Background
Dengue has become a global public health concern, especially in most tropical and subtropical countries. It is a mosquito-borne (
Aedes aegypti and Aedes albopictus) arboviral disease caused by the dengue virus (DENV) [
1]. Over the last 60 years, the dengue virus has spread to over 130 countries, causing nearly 10,000 deaths and 100 million symptomatic cases every year [
2,
3]. Besides, more than 50% of the global population are at risk of dengue transmission, with the vast majority in Asia, followed by Africa and America [
4]. It is one of the leading causes of death among children in Southeast Asia [
2].
In Bangladesh, the first recognized dengue outbreak was reported in the capital city, Dhaka in 1964 [
5]. Subsequently, sporadic dengue cases were reported [
6] until 2000, when the first major epidemic occurred throughout the major cities and towns of Bangladesh [
7]. In 2019 the most extensive and deadliest outbreak of dengue occurred in the history of Bangladesh. Over 100,000 people were hospitalized, and 129 deaths were recorded [
8]. The unofficial number of cases and deaths might be higher as the health reporting system is poor in the country. The high mortality was suspected to be associated with a high incidence of dengue shock syndrome (DSS) and secondary dengue infections [
9].
DENV has four different serotypes (DENV-1 to 4). A peculiarity of DENV infection is that infection by one serotype produces serotype-specific lifelong immunity; however, contrary to giving protection or remaining neutral against other serotypes, a secondary infection by a heterogeneous serotype often results in severe disease by a mechanism called antibody dependent enhancement (ADE) [
10]. All four serotypes have been isolated in Bangladesh with a predominance of DENV-3 till 2002 [
11,
12]. After 2002, DENV-1 and DENV-2 were the prevalent serotypes, which increased the susceptibility of severe secondary infection by other serotypes [
13]. In 2017, the DENV-3 serotype reemerged, causing a sharp rise of dengue cases in 2018 [
14]. Thereafter, the 2019 outbreak was predominantly caused by type I genotype of the DENV-3 serotype [
15].
Among children, nearly 95% of dengue cases are aged less than 15 years [
16]. Owing to their immature hemodynamic system, children and particularly infants, tend to develop severe dengue disease [
17,
18]. National surveillance data from Asian countries show that infants under 1 year of age and children aged 4–9 have consistently been at the highest risk for severe dengue disease [
18].
Being a systemic disease, the clinical features of dengue show wide-ranging variations. Many of the clinical features were explored, but some (like constipation, mouth sores) are seldom reported. Moreover, children with dengue show significant variations in symptoms and signs in comparison to adults. Pain symptoms of dengue infection are less common among children [
19‐
21]. Severe infection, when present, maybe preceded by so-called ‘warning’ signs [
22]. Many clinical and laboratory factors have been attributed to the severity of the disease [
23]. Early detection and appropriate management of severe dengue can reduce dengue-associated mortality [
24]. Hence characterization of typical and atypical clinical features and determinants of severity, especially in the context of an outbreak, is essential. Previous studies have described the clinical presentations of dengue in the pediatric population of Bangladesh [
25‐
28] from single centers and with a small number of samples. Globally, detailed profiling of clinical features, including usual and unusual clinical presentations of dengue in the pediatric population, is lacking. We hypothesized that the ongoing outbreak was characterized by increased severity and atypical clinical features suggestive of multi-system involvement among children with dengue. This study aimed to investigate that hypothesis, describe the typical and atypical clinical features, and model the clinical predictors of the severity of dengue among children in the setting of Bangladesh.
Discussion
This was a large multicenter cross-sectional study of dengue among children, which reports the atypical and less reported clinical presentations, and models severity predictors of dengue in Bangladesh. ‘Mouth sore’ and ‘constipation’, two previously less reported gastrointestinal features, were prevalent among our participants. Atypical features were found mainly in the form of neurological manifestations, with confusion and blurring of vision being predominant. Age, mouth sore (in the form of sore throat and/or mouth ulcers), and decreased platelet with increased hematocrit were clinical predictors of severity.
Nearly one-third of cases had severe dengue in our study, which is higher than that (one-fifth of cases) found in 2018 by both Shultana et al. [
27] and Afroze et al. [
28] in two separate tertiary care hospitals (sample size,
n = 89 and 106 respectively) in Dhaka city. Proportionately dengue shock syndrome cases were higher than any prior outbreaks recorded in the country [
25,
27,
28]. This finding supports our hypothesis that the outbreak of 2019 was relatively more intense than the previous ones in terms of clinical severity. Besides, we found six cases of expanded dengue syndrome among whom at least two had features suggestive of encephalitis- a rare manifestation of dengue infection [
34]. However, only 1.6% of children had a past history of dengue indicating that the severe cases cannot be explained by the secondary infection by a different serotype alone. The re-emergence of DEN-3 serotype in 2019 [
15] combined with the fact that severe primary infection may occur in children [
18] might explain the higher severity noted in this study. Moreover, several studies conducted within the last 5 years in India [
35], Indonesia [
36], and Mexico [
37] found respectively 35, 20, and 58% severe dengue cases among their participants, supporting our hypothesis that dengue among children is occurring with increased severity.
Compared to the 2018 outbreak, children infected with dengue were older in the 2019 outbreak in the country [
27,
28]. However, our age pattern matches the 2000 outbreak [
25,
38], indicating a temporal variation in the age of involvement among children in this region. Additionally, we found that younger age is significantly associated with severe disease in multivariate analysis. This finding is also supported by surveillance studies from Asian countries [
18].
A higher proportion of males among dengue-infected children of this study has been persistently noted in Bangladesh [
27,
28] and several South-East Asian countries [
32‐
35]. Anker and Arima [
39] explored the sex-related differences in the prevalence of dengue in more detail. They noted that the magnitude of the difference is small and is not consistent in pediatric patients. However, male-female differences in the use of health services, the use of fully covered dresses by female children, and prioritizing provisions of male children in the society might be reasons for the differences noted in our country [
40,
41].
Approximately three-fourths of `the families in this study came from the lower economic categories, with 33.7% having income less than 15,000 BDT per month (equivalent to USD 17.7). One explanation for this could be people with a lower monthly income are more likely to live in crowded places having a higher risk of contracting the disease from an infected person. Another possible reason is that public health facilities, from where our data was collected, are mostly visited by poor patients in the country [
42].
The most frequent blood group among dengue infected children in our study was ‘O’ positive in contrast to the most common ABO and Rh blood type (‘B’ positive) noted among healthy donors of the country [
43,
44]. Reports on the association between blood groups with dengue fever have been mixed. Khode et al. [
45] found an association of ‘O’ positive blood groups with higher prevalence of dengue. But, in the study by Ravichandran et al. [
46] it was ‘AB’ positive. Contrary to Kalayanarooj et al. [
47] we did not find any association of severity of dengue with any particular blood ABO and Rh blood type. Therefore, the high prevalence of ‘O’ positive blood in our study could be a random finding.
Children with dengue show varying clinical presentations throughout the world. But fever is the most common symptom in dengue irrespective of types and severity. Variation exists in other symptoms and signs associated with fever. Likewise, we found a high-rise fever in all cases. Gastrointestinal features were more frequent in our study than that of previous year [
48,
49]. Constipation, a seldom sought for feature in studies involving dengue [
48,
49], was found to be more frequent than diarrhea in our patients. This might be consequent of decreased appetite and vomiting found in higher proportion in our study. However, the proportion of cases presenting with diarrhea was not less than other recent studies either [
50,
51].
One unique finding of this study was independent association of sore mouth (including oral ulcers and/sore throat) with severe disease. Although, it was not found to be a factor for severity by Zhang and colleagues in their systematic review and meta-analysis of predictive symptoms and signs of severe dengue [
52]. Sore mouth is a less common feature of dengue compared to other febrile illness which is usually reported as sore throat [
51], therefore presence of it especially among younger children with dengue should raise concern regarding severity.
The most common bleeding manifestations was epistaxis and melena similar to several previous studies [
53,
54]. A significantly higher proportion of melena was noted in severe dengue than the non-severe one consistent with previous evidence [
25,
52].
The unusual features presented by children with dengue were mostly suggestive of nervous system involvement except palpitation and chest tightness which might have occurred due to hypovolemic shock associated cardiac ischemia [
55] or due to transient involvement of cardiac muscles [
32]. The neurological symptoms were suggestive of encephalopathy due to hypovolemic shock, and/or encephalitis and meningitis due to direct involvement of nervous tissue by the virus [
56].
As older children could report their symptoms properly a higher proportion of pain related symptoms was noted among them. Interestingly female children also reported a higher frequency of different aches probably as a result of a lower threshold of pain sensitivity evident in women [
57]. The high prevalence of mouth sores, rash, decreased appetite, and constipation among the youngest children might be explained by combinations of a developing immune system, decreased leucocyte counts, and a severe disease in this group.
We used conservative cut-off values for blood counts so that even slight elevation and reduction in specific investigations are not missed. Therefore, elevations in hematocrit value, and reductions in hemoglobin, leucocyte count, platelet count, were statistically similar between non-severe and severe dengue. But an elevation of hematocrit (> 20% from baseline) combined with a reduction in platelet count (< 50,000/mm
3) was found to be a significant predictor of severe disease on multivariate analysis. Also, this was the only warning sign among others included in the multivariate logistic regression that came out as a significant predictor of dengue severity. Contrary to the estimates of Zhang et al. [
52] our study did not find abdominal pain, vomiting and rash to be significant predictors of severity.
There were several limitations of our study. First, we had a small final sample size due to limited data collection and curation of incomplete data. Our data collection was limited in that we could not include the specialized centers dedicated for mother and child health in the city as because the data on pediatric dengue cases were collected in conjunction with adult cases from the multidisciplinary tertiary care hospitals. Second, we used a non-random convenience sampling, due to the rapidly evolving nature of the outbreak, to collect as many responses as possible near the end of peak time. This should be taken into consideration while interpreting our result. Third, the clinical classification of dengue was used for stratification of severity which might cause misclassification bias due to the subjective nature of the assessment. However, the initial classification of a patient during admission was subsequently verified by experts before discharge reducing the bias. Fourth, our design was cross-sectional in nature making the relations found between predictors and severity general associations.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.