OUP user menu

Patterns of disease among adults hospitalized with dengue infections

G.N. Malavige, V.G.N.S. Velathanthiri, E.S. Wijewickrama, S. Fernando, S.D. Jayaratne, J. Aaskov, S.L. Seneviratne
DOI: http://dx.doi.org/10.1093/qjmed/hcl039 299-305 First published online: 7 April 2006


Background: Recent years have seen an increase in dengue infections among adults in Sri Lanka, with similar trends seen in many other countries. Data on the natural history and outcome of dengue in adults are quite limited.

Aim: To study clinical and laboratory findings in adult dengue patients hospitalized in Sri Lanka during a recent major dengue epidemic.

Design: Prospective observational study.

Methods: Clinical, laboratory and demographic information were collected from adult patients with confirmed dengue infections (n = 108) treated in a general medical ward in Sri Lanka from 24 April to 31 July 2004.

Results: There were 68 male and 40 female patients, mean age 26.6 years. Dengue fever (DF) was seen in 33 (30.6%) and dengue haemorrhagic fever (DHF) in 75 (69.4%). Of the 37 (34.3%) with primary dengue infections, 19 (51.4%) developed DF and 18 (48.6%) developed DHF. Overall, 42 patients (38.9%) had bleeding manifestations. These adults showed differences in clinical and laboratory findings, disease severity and mortality, compared to children seen during the same epidemic. Secondary dengue infections were significantly associated with development of severe disease (OR 5.0, 95%CI 1.9–13.5, p < 0.001). Mortality was 3.7%.

Discussion: Pooling data on adult dengue patients from different regions should help us to understand the natural history of disease in this group. It would also help in developing evidence-based treatment guidelines and allocating limited and scarce health resources. Our data contribute towards this goal.


Dengue is the most prevalent mosquito-borne viral infection worldwide, with 100 million cases of dengue fever (DF) and half a million cases of dengue haemorrhagic fever (DHF) annually.1,,2 Since 1989, there have been regular epidemics of DHF in Sri Lanka,3 with the worst recorded so far in April–August 2004 (personal communication). During the past few years, the characteristics of dengue in Sri Lanka appear to have changed. For instance, a decade ago, children were predominantly affected, but in recent years clinicians have seen increasing numbers of adult dengue patients, with both significant morbidity and increasing numbers of adult deaths due to dengue.

The rise in dengue among Sri Lankan adults adversely affects Sri Lanka's developing economy. It also affects health planning, and is further compounded by the general lack of systematically collected information on the natural history of dengue in such patients. This lack often leads health planners and clinicians to base their decisions regarding resource allocation and clinical management on personal experiences, rather than on tangible evidence.

Similar recent trends of increasing numbers of adult dengue patients can also be seen in other South Asian, South-East Asian and Latin American countries.2,4–7 If we are to take effective steps to reduce this trend and treat this group optimally, pooling information from different countries is important. At present, information on adult dengue infections in South Asia is quite limited. We studied a selected group of hospitalized adult dengue patients in Sri Lanka during a recent major dengue epidemic.


We studied patients admitted to an adult general medical ward at the Colombo South Teaching Hospital, Sri Lanka, from 24 April to 31 July 2004. All patients with clinical features suggestive of dengue infections were included, following informed written consent. Ethical clearance for the study was obtained from the ethical review committee, University of Sri Jayawardanapura, Sri Lanka. Demographical data were collected from each patient, and clinical features, haematological and biochemical laboratory measurements were recorded serially until discharge from hospital. Bloods for serological investigations were collected after day 7 of the illness. Dengue-virus-specific antibodies were measured using the PANBIO Dengue duo IgM and IgG rapid strip test.8 This test has a high sensitivity (76% for primary dengue infections; 88% for secondary dengue infections) and specificity (88–99%) and thus can be used for differentiation of primary and secondary dengue viral infections. If only dengue-virus-specific IgM antibodies were detectable in the test sample, the patient was considered to have a primary dengue infection, whereas the presence of both IgM and IgG was considered to indicate a secondary dengue infection.

Dengue RT-PCR was attempted only in 24 patients who presented to hospital during the first 4 days of the study, to try to characterize the disease-causing dengue serotype. This was performed using dengue-specific universal primers and type-specific primers, as described by Velathanthri et al.9

The WHO classification and case definitions were used to categorize patients as having either DF or DHF.10 If patients had evidence of plasma leakage (pleural effusions, ascites) or shock, even in the absence of bleeding manifestations, they were categorized as having DHF. DHF was further divided into 4 grades (I, II, III, IV) according to the WHO definitions. Haemoconcentration was defined as Packed Cell Volume (PCV) >50, as a baseline value was not known for most patients.

Data were analysed using SPSS (v. 10.0). Odds ratios (OR) of possible risk factors for severe dengue infections were calculated. For this purpose, DF and DHF grade I were categorized as mild infections, and DHF grades II–IV as severe infections. χ2 tests were used to determine statistically significant differences between groups.


Of those who presented to hospital with clinical features suggestive of dengue infections, 108 were diagnosed as having dengue infections, based on serology. Our report is based on this group of 108: 64 (59.3%) males, 44 (40.7%) females, mean age 26.6 years (SD 9.9, range 13–56). There were 33 (30.6%) with a diagnosis of DF and 75 (69.4%) with DHF. Of those with DHF, grades I, II, III and IV were seen in 33 (44%), 28 (37.4%), 10 (13.3%) and 4 (5.3%), respectively. Dengue infections were categorized as primary in 37 (34.3%) and secondary in 71 (65.7%). Of the 37 with primary dengue infections, 19 (51.4%) developed DF and 18 (48.6%) developed DHF. Secondary dengue infections were significantly associated with development of severe disease (OR 4.3, 95%CI 1.8–10.3, p < 0.001).

Clinical findings

See Table 1. Patients presented after a mean 5.5 (SD 1.6) days of illness and were hospitalized for a mean 3.6 (SD 3.3) days. All patients had fever during the early phase of their illness (mean 4.7 days, SD 1.4). Mean stay in hospital was 2.7 (SD 0.6) days for patients with DF, and 4.6 (SD 4.5) days for those with DHF (p < 0.05). Two patients complained of a sore throat and had enlarged tonsils on examination. A patient with grade III DHF developed massive scrotal oedema during the recovery phase. A high proportion of patients with DHF had abdominal symptoms (diarrhoea 32%, vomiting 68%, abdominal pain 18.7%), and these were the main clinical symptom in some.

View this table:
Table 1

Clinical and laboratory findings in patients with dengue fever (DF) and dengue haemorrhagic fever (DHF)

FindingDF (n = 33)DHF (n = 75)p
Myalgia26 (78.8%)56 (74.7%)>0.05
Arthralgia19 (57.6%)43 (57.3%)>0.05
Headache26 (78.8%)45 (60%)>0.05
Flushed appearance12 (36.4%)33 (44%)>0.05
Vomiting18 (54.5%)51 (68%)>0.05
Diarrhoea7 (21.2%)24 (32%)>0.05
Abdominal pain3 (9.1%)14 (18.7%)>0.05
Enlarged liver10 (30.3%)39 (52%)<0.05
Splenomegaly0 (0%)2 (2.7%)>0.05
Hypotension0 (0%)13 (17.3%)<0.05
Pleural effusions or ascites0 (0%)11 (14.7%)<0.05
Development of shock0 (0%)14 (18.7%)<0.001
Bleeding manifestations5 (15.2%)37 (49.3%)<0.001
Petechiae5 (15.2%)12 (16%)>0.05
Haematemesis0 (0%)6 (8%)>0.05
Melena0 (0%)10 (13.3%)<0.05
Ecchymoses1 (3%)7 (9.3%)>0.05
Gum bleeding1 (3%)6 (8%)>0.05
Vaginal bleeding0 (0%)7 (9.3%)>0.05
Haemoconcentration (PCV >50)3 (9.1%)19 (25.3%)>0.05
Platelet counts <10 × 109/l8 (24.2%)71 (94.7%)<0.001
White blood cell count <4 × 109/l9 (27.3%)25 (33.3%)>0.05
  • Data are numbers (percentages).

Bleeding manifestations were seen in 42 patients (38.9%). Petechiae (15.7%) were the most frequent, followed by melaena (9.3%), ecchymoses (7.4%), bleeding gums (6.5%) and haematemesis (5.6%). Vaginal bleeding was seen in 7 (15.9%) of 44 females, none of whom were menstruating at the time of infection. Haematuria and epistaxis were present in some. Mild bleeding manifestations (petechiae, ecchymoses) were also seen in 5 (15.2%) of 33 DF patients. Of the 42 patients (all DHF) with bleeding manifestations, 13 (30.9%) needed platelet concentrates. Six (8%) patients with DHF were given FFP for profound shock that was not controlled with crystalloid solutions.

A platelet count <50 × 109/l and PCV >50 was significantly associated with the presence of bleeding manifestations. Furthermore, a significant negative correlation was seen between disease severity (grade of DHF) and platelet counts (Spearman's correlation coefficient –0.69, p < 0.001).

Laboratory findings

Eight (24.2%) patients with DF and 71 (94.7%) with DHF had platelet counts <100 × 109/l. In the 75 with DHF, platelet count was <20 × 109/l in 13 (17.3%) and 20–50 × 109/l in 30 (40%). Bleeding manifestations were seen in 27 (62.8%) of the 43 patients with platelet count <50 × 109/l. Haemoconcentration was present in 19 (25.3%) patients, all with DHF. Lowest platelet counts and highest PCVs were seen at day 5.3 (SD ± 1.3) and 5.2 (SD ± 1.4) of illness, respectively. A white blood cell count <4 × 109/l was seen in 9 (27.3%) patients with DF, and 25 (33.3%) with DHF. Dengue RT-PCR serotype results (n = 7) were: 5 DEN-3; 1 DEN-2; and 1 DEN-1.


Complications were only seen in patients with DHF grades III and IV. Four (3.7%) patients needed ICU care for management of encephalopathy (manifesting as confusion, focal seizures and altered level of consciousness as assessed using the Glasgow coma scale). These four patients also had severe bleeding manifestations (haematemesis, melaena, and extensive ecchymoses) and shock. Therefore, they needed several transfusions of platelets or other blood products (FFP, cryoprecipitate). All were ventilated and given broad spectrum antibiotics and inotropes. They also had varying degrees of hyponatraemia. Two of these four patients also had myocarditis (as evidenced by pulmonary oedema, hypotension and tachycardia and low ejection fractions on echocardiography) and two also had disseminated intravascular coagulation. Unfortunately, three of these four patients died from intractable hypotension and multi-organ failure. The initial PCVs were very high in three of the patients that died, but fell to <25 later on because of possible internal bleeding as evidenced by falling of haemoglobin levels. The overall mortality of our cohort was 3.7% (4/108 patients). Consent for a pathological post-mortem was only obtained in one of these patients, who had intractable fits and clinical signs of consolidation in both lungs. Widespread abscess formation was present in both lungs, pus was found in the pleural cavity and the brain was congested and oedematous.


Hospital clinicians in Sri Lanka have recently been treating increasing numbers of adults with dengue infections. We report on clinical and laboratory findings in a cohort of hospitalized adult dengue patients from Sri Lanka seen during a recent major dengue epidemic. As in recent years, this epidemic affected a large number of adults, causing significant morbidity and some mortality. Our study was done during the time of the largest epidemic of DF/DHF seen so far in Sri Lanka. The mortality rate in our patients (3.7%) was higher than the mortality rates seen among hospitalized patients in other studies6,11–17 (Table 2) and is also higher than those seen in previous epidemics in SL.17

View this table:
Table 2

Clinical and laboratory features in studies of adults hospitalized with dengue infections

Study …PresentSeet14Gonzalez15Wichmann6Harris16Pervin13Lai11Kularatne17
CountrySri LankaSingaporeCubaThailandNicaraguaBangladeshTaiwanSri Lanka
Number studied10834763476149713183
DF/DHF (%)31/6997/30/10037/6344/5679/1854/46NS
Primary dengue infections (%)34.397NS202747NS23
Secondary dengue infections (%)65.73NS807353NS77
Clinical features
Myalgia (%)76NS76864857774
Arthralgia (%)57NS748626861NS
Flushed appearance (%)4253NSNSNSNSNS62
Vomiting (%)64385959503615NS
Diarrhoea (%)294129NS16NS15NS
Abdominal pain (%)172149NS46654NS
Enlarged liver (%)4501.84041321NS
Splenomegaly (%)0.02NS13NSNSNSNSNS
Bleeding manifestations (%)42271003637164615
Pleural effusions/ascites (%)11NS20NSNS5383
Shock (%)1402412101150.005
Thrombocytopenia (%)7985100NSNS238590
Raised haematocrit (%)22193NSNS173991
WBC <4 × 109/l (%)318271NSNSNS61NS
Received IV fluids (%)74NSNSNSNSNSNSNS
Given platelet concentrates (%)12NSNSNSNSNSNSNS
Mortality rate (%)3.702.60.50.001NS00.02
  • NS, not stated; FFP, fresh frozen plasma; DF, dengue fever; DHF, dengue haemorrhagic fever.

A high proportion of patients with DHF had abdominal symptoms: diarrhoea (32%), vomiting (68%) and abdominal pain (18.7%). In some, their abdominal pain simulated an acute surgical abdomen. One patient, who later died, had initially been investigated and treated as having an acute abdomen in a surgical unit, until she later developed skin bleeding and severe hypotension. Dengue infections presenting with predominantly gastro-intestinal symptoms have been reported previously,6,,14 and highlight the need for medical personnel to consider the possibility of dengue when assessing patients from areas where it is common, especially during a major dengue epidemic.

During the same period as this adult dengue study, we also studied a group of children with dengue from the same district, in a tertiary care hospital.18 Table 3 shows some significant differences between the clinical features between adults and children. Some salient differences in clinical and laboratory findings were noted between the two groups. Fewer adults developed pleural effusions or ascites (adults 10.2%, children 81.4%) or shock (adults 14%, children 30%). Although, the reasons for these observations are unclear, it appears that children are more prone to fluid leakage when compared to adults. A flushed appearance, pharyngeal congestion, a runny nose and the presence of a recovery rash were common in children, but infrequent in adults. Although similar proportions of adults and children developed bleeding manifestations, vaginal bleeding was not seen in children. However, despite the higher incidence of shock in children, the mortality rate was higher among adults (adults 3.7%, children 0%).

View this table:
Table 3

Differences in clinical and laboratory features in children and adults with dengue haemorrhagic fever

FeatureChildren (n = 86)Adults (n = 75)p
Flushed appearance76 (88%)33 (44%)<0.001
Diarrhoea16 (19%)24 (32%)>0.05
Vomiting66 (77%)51 (68%)>0.05
Pharyngeal congestion11 (13%)2 (2.7%)<0.001
Runny nose15 (17%)0 (0%)<0.001
Bleeding manifestations36 (42%)37 (49.3%)>0.05
Pleural effusions or ascites70 (81%)11 (14.7%)<0.001
Development of shock30 (34.9%)14 (18.7%)<0.001
Impaired consciousness6 (7%)4 (5.3%)>0.05
Presence of recovery rash17 (20%)1 (1.3%)<0.001

In adults, a platelet count <50 × 109/l and PCV >50 was significantly associated with the presence of bleeding manifestations. However, in children with dengue, we did not find a relationship between platelet counts and bleeding manifestations. Although associations between disease severity and platelet counts were seen in both adults and children, the association was weaker in children. Previous studies on the relationship between platelet counts and bleeding manifestations in dengue have also shown varying results.19,,20 Five (15.2%) patients with DF also developed bleeding manifestations, comparable to observations by others in hospitalized patients.6,,14 However, some have reported the presence of bleeding manifestations in a very high proportion of patients with DF (78%).14 However, this figure is likely to overestimate the ‘true’ frequency of haemorrhage observed among those with DF, as those who are admitted to hospital are likely to have more severe illness.

Our study was done at the height of a recent dengue epidemic in Sri Lanka. Dengue epidemics usually occur in Sri Lanka at this time of the year, because of an increase in mosquito densities following the monsoon rains. In virological studies by us and by others21 during this epidemic, DEN-3 was the predominant circulating virus serotype. From 1989 to 2002, DEN-2 was the main circulating dengue serotype in Colombo, Sri Lanka, followed by DEN-3.9,,22 The relative proportion of DEN-2 and DEN-3 serotypes in Colombo, Sri Lanka during the period 1989–2004 is shown in Table 4. The DEN-3 serotype also appears to be the cause of other recent large dengue epidemics in South Asia.23,,24 It is possible that a change in the predominant circulating serotype may have in some way contributed to these large and severe epidemics. We were unable to perform detailed phylogenetic analysis of isolated DEN-3 viruses to assess whether they differed from those of previous years.

View this table:
Table 4

Dengue viral serotypes in Sri Lanka from 1989 to 2004

20041150Present study

Nearly 40% of adults in our cohort had a primary dengue infection. Despite this, quite severe clinical manifestations were noted in a significant proportion of our patients. This may have been because a non-immune adult population was exposed to a more virulent viral serotype. However, it is also possible that these individuals could have had simultaneous infection with one or more dengue viruses which could have accounted for the more severe disease manifestations.

Increasing numbers of economically productive adults in some South Asian, South East Asian and South American countries are developing dengue infections, resulting in significant adverse financial effects. For example, in Thailand, annual financial loss due to dengue had been estimated at about US$61 per family, a value that exceeds the average monthly income.25 Although the economic impact of dengue infections has not been formally assessed in South Asian countries, it is likely to be of a similar magnitude.

In summary, we report clinical and laboratory findings in hospitalized adult dengue patients from Sri Lanka. Disease severity and mortality were higher in these adults than in children seen during the same epidemic. This emerging pattern of dengue infection in Sri Lanka is likely to place a great strain on its economy.


We thank Dr John Aaskov, Director of the Arbovirus Reference Centre for providing the Dengue duo IgM and IgG rapid strips, Dr Hasitha Tissera of the Epidemiological unit, Dr Vathsala Jayasuriya, for assistance with the statistics and the Asian Development Bank for providing funds for this study.


View Abstract