Detection of dengue virus infection in children presenting with fever in Hawassa, southern Ethiopia

Dengue fever is a mosquito-borne viral infection, with rising incidence globally. Eastern Ethiopia has had dengue fever outbreaks in recent years. However, the extent to which the infection contributes to hospital presentation among children with fever in southern Ethiopia is unknown. We examined 407 stored plasma samples collected to investigate the aetiology of fever in children aged at least 2 months and under 13 years presenting to the outpatient of the largest tertiary hospital in southern Ethiopia. We analyzed samples for dengue virus non-structural 1 antigen using enzyme-linked immunosorbent assay. The median (interquartile range) age of the 407 children examined was 20 (10–48) months, and 166 (40.8%) of the children were females. Of 407 samples analyzed, 9 (2.2%) were positive for dengue virus non-structural 1 antigen, of whom 2 were initially treated with antimalarial drugs despite having negative malaria microscopy, and 1 of the 8 patients had a persistent fever at the seventh day of follow-up time. The presence of active dengue virus infection in the study area highlights the need for studies at the community level as well as the integration of dengue diagnostics into fever-management strategies. Further research to characterize circulating strains is warranted.

www.nature.com/scientificreports/ tests that we used for this study. We have subsequently been able to undertake dengue testing, and therefore we can now report the results in our cohort of febrile children.

Methods
Study setting and samples. The study setting and overall design have been previously reported in detail 18 .
In brief, the study was conducted at Hawassa University Comprehensive Specialized Hospital (HUCSH), which is the largest in southern Ethiopia. Hawassa City is situated on the shore of Lake Hawassa at 1,708 m above sea level, with an annual temperature range of 9-29 °C (a mean monthly temperature of 19.7 °C), and a mean annual rainfall of 961 mm 19,20 . A consecutive series of 433 children aged at least 2 months to less than 13 years presenting to the hospital were prospectively enrolled from May 2018 through February 2019 to assess aetiologies of fever. Fever was defined as a temperature of 37.5 °C or a history of fever within the preceding 48 h, lasting no longer than 7 days. Critically ill patients for whom blood or urine cultures were not performed as part of their care at admission were excluded due to ethical reasons. In July 2022, stored plasma samples from 407 children were examined for the current study to make a retrospective diagnosis of dengue virus infection. Due to inadequate blood volume, plasma samples from the remaining 26 enrolled children were not stored.
Diagnosis, clinical management, and outcome. As described in detail elsewhere 18,21 , hospital staff gathered clinical information from physical examinations and history-taking. Laboratory investigations including complete blood count (CBC), malaria microscopy, human immunodeficiency virus (HIV) testing, urinalysis and urine culture, and blood culture were performed for each participant. Other tests were performed based on relevant case presentations 18 . Clinicians managed participants on the day of enrolment in accordance with their presentation and results of malaria microscopy, CBC, and urinalysis. Within a week after enrolment, additional diagnoses and management were made based on urine or blood culture findings. Data on fever status were gathered on day 7. Records for each child were retrospectively reviewed by senior paediatricians, who determined whether antibacterial or antimalarial treatments had been indicated for the child at the time of presentation, based on Ethiopian national guidelines 22 . Case definitions for clinical and laboratory indicators are presented as supplementary information (Table S1).

Serological analysis for the retrospective diagnosis of dengue virus infection.
Plasma specimens that had been stored at − 70 °C for about 3 years were tested for the specific dengue non-structural protein 1 (NS1) at Armauer Hansen Research Institute in Addis Ababa, using the commercially available Dengue Virus NS1 ELISA (EUROIMMUN Medizinische Labordiagnostika AG, Lübeck, Germany). Briefly, the plasma sample was diluted 1:2 in sample buffer and added to microplate strip wells precoated with monoclonal anti-dengue virus NS1 antibodies against all the dengue serotypes and incubated for 1 h. After a washing step, peroxidase labelled anti-dengue virus NS1 antibody was added to the wells and incubated for 1 h followed by a washing step. Then, after the addition of the chromogenic substrate and incubation for 15 min, the stop solution was added. Shortly, a photometric measurement was made at 450 nm/620 nm and a semi-quantitative and quantitative determination of NS1 antigen was performed based on the absorbance of the samples and calibrators. In the semi-quantitative analysis, a result is considered positive when the ratio of extinction reading of a sample to a calibrator (standard) is ≥ 1.1. For the quantitative analysis, the concentration of NS1 in a sample in Relative Units (RU)/ml was determined from a standard curve plotted using extinction readings measured for the 3 calibration sera (standards) against the corresponding relative units; hence, a sample with ≥ 11 RU/ml was considered as positive. The test is claimed to have a 99% sensitivity and specificity, and is carried out in accordance with the manufacturer's instructions. An acute dengue virus infection was defined as the presence of NS1 antigen in the blood (NS1 antigenaemia).

Data analysis.
Statistical Package for Social Sciences version 20 (IBM Corp., New York, USA) was used for data entry and analysis. Z-scores for anthropometry were calculated using the WHO AnthroPlus software 23 . The detection of NS1 antigen in relation to demographic and clinical characteristics was expressed as frequencies with percentages. The crude odds ratio (cOR) for the evaluation of factors related to the prevalence of dengue virus infection was calculated using binary logistic regression analyses. An association was judged as statistically significant if the corresponding p-value was less than 0.05. Children's caregivers were given adequate information about the study and written informed consent was obtained from all caregivers, in addition to assent from children aged at least 12 years. Code numbers were used in place of identifiers to ensure the confidentiality of collected information. All methods were performed in accordance with the relevant guidelines and regulations.
Dengue was detected in 3 (3.9%) of 76 children with a fever of 5-7 days at presentation, 5 (2.3%) of 218 with a cough, 2 (2.6%) of 76 with diarrhoea, 3 (2.0%) of 148 with vomiting, and 2 (3.4%) of 58 with an axillary temperature of at least 39 °C. Moreover, 2 (4.2%) of 48 children with anaemia and 0 (0%) of 38 children with leukopenia were positive for dengue ( Table 2). The presence of NS1 antigen was not significantly associated with any demographic or clinical characteristic of the participants (Tables 1 and 2).
As shown in Table 3, of 9 children who tested positive for NS1, 4 were diagnosed with pneumonia, 1 with tonsillopharyngitis, and 2 with acute diarrhoea at the time of initial management on enrolment day. When the culture results were released following initial management, 2 of the 8 children with urine cultures and 1 of the 9 children with blood cultures had urinary tract infections (Escherichia coli, Klebsiella pneumoniae) and bacteraemia (Staphylococcus aureus), respectively.
Although all 9 NS1-positive participants were negative for malaria, 2 of the 9 children had initially been given antimalarial medication. Based on the empiric guidelines, 6 of the 9 children were judged to have initial clinical indications for antibacterial drugs, but 7 of the 9 children received this treatment (Table 4).
On day 7 of the follow-up, 8 of the 9 NS1-positive children were contacted. Of these, the caregiver of one child reported persisting fever, and 7 reported that fevers had subsided within 3 days following initial management (Table 4).

Discussion
To our knowledge, this is the first report of acute dengue virus infection (NS1 antigenaemia) in southern Ethiopia. We found that 2.2% of febrile children presenting to a tertiary hospital in southern Ethiopia were positive for dengue NS1 antigen, an early marker of acute dengue virus infection. Compared to our finding, dengue IgM, a sign of recent infection, was seen in all ages of patients in southern (8.1%) and Northwest Ethiopia (19%) 6,7 . The discrepancy seen may be because previous studies tested for dengue-specific IgM, which is detectable for about 3 months after the onset of illness as opposed to the NS1 antigen, which is detectable for 1-2 weeks 9-11 . Moreover, the participants in the current study were children with localized or non-localized febrile illnesses who consecutively presented to a tertiary hospital, as opposed to earlier studies when all ages of patients without any clear clinical focus of infection or suspected of dengue infection were examined. www.nature.com/scientificreports/ Studies in non-outbreak settings showed higher percentages of NS1 detection than we did, with NS1 antigen being found in 4.4% of all ages outpatients with undifferentiated fever in the Democratic Republic of the Congo 24 , and 6.1% of children under 15 years of age with dengue-related symptoms in Cameroon 25 . While a prevalence of NS1 antigenaemia as high as 35% is seen in all age febrile patients in Nigeria 26 , a prevalence as low as 2.2% in the same nation 27 is reported that is compatible with our observation. An NS1 antigen prevalence of 24.1% is shown in a systematic review among dengue suspected cases in Southeast Asia 28 . Overall, the findings suggest that dengue, as an endemic and epidemic-prone arboviral infection, likely plays a varying role as a cause of illness in both place and time.
The ecological environment and climatological changes and patterns such as seasonality influence the breeding and abundance of Aedes mosquito vectors and contribute to the heterogeneity of the infection among African residents 29 . However, the limited studies that are available in Ethiopia make it difficult to comprehend the dynamics of dengue and urge further investigations into its epidemiology. The reported low proportion of dengue at the study hospital, which is located close to Lake Hawassa where mosquito-borne infections are expected to be common, may be explained by the hospital's regional referral function, which draws severe cases such as severe pneumonia, as shown in our initial report 21 . It is also possible that caregivers were more likely to seek care from lower-health facilities than the hospital when they suspected malaria and similar acute febrile illness, as observed in our recent report 30 . As a result, it would likely be more representative to expand the scope of such studies to include lower-level health facilities to capture mild acute febrile illnesses.
Early diagnosis of dengue is essential to reduce the risk of complications and prevent the spread of the virus 9 . Given the growing body of evidence showing dengue outbreaks in Ethiopia 4,5 , introducing rapid dengue testing Table 2. Distribution of dengue virus infection by clinical characteristics and laboratory findings of children attending HUCSH, 2018-19. cOR crude odds ratio, CI confidence interval, WBC white blood cell. † Percentages within categories of the characteristics (raw total), c (N = 404).

Dengue virus infection n (%) blood tested (N = 407) n (%) † positive n (%) † negative cOR (95% CI)
Clinical history    www.nature.com/scientificreports/ aids to make appropriate fever management. Specifically, determining a viral aetiology of fever may minimize the unnecessary prescription of antibacterial and antimalarial drugs. Despite the small number of dengue cases in the current study, we were still able to observe antimalarial treatment being given to children who had tested negative for malaria but later found positive for dengue. One of the limitations of this study was that dengue markers other than the NS1 antigen were not examined. The analysis of the samples for dengue RNA using more sensitive techniques such as RT-PCR might have recovered additional dengue cases and may have allowed for the identification of circulating dengue serotypes 9,12 . Antidengue IgM and IgG detection could have provided information on the level of recent and previous dengue infections circulating in the community 10 . The small number of dengue cases detected precluded conclusions about associated clinical characteristics. Dengue was not attempted to be categorized by severity. The generalizability of results to all febrile children may be constrained by selection bias arising from an institution-based study.

Conclusion
A low prevalence of acute dengue virus infection was observed in febrile children presenting to a tertiary hospital. However, dengue was commonly misdiagnosed as other febrile illnesses and goes unnoticed, leading to an overuse of antimalarial and antibacterial medications. Our findings suggest that dengue needs to be adequately highlighted, including the importance to integrate dengue diagnostics into fever-management approaches. Further study is urgently required into the level of transmission at the community level and ecological factors associated with dengue fever to lessen the threat of outbreaks and characterise circulation strains.

Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author, upon reasonable request and with the Institutional Review Board of the Hawassa University College of Medicine and Health Sciences. Restrictions apply to the availability of these clinical data, which caregivers had consented for the collected information to be used for our research study only, and so are not publicly available.