Based on the descriptive results, DHF patients have a slightly higher number of patients above ten years old. This may result from higher outdoor activities in these children with an increased risk of being bitten by Aedes mosquitoes. Gender distribution shows that the majority of DF and DHF patients are male. Male preponderance may be due to socio-cultural environment differences in which males are more exposed to outdoor activities, and females are donned in more covering clothing[17, 18].
From multivariate analysis, prior dengue infection, needing a transfusion, presence of hepatomegaly, and other bleeding manifestations are significant predictors of DHF. In DHF patients with different serotypes, the secondary infection produces more severe manifestations than primary infection through antibody-dependent enhancement (ADE)[19, 20]. This phenomenon enhances T-cell activation in secondary infection with excessive inflammatory cytokines, which promote vascular leak21. A meta-analysis also concludes that antibodies produced from primary infection are incapable of neutralizing the virus, forming immune complexes, and enhancing viral entry. This mechanism allows patients with prior dengue infection have more tendency to develop DHF. Previous multivariate analysis has also reported prior dengue infection as one of the risk factors of DHF.
Hepatomegaly shows significant association with DHF and is included in this study to distinguish between DF and DHF. Direct viral toxicity to the liver or dysregulated immunologic injury results in hepatic manifestations. Hepatocytes and Kupffer cells are prime targets for DENV infection, which causes cellular apoptosis. DENV infected cells then induce the production of proinflammatory cytokines and chemokines, which mediate the increase in vascular permeability leading to plasma leakage in DHF. Following decreased liver perfusion as a result of plasma leakage1,25, hepatomegaly may occur in DHF patients. Under WHO 2011 dengue criteria, hepatomegaly is one of the warning signs of severe dengue. This finding is supported by a previous study of Ferreira et al., concluding that hepatomegaly is a significant predictor of DHF. Disease course of DF patients has not progressed to plasma leakage. Therefore, early signs such as hepatomegaly have not been present in DF patients.
The presence of other bleeding manifestations is also found to be significant to distinguish between DF and DHF. Bleeding may decrease thrombocytopenia, contributing to decreased bone marrow function, shortened platelet survival, and escalation of platelet consumption in DHF patients27. In addition, as the disease progresses to DHF, the coagulation system may also be impaired. Regarding this condition, DHF patients have a higher tendency and occurrence of other bleeding manifestations, hence significant in differentiating DF and DHF. Several multivariate studies have also reported hepatomegaly, and bleeding manifestations as significant clinical signs of dengue shock syndrome (DSS) encompassing grade 3 and grade 4 DHF[29–31]. Bleeding manifestations may exacerbate plasma volume loss due to leakage, thus accelerating the occurrence of shock, which result in mortality.
The need for transfusion is another significant parameter to distinguish DHF and DF. In this study, several dengue patients received fresh frozen plasma transfusion, thrombocyte concentrate transfusion, or both. In conjunction with the presence of other bleeding manifestations as a significant factor in DHF, this event leads to requirements of platelet transfusion therapy in DHF patients compared to DF as a consequence of the bleeding. DHF patients have an increased tendency to undergo severe hemorrhage, associated with abnormal immune response. Transfusion are given with massive bleeding or very low platelet counts to prevent bleeding complications. Hence, the risk of major bleeding in DHF patients is greater with profound shock encountered in DSS. Haemostatic system is also impaired in DHF and increased endothelial dysfunction leading to capillary fragility. Cytokines and chemokines release also cause systemic effects to induce plasma leakage, therefore DHF patients may require plasma transfusion to manage shock with close monitoring of hematocrit. On the other hand, severe hemorrhage and plasma leakage are not found in DF patients. Thus, transfusion is unnecessary.
Based on DHF criteria of WHO 2011, the main hematological parameters to differentiate DHF and DF are hematocrit rise ≥20% and thrombocyte level of <100,000/µl. These findings are constant features of DHF. An increase in hematocrit describes the condition of hemoconcentration. Plasma leakage through damaged blood vessels to the extravascular leads to an increased percentage of hematocrit consequent to deficiency in blood plasma related to blood viscosity[37, 38]. Thrombocytopenia results from an immunological reaction as DENV binds to platelets. These events enhance platelet aggregation and platelet destruction through apoptosis[34, 35]. Proliferative capacity of hematopoietic cells is also suppressed as a result of DENV infection. However, in this study, thrombocytopenia is not a significant variable to distinguish DHF and DF through multivariate analysis. This may be consequent to some DHF patients who had not yet reached thrombocytopenia below 100.000/µl but already had evidence of plasma leakage. Thus, they were still categorized as DHF patients. One explanation is that platelet counts gradually fall and then reach the nadir later in the disease. Furthermore, the patients may be admitted at the late stage of the disease in which the platelet count has increased gradually. This causes thrombocytopenia is not found to be a significant variable to distinguish DHF and DF. Moreover, this study could not analyze the significance of hematocrit as a factor to differentiate DHF and DF as all patients with DHF had a rise of hematocrit ≥20%. Therefore we could not obtain the p-value of this variable by logistic regression.
Through ROC analysis of Youden’s index, the predictor model in this study has good specificity of 86.3% yet is less sensitive with a sensitivity value of 55.6%. This model has less ability to differentiate DHF and DF patients. However, this model is specific enough to reduce false positives, in which by fulfilling at least two significant variables, it is specific to identify the patients to have dengue infection. This is because these significant variables can be encountered in other diseases. Patients with coagulation or hematologic disorders may have bleeding manifestations37 thus requires transfusion therapy of thrombocyte and/or FFP, or patients with hepatitis, hepatic abscess, or other infections may result in hepatomegaly. As various diseases with similar manifestations may overlap with significant variables found in this study which causes this model to be less sensitive.
This study has several limitations. As the data was collected from medical records, some laboratory values and demographic data were incomplete. Patients in this study did not receive reverse transcriptase polymerase chain reaction (RT-PCR) examination therefore, we were unable to identify serotypes of current DENV infection in patients with a history of previous dengue infection. Hence, we could not verify the significant association between prior dengue infection and DHF. The model in this study was also less sensitive to different DHF with other febrile illnesses commonly found in tropical countries. Therefore further studies with prospective methods in different populations are needed to find more sensitive and specific variables. However, this study included a large sample representing the clinical and laboratory characteristics of DF and DHF patient populations. In addition, this study yielded a good predictive model shown by the AUC value. Thus this study may benefit countries and regions with few populations with limited advanced laboratory methods for rapid identification to prevent disease progression into a severe form of DHF.