A cross-sectional study of enteric fever among febrile patients at Ambo hospital: prevalence, risk factors, comparison of Widal test and stool culture and antimicrobials susceptibility pattern of isolates

The study was conducted in Ambo town at Ambo hospital from February 2016 to December 2016. Ambo town is located in West Shewa Zone of Oromia Regional State, Ethiopia, 114 Km from Addis Ababa. Ambo town (latitude 8⁰59′N and longitude 37,051′E) has an elevation of 2100–2200 m above sea level. The mean annual temperature and annual rainfall of the area is 18 °C and 800–1200 mm respectively. The total population of Ambo town is 48,171 (24,634 males and 23,537 females. Ambo hospital gives service to patients coming from both urban and rural areas approaching 1357, 833 (677,599 males and 680,234 females) per year.

The study was a health facility-based cross-sectional survey.

Patients visiting Ambo hospital during the study period were the population for the study. Data from the hospital for the period from July 2014 to June 2015 indicated that out of the total patients attending the hospital, 6720 febrile cases were diagnosed as enteric fever. Presumptive febrile patients from outpatients’ departments (OPDs), wards (inpatients) and/or antiretroviral therapy (ART) clinic with symptoms clinically similar to enteric fever were the target population of the study.

The required sample size was calculated using the single proportion formula N = (Z) ² P (1-P)/d2 [18] by considering expected prevalence (P) of 32.6% [8], 95% confidence interval (Z = 1.96) with a 5% margin of error or precision (d) and 10% non-response. Accordingly, the total sample size required for the study was calculated as 372.The representative groups were selected by simple random sampling.

Febrile patients who were willing to participate in the study by giving informed consent; those who did not take antibiotic treatment for the enteric fever or for any other bacterial infections, and those who had fever for 2 or more days before visiting Ambo hospital accompanied by other clinical symptoms of enteric fever were included in the study. Those febrile patients who had received antibiotic treatment within 2 weeks before coming to the hospital and those who were diagnosed for other known febrile illness were excluded from the study.

Participants of the study were interviewed face to face by trained laboratory technicians using pre-tested structured questionnaire aimed to gather information on potential risk factors such as agro-ecology, age, sex, level of education, hand washing practice after visiting bath room (yes/no), use of toilet (yes/no), source of drinking water (pipe, stream/river), habit of drinking raw milk (yes/no), habit of eating raw meat (yes/no), habit of eating raw vegetables (yes/no), family size, marital status (single, married, widowed, divorced), occupation (government employee, self-employee, house wife, student) history of HIV/AIDS sero-status (positive/negative), previous history of typhoid treatment (yes/no). Other risk factors like, malaria, diabetes, tuberculosis (TB), cancer and organ transplantation were included in the questionnaire.

The blood and stool samples were collected from those patients sent to the laboratory in the morning session (9 AM to 12 AM) of patient examination. Venous blood sample (2–3 ml) was collected aseptically using 70% alcohol in to a sterile test tube and centrifuged at 2500 revolution per minute for 5 min to separate the serum. Fresh stool specimen provided by febrile patient was put into screw capped container, labeled and transported to Microbiology Laboratory of Ambo University for culture.

As a means of quality control, the questionnaire prepared to assess potential risk factors was pre-tested before the main study on febrile patients. The data was collected by a trained laboratory technicians and nurses. The quality of culture media was tested for sterility and performance. Sterility of culture media was checked by incubating overnight at 35–37 °C without specimen inoculation. Any physical change like cracks, excess moisture, color, dehydration, and contamination were assessed and expiration date also checked. As a positive control standard strains of E. coli ATCC 25922 and S. aureus ATCC 25923 obtained from Ethiopian Public Health Institute were used during culture and antimicrobial susceptibility testing.

Data generated from questionnaire survey and laboratory investigation were entered into Excel spread sheet (Microsoft Corporation). Then coded data was analyzed using STATA version 11.0 for Windows (Stata Corp. College Station, TX, USA). The data were summarized using descriptive statistics. The apparent prevalence of enteric fever was calculated as the number of seropositive/culture positive patients divided by the total number of patients examined multiplied by 100. True prevalence (TP) of enteric fever exposure was calculated by adjusting the apparent prevalence for specificity and sensitivity of the test using the formula, TP = AP+(Sp-1)/Se + (SP-1), [21], where AP is apparent prevalence, Sp is specificity and Se is sensitivity of the Widal test used [22]. Confidence limits for the proportions were established by exact binomial test with 95% confidence intervals (CI). The variables assessed were categorical and variables with more than two categories were transformed into indicator (dummy) variables. To identify explanatory variables of enteric fever seropositivity, first, the Pearson’s Chi-square test was used followed by univariable and multivariable logistic regression analyses. All non-collinear variables with P-value ≤0.25 in univariable analysis were considered for the multivariable logistic regression model to construct the likely model (P < 0.05). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) at 95% CI of Widal test was calculated by considering culture as gold standard. All statistical tests were considered significant if the P-value is ≤0.05.

A total of 372 febrile enteric fever suspected patients were enrolled during the study period. Majority (59.1%, 220/372) of the participants were females and the rest males (40.9%, 152/372). The age of the participants ranged from 7 to 82 years with mean + standard deviation (SD) of 31.4 + 13.4 years. Over 75.5% of the study participants were able to read and write and 10.7% had history of malaria medication. Overall, 209 patients were positive for enteric fever using Widal test giving an apparent or test prevalence of 52.6% (95% confidence interval [CI] = 50.97–61.29%). The true seroprevalence of enteric fever after correction for sensitivity (73.5%) and specificity (75.7%) of the testwas57.52% (95% CI:52.3–62.6%).The reactions of patients’ sera to O and H antigens using Widal test was presented in Table 1. Chi-square test showed that seropositivity to Widal test was significantly different with respect to age, sex, religion, source of water and meat consumption. The results of Widal test and its association with socio-demographic factors is presented in Table 2.

Diagnosis of typhoid fever using Widal slide agglutination test was common in the current study area and subsequent treatment of patients relies on this test result due to the absence of culture facilities [23].The serological prevalence of enteric fever in this study was 56.2%, while only 10 (2.7%) febrile patients were culture proven to have enteric fever agents. The government of Ethiopia has achieved coverage of safe water to 54% of households, however, nearly 39 million people, most of them in rural areas, don’t have access to safe water and nearly 48 million lack access to basic sanitation [24]. Considering the above fact, the high seropositivity found in the current study is justifiable. Inadequate food and personal hygiene; and lack of potable water have been frequently cited as major risk factors contributing for high seropositivity of enteric fever [1].Moreover, people who live in enteric endemic areas like Ethiopia are supposed to have background antibody due to old infection or repeated infection with agents causing enteric fever or other Salmonella serotypes, that can react with Widal test. Therefore, a single Widal test might not be of great diagnostic relevance in adolescents and adults residing in endemic areas as it leads to over diagnosis and over-reporting. The background or baseline level of antibody to enteric fever determined in healthy population in Ethiopia showed that almost all the blood tested showed some titer of the antibody reactive to slide agglutination tests [25]. This shows low clinical diagnostic reliability of Widal test [12, 23].False positivity in Widal test could also exist due to infection with other Enterobacteriacae possessing cross-reacting epitopes and cross reaction with malaria [1, 26].False negative Widal test result might arose due to testing of patients at early stage of the disease (acute phase) before full antibody response or prior antibiotic treatment or ingested bacterial load is inadequate to induce the antibody production. Such negative Widal test result makes exclusion of enteric fever difficult in a patient with history and signs matching with enteric fever. Thus, the inadequate sensitivity and low specificity of Widal test kits coupled with treatment of patients using single Widal test result alone might have resulted in mistreatment or unnecessary treatment or undesirable treatment outcome or missed appropriate treatment opportunities [12, 23, 26].

The present study showed that the serological prevalence of enteric fever (56.2%) is high while culture prevalence (2.7%) is relatively low. Similar results have been reported previously from Khartoum [13], Nigeria [27] and Ethiopia [23]. As opposed to our study, low seroprevalence of typhoid fever using Widal test, i.e. 19% (38/200) [26] and14.1% [12] have been reported in febrile patients at a Rural Health Center in Northwest Ethiopia and in a 5-year (2007–2011) retrospective study in Addis Ababa, respectively. Similarly, Animut et al. [28] reported low seroprevalence of enteric fever (5.8%) in febrile patients of Quarit, Dembecha and Jabitehnan districts of Northwest Ethiopia.

The sensitivity, specificity, positive and negative predictive values of Widal test as compared to the gold standard test (culture) were 80.0, 44.5, 3.8 and 98.8% respectively. This show that the Widal slide agglutination test had low specificity and low positive predictive values but high sensitivity and negative predictive values. A negative Widal test result is a good indicator of the absence of enteric fever. Similarly, a study conducted in Ethiopia indicated that a negative result of Widal test would have a good predictive value of the absence of the disease (NPV = 98.9%) while a positive result would have a very low value for enteric fever (PPV = 5.7%) [8].

The overall culture prevalence of Salmonella Typhi and Salmonella Paratyphi isolates in this study was 2.7% suggesting that these positive patients might be source of infection for healthy individuals through food handling and contamination of the environment. This low prevalence might be attributed to irregular shedding of the bacteria [1] and to the fact that the majority of the patients might have come from urban areas with relatively better accesses clean water and health care settings. On the other hand, attribution of the isolation rate to clinical disease should be done cautiously since enteric fever is endemic in Ethiopia and carriers do exist. This is similar with the study conducted in Bahir Dar University among food handlers (2.7%) [29]. A relatively higher isolation rate of typhoidal Salmonella was reported from other regions of Ethiopia such as Addis Ababa (3.5%) [30], Arba Minch University (6.9%) [31] and Jimma (6.2%) [32]. However, compared to the present isolation rate, much lower percent of isolation (0.5%, 1/200) was reported through culturing of blood of febrile patients in Northwest Ethiopia [26].This varied prevalence might be due to difference in year of study, sampling technique used, laboratory methodologies used, ecological variations among study sites, difference in cultural and hygienic practices, volume of stool examined, time of sample collection (patients with a history of fever for 7–10 days being more likely than others to have positive culture) and the type of sample cultured (stool versus blood) [1, 19].

Considering the availability of drugs without prescription and the practice of initiation of treatment at home and health-seeking behavior of people when such treatments fail, it is difficult to trust the response of all respondents interviewed aimed at exclusion of those who started antimicrobial usage before coming to hospital. Thus, the low isolation rate might also be partly attributed to initiation of self-prescribed antimicrobial therapy before visiting the hospital.

The high rate of S. Paratyphi isolation (7/10) as compared to S. Typhi (3/10) in the present study, in the absence of S. Typhi vaccination programs in Ethiopia, worth mentioning. This might show S. Paratyphi is the dominant etiological agent of enteric fever circulating in the study area. However further large-scale studies are required to ascertain our finding before considering S. Paratyphi as emerging agent of enteric fever. In accord with our finding, recently, Mahapatra et al. [33] in India isolated 167 Salmonella Spp in a 5-yearretrospective study and reported emerging enteric fever caused by S. Paratyphi due to a dramatic increase in the isolation of Salmonella Paratyphi A (83.8%) than S. Typhi (16.6%) during 2008–2011.

The interesting findings of the present study are that out of the 10 culture positive cases 8 were Widal positive and two febrile patients culture positive for S. Typhi were Widal negative (data not shown). This shows that the humoral immune response confers incomplete protection against relapse or re-infection [1]. Similar to our finding Shlim et al. [34] and Tankhiwale et al. [35] reported that increasing proportions of enteric fever is due to paratyphoid fever. The high rate of isolation of S. Paratyphi than S. Typhi together with antimicrobial resistance of S. Paratyphi might suggest important health challenge in the absence of effective vaccine for paratyphoid fever [3].

In this study, Salmonella isolates showed high rate of resistance to a number of commonly used antimicrobial drugs in Ethiopia such as amoxicillin (100%), bacitracin (100%), erythromycin (100%), cefotaxime (80%), streptomycin (80%) and chloramphenicol (20%). This is in agreement with the findings of Aklilu et al. [30] who reported 100% resistant Salmonella isolates to amoxicillin and erythromycin. In our study, all enteric fever isolates were susceptible for ceftriaxone and gentamicin (100%), hence these drugs could be considered for treatment of enteric fever. All three S. Typhi isolate were susceptible for chloramphenicol, 28.6% of S. Paratyphi isolates were resistant for chloramphenicol while 33.3% of S. Typhi and 100% of S. Paratyphi isolates were resistant to cefotaxime. The relatively lower resistance (20%) to the conventional antibiotic, chloramphenicol, that has been used for so long is not to our expectation. Similarly, Mahapatra et al. [33] from India reported low level of resistance to chloramphenicol. Contrary to our study, high resistance rate of typhoid agents to ceftriaxone (75%), chloramphenicol (83.7%) and gentamicin (75.6%) have been reported in other part of Ethiopia [36]. The detection of one S. Typhi isolate (1/10, 10%) resistant for ciprofloxacin is of concern because this antibiotic is commonly used for treatment of other bacterial infections. It is also used to treat enteric fever due to emergence of MDR to first line drugs like chloramphenicol, ampicillin and cotrimoxazole.

The high MDR among enteric fever isolates suggests that patients treated with those antimicrobial drugs (like amoxicillin) might have had high frequency of poor clinical response, severe disease with potential complications and adverse outcomes [1, 6] as compared to drug sensitive enteric fever isolates. The high rate of MDR also leads to limited therapeutic options and high financial implications. Thus, rational use of antimicrobial drugs and occasional antimicrobial drug resistance surveillance and monitoring are essential to limit the further evolution of multi-drug resistance [6].

Factors which contribute for emergence of AMR are often complex involving the interplay of human, environmental and pathogen-related factors [37, 38].Drug resistance to etiological agents of enteric fever in the present study might probably be due to over- prescription, indiscriminate prescription [23] ease of access (cheap, widely available) to antibiotics [6, 36], antibiotic use practices including self-medication, high frequency of antibiotic use, sub-therapeutic use or indiscriminate use [15] and substandard antibiotic preparations [6].It might also be due to the transfer of resistance genes (plasmid mediated) between enteric bacteria or through chromosomal mutation, that make the bacteria, impermeable to the drugs. In recent decades emergence of plasmid-encoded MDR, particularly to the quinolones has been reported as a major problem [33].

The risk factors identified in the study area showed that seroprevalance is significantly associated with age, religion, level of education, source of water, raw milk and raw meat consumption which might play a major role in the endemicity of typhoid fever in the area.

The high rate of urbanization in Ethiopia and its consequent increasing number of people living in slum where sanitary facilities and provision of clean water are not fulfilled coupled with the poor municipal waste management, might predispose people to use fecal contaminated stream water for food preparation and drinking purpose Discharge of inadequately treated sewage and run-off storm water into the environment can lead to deterioration of quality of water sources [39]. In many studies unsafe contaminated water sources have been linked to enteric fever [40, 41].

Human beings are the major reservoir for enteric fever causative organisms [1] thus milk from healthy cows is not expected to contain agents of enteric fever. However, inadequate hygiene of milkers’ who might be carriers, the use of contaminated water for washing hands, udder and teats and milking utensils might contribute for contamination of milk [18] and high seropositivity in patients who consumed raw milk as compared to those who consume pasteurized or boiled milk. In a review of 68 epidemiological investigations of typhoid fever outbreaks Glynn and Bradley [42] reported milk and ice cream to account for 11 (16%) and 3 (4%) of the outbreaks, respectively. Nevertheless, studies on epidemiology and environmental drivers of enteric fever infection in Africa is limited [43] for further comparison.

Consumption of raw meat, identified as risk factor in the present study, is a well-known and widespread traditional practice in Ethiopia. Carcasses from apparently healthy animals are generally assumed to be free of Salmonella. Contamination of carcass with Salmonella Typhi and Salmonella Paratyphi from food handlers, who might be chronic carriers [29], could occur at different points of the meat chain from abattoir until consumption. Inadequate carcass handling, and inadequate hygienic standards of the people working in the slaughter houses, butchers and restaurants have been previously reported as source of contamination in Ethiopia [29, 44].The use of the same knives and cutting boards for infected and uninfected meat in butcheries and restaurants contribute for cross contamination. The use of unsafe water for washing of carcasses and food contact surfaces might also expose carcass surface and susceptible fleshy parts for contamination [45]. Similar to our finding, Tassew et al. [46] also reported consumption of raw and inadequately cooked meat to be potential risk factors of enteric infection.

The risk of typhoid infection was high in the age group of 19–35 years which is in accord with a study conducted in another region of Ethiopia [12, 30, 31]. This might be due to demographically and economically active and productive individuals in the communities, who are more exposed to occupational hazards of farming related water contact activities, contaminated environment, food and drinks that might be eaten out doors and the nature of the environment in which the food and drink are prepared [47].

People from elementary education are at higher risk (adjusted OR = 2.60, 95% CI: 1.27–5.28; P = 0.009) of acquiring typhoid infection perhaps due to their low level of knowledge and awareness which predispose them to various sources of infections such as from contaminated soil, water, hands and food. In accord to our findings Sutiono et al. [41] from Indonesia reported education as significant risk factor for typhoid fever prevalence. Unlike our findings, Birhanie et al. [26] reported absence of significant association between typhoid fever and source of water as well as educational background. The same authors also reported significant association between hand washing habit and typhoid fever infection, which is different from our finding.

Febrile patients with a Religion categorized as others (Muslims, Abageda, Wakeyfeta) are more likely to acquire enteric fever (adjusted OR = 15.57, 95% CI: 3.01–80.64; P = 0.001) as compared to febrile patients of Protestant Religion. No logical explanation could be given for the significantly high seropositivity in Muslim and Orthodox as compared to Protestant Religion followers. It was difficult to give valid conclusion on this risk factors as it might be due to chance or the small sample size of other Religion followers in the present study. Therefore, further investigation is required to give a more valid conclusion on the role of Religion as risk of enteric fever seroprevalence. Nevertheless, Sur et al. [40] from India reported more frequent case of enteric fever among Muslim than Hindu communities..

Results of Widal seropositivity was significantly high in females (61.36%) than males (48.68%) using univariable logistic regression analysis (p = 0.016), however, sex was not independent predictor in the multivariable model. The household activities of women in food preparation, child and patient care might expose them more to typhoid infection than males [26].

One of the strength of our study is that we assessed about 17 putative factors for their association with enteric fever seropositivity using appropriate model. Findings from this model could be used for improving the control strategies. Nevertheless, it is worthy to consider our findings in light of some limitations such as failure to determine end titer of antibodies, failure to screen blood films to rule out malaria and other febrile illnesses, failure to culture from blood and failure to perform antimicrobial susceptibility testing using all recommended drugs due to shortage. Moreover, the findings are from hospital based cross-sectional study and interpretation of the results to the general population should be cautiously done. Rather our data provide a platform for population based studies in order to ascertain the burden of typhoid fever.