widespread infections associated with drug-resistant uropathogens such as the extended-spectrum beta-lactamase and carbapenemase-producing strains [10].

Individuals with diabetes mellitus (DM) are highly affected by bacterial UTIs due to various factors. The poor glycemic control results in high glucose in urine, which favors the growth of bacteria [11, 12]. Additionally, weak immune response and low bladder emptying may increase persistent bacterial UTIs due to low physical or immune-mediated bacterial clearance [13, 14]. However, in the study area, research findings are scarce on bacterial UTI and antimicrobial resistance profiles among individuals with DM.

Due to cost and time implications associated with urine culture, the laboratory diagnosis of UTI mostly depends on urine chemical and microscopic analysis in developing countries, including Ethiopia [15, 16]. Nevertheless, there is a scarcity of routine assessments aimed at verifying the accuracy of urine dipstick tests as an indicative tool for UTI in the study area. The lack of verification raises the possibility of misinterpretation stemming from the utilization of inaccurate proxy screening tests, potentially resulting in the inappropriate use of antibiotics. To date, no prior research has investigated the diagnostic efficacy of urine dipsticks in predicting UTIs among diabetic individuals in this particular geographical area. Therefore, the current facility-based cross-sectional study was conducted to ascertain the evidence gap in the prevalence of bacterial UTI, antimicrobial susceptibility patterns of bacterial isolates, as well as to evaluate the diagnostic performance of urine dipstick parameters among people with diabetes mellitus. The findings generated in this regard are important baseline evidences in the Tigray region, Northern, Ethiopia.

Materials and methods

Study area

The study was conducted at Ayder Comprehensive Specialized Hospital (ACSH) and Mekelle General Hospital, situated in Mekelle, Tigray, northern Ethiopia. Mekelle, Tigray Regional State’s administrative center, is located 787 km north of Ethiopia’s capital, Addis Ababa. As per the 2007 census program, the town’s total population was approximately 258,258 [17]. ACSH stands as the largest university medical facility in the area and the second most extensive hospital in Ethiopia, boasting a collective capacity of around 500 inpatient beds across various departments and specialized units. Catering to a population of 5 million, the hospital accommodates both referral and non-referral patients hailing from diverse parts of the Tigray region and other neighboring states, such as Afar and Amhara regional states, including the Eritrean refugees. Ayder Comprehensive Specialized Hospital offers DM follow-up services to approximately 1620 individuals. In parallel, Mekelle General Hospital extends DM follow-up services for about 2170 individuals diagnosed with diabetes, as indicated in the 2020 unpublished reports provided by each healthcare establishment.

Study design and study period

A hospital-based cross-sectional study was carried out in a hospital setting from June 1, 2020 to September 20, 2020.

Source populations

The population was all diabetic follow up patients at Ayder Comprehensive Specialized Hospital and Mekelle hospital.

Study populations

The study population was individuals with diabetes who were receiving medical care at Ayder Comprehensive Specialized Hospital and Mekelle General Hospital throughout the research duration, meeting the specified inclusion criteria.

Inclusion and exclusion criteria

Participants with diabetes who provided consent/assent were enrolled in the research investigation. Diabetic participants with antibiotic treatment, defined as within the 14 days prior to interview/screening) therapy, either for therapeutic or preventive purposes (self-reported), were excluded from the study. Additionally, those with diabetes suffering from critical illness, as well as female participants in their menstrual phase, were considered excluded from study.

Sample size determination and sampling techniques

Sample size determination

Sample size was computed using the following formula:

Where n = sample size

z = statistic for level of confidence

p = estimated prevalence

d =represents the considered margin of error

Considering 50% because there was no previous study, 5% precision (d = 0.05) and 95% confidence interval, Z is 1.96. Therefore, the sample size was estimated to be:

n = 384; however, to account for non-responses, reach the targeted minimum sample size, and to ensure that the minimum number of analyzable respondents meets the minimum requirements for statistical power, a 10% buffer was added to the initial sample size. Therefore, the final sample size turned out to be 422 (n=384+38=422).

A proportionate allocation was used to determine the sample size from both study areas (Ayder Comprehensive Specialized Hospital and Mekelle General Hospital). Considering both study areas as separate strata, a separate sample was taken independently from both study areas.

Proportionate allocation:

The formula above works based on the assumption that “k” indicates the number of strata, n_j indicates the j^th stratum’s sample size, and Nj stands for the j^th stratum’s population size.

This means that the entire sample size is n = n1 + n2+…+nk, while the total population size is N= N1 + N2+…+Nk.

The sample size from each health facility was calculated as:

Let n1 and n2, are sample sizes to be determined from Mekelle general hospital and ACSH. N is the total population size, which is 3790 diabetic follow-ups in a recent year. The total sample size for this study was 422. The population of both study area (Nj) was 3790, 1620 and 2170 for ACSH and Mekelle general hospital, respectively. Based on this data and the formula for proportional allocation to the sample size for both study areas was:

Sampling technique

A consecutive sampling technique was employed to recruit individuals with diabetes attending Ayder Comprehensive Specialized Hospital and Mekelle General Hospital during the study period, who participated in the study. Started from the first comer during the study period and continued until the sample size was fulfilled.

Study variables

Independent variables

Age, Gender, Residence, Type of occupation, and Educational level of the study participants were considered as independent Socio-demographic variables. Additionally, clinical characteristics and risk factors (Type of DM, duration of DM, History of UTI, symptoms of UTI, history of hypertension, history of antibiotic use, pregnancy, Fasting Blood Sugar (FBS) value and catheterization history) of the participant were considered.

Dependent variables

Prevalence of uropathogens.

Antimicrobial susceptibility patterns.

Data collection procedures

Demographic and clinical profile

Clinical histories of the study participants were collected from each diabetic patient through direct interviews and review of patient history records. Additionally, trained professionals using a structured questionnaire developed for this research, pretested and translated to the vernacular language, collected the socio-demographic data of the study participants (supplementary file 2). Therefore, the sociodemographic variables (age, gender, residence, type of occupation, and educational level) were assessed using yes/no recall questions. In contrast, the clinical parameters, including DM type and duration, FBS level, history of UTI, symptoms, antibiotic use, catheterization and hypertension histories were captured from medical records.

Specimen collection and management

Each diabetic participant provided two urine samples, collected in sterile containers of 10–20 ml capacity, which were accurately labelled. The samples were refrigerated at 2 and 8 °C prior to their transportation to ACSH for analysis [18].

Laboratory investigation

Urine dipstick, culture, biochemical tests, gram-stain and antimicrobial susceptibility testing were done according to the standard operating procedures of the laboratory.

Chemical (dipstick) analysis of urine

The study used Laboquick URS (urine reagent strip) 10-T (10-test) parameter (Bornova, Izmir, Turkey) for dipstick urinalysis at Mekelle General Hospital and Ayder Medical Microbiology Laboratories. The dipsticks assessed nitrite, leukocyte esterase, pH, specific gravity, protein, glucose, ketone, urobilinogen, bilirubin, and blood. Leukocyte esterase, blood, and nitrite reactions were evaluated as predictive parameters for UTI [19]. The urine dipstick was used following the guidelines provided by the manufacturers. The procedure began by collecting fresh, well-mixed urine in a clean, dry container. Next, a strip was taken from the strip container, resealed, and all reagent pads were immersed in the urine for approximately one second. After removing the strip, its edge was run along the rim of the container to eliminate excess urine; the side of the strip was blotted on absorbent paper to avoid cross-contamination between pads. Finally, the results were read by comparing the reagent pads to the color chart on the bottle label at the specified recommended intervals. Specific timing of reading each parameter’s result was set at 60 s for blood and nitrite, as well as 120 s for leucocyte esterase. The results of blood and leucocyte esterase tests were reported as “Negative, “ Trace, or 1⁺ to 3^+”, whereas the nitrite test results were reported as “Negative” or “Positive” as per the kit’s recommendations.

Urine culture and biochemical tests

Urine specimens dedicated for a culture were inoculated into Cysteine-Lactose-Electrolyte-Deficient agar (CLED) medium plates (Oxoid, Hampshire, UK), Blood Agar (Biomark labs, India), and MacConkey agar (Liofilchem, Italy) under sterile conditions. After successful inoculation, the streaked plates were incubated at 37 °C for a period of 18 to 48 h, and the growth of bacterial colonies was observed afterwards. In the context of DM, bacterial counts exceeding ≥ 10⁵ CFU/ml (colony forming unit per milliliter) in midstream urine for asymptomatic UTI and 10³ CFU/ml for symptomatic UTI were regarded as indicative of significant bacteriuria. In the case of mixed growth, the uniform and predominant type of colonial growth was used to judge the CFU threshold. However, after careful enumeration of the growth of all morphological types, unique-looking colonies of two or more types appearing in significant numbers ( 10⁵ CFUs) were sub-cultured independently for identification. This means all qualifying growth types were considered for further steps, indicating mixed infection types [18, 20–22]. S. aureus growth was considered substantial regardless of the number of colony-forming units detected. Colony count was made using a colony counter in order to appreciate a significant bacterial growth of the urine culture [23].

Positive urine cultures exhibiting substantial bacteriuria were further recognized by their distinct appearance and pattern of biochemical reactions. Subcultures of gram-positive bacteria were derived from blood agar, and transferred to nutrient agar and mannitol salt agar. Further discrimination was made following the catalase and coagulase test results. For the differentiation of gram-negative bacteria, biochemical testing involved catalase, oxidase, fermentation patterns on triple sugar iron agar (TSI), hydrogen sulfide (H₂S) production, lysine decarboxylase, indole, motility, citrate utilization, urease, and lysine iron agar results (HIMEDIA, India). Results of the biochemical test were used in combination with colonial morphology characteristics (size, shape, color, pigmentation, swarming, hemolysis), and Gram reaction for final identification of the isolates [22, 24].

Antimicrobial susceptibility testing

After the identification and isolation, a minimum of three to five colonies exhibiting similar morphological characteristics were chosen from each subculture. Then, the colonies were gently added into an adequate amount (four to five milliliters) of normal saline with the help of a sterile microbiological wire loop. The turbidity level was standardized to a standard of 0.5 McFarland. Then, using a sterile cotton swab, the bacterial suspension was uniformly introduced across the surface of the Mueller-Hinton agar [25]. In this study, the antibiotic susceptibility profile of identified bacteria was evaluated using the disc diffusion method on Mueller-Hinton agar sourced from Oxoid, Hampshire, UK [26].

The following antimicrobial agents were included during the antimicrobial susceptibility: amikacin, ampicillin (10 µg), tetracycline (30 µg), chloramphenicol (30 µg), ciprofloxacin (5 µg), gentamicin (10 µg), penicillin G (10IU), erythromycin(15 µg), amoxicillin-clavunilic acid (30 µg), trimethoprim sulphamethoxazole (25 µg), cefotaxime (30 µg), cotrimoxazole, norfloxacilin (5 µg) ceftazidime and nitrofurantoin (300 µg). The selection of these antibiotics was influenced by their accessibility and the prevailing trends in local prescribing practices. The Mueller-Hinton agar plates that were inoculated underwent aerobic incubation at 37 °C for a duration of 16 to 18 h. Afterwards, the diameter of inhibition zones in millimeters was determined for each antibiotic disc manually using a ruler and documented as “susceptible”, “Intermediate” or “resistant” in accordance with guidelines set by the Clinical and Laboratory Standards Institute (CLSI), M00, 2022 [25]. The workflow is as indicated below (Fig. 1).

Fig. 1.

Schematic laboratory procedure flow chart. Note: AST=Antimicrobial susceptibility testing; BA=Blood agar; CLED=Cysteine-Lactose-Electrolyte-Deficient agar; DM=Diabetic mellitus; MAC=MacConkey agar, MSA=Mannitol salt agar; NA: Nutrient agar

Quality control

Laboratory analyses were conducted following standard operating procedures (SOPs). Before commencing the laboratory investigation, reagents underwent inspection to confirm their expiration date and appropriate storage conditions. The sterility of the media was assessed through overnight incubation. Strains of reference bacteria of American Type Culture collection (ATCC) E. coli (ATCC-25922), Staphylococcus aureus (S. aureus) (ATCC- 25923) and Pseudomonas aeruginosa (P. aeruginosa) (ATCC-27853) were utilized to control the quality of the culture positivity [25].

Data management, analysis and interpretation

The completeness of the questionnaire was assessed, with manual verification of unrecorded data. Each participant’s results were tagged and entered into the laboratory data recording forms. For statistical analysis, the collected data were imported into the EPiData version 3.1 program and later exported to the Statistical Package for Social Sciences (SPSS) version 26. Frequencies, mean, and standard deviation were calculated using descriptive statistics to ascertain the patterns of antimicrobial susceptibility tests and the prevalence of prevalent uropathogens. The Hosmer-Lemeshow test was utilized to evaluate the fit of the logistic regression model, yielding a p-value greater than 0.05. This result indicates that the model is well-calibrated for the analysis. The relationship between bacterial uropathogen positivity and various independent variables, including age, gender, residence, occupation, educational level, type and duration of diabetes mellitus (DM), history of urinary tract infections (UTIs), UTI symptoms, hypertension, antibiotic use, pregnancy, fasting blood sugar (FBS) values, and catheterization history, was analyzed using a bivariate logistic regression model. This analysis involved calculating odds ratios (OR) at a 95% confidence interval (CI).In this study, variables with a p-value of ≤ 0.25 from bivariate logistic regression were selected for multivariate analysis to control for confounding factors and calculate adjusted odds ratios (AOR) that indicated the strength and direction of statistical associations. A p-value of less than 0.05, along with the corresponding 95% confidence interval not crossing “1”, was considered statistically significant in the multivariate logistic regression analysis.

Furthermore, using urine culture as the confirmatory (gold standard) test, the specificity, sensitivity, positive predictive value, and negative predictive value of the dipstick (Laboquick URS-10T) parameters were examined and computed. The likelihood ratio was also determined to assess the accuracy of each urine dipstick parameter. Finally, findings are presented as text narratives, charts, and tables, and the write-up of the entire paper has complied with the “Strengthening the reporting of observational studies (STROBE)” checklist/recommendations for cross-sectional studies [27].

Operational definitions

UTI: Urinary tract infections are bacterial infections in humans that can happen anywhere along the urinary tract. The urinary tract includes the Bladder, Kidneys Ureters (the tubes that take urine from each kidney to the bladder), and Urethra (the tube that empties urine from the bladder to the outside). Asymptomatic UTI: is the presence of significant bacteria (≥ 10⁵ CFU/ml) in an individual‘s urine without sign and symptoms of UTI. Symptomatic UTI: defined when a patient has two or more of the following signs or symptoms with other recognized causes: fever (temperature>38 °C), urgency to urinate, frequency, dysuria, or suprapubic tenderness and a urine culture positive for greater than 10³ or more microorganisms per milliliter. Significant bacteriuria is defined as the presence of ≥ 10³ colony-forming units (CFU) per milliliter of urine for symptomatic UTI patients and ≥ 10⁵ CFU per milliliter of urine for asymptomatic UTI respectively [28].

Results

The sociodemographic and clinical features of the respondents

The study involved 422 diabetic patients aged two to 79 years, with a mean age of 43.3 years. Among the participants, 222 (52.6%) were men, with 113 (72.8%) residing in rural areas and 309 (73.2%) in urban areas. A majority, 368 (87.2%), exhibited no symptoms of urinary tract infection (UTI). Notably, 348 (82.5%) were diagnosed with type 2 diabetes mellitus. Blood glucose levels revealed that 176 individuals (41.7%) had levels below 126 mg/dl, while 246 individuals (58.3%) had levels at or above this threshold. Among the respondents, 222 individuals (52.6%) had been living with diabetes for five years or more. A significant majority of respondents, specifically 361 individuals or 85.5% reported no history of hypertension, while 72% (304 individuals) indicated they had no UTIs (Table 1).

Table 1.

Clinical and socio-demographic characteristics of individuals with diabetes, and bacterial uropathogens in ACSH and Mekelle General Hospital, northern Ethiopia, 2020 (n = 422)

Characteristics	Classification	Frequency	Percent (%)
Age (years)	0–19	36	8.5
	20–35	94	22.2
	36–45	98	23.2
	46–55	93	22.0
	≥ 56	101	23.9
Gender	Female	200	47.4
	Male	222	52.6
Marital status	Single	120	28.4
	Married	237	56.2
	Widowed &divorced	65	15.4
Residence	Urban	309	73.2
	Rural	113	26.8
Educational status	Can’t read and write	61	14.5
	Elementary	137	32.5
	Secondary and preparatory	123	29.1
	Colleague and above	101	23.9
Occupation	Employed	211	50.0
	Un employed	211	50.0
DM Type	Type 2	348	82.5
	Type 1	74	17.5
DM duration in years	< 5 years	200	47.4
	≥ 5 years	222	52.6
BMI	Low	59	14.0
	Normal	179	42.4
	Fat and obesity	184	43.6
Hypertension history	No	304	72.0
	Yes	118	28.0
Previous UTIs history	No	361	85.5
	Yes	61	14.5
UTI symptom	No	368	87.2
	Yes	54	12.8
Antibiotic use history earlier than 14 days	No	318	75.4
	Yes	104	24.6
DM family history	No	374	88.6
	Yes	48	11.4
FBS (mg/dl)	Below 126	246	58.3
	≥ 126	176	41.7
Pregnancy presence Catheter history	No	114	95.0
	Yes N0 Yes	6	5.0
		401 94.8 21 5.2

Abbreviations: BMI: Body mass index; DM: diabetic mellitus FBS: Fasting blood sugary; mg/dl: milligrams per deciliter; UTI: Urinary tract infection

The prevalence of isolated bacterial uropathogens among diabetic patients

Of the total 422 participants, 56(13.3%) individuals with diabetes were positive for both gram-positive and gram-negative bacteria. Of the 56 individuals with diabetes infected with bacteria, only two (3.6%) study participants had mixed infection with more than two bacterial species. The prevalence of UTI was significantly higher (p < 0.001) in symptomatic UTI patients compared to asymptomatic UTI patients. The most predominant bacterial isolates were E. coli 24(42.9%), followed by Coagulase-negative Staphylococci (CoNS) 12(21.9%) and Klebsiella pneumoniae (K. pneumoniae) 10(17.4%). Generally, from this study, 41(73.2%) and 15(26.8%) were gram-negative and gram-positive bacteria, respectively (Fig. 2).

Fig. 2.

A pie chart showing the culture results of urine samples of individuals with diabetes mellitus and suspected for urinary tract infections in Ayder Comprehensive Specialized Hospital and Mekelle General Hospital, northern Ethiopia, 2020 (n = 422). Note: CoNS=Coagulase-negative Staphylococci

Gender-based distribution of isolated bacteria in symptomatic and asymptomatic UTIs

Among the 422 study participants, 54 individuals (12.8%) reported experiencing UTI symptoms, with 53.7% of them (29/54) exhibiting significant bacteriuria. Among those with symptomatic UTI [28], 6 20.7% (6/29) were male and 79.3% (23/29) were female. Among the remaining 368 participants without current UTI symptoms, 27 (7.3%) were found to carry bacterial uropathogens. Overall, bacterial uropathogens were detected in 42 females (75%) and 14 males (25%) among the symptomatic group (Table 2).

Table 2.

Gender-based distribution of bacterial isolates in symptomatic and asymptomatic UTIs among individuals with diabetes in ACSH and Mekelle General Hospital, north Ethiopia, 2020 (n = 422)

Bacterial Gram reaction	Bacterial species	Symptomatic UTIs		Asymptomatic UTIs		Total N (%)
		Male N (%)	Female N (%)	Male N (%)	Female N (%)
Gram–negative bacteria	Escherichia coli	1(4.2)	11(45.8)	2(8.3)	10(41.7)	24(100)
	Klebsiella pneumonia e	1(10)	4(40)	1(10)	4(40)	10(100)
	Pseudomonas aerugonosa	0(0)	2(50)	2(50)	0(0)	4(100)
	Enterobacter species	0(00	2(100)	0(0)	0(0)	2(100)
	Acinetobacter	0(00	1(100)	0(0)	0(0)	1(100)
Gram-positive bacteria	CoNS	4(33.3)	1(8.4)	3(25)	4(33.3)	12(100)
	Staphylococcus aureus	0(0)	2(66.7)	0(00)	1(33.3)	3(100)
Over all	7 species	6(10.7)	23(41.1)	8(14.3)	19(33.9)	56(100)

Abbreviations: CoNS= Coagulase-negative Staphylococci; N=number; UTIs=Urinary tract infections

Factors associated with urinary tract infection among diabetic patients

In the final multivariate logistic regression analysis, the following factors were significantly associated with positivity to bacterial uropathogens. The odds of bacterial uropathogens positivity were significantly higher among females (AOR = 3.21 (95% CI: 1.43–7.17) and individuals with diabetes mellitus (DM) for five or more years (AOR = 2.27 (95% CI: 1.06–4.63). Furthermore, individuals presenting current UTI symptoms (AOR = 17.80 (95% CI: 7.38–42.93), those with a history of antibiotic prior to 14 days of screening (AOR = 4.45 (95%CI: 2.11–9.39), and those 56 years of age or older (AOR = 4.3 (95% CI: 1.01–18.48) were significantly more likely to have positive cultures for bacterial uropathogens. After adjusting for confounders, the odds of testing positive for bacterial uropathogens were 1.92 times higher among individuals with a FBS level of 126 mg/dl or greater (AOR = 1.92 (95%CI: 0.85–4.38) compared to those with a FBS level of 125 mg/dl. However, this result is not statistically significant at the 5% level, as the 95% CI includes “1.0” (Table 3).

Table 3.

The factors associated with risk of uropathogens among individuals with diabetes in ACSH and Mekelle General Hospital, north Ethiopia, 2020 (n = 422)

Characteristics	Category	Uropathogen (+)	COR(95%CI)	P-value	AOR(95%CI)	P-value
Age (in years)	0–19	2	1.00		1
	20–35	7	2.37 (0.97-6 5.81)	0.057	2.01(0.65–6.19)	0.220
	36–45	13	3.04(1.26–7.28)	0.013	1.90(0.62–5.83)	0.262
	46–55	15	3.83 (1.57–9.66)	0.004	2.55(0.72–8.78)	0.149
	≥ 56	19	3.75 (1.28–10.91)	0.015	4.3(1.01–18.48)	0.049
Gender	Female	42	3.94(2.08–7.48)	0.000	3.21 (1.43–7.17)	0.004*
	Male	14	1.00
Residence	Urban	46	1.80 (0.87–3.70)	0.109	1.76(0.69–4.46)	0.233
	Rural	10	1.00
Educational status	Unable to read and write	12	2.50 (0.98–6.35)	0.053	1.46(0.422–5.04)	0.550
	Elementary	15	1.25 (0.52–2.99)	0.606	1.07(0.36–3.16)	0.896
	Secondary &preparatory school	20	1.98 (0.86-4.57)	0.108	01.37(0.48–3.94)	0.551
	Colleague & above	9	1.00		1
DM type	Type 2	50	1.80(0.74–4.37)	0.195	0.808(0.27–2.40)	0.702
	Type 1	6	1.00	.	1
Duration of DM	Below 5 years	17	1.00
	≥ 5years	39	2.42 (1.22–4.43)	0.004	2.27 (1.06–4.63)	0.034*
Hypertension history	No	36	1.00
	Yes	20	1.51 (0.83–2.75)	0.167	046(0.20–1.04)	0.064
Previous UTI	No	42	1.00
	Yes	14	2.246(1.26–4.45)	0.018	0.59(0.23–1.52)	0.279
Current Symptoms of UTI	No	29	1.00		1
	Yes	27	14.65 (7.54–28.43)	0.000	17.80(7.38–42.93)	< 0.001*
FBS in mg/dl	125	12	1.00		1
	≥ 126	12	2.97(1.52–5.82)	0.001	1.92(0.85–4.38)	0.116
History of antibiotic use earlier than 14 days of screening	No	29	1.00		1
	Yes	27	3.49(1.95–6.24)	0.000	4.45(2.11–9.39)	< 0.001

Abbreviations: 1.00: Reference Category AOR: Adjusted Odds Ratio, COR: Crude Odds Ratio, *: Statistically significant at p < 0.05; DM: Diabetic Mellitus; FBS: Fasting Blood Sugar;

Antimicrobial susceptibility patterns of isolated bacterial uropathogens

The majority of gram-negative bacteria in this study were shown to be susceptible to ciprofloxacin, gentamicin, amikacin, and nitrofurantoin. On the other hand, the majority were resistant to tetracycline, augmentin, and ampicillin. High percentage of Tetracycline and Trimethoprim-sulfamethoxazole resistance was observed in E. coli (79.1%, 66%) and K. pneumoniae (80%, 80%) among the gram-negative organisms, respectively. Furthermore, among gram-positive bacteria, CoNS exhibited better sensitivity to amoxicillin-clavulanic acid (83.3%), ciprofloxacin (91.7%), and nitrofurantoin (91.7%) (Supplementary file 1). Both gram-positive and gram-negative bacteria showed multiple resistances.

Multidrug-resistant (MDR) isolates

Of the 56 bacterial isolates, 24 (42.8%) were resistant to at least one antibiotic in three classes of antimicrobial agents (MDR). Specifically, 50% of E. coli (12/24), K. pneumoniae (5/10), and Enterobacter species (1/2), 25% of P. aeruginosa (1/4) and CoNS (3/12), and 33.3% of S. aureus isolates were MDR. The single isolate of Acinetobacter species was also MDR.

Current UTI symptoms, urine dipstick and urine culture findings

Urine culture positivity showed a signifiant association with every UTI symptom, excluding fever and patient’s temperature. Urgency during urination (44.6%), abdominal pain (44.6%), burning sensation (42.9%), painful urination (32.9%), and dysuria (28.5%) have been significantly correlated with culture positivity (p < 0.05). Furthermore, the positivity of nitrite and leukocyte esterase in the urine chemical analysis displayed a stronger association (p < 0.05) with UTI symptoms (Table 4).

Table 4.

Current UTI symptoms, urine dipstick and urine culture findings among individuals with diabetes in ACSH and Mekelle General Hospital, northern Ethiopia, 2020 (n = 422)

Current symptoms of Urinary tract infection	Urine culture		P-values	Urine dipstick analysis
	Positives (n = 56) N(%)	Negatives (n = 366) N(%)		Nitrite positive (n = 25) N(%)	Leukocyte Esterase positive(n = 78) N(%)	Blood positive(n = 81) N(%)
Dysuria	16(28.5%)	20(5.5%)	0.000	6(24%)	16(20.5%)	13(16.0%)
Urgency	25(44.6%)	20(5.5%)	0.000	10(40%)	26(33.3%)	18(22.2%)
abdominal pain (lower)	25(44.6%)	19(5.2%)	0.000	12(48%)	28(38.9%)	18(22.2%)
Burning sensation	24(42.9%)	21(4.3%)	0.000	10(40%)	24(30.8)	15(18.5%)
Painful urination	19(33.9%)	12(3.3%)	0.000	7(28%)	18(23.0%)	12(14.8%)
Temperature	12(21.4%)	45(12.3%)	0.063	4(16%	12(15.4)	6(7.4%)

Note: N=number

The results of Laboquick URS 10-T (urine dipstick) parameters versus urine culture results among diabetic patients

The analyis of Laboquick URS 10-T parameters revealed positive results for nitrite, 26/422(6.2%), leukocyte esterase, 78/422(14.5%), and blood, 81/422(19.2%). Notably, a significant percentage of those with bacteriuria tested positive for leukocyte esterase, 94.6% (53/56), and blood, 51.7% (29/56) (Table 5).

Table 5.

The consistency of urine dipstick parameters results with urine culture among individuals with UTI and diabetes in ACSH & Mekelle General Hospital, northern Ethiopia, 2020 (n = 422)

Urine dipstick results (Laboquick URS 10-T)		Urine culture result as gold standard (confirmatory) test
		Significant uropathogens	Non-significant uropathogens	Total
Nitrite	Positive	24	2	26
	Negative	32	365	366
	Total	56	366	422
Leukocyte Esterase	Positive	53	25	78
	Negative	3	341	344
	Total	56	366	422
Blood	Positive	29	52	81
	Negative	27	314	341
	Total	56	366	422

Note: Laboquick URS 10-T=Laboquick urine reagent strip 10-test parameter Sensitivity, Specificity, Positive predictive value (PPV), Negative predictive value (NPV) and accuracy of urine dipstick parameters

In terms of the sensitivity and specificity of each urine chemical parameter, the leukocyte esterase test exhibited high sensitivity (94.6%). Whereas nitrite test exhibited low sensitivity (42.85%) and high specificity (99.7%). In the evaluation of diagnostic tests, the leukocyte esterase test demonstrated a positive predictive value (PPV) of 67.9% and a negative predictive value (NPV) of 99.1%. Conversely, the nitrite test exhibited a PPV of 96% and an NPV of 99.7% (Table 6).

Table 6.

Accuracy of urine dipstick parameters towards the diagnosis of UTI among individuals with diabetes in ACSH &Mekelle General Hospital, northern Ethiopia, 2020 (n = 422)

Screening test	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)	Accuracy (%)
Nitrite	42.85%	99.7%	96%	99.7%	92.4%
Leukocyte esterase	94.6%	93.2%	67.9%	99.1%	96.4%
Blood	51.8%	85.8%	37.1%	92.08%	80.8%

Urine culture was used as a confirmatory test to assess the correctness of the urine dipstick parameters, and the likelihood ratio was computed based on the following formula: (1) Positive likelihood ratio (+ LR) = Sensitivity/1-Specificity; (2) Negative likelihood ratio (-LR) = 1-Sensitivity/Specificity. Therefore, the positive likelihood ratio (+ LR) and negative likelihood ratio (-LR) of leukocyte esterase were 13.91 and 0.057, respectively. The Nitrite test yielded positive likelihood ratios (+ LR) of 142.8 and negative likelihood ratios (-LR) of 0.57. Furthermore, blood’s positive likelihood ratio (+ LR) was 3.64, while its negative likelihood ratio (-LR) was 0.57 (Table 6).

Discussion

Investigating bacterial urinary tract infections (UTIs) among diabetic individuals in low-income countries is an essential public health concern due to the widespread prevalence of diabetes and the heightened risk of severe, often unnoticed, infections and death. This subject has significant implications, particularly concerning the heightened vulnerability of individuals with diabetes to severe infections, the rising occurrence of urinary tract infections (UTIs) due to drug-resistant bacteria, and the substantial costs linked to managing long-term complications [29, 30]. In addition to providing information on drug susceptibility, evaluating the accuracy of UTI diagnostic parameters is crucial for accurate diagnosis and rationalized therapy, particularly in low-resource settings [31, 32]. Therefore, the current study offers insights into the prevalence of bacterial uropathogens, contributing risk factors, antimicrobial resistance patterns, and the effectiveness of urine dipstick tests within the diabetic population.

In the current investigation, the total prevalence of urinary tract infection among individuals with diabetes was 13.3% (95%CI = 10.4–17%), including both symptomatic and asymptomatic cases. This prevalence is comparable with prevalence reported in prior researches conducted in other sections of Ethiopia: Jigjiga (10.5%) [33], Dessie (11.6%) [34], Addis Ababa (14.7%) [35], Hawassa University Referral Hospital (13.8%) [36], Harar, eastern Ethiopia (15.4%) [37], St. Paul Specialized Hospital Millennium Medical College, Addis Ababa (14.9%) [38], Metu Karl Heinz Referral Hospital, southwest Ethiopia (16.7%) [39], Nekemte Regional Laboratory, Ethiopia (16.5%) [40], and African countries: Uganda (13.3%) [41], and Malawi at 11% [42]. However, the current finding exceeds the results reported in other studies in Ethiopia: Jimma (9.2%) [43], Zewditu Memorial Hospital, Addis Ababa, Ethiopia (9.8%) [44] and other countries, such as the United States (8.2%) [45], and Pakistan (8.2%) [46].

The study indicates that the prevalence of UTIs among individuals with diabetes is lower compared to various other studies conducted in countries including Sudan (19.5%) [47], Kenya (21.6%) [48], Uganda (22%) [49], Nigeria (26.1–55%) [50–52], Gahan (19.4%) [53], Nepal (27%) [9], United Arab Emirates (40.2%) [54], India (34.4%) [55], and Romania (29.8%) [56]. Factors such as socio-economic status, age, medical and treatment environments, individual host factors, community social habits, personal hygiene standards, health education practices, and sample size variations may influence the variance in UTI prevalence among diabetics across different countries [30, 57–59]. Despite these differences, the prevalence of bacterial uropathogens observed in the current study aligns with the pooled estimates reported in Ethiopia (12.7–19.2%) [30].

In this study, gram-negative bacteria were recovered more frequently than gram-positive bacteria as etiological agents of UTI in diabetic patients, which aligns with findings from previous research conducted in Ethiopia [60, 61] and other parts of the world [41, 62], but contrasts with data from Uganda [63]. The differences between the reports and our findings may stem from variations in the immune status of the individuals, and or other interventions such as history of catheterization and preoperative prophylactic measures [64, 65].

In the current study, E. coli was identified as the most prevalent organism, accounting for 42.9% of isolates, consistent with findings from multiple international studies [35, 64, 66]. This prevalence can be attributed to E. coli’s significant presence in fecal microbiota, which facilitates its migration through the genital tract and contributes to urinary tract infections. Additionally, the bacteria’s virulence factors, notably P-fimbriae or pili adherence factors, play a critical role in adhesion to uroepithelial cells during the processes of colonization and invasion [4].

The study findings reveal a significant association between bacterial uropathogen positivity and independent variables/factors including diabetic duration of five years or more, elevated fasting blood sugar levels, current UTI symptom, previous antibiotic use, older ages (56 years or greater), and female gender. These results align with previous research conducted in Gondar [34] and Harar [37], Ethiopia. The link between prolonged diabetes and bacterial uropathogen positivity may stem from declined immune responses and physiological changes characteristic of diabetic conditions. Specifically, reduced recruitment of phagocytic cells, lower leucocyte infiltration to infected tissues, and decreased production of pro-inflammatory cytokines contribute to poor specialized and generalized response to bacterial infections. Furthermore, glycemic control and recurrent UTIs can elevate the risk of serious complications, including bloodstream infections, hospitalization, and death [67]. Hence, the significant correlation between high fasting blood sugar levels and the presence of bacterial uropathogens aligns with prior findings in Ethiopia [68]. This relationship may stem from inadequate glycemic control and elevated levels of glycosuria. Consequently, prolonged diabetic conditions have serious implications for the prevalence of bacterial uropathogens and their resulting complications [69].

In developing countries, the risk of antibiotic resistance in diabetes -related UTI is a serious concern, particularly where urine cultures are lacking [35, 36]. Hence, the higher odds of bacterial uropathogen positivity and previous history of antibiotic use in the current study might be a manifestation of infection with drug-resistant pathogens. In light of this, recurrent infections with drug-resistant bacterial uropathogens were significantly linked to prolonged diabetic duration, poor glycemic control and diabetic nephropathies [69]. Furthermore, the association between older age and higher positivity rates of bacterial uropathogens indicated in the current study and elsewhere [70] suggests diminished immunity from ageing, potentially exacerbated by the duration of diabetes.

The study found that female individuals had significantly higher odds of testing positive for bacterial uropathogens, a finding that aligns with previous research from Ethiopia [66, 71], Kuwait [11] and Pakistan [69]. This contrasts with an earlier investigation at Gondar Dessie, Ethiopia University Hospital, which found no statistical link between sex and bacteriuria [27], potentially due to variations in male-to-female ratio. Contributing factors may include a higher incidence of UTI in females, likely due to reduced normal vaginal flora and poor hygiene [40, 64, 72]. Additionally, anatomic differences such as shorter urethra and proximity of the urethral opening to the anus may facilitate bacterial access, along with the absence of prostatic secretions that aid in bacterial survival [73].

The current study indicates a significant association between, current UTI symptoms and of the likelihood of bacterial uropathogen positivity. While a research from Dessie supported this assertion, previous studies in Gondar, Harar, and Hawassa found no statistical correlation [35, 39, 74, 75]. Symptoms of UTI may arise from bacterial colonization and invasion of the urethra and bladder aided by specific adhesions, suggesting ongoing bacterial proliferation linked to poor bladder function and urine incontinence. Additionally, diabetes can negatively impact bladder function do to s factors like neuropathies, and abnormalities in the detrusor muscle, and urethra, facilitating bacterial survival in both sterile and non-sterile anatomical sites of the genitourinary system [14, 76].

The current investigation revealed significant resistance to various antimicrobial drugs in both Gram-negative and Gram-positive bacteria. Notably, high resistance levels were recorded for tetracycline, ampicillin, and trimethoprim-sulfamethoxazole. Among Gram-negative bacteria, E. coli exhibited resistance rates of 79.1%, 66%, and 70.8% to tetracycline, trimethoprim-sulfamethoxazole, and ampicillin, respectively, while K. pneumoniae showed 80% resistance across the same antibiotics. Furthermore, Enterobacter species demonstrated complete resistance (100%) to tetracycline, trimethoprim-sulfamethoxazole, gentamicin, augmentin, and cefotaxime. These findings align with similar high-resistance trends reported in previous studies, particularly concerning the last two antibiotics [35, 66, 77, 78].

The study highlights a significant prevalence of multi-drug resistant (MDR) uropathogens, at 42.8%, which aligns with similar findings in southern Ethiopia and Dessie, Ethiopia [66, 68]. This level of resistance could potentially stem from widespread availability and indiscriminate utilization of antimicrobial agents in treating urinary tract infections. Contributing factors may include overuse and misuse of antibiotics [7, 79], incorrect administration of these medications, and insufficient infection control measures, leading to an increased presence of resistant bacteria in the community [40, 41]. Additionally, diabetes increases the risk of infection by MDR uropathogens compared to the general population [67], necessitating targeted treatment approach involving culture and antimicrobial susceptibility testing to prevent further drug resistance arising from empiric therapy [44, 80].

The accuracy of urine dipstick tests was evaluated against urine culture through metrics such as sensitivity, specificity, and predictive values [81–83]. Notably, the nitrite test exhibited a high specificity of 99.7%, suggesting its role as an adjunct confirmatory test for bacterial UIT following screening tests, including the leucocyte esterase. Additionally, the current findings align with previous research that indicated the notable specificity of nitrite test compared to its sensitivity [23]. The higher specificity of the nitrite test may be attributed to the fact that the gram-negative bacteria mostly causing UTI are nitrate reductase producers [84]. However, its sensitivity is lower, which can be due certain UTI-causing bacteria, such as Enterococcus and Staphylococcus species lack nitrate reductase.

The leukocyte esterase test demonstrates a sensitivity of 94.6%, surpassing that of nitrite and blood parameters, indicating its preferred role in screening for bacterial urinary tract infections. This finding is consistent with previous reports from Kenya [85]. There are also recommendations in that sensitivity can be further enhanced by the combinational use of the leukocyte esterase and nitrite test [85]. Furthermore, combining the Gram stain, leukocyte esterase and nitrite test can substantially avoid unnecessary urine cultures [19].

Blood parameter exhibited low sensitivity and positive predictive values (51.8%, 37.1%), compared to its specificity and negative predictive values (85.8%, 92.0%). Additionally, a negative likelihood ratio of 0.57 was observed. Therefore, this implies the negligible (weak) role of the blood parameter against UTI. Generally, a test with an LR⁺>10 is considered to have a strong prediction or diagnostic potential. From this perspective, the leukocyte esterase (positive LR⁺=13.9) and nitrite test (positive LR⁺=142.8) show superior and strong power to diagnose UTI over the blood parameter. This may be because blood test can be positive in other conditions, including kidney or bladder stones, traumatic injuries and intravascular hemolytic conditions.

Strengths and limitations of the study

The study generated evidence of great clinical significance by employing descriptive and analytic methods. Beyond the prevalence of UTI, the assessment of antimicrobial resistance and diagnostic accuracy of urine dipstick tests is an asset, particularly for developing regions. However, the study was limited to health-facilities and lacks molecular testing. It also shares the limitations of cross-sectional studies, including the difficulty of establishing temporal order for cause and effect.

Conclusion

In conclusion, E. coli was identified as the predominant bacterial uropathogen among individuals with diabetes, followed by coagulase-negative Staphylococci, and K. pneumoniae. Longer diabetes duration and female gender were significantly associated with the presence of bacterial uropathogns. The majority of isolated bacteria exhibited resistance to tetracycline, and trimethoprim-sulfamethoxazole, while amikacin, nitrofurantoin, and ciprofloxacin demonstrated effectiveness in treating urinary tract infections (UTIs). Overall, multidrug resistance was high among uropathogens. Among diagnostic tests, the nitrite test demonstrated superior potential compared to the leukocyte esterase and blood tests. Both leukocyte esterase and nitrite tests are significant for confirming bacterial UTIs. Nonetheless, in diabetic patients, the diagnosis and management of bacterial UTIs must advance to include the identification of multidrug-resistant pathogens, tailored treatment strategies, and effective glycemic control.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ACSH

Ayder Comprehensive Specialized Hospital

AOR

Adjusted odds ratio

ATCC

American type culture collection

CFU

Colony forming units

CI

Confidence interval

CLED

Cysteine-Lactose-Electrolyte-Deficient agar

CLSI

Clinical and Laboratory Standards Institute

CoNS

Coagulase negative Staphylococci

DM

Diabetic mellitus

FBS

Fasting blood sugar

LR

Likelihood ratio

MDR

Multi-drug resistant

NPV

Negative predictive values

PPV

Positive predictive value

SPSS

Statistical Package of Social Sciences

STROBE

Strengthening the reporting of observational studies in epidemiology

TSI

Triple sugar iron agar

UTI

Urinary Tract Infections

Author contributions

HDT: Conceptualization, methodology, data collection, formal analysis, writing original &final draft; SM: Conceptualization, methodology, resource, data curation, and supervision; LNW: Conceptualization, methodology, data curation, and supervision; GKA: Conceptualization, methodology, data curation, and supervision; TA: Resource, data curation, software, and formal analysis; MTS: Supervision, data curation, formal analysis, manuscript preparation and follow-up revisions. All contributors (authors) reviewed and consented to the final draft of the manuscript.

Funding

Mekelle University College of Health Sciences, postgraduate research fund, funded this research.

Data availability

Data available on request from the authors.

Declarations

Ethics approval and consent to participate

The Health Research Ethics Committee (HRERC) (ERC 1487/2020) of the College of Health Science, Mekelle University, granted ethical approval for the study. Permission was obtained from Ayder Comprehensive Specialized Hospital and Mekelle General Hospital’s administration. Written informed consent was obtained from adult patients, and assent was taken from the children’s family or guardians. A named institutional and/or licensing committee approved all experimental protocols, and in compliance with the Helsinki declarations. The confidentiality of the patient’s information was safeguarded throughout the process using codes instead of names.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.