Assessment of Clinical Presentation, Performance of Diagnostic Methods and Antibiotic Susceptibility Testing for Salmonella among Patients Attending Kangema Sub-County Hospital, Kenya ()
1. Introduction
Salmonella, a Gram-negative rod-shaped bacterium, of the family Enterobacteriaceae is attributed to cause Salmonellosis [1]. More than 2500 Serotypes of Salmonella have been identified but only less than 100 Serotypes are associated with to cause the disease in humans [2]. Typhoidal Salmonella (Typhi, Paratyphi A-C) are grouped as Salmonella enterica serovars that cause enteric fever in humans, whereas nontyphoidal Salmonella (NTS) colonize a range of vertebrates or non-human animal species mostly leading to gastroenteritis [3]. The most reported serovars in Africa are S. Typhimurium and S. Enteritidis [4].
The global burden of typhoidal Salmonella and NTS is not well defined. Yearly, an estimated 11 - 20 million people get sick from typhoid and between 128,000 and 161,000 people die from it worldwide [5]. In 2006, NTS was estimated to cause 155,000 deaths [6], while in 2010, 190,200 deaths were attributed to typhoid and Paratyphoid fever with as high as 12.2 million disability-adjusted life years which initiated the disease inclusion in the Global Burden of Disease project [7] [8]. A decade earlier, typhoid fever caused 21.7 million illnesses and 216,000 deaths while Paratyphoid fever was linked to causing 5.4 million illnesses [9].
Invasive NTS (iNTS) disease is elusive in sub-Saharan Africa and is attributed to cause high mortality and morbidity. The global estimates on iNTS disease, included in the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2017, indicated that the highest incidences occurred in SSA [10]. NTS serovars and sequence types that are endemic in SSA include the highly invasive S. Typhimurium Type (ST) 313 associated with the cause of large outbreaks and highly Multi-Drug Resistant (MDR) [11] [12]. S. Typhimurium ST313 and S. enteritidis ST11 are common isolates from studies that employed NTS genome-sequencing between 2007 and 2014 in Kenya [13]. Typhoid is now estimated to have an average annual incidence of 263 per 100,000 person-years of observation (95% CI: 199 - 347) in all age groups in Kenya and causes more illness among older children compared to NTS [14].
Salmonella is transmitted through the faecal-oral route through contaminated food, water, and poor sanitation. The acquisition of typhoidal Salmonella is higher in endemic regions of developing countries, especially where sanitation and hygiene are not well observed [5]. In high-income countries, enteric fever is often associated with travel to endemic areas [15] or acquired from food handlers who are chronic carriers of S. Typhi [16]. There is now a considerable body of research that asymptomatic patients may shed Salmonella through faecal matter [17]. Shedding of viable bacteria into the environment leads to its transmission and spread to new hosts.
Symptoms of Salmonella infection vary and are clinically difficult to distinguish from other febrile illnesses [4]. Often, during outpatient visits, symptoms are limiting but if not treated promptly complications such as intestinal perforation, typhoid encephalopathy, and intestinal bleeding severe anaemia may develop and must require hospital admission [4]. Symptomatic illness is associated with high fever in over 80% of cases, abdominal discomfort, headache, and malaise [18]. The laboratory infrastructure is limited in many rural and semi-urban settings to accurately and timely diagnose invasive and noninvasive typhoid disease. Preventive options have been explored and vaccines against typhoid disease have shown promising outcomes. Attenuated and inactivated vaccines against typhoid are available and have significantly reduced typhoid fever cases [19]. However, despite the availability of the typhoid vaccine, vaccination coverage is limited in most developing countries, lacks inclusion in the national immunization programs and is faced with weak surveillance and laboratory systems [20].
Antibiotic resistance-associated infections are currently a global catastrophe that is directly linked to poor clinical outcomes and high case fatality rates [21]. In the USA, Salmonella susceptibility was accessed across all states to provide surveillance data for the National Antimicrobial Resistance Monitoring System for Enteric Bacteria (NARMS) 2012. NARMS highlighted Salmonella strains resistant to Ampicillin, Chloramphenicol, Streptomycin, Sulfonamide (Sulfamethoxazole/Sulfisoxazole), and Tetracycline [22]. Drug resistance to the first-line treatment of Salmonella enterica infections has been reported and is widespread. S. Typhimurium’sresistance to an array of 15 antibiotics comprising 6 or 7 Clinical and Laboratory Standards Institute (CLSI) drug classes has been reported [23]. Extensively drug-resistant ST 313 sublineage associated with a combined MDR, extended-spectrum beta-lactamase (ESBL) production and resistance to Azithromycin was reported in the Democratic Republic of the Congo [24]. Ceftriaxone-resistant S. Typhimurium ST313 poses a major threat to the management of iNTS in Kenya and SSA [25].
This study, therefore, aimed to determine the clinical presentation, performance of diagnostic tests and antibiotic susceptibility testing of Salmonella among adults attending Kangema Sub-County Hospital in Kenya.
2. Methods
2.1. Study Population
The study took place between April and June 2021. The study participants were recruited from a population of Kangema Sub-County residents in Murang’a County. Kangema Sub-County is one of the 8 sub-Counties in Murang’a County with a population density of 80,447 as of the 2019 Kenya Population & Housing Census while the County’s overall population was 1,056,640 [26]. All the study participants presented with typhoid-like symptoms at Kangema Sub-County Hospital, which is the main Government healthcare facility in the area.
2.2. Study Design
This cross-sectional study was conducted among patients who presented with typhoid-like symptoms at Kangema Sub-County Hospital in Murang’a County. Upon obtaining informed written consent, 97 study participants were enrolled at the outpatient department (OPD).
2.3. Eligibility Criteria
Inclusion criteria used included patients presenting with typhoid-like symptoms at Kangema Sub-County Hospital OPD, ≥18 years, and those who were willing and able to give written informed consent. Patients unable to give a stool sample and those who were not referred to the laboratory by a clinician for typhoid diagnosis and confirmation were excluded from the study.
2.4. Sample Size Determination
The formula for estimating the population proportion with specified absolute precision by Daniel [27] was used to determine the number of patients recruited in this study. Setting α at 0.05 and typhoid prevalence of 6.3% [28], a total of 97 patients were recruited to achieve the 0.95 power.
2.5. Sampling Procedure
A systematic sampling procedure was employed by picking every kth adult patient sent to the Laboratory for a typhoid test until the sample size was achieved. To obtain the kth adult patient, the previous year’s three months data (April, May, and June) for patients referred for typhoid tests were abstracted, and the total number of adult patients was divided by the sample size.
2.6. Data Collection
Patients were referred to the laboratory department by a clinician for test confirmation of Salmonella disease. All the study participants presented with typhoid-like symptoms that included headache, fever, myalgia, malaise, abdominal pain, abdominal distention, diarrhoea, nausea, and cough. The general physical condition was graded as either mild, moderate, or severe. Enrollment of the study participants was carried out at the laboratory department and patients who met the inclusion criteria were recruited upon giving informed consent. The study participants were interviewed with the help of a structured questionnaire and a stool sample was obtained in a sterile container. The questionnaire contained four sections that evaluated the socio-demographic, clinical, sanitation & hygienic characteristics and knowledge of typhoid fever among the study participants. The appearance of the stool sample was documented and immediately inoculated in Cary Blair transport media (Thermo Scientific, Loughborough, UK). The cultured sample was stored at 4˚C - 8˚C for 24 hours before transporting samples to Murang’a County Referral Hospital laboratory at the Department of Medical Microbiology following strictly the institutional collection, storage, and shipment of human Biological samples. Immediately, approximately 1 gram of stool sample was inoculated into 10 mL of Selenite F broth and incubated at 37˚C for 18 - 48 hours. Culture, isolation, and identification of Salmonella species were performed following recommendations published in the Bacteriological Analytical Manual as provided by Feng [29]. Antibiotic susceptibility testing was performed following the CLSI 2020 guidelines [30]. Rapid Serology testing was conducted using the SAT kit. Positive results were interpreted qualitatively by a visual clear coloured red line, whereas a negative result had no visible line on the test area.
2.7. Study Validity & Reproducibility
A pre-test was conducted in February and March 2021, 2 months before the actual enrollment study process. The pre-test was meant to assess the feasibility of data collection tools and patients flow at the different OPD service points for corrective action. The final data capture tools were adjusted accordingly to enhance the study validity. All laboratory tests were conducted following approved protocol, Standard Operating Procedure guidelines and setting controls whenever necessary.
2.8. Data Analysis & Interpretation
Categorical variables were analyzed and interpreted using descriptive statistics. Frequency (%), mean, standard deviation, and median (interquartile ranges at 25% and 75%) were used to describe the qualitative and laboratory parameters. Chi-square tests were used to test for significance where applicable. In bivariate and multivariate analyses, odds ratios (OR) and 95% confidence intervals (CI) for the association between Salmonella infection among study patients and socio-demographic, clinical presentation variables were calculated using regression analyses, including factors that were associated with pathogenic bacterial isolates at the significance level of P < 0.05.
The test sensitivity was calculated using the formula; Sensitivity = number of true positive (TP)/sum of the number of TP and number of false-negative (FN). Specificity was calculated as follows; Specificity = number of true negatives (TN)/sum of TN and the number of false positives (FP). The positive predictive value (PPV) which is the proportion of patients with positive test results who are correctly diagnosed was determined as follows; PPV = TP/sum TP + FP; while the negative predictive value (NPV) is defined as the proportion of patients with negative test results who are correctly diagnosed was determined as follows; NPV = TN/TN + FN. All statistical analyses were performed using STATA v 13 (StataCorp LP, College Station, TX, USA).
2.9. Ethical Consideration & Approval
This study was conducted as per the Declaration of Helsinki and the International Conference on Harmonization Guideline on Good Clinical Practice (ICH-GCP). The protocol and informed consent form were reviewed and approved by the Mount Kenya University Ethics Research Committee (MKU-ERC) before the commencement of the study (Ref no: MKU/ERC/0614). A research permit was obtained from the National Commission for Science, Technology, and Innovation (NACOSTI) (Ref no: NACOSTI/P/18/95130/20871). At the same time, authorization was granted from the Director of Health, Ministry of Education (Ref no: MGA/CTY/GEN/64/VOL: II/79) and County Commissioner (Ref no: PUB. 24/11/VOL.11/253), Murang’a County. Informed consent of the client was obtained by explaining the purpose of the study. Unique identification numbers were used to ensure the confidentiality of the study participants. The researcher treated all the information acquired with utmost privacy and confidentiality
3. Results
In this study, all the 97 patients enrolled provided the required samples and responded to the structured questionnaire. Out of the 11 patients who presented with fever, 5 (45.5%) of them had Salmonella infection. Out of the 6 patients who had headache 2 (25%) were infected with Salmonella. The distribution of Salmonella by other clinical presentations was as follows; 4 out 20 (20%), 5 out 82 (6.2%), 5 out 85 (5.9%), 4 out 9 (44.4%) and 5 out 34 (11.8%) of Salmonella infections were among patients with myalgia, malaise, abdominal pain, abdominal distention, and diarrhoea respectively. Further, 4 out 34 (11.8%), 4 out 11 (28.6%), 4 out 10 (40%) and 3 out 10 (30%) of Salmonella infections were among patients with nausea, cough, moderate general physical condition and had loose and water stool presentations respectively (Table 1).
In the bivariate analysis, patients who had fever (uOR 39.1, 95% CI 4.6 - 334; p = 0.001), had headache (uOR 5.6, 95% CI 1.1 - 3.4; p = 0.048), had myalgia (uOR 7.7, 95% CI 1.4 - 42; p = 0.018), had abdominal distention (uOR 19.1, 95% CI 3.6 - 106; p = 0.028), had diarrhoea (uOR 9.6, 95% CI 1.1 - 83; p = 0.038), had
![]()
Table 1. Association with clinical presentation and Salmonella infection.
OR—Odds ratio; CI—confidence interval; u—unadjusted and a—adjusted odds ratio; ND—Not done; P-value—significance level.
cough (uOR 11.9, 95% CI 2.2 - 64.7; p = 0.004) and those who had loose or watery stool (uOR 14.4, 95% CI 1.5 - 138; p = 0.021) were more likely to be infected with Salmonella. Patient who presented with mild general physical condition were less to be infected with Salmonella compared to those who were stable (uOR 0.02, 95% CI 0.01 - 0.4; p = 0.008) (Table 1).
In multivariate analyses, patients who had fever (aOR 12.1, 95% CI 3.6 40.1; p = 0.0001), abdominal distention (aOR 4.3, 95% CI 1.2 - 16.3; p = 0.028), had diarrhoea (aOR 13.8, 95% CI 1.5 - 126.5; p = 0.04), had loose or watery stool (aOR 16.7, 95% CI 1.6 - 177.8; p = 0.02) and those who presented with mild general physical condition (aOR 0.03, 95% CI 0.01 - 0.7; p = 0.02) remained independently associated with Salmonella infection among study patients (Table 1).
3.1. Prevalence of Typhoid Fever
3.1.1. Prevalence of Salmonella Infection by Test
Prevalence of Salmonella infection among the patients varied depending on the test used: Using SAT and culture to detect Salmonella infection the following prevalence were observed: 7/97 (7.2%) 95% CI by SAT and 6/97 (6.2%) 95% CI by culture as shown in (Figure 1).
3.1.2. Distribution of Salmonella Infection by Isolates
Out of these 6 Salmonella isolates, were S. Typhi (n = 1; 16.7%) while the rest were S. Paratyphi B (n = 5; 83.3%) (Figure 2).
3.2. Sensitivity and Specificity of SAT against Culture
Because SAT is routinely used to detect Salmonella infection in Kangema Sub County Hospital, Murang’a County, Kenya, the test performance was compared using culture as the gold standard. Data were used for performance analyses only if the results were definitive. Results concordant with those of culture score were obtained in 96/97 (98.9%; 95% CI 94.4% - 99.9%) of the patients by SAT. The kappa of tests which measures the level of agreement showed almost perfect agreement between SAT and culture to detect Salmonella infection in this study kappa (0.9641—almost perfect agreement).
![]()
Figure 1. Prevalence of Salmonella infection by test.
![]()
Figure 2. Prevalence of Salmonella infection by isolates.
Based on the culture results as the gold standard, the sensitivity of SAT was 6/6 (100%; 95% CI 54.1% - 100%) with a specificity of 91/92 (98.9%; 95% CI 94.1% - 99.9%). The positive predictive value (PPV) of SAT was 6/7 (85.7%; 95% CI 42.1% - 99.6%) with a negative predictive value (NPV) of 91/91 (100%; 95% CI 96% - 100%) (Table 2).
3.3. Antibiotic Susceptibility Profile of Salmonella Strains
Isolates were tested for their antimicrobial susceptibility by the Kirby-Bauer disk diffusion method according to CLSI guidelines (CLSI, 2016). The drugs tested included Ampicillin (AMP), Nalidixic acid (NA), Chloramphenicol (CRO), Gentamicin (GEN), Ciprofloxacin (CIP), Trimethoprim-sulfamethoxazole (SXT), Tetracycline (TET). The antimicrobial susceptibility was classified using the CLSI guidelines, as susceptible, intermediate, or resistant to each antibiotic. In addition, we also classified the isolates as either non-susceptible (including both intermediate and resistant isolates) or susceptible.
Resistance to AMP, TET, GEN, CRO, NA and SXT was 100%, 100%, 83.3%, 50%, 33.3% and 33.3% respectively. No isolate showed resistance to CIP. All the S. Typhi and S. Paratyphi B were Multidrug-resistant (Table 3).
4. Discussion
Salmonella enterica serovars Paratyphi B isolated in this study were responsible for typhoid fever and associated with symptoms of fever, headache, myalgia, diarrhoea, cough, and abdominal distension. Studies have shown that upon ingestion and incubation phase, some patients develop subclinical symptoms or may be asymptomatic during the primary phase, and faecal shedding of the bacteria can occur [18]. Fever dominates, whereby, the body temperature rises gradually
![]()
Table 2. Comparison of SAT against culture (N = 97 samples).
N—Number; %—Percentage; CI—Confidence Interval; NPV—Negative Predictive Value; PPV—Positive Predictive Value; K— Kappa.
![]()
Table 3. Antibiotic susceptibility of Salmonella isolates.
S—Susceptible; R—Resistant; I—Intermediate-resistant; AMP-Ampicillin; NA—Nalidixic acid; CRO—Chloramphenicol; GEN—Gentamycin; CIP—Ciprofloxacin; SXT—Trimethoprim-sulfamethoxazole; TET—Tetracycline.
during the first and second week followed by influenza-like symptoms, headache, malaise, dry cough, anorexia, and sometimes abdominal pains and diarrhoea [4]. From this study, the disease presentation among the study participants was a mild infection and was mainly managed within the outpatient department of the hospital. Most of the study participants presented with high fever and these findings are broadly similar to the findings from a multicounty study in Bangladesh, Nepal, and Pakistan that indicated fever was a prevalent symptom in patients whose samples tested positive for S. Typhi or S. Paratyphi [31]. Although the typhoid causative agents were isolated from stool samples in this study indicating typhoid fever, NTS should always be considered because much recent work within the region and elsewhere has indicated nontyphoidal outbreaks and MDR strains [14] [26].
In the current study, the test performance of immunochromatographic SAT was compared to culture as the “gold standard” from stool samples collected from patients reporting to Kangema Sub-County hospital with typhoid-like symptoms. The sensitivity of SAT (100%) and a specificity of 98.9% concurred with the findings from a study that evaluated the sensitivity performance of SD Bioline RDT (100%), near sensitivity rate using Creative Diagnostics (98.1%) [32] and 96.7% in Laos [33]. Although the researchers from the former study used spiked blood culture broth to evaluate the validity of Salmonella RDTs, the current study used fresh stool samples from symptomatic patients with typhoid-like symptoms to compare test performance of the rapid test. Although microbiological examination and isolation of the bacteria from blood or bone marrow samples remains the standard reference laboratory test for the confirmation of salmonellosis, a rapid serologic test based on antibody detection may provide a convenient supplementation. The use of a rapid and sensitive test is vital to bacterial control and its spread [34]. However, some of the commonly used rapid tests are faced with limitations and are no longer in use in the national laboratory diagnostic programs in some countries. The diagnostic value of the Widal agglutination test that detects antibodies against O (surface) and H (flagellar) antigens is often questioned because of its false-negative and false-positive, poor test performance, cutoff titer levels, and poor agreement with culture tests. The more sensitive and specific assay, Enzyme-Linked Immunosorbent Assay (ELISA), that identifies antibodies to the capsular polysaccharide Vi antigen is superior to identifying carriers but limited to diagnosing acute enteric fever due to lack of specificity [35]. Polymerase chain reaction (PCR) and proteomic assays have a high degree of sensitivity and specificity to detect enteric fever but are cost-prohibitive especially in developing countries [36].
For these reasons, a sensitive, simple, and cost-effective SAT diagnostic method for the diagnosis of Salmonella infection may result not only in rapid salmonellosis control and surveillance measures but also in prompt diagnosis followed by appropriate treatment. In this study, the kappa statistic of almost perfect agreement (0.9641) with culture provides some significant evidence to suggest that the antigen test is applicable in the diagnosis of S. Typhi among the study population although the problem of antibiotic selection is limited without culture testing. Confirmation by culture (or validated molecular methods, as available) is essential as typhoid fever, Paratyphoid fever and another invasive salmonellosis can present as a non-specific febrile illness, and current serological tests lack diagnostic specificity. It can be argued that those patients who sought laboratory confirmation had clinical signs and symptoms of enteric fever. Confirmation is essential to assess the proportion of enteric fever caused by these different organisms, determine antimicrobial susceptibility, and perform molecular epidemiology studies [37].
Antibiotic susceptibility profiles of Salmonella strains in this study conformed with the findings from a previous study from stool samples of patients at the Aga Khan University Hospital in Kenya that highlighted high resistance of Salmonella against NAL, CRO, AMP, CHL, SXT [26]. We tested the susceptibility of the commonly used antibiotics prescribed to patients when they present with typhoid-like symptoms or during a confirmed case of typhoid disease within the study region. Strains of S. Paratyphi B and S. enterica isolated in this study were MDR to AMP, NA, CRO, GEN, SXT and TET which were also similar antibiotics highlighted in the NARMS that captured Salmonella surveillance data across all the USA states [22]. Salmonella strains exhibited high resistance rates when subjected to an array of these antibiotics. The current study illuminates a light on the current response of commonly prescribed antibiotics even as major concerns have been expressed about AMR. For years, CRO demonstrated a good efficacy against Salmonella infection and was used as a first-line drug, but widespread resistance was reported shortly. Resistance in the self-transmittable plasmid H1 incompatibility group was reported [38], in addition to the ability of genes in the plasmids to confer resistance to other antibiotics [39]. For this reason, AMP, SXT and TET became common as first-line antibiotics to treat enteric fever but eventually gradually reduced efficacy against resistant Salmonella strains [22]. The bacterial resistance data generated from this study affirms the findings of most researchers that AMP, SXT and TET are of less use in the treatment of typhoid disease [40] with these drugs demonstrating 100% resistance rates in this study.
In the current study, Ciprofloxacin demonstrated the best MIC and achieved a 100% susceptibility rate against all Salmonella isolates. The use of Fluoroquinolones in the treatment of enteric fever is common, especially in regions reported to exhibit low resistance rates to this class of antibiotic. Earlier studies showed that Ciprofloxacin and Ofloxacin concentrations were above the MIC [41], achieved an optimal drug concentration at the site of infection [42] and employs a bactericidal mode of action both in vivo and in vitro [43] . Ciprofloxacin being the most ideal antibiotic of choice in this study chimes with recent findings on its use in the treatment of S. Typhi infection, whereby, it surpassed the MIC in a controlled human infection model. [44].
However, over time, in vitro MIC of Ciprofloxacin against strains has gradually reduced and some studies have recommended its withdrawal in the treatment of enteric fever [45]. In recent data from Pakistan published as part of the surveillance for enteric fever in Asia Project (SEAP), Fluoroquinolone resistance was noted in nearly 90% of S. Typhi and S. Paratyphi isolates [46]. The rapid resistance against Fluoroquinolones is linked to molecular evolutionary biology and direct response toward drug pressure (Redgrave et al.,2014). Fluoroquinolones targets DNA gyrase and topoisomerase IV, thereby inhibiting bacterial DNA synthesis, but chromosomal mutations in the quinolone resistance-determining regions of gyrA, gyrB, parC and parE genes have been determined [47]. Plasmid-mediated resistance is also a mechanism acquired by the bacterium against the action of quinolones [48].
The findings of this study are restricted to the study population and researchers were primarily concerned with test performance of SAT usage in the diagnosis of typhoid disease and antibiotic susceptibility. There has been much interest lately to develop tools that are more sensitive and specific to rapidly diagnose typhoidal Salmonella and NTS such as genomic studies targeting nucleic acid, bacterial gene expression, proteomic and immunoscreening strategies, removal of human DNA contaminant in PCR [4].
5. Conclusion
In conclusion, our study findings offer suggestive evidence that it is important to further investigate the typhoid-like symptoms by conducting SAT to confirm enteric fever accurately and rapidly. SAT performance is comparable to culture and might be used to accurately detect Salmonella infection where culture is typically not available. The antibiotic sensitivity tests offer the best guidance for the selection of antibiotics to guide treatment, especially with the MDR Salmonella strains.
Limitations of the Study
Although we reported high carriage of MDR Salmonella, we cannot conclusively predict the source of exposure to these multidrug-resistant isolates; whether it is due to the modern food-animal production characterized by densely concentrated animals and routine antibiotic use or is due to misuse of antibiotics in the human population, which is a common phenomenon in Kenya.
Acknowledgements
The authors acknowledge the research participants and Murang’a County Government officers who were directly or indirectly involved during the study period. The authors also thank Mount Kenya University-ERC for giving the approval to conduct the study.
Authors’ Contributions
All authors listed have made a substantial, direct, and intellectual contribution to the article, and approved it for publication. Saweria W. Mbuthia was involved in conception and design, data collection, and data analysis and assisted in the drafting of the manuscript. Eliab S. Some and Mbaruk Suleiman were involved in the supervision of the project. Oliver W. Mbuthia assisted in the review of data analysis and manuscript writeup while Musa O. Ngayo assisted in the data analysis.
Appendix I: Questionnaire