Clinical observations on virologically confirmed fatal dengue infections in Jakarta, Indonesia*

Bulletin ofthe WorldHealth Organization, 61 (4): 693 - 701 (1983) (©) World Health Organization 1983 Clinical observations on virologically confirmed fatal dengue infections in Jakarta, Indonesia* SUMARMO,' H. WULUR,2 E. JAHJA,3 D. J. GUBLER,4 W. SUHARYONO,5 & K. SORENSEN6 Thirty virologically confirmed cases of dengue infection with a fatal outcome were studied clinically in Jakarta, Indonesia, from 1975 to 1978. All 4 dengue virus serotypes were isolated from fatal cases, but dengue type 3 was responsible for 21 (70%o) of these isolates, compared to only 47% of isolatesfrom all cases ofdengue infection. The majority (60%) of these 30 cases were males in the 5-9-year age group. Nonspecific signs and symptoms in thefatal cases were no differentfrom those in patients who survived dengue infection, but 70% ofthepatients withfatal outcome had one or more signs ofencephalitis, primarily convulsions and somnolence; 3 of them developed spastic tetraparesis before death and2 died ofan illness clinically compatible with viral encephalitis. Other unexpected observations were that only 63% of the patients had classical dengue shock syndrome with haemoconcentration, thrombocytopenia and shock. A high percentage (80%) had gastro- intestinal haemorrhage, and in 9 patients (30%) this was severe enough to cause shock and death. In these 9 cases, the gastrointestinal haemorrhage and haematemesis began before the onset ofshock and there was no evidence ofhaemoconcentration orpleural effusion at any time during hospitalization. According to certain widely accepted criteria, these patients would not be diagnosed as dengue haemorrhagicfever (DHF). But as they made up nearly one-third of the confirmed fatal dengue infections in this study and had massive gastrointestinal haemorrhages with thrombocytopenia, the definition ofDHF should be changed to include this type of patient. It is proposed that the disease should be more realistically classified as dengue fever with or without haemorrhage and dengue shock syndrome. Dengue haemorrhagic fever (DHF) is an important public health problem in south-east Asian and western Pacific countries and is one of the leading causes of hospitalization and death among children in * This study was supported by funds provided by the Ministry of Health, Indonesia, and the U.S. Naval Medical Research and Devel- opment Command, Navy Department, for Work Unit MR041.01- 0151. The opinions and assertions contained herein are the private ones of the authors and are not to be construed as official or as reflecting the views of the Indonesia Ministry of Health, Navy Department, or the Naval Service at large. ' Head, Division of Tropical Medicine, Department of Child Health, Medical School, University of Indonesia, Jakarta Pusat, Indonesia. 2 Professor, Department of Pediatrics, Sumber Waras Hospital, University of Taramanagara, Jakarta, Indonesia. 3 Paediatrician, Husada Hospital, Jakarta, Indonesia. 4 Head, Department of Virology, Naval Medical Research Unit No. 2, Jakarta Detachment, Jakarta, Indonesia. Present address: Chief, Dengue Branch, and Director, San Juan Laboratories, Vector-Borne Viral Diseases Division, Center for Infectious Diseases, CDC, GPO Box 4532, San Juan, Puerto Rico 00936, USA. All correspondence should be addressed to this author. s Chief, Virology Division, Biomedical Research Center, National Institute for Health Research and Development, Ministry of Health, Jakarta, Indonesia. 6 Officer-in-Charge, Naval Medical Research Unit No. 2, Jakarta Detachment, Jakarta, Indonesia. Present address: NRMC Branch Clinic, Marine Corps Recruit Depot, San Diego, CA, USA. many tropical Asian countries (1 ).' Detailed clinical observations have shown the spectrum of disease associated with dengue infection to range from non- specific febrile illness to severe haemorrhage, shock, and death (1-7).a Severe and fatal dengue infections are frequently reported and mortality rates associated with DHF have varied greatly from one country to another and from epidemic to epidemic. A major problem has been that death usually occurs before a second blood sample can be taken, and serological diagnosis is usually not possible. Virological confir- mation, when attempted, has been unsuccessful for the most part, leading to speculation that the absence of virus in fatal cases may be due to antigen-antibody complexing (8, 9). As a result, the clinical description of fatal dengue infection has been based largely on un- confirmed cases that conformed to the clinical picture of patients who survived and whose diagnosis could be serologically confirmed (8). The development of the mosquito inoculation tech- nique provided a highly sensitive method for isolation a Guidefor diagnosis, treatment and control of dengue haemor- rhagic fever. Second edition, 1980. Manila, WHO Regional Office for the Western Pacific (unpublished document). 4329 -693- 694 SUMARMO ET AL. of dengue viruses (10). Using this technique in Indo- nesia during the period from October 1975 through June 1978, we confirmed 33 fatal dengue infections by virus isolation. This report describes the clinical aspects of 30 fatal dengue infections from Jakarta. MATERIALS AND METHODS All patients described here were admitted to one of three study hospitals in Jakarta and were seen by one of us. Criteria for the diagnosis of DHF followed the procedures outlined in the technical guide for diag- nosis, treatment, and control of DHF.b Severity of ill- ness was graded according to the following criteria. Dengue fever (DF): fever accompanied by a variety of nonspecific constitutional symptoms, with or without a positive tourniquet test. Grade II: like DF, but accompanied by skin haemorrhage or other bleeding, such as epistaxis, gum bleeding, or gastrointestinal haemorrhage. Grade III: circulatory failure manifest by a rapid, weak pulse, narrowing of pulse pressure (< 20 mmHg (2.67 kPa), or hypotension (systolic pressure < 80 mmHg (10.67 kPa)). Grade IV: mori- bund patients with undetectable blood pressure and pulse. A detailed record was kept for each patient includ- ing clinical history, name, age, sex, date of onset, date of admission, and address. On admission, all patients were examined thoroughly for signs and symptoms suggestive of DHF. These included vital signs, exam- ination for petechiae or other skin haemorrhages, examination for signs of circulatory failure and presence of pleural effusions. Routine haemogram and urinalysis, bleeding time, whole blood clotting time or recalcification time and tourniquet test were performed on all patients. In addition, fibrinogen levels, fibrin degradation products, and partial thromboplastin times were determined on some patients and careful examination of blood smears was made for the presence of fragmented and distorted red blood cells in patients with suspected dis- seminated intravascular coagulation. The clinical course of each patient was monitored continuously and serial determinations of the haemoglobin, haem- atocrit and platelets were made daily or as frequently as every 4 hours in the case of severely ill patients. Blood samples were taken as soon as possible after admission. The blood was allowed to clot for 4-6 hours at ambient temperature, after which it was centrifuged and the serum transferred to 3-ml screw- cap vials and stored in a freezer ( - 60 °C or lower). When possible, tissue specimens (generally liver), were taken immediately after death using a Vim- Silverman 14-gauge biopsy needle. Specimens were placed in 3-ml screw-cap vials containing 0.5 ml b See footnote a, page 693. of phosphate-buffered saline (PBS) with 30%7o heat- inactivated calf serum and stored in a freezer (-60 °C or lower). To process for virus isolation, biopsy specimens were thawed, transferred to 15-ml conical centrifuge tubes (along with the PBS), dis- integrated by sonic energy, and centrifuged at 1580 g for 30 minutes at 4 'C. The supernatant was collected and inoculated into mosquitos. Sera were tested for dengue antibodies by the haemagglutination-inhibition (HI) test adapted for microtitration plates using 8 units of dengue type 1 or 2 antigen (11). Paired sera, if available, were always tested together in the same test using serial twofold dilutions. All attempts at virus isolation were by the mosquito inoculation technique (10). Sera from acute cases and any tissue biopsy specimens from fatal cases were inoculated intrathoracically into female Aedes aegypti or A. albopictus, both undiluted and at a dilution of 1:5 in PBS with 5% heat-inactivated calf serum. Inoculated mosquitos were held at 30-32 "C for at least 14 days and killed by freezing. Infection was determined by the presence or absence of viral antigen in the brain and salivary glands using the direct fluorescent antibody test (12, 13). Virus isolates were identified according to serotype by the complement fixation test employing antigen prepared in male mosquitos (14). RESULTS During this study, virus isolation was attempted on sera from 148 fatal cases admitted to hospitals in Jakarta with a suspected diagnosis of dengue. Virus was isolated from the sera of 30 of these patients (20.3%), from only one of whom a liver biopsy speci- men was obtained; biopsy specimens were not avail- able from the other 29 virologically confirmed cases. Liver biopsies were performed on another 16 patients, from whom virus was not isolated from the serum and all were negative for dengue virus. Table 1 shows the total number of virologically confirmed dengue patients including fatal cases, by virus serotype, during the 1975-78 study period. Dengue type 3 was the predominant virus isolated in both groups; but whereas less than one half of all the patients (47%) had a dengue type 3 infection, over two-thirds of the fatal cases (70%0) were associated with this serotype. The fatality rates for virologically confirmed cases were nearly three times higher for dengue type 3 than for the other serotypes (X2 = 8.3, P < 0.002). The age distribution of virologically confirmed fatal dengue cases was similar to that for all confirmed dengue infections (Table 2). The majority of patients (600/o of fatal cases and 54% of all cases) were in the 5-9-year age group. The male-to-female FATAL DENGUE INFECTIONS Table 1. Serotypes of virologically confirmed dengue infections, including fatal cases, Jakarta, 1975-78 All infectionsa Fatal infections Dengue Fatality serotype No. % No. % rateb Den.1 28 18.2 3 10.0 11 Den. 2 44 28.6 5 16.7 11 Den. 3 73 47.4 21 70.0 29 Den. 4 9 5.8 1 3.3 11 Total 154 100.0 30 100.0 19 a Includes fatal infections. b For virologically confirmed cases only. Table 2. Age and sex distribution of confirmed dengue infections, Jakarta, 1975-78 Males Females Total Age group (yrs) Fatal All Fatal All Fatal All infections' infections b infections a infections b infections a infections b 0-4 7 69 3 86 1 0 155 5-9 10 155 8 182 1 8 337 10-14 1 60 1 69 2 129 15 0 4 0 4 0 8 Total 18 288 12 341 30 629 a Confirmed by virus isolation. b Includes both serologically and virologically confirmed cases. ratio for fatal infections was 1.5:1 or 60/o and 4007, respectively, whereas the male-to-female ratio for all dengue infections was 1:1.2 or 46%/o and 54%, respectively. The duration of illness before admission to a hospital ranged from 1 to 5 days, with an average of 3 days. The time between admission and death ranged from 2 hours to 5 days with an average of 32 hours, with no differences between serotypes. Also, there were no differences in dengue HI antibody titres in acute sera or in the day of illness on which viruses were isolated among persons with virologically confirmed infections with different serotypes. The seasonal distribution of confirmed fatal dengue cases resembled the distribution of sero- logically confirmed cases. Thus, the majority of confirmed cases and fatalities occurred from October through March each year, which corresponds with the rainy season in Indonesia. Nonspecific signs and symptoms Table 3 shows the nonspecific constitutional signs and symptoms associated with fatal dengue infections in Jakarta by virus serotype: 53% had a palpable liver, 43 % had cough or sore throat, 43% had vomit- ing, and 307o had abdominal pain. There were no obvious differences in the frequency of these findings that could be related to the different dengue virus serotypes. Encephalitic signs A surprisingly high number of patients who died had encephalitic signs. Table 4 shows that, overall, 70% of the dengue patients who died had one or more of these signs. Convulsions and somnolence (semi- coma) were the most common signs, and the latter was observed in patients infected with all serotypes. Stiff neck and paresis, however, were observed only in patients infected with dengue type 3. All three of the patients with paresis developed spastic tetra- paresis before death; the CSF in these patients was clear, with a few white blood cells (9-27/1l) and normal protein and sugar levels, but the CSF pressure was not measured. CSF for virus isolation was obtained from one of the patients with paresis; the result was negative. 695 SUMARMO ET AL. Table 3. Frequency of nonspecific constitutional signs and symptoms associated with virologically confirmed fatal dengue infections, by serotype, Jakarta, 1975-78 Dengue serotype Sign/ Den. 1 Den. 2 Den. 3 Den. 4 Total % symptom (N = 3)a (N = 5)a (N =21)a (N =)a (N = 30)a Hepatomegaly 2 2 11 1 16 53,3 Abdominal pain 0 2 6 1 9 30.0 Vomiting 2 2 8 1 13 43.3 Cough 0 0 3 0 3 10.0 Sore throat 2 2 5 1 1 0 33.3 Headache 0 1 1 0 2 6.7 a N = number of cases in each group. Table 4. Frequency of encephalitic signs associated with fatal dengue infections, by serotype, Jakarta, 1975-78 Dengue serotype Encephalitic Den. 1 Den. 2 Den. 3 Den. 4 Total % signs (N = 3)a (N = 5)a (N = 21)a (N = 1) (N = 30)a Convulsions 0 3 7 0 10 33.3 Coma/semicoma 2 3 1 1 1 17 56.7 Stiff neck 0 0 2 0 2 6.7 Paresis 0 0 3 0 3 10.0 Total of patients with one or more signs b 2 5 13 1 21 70.0 e N = number of cases in each group. b Nonadditive because some patients had more than one encephalitic sign. Haemorrhagic manifestations The haemorrhagic manifestations associated with virologically confirmed fatal dengue infection are shown in Table 5. Only 43% had a positive tourniquet test; but as death generally occurred within a few hours after admission, this test was frequently not repeated. Eighty percent of the patients had gastro- intestinal haemorrhage, many of them severe enough to result in shock (due to blood loss) and death. Signs of haemodynamic failure (shock) were observed in 26 of 30 patients (87%o) and ranged in severity from a narrowed pulse pressure (.< 20 mmHg (2.67 kPa)) or hypotension to profound shock. Some differences were apparent in haemorrhagic manifestations according to serotype. Thus, 90%7o of dengue type 3 infections had gastrointestinal haemor- rhage compared to 60%1o of dengue type 2 and 33%1o of dengue type 1 infections. On the other hand, 1000%o of dengue types 1 and 2 infections had shock compared to 81 0 of type 3 infections. Because of the small numbers involved, none of these differences are significant statistically. The cause of death is shown in Table 6. Only 19 (637o) of the patients died of classical dengue shock syndrome (DSS) with haemoconcentration, and 3 of these patients died of shock complicated by massive haemorrhage. The cause of death in another 9 patients (300o) was considered to be shock due to blood loss caused by massive gastrointestinal haemor- rhage; it may be noted from Table 5 that 24 patients (80%o) had such a haemorrhage. Screening laboratory tests to examine for coagulation disorders were performed on 6 of the patients who died of shock due to severe gastrointestinal haemorrhage. All showed depleted fibrinogen, thrombocytopenia and pro- longed bleeding time, recalcification time and partial thromboplastin time, and five had prolonged pro- 696 FATAL DENGUE INFECTIONS Table 5. Frequency of haemorrhagic manifestations associated with virologically confirmed fatal dengue infections, by serotype, Jakarta, 1975-78 Dengue serotype Haemorrhagic Den. 1 Den. 2 Den. 3 Den. 4 Total % manifestations (N = 3)° (N = 5)0 (N = 21)° (N = 1)° (N = 30)° Positive tourniquet 2 2 9 0 13 43.3 test Petechiae 2 2 12 0 16 53.3 Epistaxis 1 1 3 0 5 16.7 Ecchymoses 1 0 1 1 3 10.0 Bleeding gums 0 1 1 0 2 6.7 Haematemesis/ 1 3 19 1 24 80.0 melaena Shock 3 5 17 1 26 86.7 ° N = number of cases in each group. Table 6. Causes of death in patients with dengue infections, by serotype, Jakarta, 1975-78 Dengue serotype Cause of death Den. 1 Den. 2 Den. 3 Den. 4 Total % Profound shock with haemo- 3 4 11 1 19 63.3 concentration Severe bleeding followed - 1 8 - 9 30.0 by shock Encephalopathy - - 2 - 2 6.7 Total 3 5 21 1 30 thrombin times. Eight deaths caused by severe bleed- ing were associated with dengue type 3 infections. Finally, two deaths were associated with signs and symptoms compatible with viral encephalitis, but the virus was isolated only from the serum, not the brain or CSF. Both of these were dengue type 3 infections. Case histories Clinical details of four fatal cases are given below. Case S (a case of dengue shock syndrome, DSS). A 7-year-old Indonesian male was admitted on 12 March 1976 with a history of 4 days of fever and vomiting. On admission, he was restless and pale with cold extremities. The rectal body temperature was 37 °C, and the child was in profound shock with un- detectable pulse and blood pressure. The liver was palpable. Laboratory examination revealed the following: erythrocyte volume fraction (haematocrit) 0.51, haemoglobin (Hb) 135 g/l (or haemoglobin (Fe) 8.37 mmol/l), and thrombocytes 13.4 x 103 per Al. Lactated Ringer's solution followed by 300 ml of plasma were given but shock could not be reversed. Erythrocyte volume fraction, Hb, and thrombocyte determinations performed 6 hours later were 0.46, 150 g/l, and 18 x 103 per Al respectively. The child died after 11 hours in hospital. Two hours before death, he had haematemesis and melaena. The dengue HI antibody titre in serum taken on the fourth day of illness was 80. Dengue type 3 virus was isolated from that serum. CaseP (a case of DSS complicated by severe gastro- intestinal haemorrhage). A 2-year-old Indonesian male was admitted on 2 April 1977 with a history of 3 days fever. Approximately 1 hour before admission, he had melaena followed by general convulsions 697 698 SUMARMO ET AL. which lasted for about 10 minutes. On admission, he was somnolent and had cold extremities, with a rectal body temperature of 40 'C. Liver was enlarged to 4 cm below the costal margin. Blood pressure and pulse were undetectable. Laboratory examination revealed Hb 120 g/l, erythrocyte volume fraction 0.37, and thrombocytes 73 x 103 per ul. Shock was treated with lactated Ringer's solution and plasma. Cerebrospinal fluid was clear with 9 leukocytes per ul, glucose 2.55 mmol/l (46 mg/dl), and total protein 0.62 g/l. Six hours after admission, the condition of the patient deteriorated. He was semicomatose, had profuse haematemesis and melaena, and was in pro- found shock. Laboratory findings were as follows: Hb dropped to 85 g/l, erythrocyte volume fraction 0.3, and thrombocytes 12 x 103 per 1d. Blood transfusion was given, but the child died 7 hours after admission. The dengue HI antibody titre in serum taken on day 3 of the illness was 10. Dengue infection was confirmed by isolation of dengue type 3 virus from that serum. Case Y (a case of severe gastrointestinal bleeding). An 11-year-old Chinese female was admitted on 15 February 1978 with a history of 3 days of fever, epigastric pain and vomiting. On admission, she was severely ill, but still conscious. Temperature was 37.5 'C, pulse 88/min and blood pressure 90/ 60 mmHg (12/8 kPa). Tourniquet test was pos- itive, the tonsils were enlarged and inflamed, and cervical lymph nodes were palpable. Liver and spleen were not enlarged. Initial laboratory examination showed Hb 102 g/l, erythrocyte volume fraction 0.31, leukocytes 2.9x 103 per il, and thrombocytes 41 x 103 per il. She was placed on fluid therapy and her condition was maintained for the next 24 hours. Blood pressure was monitored every 1-3 hours and remained between 90/60 mmHg (12/8 kPa) and 110/ 70 mmHg (14.7/9.3 kPa). Erythrocyte volume frac- tion, Hb, and thrombocytes were monitored every 2-6 hours and seven determinations showed all to remain relatively stable with erythrocyte volume frac- tion between 0.31 and 0.38, Hb between 102 and 122 g/l, and thrombocytes between 34 x 103 and 92 x 103 per gl. Approximately 28 hours after admission, the patient began to vomit blood pro- fusely. At that time, the prothrombin time was 16 seconds (s) (control, 14 s), partial thromboplastin time was 65 s (control, 40 s), recalcification time was 180 s (control,90 s), and depleted fibrinogen (0.52 g/l) was observed. A chest X-ray revealed increased pul- monary vascular markings, but no signs of pleural effusion. Although fluid therapy was continued, the follow- ing morning (03h 00) the patient became hypotensive (blood pressure, 70/40 mmHg (9.33/5.33 kPa)) and then developed shock. She was given plasma expanders and pressor-amines and the blood pressure was brought up to 100/60 mmHg (13.3/8 kPa) after several hours. At that time the erythrocyte volume fraction was 0.32, Hb 108 g/l and thrombocytes 29 x I03 per ul. However, severe gastrointestinal haemorrhage with repeated and profuse haemat- emesis began again at 12h 30. Despite transfusions of blood (300 ml) and plasma (500 ml), the patient again became hypotensive and went into shock. During this time the erythrocyte volume fraction dropped from 0.34 to 0.25, haemoglobin from 108 to 58 g/l, and thrombocytes from 28 x 103 to 19 x 103 per yd. The patient died of respiratory failure 48 hours after admission. The serum taken on the day after admission (day 4 of the illness) had a dengue HI titre of 160. Dengue infection was confirmed by isolating dengue type 3 virus from this serum. Case R (a case of viral encephalopathy). A 1-year- old Indonesian male was admitted on 20 May 1976 with a history of 3 days of fever. Before admission, he had had general convulsions 3 times at intervals of about 15 min. On admission, he was somnolent and dyspnoeic, with a body temperature of 39 'C. The liver was enlarged to about 2.5 cm below the costal margin. Spinal fluid was clear with 27 leukocytes per 1l, glucose 3.66 mmol/l (66 mg/dl), and total protein 0.64 g/l. Both upper and lower extremities were spastic. Erythrocyte volume fraction 0.35, Hb 118 g/l, thrombocytes 309 x 103 per u1, and leuko- cytes 11.8 x 103 per ul. The tourniquet test on the left arm was positive, and later that morning petechiae were observed on the right hand. The patient deteriorated; on the second day in the hospital he was semicomatose with spastic tetraparesis. Petechiae were still limited to his right hand. He died on the third day. The dengue HI antibody titre in serum taken on day 3 of the illness was 10. Dengue infection was confirmed by isolation of dengue type 3 virus from this serum. The first of the above four cases died of classical DSS with haemoconcentration. The second case was DSS complicated by massive gastrointestinal haemor- rhage, and the third appeared to have died from shock due to blood loss. The fourth case showed signs and symptoms usually associated with acute viral enceph- alitis, but few that are associated with DHF. DISCUSSION Previous descriptions of unconfirmed, clinically diagnosed, fatal DHF have attributed the cause of death primarily to hypovolaemic shock caused by increased capillary permeability (9).c Based on these ' See footnote a, page 693. FATAL DENGUE INFECTIONS 699 studies, primarily in Thailand, strict criteria have been outlined by the World Health Organization which all DHF and DSS cases should meet. These include thrombocytopenia with concurrent haemo- concentration (9).d As noted in Table 6, many of our confirmed fatal dengue infections do not conform to these criteria. Thus, of 30 fatal cases with adequate data, only 19 (6307o) had classical DSS with haemo- concentration. Of these, 3 died of shock compli- cated by- massive gastrointestinal haemorrhage. Surprisingly, 9 (30%o) of our cases died as a result of severe gastrointestinal haemorrhage. These patients did not exhibit haemoconcentration before death, and shock was determined to be the result of blood loss. According to the definition proposed by some (7, 9)d these latter patients should not be called DHF. Since they make up nearly one-third of our confirmed fatal dengue infections, however, and since they had massive gastrointestinal haemorrhage with thrombocytopenia, the definition of DHF should be broadened to include this type of patient. Thus, we propose that the disease should more realistically be classified as dengue fever with or without haemor- rhage and dengue shock syndrome. The actual cause of the severe gastrointestinal haemorrhage is not known. It should be noted, how- ever, that all patients tested had laboratory results suggestive of disseminated intravascular coagulation. Although it is unlikely that the latter was the primary cause of the haemorrhage in these patients, the data suggest that it certainly could have been a contri- buting factor. Another potential contributing factor could have been the infecting virus strain since most of these patients had dengue type 3 infections. Two of our patients died with signs and symptoms compatible with viral encephalitis. In addition, 21 (70%o) had one or more encephalitic signs before death (Table 4). The most common signs were coma (57%o) and convulsions (33%o), but significant numbers had nuchal rigidity and/or paresis (17%7o). Although high fever is a recognized cause of convul- sions in young children, it is not a cause of coma, nuchal rigidity and paresis. Encephalitic signs, other than convulsions, have not been reported routinely in clinical descriptions of DHF from other parts of Asia. Recent reports of virologically confirmed dengue with encephalitic signs from Burma (15) and our data from Indonesia, however, suggest that dengue should be included in differential diagnosis of children admitted for viral encephalitis. To date, there have been no dengue virus isolations from brain or cerebral spinal fluid (CSF). We attempted several isolations from brain biopsies and from CSF, but all were negative. In one patient, from whom dengue type 3 virus was isolated from the serum, no isolate was made from the CSF. Therefore, d See footnote a, page 693. until pathogenetic studies can be carried out to determine whether the encephalitic signs are the result of invasion of the virus into the central nervous system or not, the term encephalopathy should be used to describe the condition associated with dengue infection. Other possible causes of encephalitic signs include anoxia of the cerebrum and accumulation of metabolic products as a consequence of the shock and disseminated intravascular coagulation, cerebral edema, microcapillary haemorrhages in cerebral tissue, or the release of toxic products such as hist- amine, serotonin and bradykinin. There have been few attempts to isolate dengue virus from tissues of fatal cases. In the only published account of a detailed virological study of fatal DHF, dengue virus was isolated from only 6 out of 169 fatal cases (8); 3 had dengue type 2 virus and 1 had dengue type 1 virus from serum and dengue type 2 virus from bone marrow and lymph nodes. In previous studies, dengue viruses had been isolated from the liver, heart and lung (16, 17). In the present study, we isolated dengue type 1 virus from a liver biopsy specimen and serum of one patient, but tissue biopsy specimens were not obtained from most patients. Unfor- tunately, post-mortem examination is seldom per- formed in Indonesia because of religious and other reasons. More detailed virological and pathological studies on tissues from fatal cases are badly needed using the new and more sensitive isolation techniques available. It will be noted from Table 6 that the majority of patients who died with atypical dengue had dengue type 3 infections. Thus, 8 of the 9 patients who died of massive gastrointestinal haemorrhage had dengue type 3 infections, and both of the patients with symp- toms of viral encephalitis were infected with dengue type 3 virus. Whether this represents a difference in virulence among the four dengue serotypes is not known. Dengue type 3 was the most predominant serotype transmitted during the last 2 years of this study, both in Jakarta and other parts of Indonesia (18). Other data suggest, however, that the virulence of the dengue type 3 virus in Indonesia may have changed between 1975 and 1976 (13, 19). Thus, increased numbers of fatal DHF cases in the spring of 1976 were nearly all associated with dengue type 3. This was followed a few months later, after the begin- ning of the rainy season, by epidemics of DHF in Jakarta, central and east Java, and west Kalimantan. Dengue type 3 was the predominant or only virus transmitted in all of the epidemics studied (13). The earlier epidemics, except for the one in Jakarta, were explosive and all were associated with severe and fatal disease. A late outbreak (1978) of dengue type 3 in Sleman, central Java, however, was less explosive and was associated with milder disease, and low viraemia (19). SUMARMO ET AL. This is reminiscent of apparent changes in dengue type 2 in the South Pacific where some islands of similar ecology experienced explosive epidemics associated with severe disease and high viraemia, while other islands experienced only sporadic trans- mission with mild disease and low viraemia (20). We think these data support the hypothesis that virus virulence is important in the pathogenesis of severe and fatal dengue infection (21). Although dengue type 3 was responsible for the majority of confirmed fatalities associated with dengue infection, it should be pointed out that the other serotypes can also cause fatal infection. Thus, 10Gb of the fatalities were associated with dengue type 1, 17% with dengue type 2, 70% with dengue type 3, and 3% with dengue type 4. Without accurate data on the background transmission of all four serotypes in the Jakarta population, however, it is not possible to say whether the higher fatality rate associated with dengue type 3 infection was due to increased virulence or merely to a higher transmission rate. An interesting difference noted between our patients and fatal DHF described previously is the sex ratio. In Thailand, the male-to-female ratio for all DHF cases was 1:1.4 and in unconfirmed fatal cases, 1:1.2 (6, 8). These ratios are similar to those observed for all confirmed dengue infections in Jakarta (Table 1), but are the reverse of those observed for viro- logically confirmed fatal dengue infections. In ana- lysing our data by sex, we found no differences between males and females in either age or antibody response. Thus, the median age of 18 males was 7 years compared to 6 years for 12 females. The geo- metric mean antibody titres for males and females were 10 and 13, respectively. These data do not support the hypothesis that females might produce more antibody than males (9). It should be noted that in the Philippines, the fatality rate was also higher in males (22). Some of these patients, both male and female, had no detectable HI antibody in sera taken on days 3-5 of the illness. These could have been primary dengue infections. Studies are now in progress to determine which are primary and which are secondary. This will be reported at a later date. Finally, it is quite obvious from our studies that children with severe and fatal dengue infection, in- cluding those with DSS, are not virologically sterile (9). We think our virus isolation technique was more sensitive than those used previously, and we were therefore able to isolate dengue viruses from fatal cases more frequently. Thus, in 15 of the 30 fatal cases, the blood sample from which virus was isolated was taken on the day of death and, in another 9 cases, on the day before death. Titration of virus in the serum of fatal dengue infections has revealed titres ranging from barely detectable (3.8 logio MID5o/ml) to over 8.3 logio MID5o/ml (23). ACKNOWLEDGEMENTS We thank Ms R. Tan, M. Abidin, A. Sie, Chairin Maroef, Woro Djuarti, Masran Masri, and Hardianto Saipan for excellent technical assistance. We are also grateful to Dr Julie Sulianti Saroso, previous Director, and Dr A. A. Loedin, present Director, National Institute of Health Research and Development, Ministry of Health, Indonesia, for making this study possible. RtSUME OBSERVATIONS CLINIQUES SUR DES INFECTIONS FATALES, VIROLOGIQUEMENT CONFIRMtES COMME CAS DE DENGUE, A JAKARTA (INDONESIE) Entre 1975 et 1978, 30 cas de dengue, virologiquement confirmes et ayant entrain6 une issue fatale, ont fait l'objet d'une etude clinique A Jakarta (Indonesie). Pour isoler et identifier le virus, on a utilise la technique de l'inoculation au moustique, l'epreuve directe des anticorps fluorescents sur cerveau et glande salivaire de moustique, et l'epreuve de fixation du complement avec un antigene prepare A partir de moustiques infectes. Si les quatre serotypes du virus de la dengue ont et retrouves dans les cas fatals, il est A noter que le serotype 3 etait en cause dans 21 (70%0) des cas d'issue fatale, alors qu'il est a l'origine de 47%o seulement de l'ensemble des cas de dengue. Dans la majorite (60%) des cas fatals, il s'agissait de garqons de 5 A 9 ans. Les signes et sympt6mes non speci- fiques enregistres ne differaient pas chez les patients qui sont morts et ceux qui ont survecu; les plus frequents etaient la fievre, les vomissements et les douleurs abdominales. Chez 21 (70%) des patients on a note un ou plusieurs signes d'en- cephalite, essentiellement somnolence (57/o) et convulsions (337o). Chez trois malades, la mort a et precedee d'une tetraparesie spastique, et deux patients (7%) sont decedes d'une maladie qui pouvait etre une encephalite virale. Ces cas ont toutefois et designes sous le nom d'oenc6phalo- pathie associee A la dengue>>, l'infection virale du systeme nerveux central n'ayant pu etre prouvee. 700 FATAL DENGUE INFECTIONS 701 Seuls 19 (630/o) des patients presentaient le syndrome de choc classique de la dengue, avec hemoconcentration, thrombocytopenie et choc; 3 d'entre eux ont fait en outre une hemorragie gastro-intestinale massive. Une hemorragie gastro-intestinale s'est produite chez 24 (80%) patients, et chez 9 d'entre eux (30%) elle a e assez grave pour causer choc et deces. Dans ces 9 cas, l'hemorragie gastro-intestinale et une franche hematemese ont precede l'apparition du choc; aucun signe d'hemoconcentration et d'effusions pleu- rales ne s'est manifeste A aucun moment durant l'hospitali- sation. D'autres manifestations hemorragiques, semblables A celles qui ont e rapportees dans d'autres pays, ont et observees. C'est la premiere description clinique detaillee d'infection fatale, confirmee, par le virus de la dengue. 11 est surprenant de constater que 63% seulement des cas correspondent aux criteres anterieurement decrits pour la dengue hemorra- gique. C'est ainsi que 30% de nos patients sont decedes par suite d'une hemorragie gastro-intestinale grave et que leur cas ne peut etre classe sous la denomination de dengue hemorragique puisqu'ils ne presentaient pas d'hemoconcen- tration. Du fait cependant qu'ils constituent pres du tiers de nos cas fatals, confirmes, d'infection par le virus de la dengue, il conviendrait d'elargir la definition de la dengue hemorragique pour y faire entrer ce type de patient. II est propose de classer la maladie, de facon plus realiste, comme dengue avec ou sans hemorragie et syndrome de choc de la dengue. REFERENCES 1. HAMMON, W. McD. Dengue hemorrhagic fever-do we know its cause? Am. j. trop. med. hyg., 22: 82-91 (1973). 2. HAMMON, W. McD. ET AL. New hemorrhagic fevers of children in the Philippines and Thailand. Trans. Assoc. Am. Physicians, 73: 140-155 (1960). 3. HAMMON, W. McD. Observations on dengue fever, benign protector and killer: A Dr Jekyll and Mr Hyde. Am. j. trop. med. hyg., 18: 159-165 (1969). 4. VENZON, E. L. ET AL. The greater Manila dengue hemor- rhagic fever epidemic of 1966. Philipp. j. sci., 48: 297-313 (1972). 5. NIMMANNITYA, S. & MANSUWAN, P. Comparative clinical and laboratory findings in confirmed dengue and chikungunya infections. J. Dept. Med. Serv. 15: 233-251 (1966). 6. NIMMANNITYA, S. ET AL. Dengue and chikungunya virus infection in man in Thailand, 1962-1964. I. Obser- vations on hospitalized patients with hemorrhagic fever. Am. j. trop. med. hyg., 17: 954-971 (1969). 7. HALSTEAD, S. B. Different dengue syndromes -the perspective from a pathogenetic point of view. Asian j. infect. dis., 2: 59-65 (1978). 8. NISALAK, A. ET AL. Observations related to pathogenesis of dengue hemorrhagic fever. III. Virologic studies of fatal disease. Yalej. biol. med., 42: 293-310 (1970). 9. HALSTEAD, S. B. Observations related to pathogenesis of dengue hemorrhagic fever. VI. Hypothesis and discussion. Yalej. biol. med., 42: 350-362 (1970). 10. ROSEN, L. & GUBLER, D. J. The use of mosquitoes to detect and propagate dengue viruses. Am. j. trop. med. hyg., 23: 1153-1160 (1974). 11. CLARKE, D. H. & CASALS, J. Techniques for hemag- glutination and hemagglutination-inhibition with arthropod-borne viruses. Am. j. trop. med. hyg., 7: 561-573 (1958). 12. KUBERSKI, T. T. & ROSEN, L. A simple technique for the detection of dengue antigen in mosquitoes by immuno- fluorescence. Am. j. trop. med. hyg., 26: 533-537 (1977). 13. GUBLER, D. J. ET AL. Virological surveillance for dengue haemorrhagic fever in Indonesia using the mosquito inoculation technique. Bull. Wld Hith Organ., 57: 931-936 (1979). 14. KUBERSKI, T. T. & ROSEN, L. Identification of dengue viruses using complement fixing antigens produced in mosquitoes. Am. j. trop. med. hyg., 26: 538-543 (1977). 15. THAUNG, U. Conference on dengue hemorrhagic fever: new developments and future research (Part I). Asian j. infect. dis., 2: 78 (Discussion) (1978). 16. DASANEYAVAJA, A. & CHARANSRI, U. First known isolation of a dengue virus from other human source than blood. In: SEATO Medical Research Monograph No. 2 (Symposium on Hemorrhagic Fever, 10-11 August 1961, Bangkok, Thailand). Bangkok, 1962, p. 61. 17. CHAN, Y. C. ET AL. Recent epidemics of hemorrhagic fever in Singapore. Jap. j. med. sci. biol., 20: 81-88 (1967). 18. SUHARYONO, W. ET AL. Dengue virus isolations in Indo- nesia 1975-1978. Asian j. infect. dis., 3: 27-32 (1979). 19. GUBLER, D. J. ET AL. Epidemic dengue 3 in Central Java associated with low viremia in man. Am. j. trop. med. hyg., 30: 1094-1099 (1981). 20. GUBLER, D. J. ET AL. Epidemiologic, clinical, and viro- logic observations on dengue in the Kingdom of Tonga. Am. j. trop. med. hyg., 27: 581-589 (1978). 21. ROSEN, L. The Emperor's new clothes revisited, or reflections on the pathogenesis of dengue hemorrhagic fever. Am. j. trop. med. hyg., 26: 337-343 (1977). 22. DIZON, J. J. The occurrence of dengue hemorrhagic fever in the Philippines. Asian j. infect. dis. 2: 15-17 (1978). 23. GUBLER, D. J. ET AL. Viraemia in patients with naturally acquired dengue infection. Bull. Wid Hlth Organ.,59: 623-630 (1981).