Background
Two-fifths of the world’s populations in tropical and subtropical countries are at risk of dengue. An estimated 500,000 people with dengue infection require hospitalization each year [
1,
2]. Based on a 20-year registry of dengue infection age based-epidemiology in Indonesia, there appears to be an epidemiological shift in the age of affected individuals. Since 1999, a steady decline in incidence was observed for children aged 5 to 14 years (the age group with highest DHF incidence, historically). The decline surpassed the increasing incidence of affected individuals aged over 15 years [
3].
The presence of plasma leakage, one of the characteristics of dengue hemorrhagic fever (DHF), is the prominent cause of severe dengue [
4,
5]. The timely diagnosis and management of plasma leakage are very important [
4]. Fujimoto et al. reported that 5.7 % of patients with plasma leakage developed cardiorespiratory dysfunction, and the mortality rate reached 7.3 % [
6]. Chairulfatah et al. reported 6 % severe bleeding in 1300 pediatric and adult DHF cases in Indonesia [
7].
The World Health Organization (WHO) guidelines defined plasma leakage with the occurrence of hemoconcentration and/or hypoalbuminemia and/or serous effusion [
2,
5,
8‐
10]. However, there are several limitations in diagnosing plasma leakage using these criteria. In clinical practice, clinicians often detect pleural effusion and/or ascites in patients with elevated hematocrit values of less than 20 %, a cut off used to define hemoconcentration as recommended by WHO (1, 8, 9). Moreover, previous studies showed albumin level <3.5 g/dL in dengue fever (DF) and DHF patients, which is in contrast with WHO criteria that use 3.5 g/dL to differentiate patients with and without plasma leakage [
2,
11‐
14]. Compared with hematocrit and albumin level, pleural effusions and/or ascites as visualized by ultrasonography (USG) are highly sensitive and specific for determining plasma leakage [
15‐
18]. However, USG is not widely available in resource limited areas [
19]. Despite the currently accepted laboratory parameters for determining plasma leakage, pathophysiologically there are associations between plasma leakage identified by the presence of pleural effusions and/or ascites with thrombocytopenia and elevated hepatic transaminase levels [
20,
21]. To the best of our knowledge, there are no studies reporting the cut off point of the lowest platelet count and elevated ratio of transaminase levels at the critical phase for diagnosing pleural effusion and/or ascites. Similarly, no reports regarding the cut off point of hyponatremia, which is another common manifestation of plasma leakage [
10], is available.
Here, we conducted a study to develop a dengue scoring system to predict pleural effusion and/or ascites using laboratory parameters, such as the degree of hemoconcentration, lowest albumin concentration at the critical phase, degree of hypoalbuminemia, lowest platelet count, elevated ratio of aspartate aminotransferase (AST), alanine aminotransferase (ALT), and sodium concentration at critical phase. This Dengue Score can be used to identify pleural effusions and/or ascites, which are better indicators of plasma leakage, to stratify dengue-infected patients at risk of developing severe dengue.
Discussion
This is the first Indonesian study to determine a scoring system to predict pleural effusion and/or ascites in adult patients with dengue infection. Michels et al. and Balasubramanian et al. showed that pleural effusion and/or ascites detection by USG is superior to clinical and laboratory parameters for diagnosing plasma leakage [
15,
16]. In this study, we tried to determine independent diagnostic predictors of pleural effusion and/or ascites and to convert the prediction model into a scoring system that could be practically used because USG is not routinely available in all healthcare facilities. Hematocrit, albumin, hepatic transaminases, platelet count, and sodium concentrations are laboratory parameters that are routinely checked in the management of dengue patients as recommended by the WHO [
2,
5,
10]. Therefore, the scoring system to predict the plasma leakage that was represented in this study by the detection of pleural effusion and/or ascites would be highly applicable in various healthcare settings where USG is not available.
As reported in other study, there is a significant difference in the degree of hemoconcentration in dengue infected patients with and without pleural effusions and/or ascites [
28]. We found a similar result, with a degree of hemoconcentration of 12.67 % (inter-quartile range - IQR 7.03 %) in patients without pleural effusion/ascites and 18.92 % (IQR 10.81 %) in patients with pleural effusion/ascites. Translating this finding, DHF would be underdiagnosed by physicians relying only on hematocrit as a diagnostic criterion. The ROC analysis showed that compared with classically used cut off point of degree of hemoconcentration ≥20 % as suggested by the WHO [
5,
10], a cut off point ≥15.1 % gives a better performance for predicting the presence of pleural effusion and/or ascites. Our study suggests the use of the lowest cut off point of hemoconcentration to define plasma leakage to minimize the risk of under-diagnosing patients at risk of severe dengue. A degree of hemoconcentration ≥15.1 % was given a score of 1 in the final Dengue Score.
The WHO defines a significantly decreased albumin >0.5 g/dL from baseline or <3.5 g/dL as indirect evidence of plasma leakage [
2,
10]. However, two publications reported low albumin levels (<3.5 g/dL) in patients with DF and DHF [
11,
12]. Jagadishkumar et al. [
11] and Itha et al. [
12] found median albumin levels of 3.1 g/dL to 3.37 g/dL in DF and 2.7 g/dL to 3.23 g/dL in DHF groups based on the WHO criteria. Roy et al. [
13] reported that mean albumin levels in dengue-infected patients without and with warning signs were 3.48 and 3.34 g/dL, respectively. We observed that the use of two albumin-related parameters, i.e., the lowest albumin concentration at the critical phase and the degree of hypoalbuminemia, for predicting a diagnosis of pleural effusion and/or ascites may not be conclusive. We found that ≤3.49 g/dL and ≥15.6 % were the best cut off points for the two parameters, respectively. We suggest using degree of hypoalbuminemia as a percentage to detect plasma leakage, similar to the concept of degree of hemoconcentration. However, due to the similarity in AROC between the 2 parameters, we selected the lowest albumin concentration at critical phase, which is more practical in daily practice, as albumin is only measured once at critical phase. In DHF, hypoalbuminemia results from the loss of albumin due to the occurrence of plasma leakage. This phenomenon is similar with the pathophysiology of pleural effusion and/or ascites in DHF, thereby explaining the association between the two conditions [
29]. Similar with degree of hemoconcentration, lowest albumin concentration at critical phase ≤3.49 g/dL was given a score of 1 in the final Dengue Score.
Thrombocytopenia has been accepted as a major sign of plasma leakage in dengue-infected patients. Damage to platelets, which possibly contribute to thrombocytopenia in dengue patients, results in the release of vascular endothelial growth factor (VEGF) which in turn is responsible for the occurrence of pleural effusions and/or ascites [
30,
31]. As shown in Table
1, we found a significant difference in the medians between patient with and without pleural effusion and/or ascites. Using cut off point of ≤49,500/μL, it gave the best AROC among all the variables tested. This variable was given a score of 1 in the final Dengue Score.
Our data show that liver injury, manifested by an elevated ratio of hepatic transaminases, was frequently present in adult patients with dengue infection. Liver cells are one of the targets of dengue viruses. The liver dysfunction was mild to moderate, presenting primarily as elevations of hepatic transaminases, with significantly higher AST elevation in patients with pleural effusion and/or ascites [
32,
33]. Consistent with the findings of Mahmuduzzaman et al. [
34], we found a greater increase of AST compared with ALT in dengue patients. In the multivariate analysis, an elevated ratio of ALT failed to show a significant association with pleural effusion and/or plasma leakage. The exact significance of the greater elevation of AST compared with ALT in dengue might be because of the excess release of AST from damaged myocytes during dengue infections [
35]. Further research is needed to confirm this hypothesis. An elevated ratio of AST is a laboratory marker of dengue severity that has been proposed by several researchers but has not gained much attention, as reflected by its absence in the plasma leakage criteria of the WHO [
2,
5,
10]. By calculating the transaminase elevation ratio, we are attempting to address the problem of the difference between laboratories in the reference limits of the AST [
36]; therefore, this diagnostic predictor variable could be universally used across laboratories.
This study has limitation that should be considered. This study was conducted in patients without comorbidities; therefore, the generalization should be re-tested in population with comorbidities, especially in those with comorbidities that potentially influencing the levels/concentration of predictors’ variables, e.g., chronic kidney disease and chronic liver disease. Future study is needed to validate the Dengue Score and to investigate the effect of the application of this Dengue Score in various healthcare facilities managing dengue-infected patients.
Abbreviations
ALT, alanine aminotransferase; AROC, area under receiving operating characteristic curve; AST, aspartate aminotransferase; DF, dengue fever; DHF, dengue hemorrhagic fever; LR-, negative likelihood ratio; LR+, positive likelihood ratio; NPV, negative predictive value; NS, nonstructural protein; OR, odds ratio; PPV, positive predictive value; ROC, receiving operating characteristic; Sens, sensitivity; Spe, specificity; USG, ultrasonography; VEGF, vascular endothelial growth factor; WHO, World Health Organization
Acknowledgments
The authors thank Kresna Dharma Suryana, MD, Eleanor Louana Urfa, M.Biomed, Beti Ernawati Dewi, Ph.D, and Benediktus Yohan, M.Biomed for helps in collecting data and performing dengue serotyping.