Classifying dengue: a review of the difficulties in using the WHO case classification for dengue haemorrhagic fever

Transcription

1 Tropical Medicine and International Health doi: /j x volume 11 no 8 pp august 2006 Classifying dengue: a review of the difficulties in using the WHO case classification for dengue haemorrhagic fever Shibani Bandyopadhyay 1, Lucy C. S. Lum 2 and Axel Kroeger 3 1 Public Health Consultant, Geneva, Switzerland 2 Department of Paediatrics, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia 3 Special Programme for Research and Training in Tropical Diseases, WHO, Geneva, Switzerland Summary background The current World Health Organisation (WHO) classification of dengue includes two distinct entities: dengue fever (DF) and dengue haemorrhagic fever (DHF)/dengue shock syndrome; it is largely based on pediatric cases in Southeast Asia. Dengue has extended to different tropical areas and older age groups. Variations from the original description of dengue manifestations are being reported. objectives To analyse the experience of clinicians in using the dengue case classification and identify challenges in applying the criteria in routine clinical practice. method Systematic literature review of post-1975 English-language publications on dengue classification. results Thirty-seven papers were reviewed. Several studies had strictly applied all four WHO criteria in DHF cases; however, most clinicians reported difficulties in meeting all four criteria and used a modified classification. The positive tourniquet test representing the minimum requirement of a haemorrhagic manifestation did not distinguish between DHF and DF. In cases of DHF thrombocytopenia was observed in %, plasma leakage in 6 95% and haemorrhagic manifestations in 22 93%. The low sensitivity of classifying DHF could be due to failure to repeat the tests or physical examinations at the appropriate time, early intravenous fluid therapy, and lack of adequate resources in an epidemic situation and perhaps a considerable overlap of clinical manifestations in the different dengue entities. conclusion A prospective multi-centre study across dengue endemic regions, age groups and the health care system is required which describes the clinical presentation of dengue including simple laboratory parameters in order to review and if necessary modify the current dengue classification. keywords dengue case classification, dengue haemorrhagic fever, plasma leakage, tourniquet test Introduction Dengue, including dengue fever (DF), dengue haemorrhagic fever (DHF) and dengue shock syndrome (DSS), is among the major causes of morbidity and mortality in children in many endemic Asian and South American 2,3 countries (Gubler 1998; Guha-Sapir & Schimmer 2005). Based on data from 112 national vital registration systems, deaths in South-east Asia, 4000 deaths in the Western Pacific and 2000 deaths in America for the year 2002 have been estimated to be due to dengue 4 (World Health Organization 2004). The current World Health Organisation (WHO) case classification into DF/DHF/DSS, formulated by the Technical Advisory Committee at its meeting in Manila, Philippines in 1974 (WHO Technical Report Series 1975) was, to a large extent, based on the pioneering studies at the Children s Hospital, Bangkok, Thailand in 1960s that defined the pattern of disease of that time. Although some minor modifications have been suggested, essentially the case definition and case classification of dengue have remained the same. (The term Dengue case definition for having or not having a case with dengue disease and the term Dengue case classification as the grouping of dengue disease manifestation into DF/DHF/DSS and others have been used throughout the article.) These were published by WHO headquarters in 1980, ,6and 1997 and also by its regional offices (WHO 1993, 1995a, 1995b, 1998, 1999; Pan American Health 5,6 Organisation 1994). The centrepiece in dengue disease classification is the differentiation between DF, a non-specific febrile illness 1238 ª 2006 Blackwell Publishing Ltd

2 with classical constitutional signs and symptoms accompanied by occasional bleeding manifestations, and DHF. The hallmark of DHF that differentiates it from DF is not haemorrhage as its name suggests, but rather the increased vascular permeability that leads to a capillary leak syndrome that may insidiously or rapidly progress to DSS. The term DHF is justified by the presence of some form of haemorrhagic manifestations that, according to the classification, always accompany the phenomenon of increased vascular permeability. According to WHO guidelines, DHF cases must fulfil all the following four criteria: (1) Fever or history of acute fever lasting 2 7 days. (2) Hemorrhagic tendencies evidenced by at least one of the following: a positive tourniquet test (TT) (The tourniquet test is performed by inflating a blood pressure cuff on the upper arm to a point midway between the systolic and diastolic pressure for 5 min. A test is considered positive when 20 or more petechiae per square inch are observed. The test may be negative or mildly positive during the phase of profound shock. It usually becomes positive, sometimes strongly positive, if the test is conducted after recovery from shock.); petechiae, purpura, ecchymoses; bleeding from mucosa, gastrointestinal tract, injection sites or other location haematemesis, melena. (3) Thrombocytopenia ( platelets/ll or less) [This number represents a direct count using a phase contrast microscope (normal is /mm 3. In practice, for outpatients, an approximate count from a peripheral blood smear is acceptable. In normal persons, 4 10 platelets per oil-immersion field (100 : the average of the readings from 10 oil-immersion fields is recommended) indicates an adequate platelet count. An average of 3 platelets per oilimmersion fields is considered low (i.e. < /mm 3 )]; (4) haemoconcentration (20% or more rise in the haematocrit (Hct) value relative to baseline average for the same age, sex and population) or evidence of plasma leakage (i.e. pleural effusion, ascites and/or hypoproteinaemia). The DHF is classified into four grades of severity. The only haemorrhagic manifestation in grade I is a positive TT. In grade II, spontaneous bleeding, usually in the skin, through the nose or in the internal organs is present. Hypotension for age, narrow pulse pressure of 20 mm Hg or less, restlessness and rapid weak pulse constitute grade III. Profound shock with undetectable blood pressure or pulse is considered grade IV. Grades III and IV constitute DSS. The original description of dengue and its manifestations were largely confined to children in Southeast Asia. With dengue entering new geographical areas and the older age groups, the clinical manifestations seem to be changing. More patients with DHFs I and II are currently getting admitted into hospitals in comparison with DHF III and IV (DSS). More patients are having fever at the time of shock compared to being afebrile at the time of shock (Kalayanarooj et al. 2002). Furthermore, a recent expert survey by TDR/WHO in 11 countries presented at the second International Congress on Dengue and Yellow fever in Cuba (Guzman et al. 2004) showed that four in five clinicians had difficulties applying the current case classification scheme. The major problems identified were the rigidity of the definitions, the low sensitivity and the assumption by some clinicians that DF means mild disease. We undertook this review of literature to understand how consistently the WHO dengue classification scheme has been applied for reporting DHF and DSS, to identify the major difficulties in using the classification, and to make recommendations for future research. Method A systematic review was undertaken by two independent reviewers (first and second author) and checked for inconsistencies by a third to analyse the experience of clinicians, public health personnel and scientists in the use of DF/DHF/DSS case classification. Primarily, it focused on the challenges faced by clinicians in applying the criteria for the diagnosis of DHF and severe dengue-related complications in routine clinical practice. Search strategy We searched Pub-med, Medline and personal medical databases using the following standard Medical Subject Headings (MeSH) terms. DHF, Classification of DHF, WHO case definition, Clinical experience. MeSH terms were grouped and linked by and or or to make it reproducible. References cited in the selected articles and reviews were hand-searched for the full text. Relevant documents and articles were obtained electronically and in paper form from different libraries, authors and other sources. Selection criteria All full published studies available in English between 1975 and May 2005 that mentioned the classification of dengue disease during the analysis of clinical data were selected. Studies analysing predictors of severe disease or risk factors where only DHF cases fulfilling the WHO classification criteria were enrolled were generally excluded. Special age group studies (e.g. on infants) were also not included. No study was discarded for poor quality, incomplete reporting or lack of laboratory data. ª 2006 Blackwell Publishing Ltd 1239

3 Extraction of data The selected articles were grouped into the following categories: outbreak studies (retrospective and prospective), reviews of medical records, prospective clinical studies and evaluation studies of the classification system. The number of cases enrolled and analysed, analysis of clinical and laboratory confirmed cases, methodology of the case classification scheme of DHF and any modification(s) of the classification criteria used were examined. Analysis of DHF cases with particular reference to the classification criteria and the authors comments regarding the difficulties were reviewed and presented in a table format. Search restrictions and limitations of the search were: 1. Articles written in languages other than English. 2. Articles whose full transcripts were not available. 3. Abstracts of presentations at scientific meetings and Pubmed/Medline abstracts where the full article could not be reviewed. The search was further extended by the use of bibliographic database and extensive hand-searching of journals to identify studies which were not indexed or unavailable in Pubmed. Several authors were also contacted individually for their articles and comments. Results A total of 38 studies with 21 outbreak case studies (14 retrospective, seven prospective), seven reviews of records from dengue surveillance, medical and laboratory forms, four evaluation studies and six prospective clinical studies were reviewed. The methodological shortcomings and difficulties in complying with the WHO guidelines were noted but no attempt was made to categorise the studies based on scientific merit such as quality of the study design. 7 The findings are presented in Tables 1 4. Criteria used in the existing case classification scheme Fever Fever is universally present. In almost all studies fever forms the starting point of the differential diagnosis leading to a diagnosis of dengue (DF/DHF/DSS). Haemorrhagic manifestations: tourniquet test A positive TT is a reflection of haemorrhagic tendency and capillary fragility. It is the only evidence of bleeding manifestation in grade I DHF. Application of TT varied widely among the studies. In a study by Richards et al. (1997) a positive TT was an enrolment criterion, while in a study by Kalayanarooj et al. (1997), the TT was performed daily until the end of the study, which was a day before defervescence. As most studies were observational and/or retrospective in nature, the frequency of performing the TT depended on the clinician s conviction of its usefulness in diagnosis and management. The sensitivity of the TT in DHF varied from as low as 0% (Pushpa et al. 1998) in observational outbreak studies to 57% (Kalayanarooj et al. 1997) in well-planned prospective studies. Phuong et al. (2004) and Lucas et al. (2000), demonstrated the poor differentiation of the positive TT between DHF (47% and 27% positive, respectively) and DF (39% and 26% positive, respectively). In addition, the TT was positive in cases of dengue-like febrile illnesses, which were subsequently non-dengue by serology, e.g. 21% (Kalayanarooj et al. 1997), 12% (Lucas et al. 2000) and 5% (Phuong et al. 2004). Seven per cent of 76 patients with fever and positive TT were found not to have dengue (Richards et al. 1997). Reducing the cut-off point to 10 petechiae improved the sensitivity but at the expense of specificity (Chairulfatah et al. 1995; Kalayanarooj et al. 1997). The low sensitivity of the TT could be explained by the timing of the test with respect to the day of illness. In a retrospective review ( ) of 5332 paediatric dengue patients in Bangkok a positive TT was observed in 88% of DSS, 94% of DHF and 90% in DF cases (Kalayanarooj et al. 2002). Except for Kalayanarooj s studies, all other studies did not mention a repeat of the TT when a negative result was obtained. The absence of haemorrhagic manifestations in DHF/ DSS could be attributed to the exclusion of TT as one of several haemorrhagic manifestations of DHF (Guzman et al. 1987; Diaz et al. 1988; Srivastava et al. 1990). In Malaysia, 7% (Wallace et al. 1980) and in India, 32 46% (Wali et al. 1999; Shah et al. 2004) of DHF cases, no haemorrhagic tendencies have been observed. Richards et al. (1997) included all fever cases with a positive TT and 100% of them fulfilled the criteria. In Cuba, Guzman et al. (1987) had observed that 31% of paediatric DHF/DSS (grades III and IV) cases did not have any bleeding. About 7% of 310 DSS cases in a large prospective clinical study in Vietnam (Phuong et al. 2004) did not exhibit any haemorrhagic tendency. These patients who presented with obvious signs of shock with evidence of plasma leakage and thrombocytopenia were classified as DSS whether or not they satisfied the criteria of haemorrhage. It has been noticed in Malaysia and Puerto Rico that clinicians do not perform the TT (L. Lum and J. Rigau 1999, personal communication) and that it is difficult to apply the test in sick and irritable children (Kabra et al ª 2006 Blackwell Publishing Ltd

4 Table 1 Fulfilment of WHO case classification criteria for DHF (additional to fever) as evidenced in the published literature. Outbreak studies (retrospective review of cases and prospective follow up of admitted cases) Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? Ahmed et al. (2001), Bangladesh Prospective Rasul et al. (2002), Bangladesh Rahman et al. (2002), Bangladesh Qiu et al. (1991), China Srivastava et al. (1990), India Kabra et al. (1992), India Pushpa et al. (1998), India Agarwal et al. (1998), India Tripathi et al. (1998), India 72 children enrolled 36 DHF I/II, 10 DSS, 26 DF 34/46 (74%) DHF had high PCV (level NM) All 46 had HM 25/46 (54%) DHF had TP No, 58% DF (15/26) also had high HcT, ST NM 151 DHF <15 years NM NM NM No, Modified Case definition used Fever with one or more of positive TT/HM/TP/HC, ST NM 174 LP DI 69 DHF I/ II; 1 DSS; 104 DF 154 LP Hemorrhagic dengue 24 paediatric DHF 7 DHF I/II; 17 DSS 21 paediatric LP DSS 69 pediatric admitted DI enrolled 49 had dengue antibody 29 DHF I/II; 20 DSS 134 paediatric DHF 92 DHF I/II; 42 DSS 560 hospitalised adult DI 14 DHF (2.5%) 12% had PE; 9% had AS 40% (62/154) had HC 9/24 (38%) had HC All DSS (17) DSS had evidence of PL 91% had HM NM No, One DF died with severe GIH, ST NM All had HM 121/154 (79%) had TP 10/24 (42%) had HM 15/24 (63%) had TP All had PL All had HM 18/21 (86%) had TP 3/27 (11%) had raised HcT. Rest had HcT 33 42% 29 of 44 CXR had PE In 54/134 (40%) HC could be documented Mucosal bleed in 39/ 49 (80%). TT negative in all. In DSS, not done ALL 14 DHF had PL All 14 DHF had HM 29 DI (5.1%) had blood transfusion (38/46) 83% had TP No, 70/154 (45.5%) had shock from GIH, ST NM No, ST NM No, ST NM No, Post Haemorrhagic shock in 15 cases 9 deaths due to massive G IH, who had no HC, 22% CFR (11/49), ST NM Most had HM All had TP No, Serial HcT done All 14 DHF had TP 168/ 560 (30%) of DI cases had TP, 23 (4%) had < cells/mm 3 Strict application of WHO criteria-not applied to eight deaths due to bleeding and classified as DHB, 23 cases with severe TP and 29 blood transfused DI cases did not satisfy the DHF classification & were reported as Dengue fever with haemorrhage. ST done in few when required ª 2006 Blackwell Publishing Ltd 1241

5 Table 1 (Continued) Authors, year, study type Patients studied Kabra et al. (1999), India 240 paediatric DI enrolled 80DHF I/ II; 113 DSS; 22 DFB Wali et al. (1999), India Prospective 110 adult LP DHF88 DHFI/II; 22 DSS Sharma et al. (1998), India 98 adults DHF 75DHF I/II; 23 DSS Narayanan et al. (2002), India Prospective 89 pediatric DI enrolled 59LP analysed 11 DHFI/ II; 5DSS; 20DF; 23 DFB Shah et al. (2004), India Prospective Richards et al. (1997), Indonesia Prospective 51 children enrolled 39 LP analysed 20DHF I/II; 18DSS; 1DF 72 children and adults enrolled 44 DI analysed 21DHF I/II; 15DSS; 8DF Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? All DHF had HC All DHF had HM All DHF had TP All cases had HC No data presented 6/71 (8%) had HC on admission; 14/ 49 (29%) had HC when serial HcT done All had HC HcT >40 was observed in two children 3/38 (8%) had HcT >40 14 had evidence of PL as ascites/pe HC present in all DHF 75/110 (68%) had HM All had TP No data present All had HM 96% (94/98). had TP Strict application of WHO criteria. Due to lack of documentation of HC, cases were reported as DFB, Serial HcT done No, Modified classification used. Fever and TP with PL or shock, ST NM No, Modified classification used. Fever with any two (HM, shock, hepatomegaly) with either TP or raised HcT, Serial HcT done in few All DHF had HM All had TP Strict application of WHO criteria. Could not be applied to 12 cases (20%) due to lack of evidence of HC, 8/13 (62%) shock cases without HC were reported as non DHF, Serial HcT done 21/39 (54%) DI had HM All DHF had positive TT, a requisite for entry into study 36/39 (92%) had TP TP present in all DHF No, ST NM Strict application of WHO criteria, 39% of DHF cases had abdominal pain, ST NM 1242 ª 2006 Blackwell Publishing Ltd

6 Wallace et al. (1980), Malaysia Hayes et al. (1988), Philippines Prospective Wichmann et al. (2004), Thailand Guzman et al. (1987), Cuba Diaz et al. (1988), Cuba Harris et al. (2000), Nicaragua Prospective 138 LP Dis 70DHF; 25 DSS 1055 haemorrhagic fever enrolled. About 517 LP Dis 108 DHF I/II; two DSS; 355 DFB; 52 DF 347 LP DIs, all age groups 219 DHF; 128 DF 40% (38/95) DHF had HC All DHF had HC 94/407 (23%) non -DHF also had HC 5% increase in the HcT level; mean HcT on admission >40 (SD 6.1) and mean max HcT >44.8 (SD 4.7) 124 DSS 103/124 (83%) had HC 104 LP adult hemorrhagic dengue survivors compared to 26 compatible DHF non-survivors 1027 cases enrolled 614 LP Dis 40 DHF I/II; 20 DSS; 17 DSAS; 267 DFB; 268DF 34% of survivors and 92% of NS had HC 88/93 (93%) had HM 88/95 (93%) DHF had TP All DHF had HM All DHF had TP.64% (261/407) non-dhf had TP 100/219 (46%) had HM 77% (65/85) DHF had positive TT 85/124 (69%) had HM 90% of DHF survivors and 65% of NS had HM Median PC of on admission, decreased significantly 97/124 (78%) had TP 80% of survivors and 72% of NS had TP All DHF had HC All had HM All DHF had TP 17 shock had no TP No, Five (5) severe dengue with shock had no HM & could not be classified as DHF, ST NM No, 94/517 (18%) had HC but did not fulfil other WHO criteria for DHF. By default these were classified as DF, with or without HM, Serial HcT done No, Case definition emphasised TP and evidence of PL, ST NM No, ST NM No, All non-survivors had shock; 50% were post haemorrhage. HC or TP could not be confirmed in Fatal & severe cases, Serial HcT done Strict application of WHO criteria. 284/614 (46%) could not be classified (43% were DFB and 17/614 (3%) were shock without HC, Serial HcT done DI, dengue infection; DHF, dengue hemorrhagic fever; DF, dengue fever; GIH, gastrointestinal hemorrhage; HM, hemorrhagic manifestation; HC, hemoconcentration; HcT, hematocrit; NM, not mentioned; PE, pleural effusion; PL, plasma leakage; PC, platelet count; LP, laboratory positive; LN, laboratory negative; ST, serial tests of HcT and PC; TP, thrombocytopenia; TT, tourniquet test. ª 2006 Blackwell Publishing Ltd 1243

7 Table 2 Fulfilment of WHO case classification criteria for DHF (additional to fever) as evidenced in the published literature. review of available records Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? Sumarmo et al. (1983), Indonesia Chan (1987), Philippines Lucas et al. (2000), Sri Lanka Guzman et al. 33 (1999), Cuba Mendez and 34 Gonzalez (2003), Colombia Reynes et al. (1994), French Guyana 3 years review of 148 fatal cases; 30 LP DI analysed 2 year review of laboratory records (number of LP not specified) Review of medical records 177 LP DIs compared with 177 non DIs analysed, 122 DHF I/II; 21 DSS; 34 DF Review of 12 fatal cases of DSS, aged >15 years 10 years review of medical records 763 clinical DHF in children 436 DHF I/II; 180DSS; 147 DF 2 years review of laboratory investigation forms for dengue 328 DIs 32 DHF I/II; Eight DSS, 40 DFB 19/30 had classical DSS with HC 76% of DHF cases with severe HM had HC All DHF had HC (HcT >40%) 9/34 (26%) DF; 53/177 (30%) of non-di had HcT >40%. 24/30 (80%) had massive GIH Severe HM mentioned 28/30 had TP No, 9/30 (30%) had shock from GIH without PL. Two deaths from encephalitis-like disease, Serial HcT done 34% of DHF with severe HM had TP All DHF had HM 48% (69/143) DHF and 1% (2/177) non-di had PC < mm 3 11/12 (8%) had HC All had hemorrhage 10/12 (83%) had TP 85% (645/763) had PL All 763 had HM 734/763 (96%) had TP All DHF had HC All DHF had HM All DHF had TP Could not be applied to an unspecified percentage of cases with severe HM and shock but without HC/TP No, Modified Case classification used, LP with HC >40% and PC < / mm 3,STNM No, ST NM Strict application of WHO criteria. 617 (81%) DHF cases satisfied the criteria. 15% clinical DHF had no HC and 4% had no TP so reported as DF, Serial HcT done Strict application of WHO criteria could not be applied to 5/6 fatal cases and 40 DFB who had no documentation of HC, ST NM 1244 ª 2006 Blackwell Publishing Ltd

8 Table 2 (Continued) Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? Torres et al. (2004), Venezuela 3 years review of medical records 112 adult LP DHF 94DHFI/II, 18DSS Total 65/112 (58%) had evidence of PL 37/112 (33%) had capillary leakage 14 had PE and 14 had acute achalculous cholecystitis 22% (25/112) had HM Data not presented No, ST NM DI, dengue infection; DHF, dengue hemorrhagic fever; DF, dengue fever; GIH, gastrointestinal hemorrhage; HM, hemorrhagic manifestation; HC, hemoconcentration; HcT, hematocrit; NM, not mentioned; PE, pleural effusion; PL, plasma leakage; PC, platelet count; LP, laboratory positive; LN, laboratory negative; ST, serial tests of HcT and PC; TP, thrombocytopenia; TT, tourniquet test. Table 3 Fulfilment of WHO case classification criteria for DHF (additional to fever) as evidenced in the published literature. Prospective clinical studies Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? Peenen et al. (1978), Indonesia Samsi et al. (1990), Indonesia Chairulfatah et al. (1995), Indonesia 104 pediatric DIs enrolled 81 LP cases analysed. (grades II, III and IV DHF) 151 pediatric LP DI analysed 128 DHF I/II, 23 DSS 306 clinical DHF <15 year. Enrolled 128 LP analysed 104 DHF I/II; 24 DSS Admission HcT ranged from 20% to 53% (mean 40.3%). 23 chest X-ray revealed nothing 8.6% on admission and 66/151 (43.7%) during observation 66/128 (52%) had HC during hospitalisation All had HM Lack of facilities for accurate measurement of PC precluded analysis 42/151 (27.8%) had HM 87/151 (57.6%) were TT positive 78% (100/128) positive for TT 8.6% on admission and 58/151 (38.4%) during observation had TP 4/128 (3%) had TP on admission, increasing to 15% (19/124) during hospitalisation No, even with, rigid inclusion criteria, Serial HcT done No, Serial HcT done No, Modified classification used. Fever with at least one sign or symptoms of (HM, shock, TP, rising HcT, positive TT) 33/174 (19%) seronegative cases had shock with four deaths, no data was presented, Serial HcT done ª 2006 Blackwell Publishing Ltd 1245

9 Table 3 (Continued) Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Were all WHO classification criteria fulfilled? Kalayanarooj et al. (1997), Thailand Prospective Lucas et al. (2001), Sri Lanka Prospective Zagne et al. (1994), Brazil Prospective 189 pediatric cases with strict inclusion criteria enrolled 60 LP acute DI analysed 32 DF, 19 DHF I/II; nine DSS 51 Paediatric cases enrolled 46 LP DHF analysed 38 DHF; 8 DF 176 DIs all age groups enrolled 56 LP DHF analysed 29 DHF; 27 DSS HC was observed in 12/25 (48%) DHF cases; others had radiological or clinical evidences of PL. Three DF cases had HC but no TP. Small PE detected in 2 (6%)DF and 17(7%) non dengue illness without any HC, TP or HM 31/38 (82%) DHF had HcT >40% All had PL 52 (93%) had HcT >20%. Four had more than 10% but also PE 33/38 had PE All DHF had HM 65% DHF and 64% DF cases were TT positive 23/28 (82%) DHF had TP.5/28 (18%) DHF cases had significant decrease in PC but never < /mm 3 All had HM 26/38(68%) had TP ( /ll) 59% (33/54) had HM 46/56 (82%) was TT positive PC ranged from 4500 to / mm 3 Strict classification criteria were used, Considerable experience and meticulous case note review is necessary for assignment of cases where evidence of PL without TP, HC or HM is noticed, Serial HcT done No, Modified classification used. LP with TP & HC >40% or 20% HcT increase over the recovery value constitute the DHF cases, ST NM No, Serial HcT done DI, dengue infection; DHF, dengue hemorrhagic fever; DF, dengue fever; GIH, gastrointestinal hemorrhage; HM, hemorrhagic manifestation; HC, hemoconcentration; HcT, hematocrit; NM, not mentioned; PE, pleural effusion; PL, plasma leakage; PC, platelet count; LP, laboratory positive; LN, laboratory negative; ST, serial tests of HcT and PC; TP, thrombocytopenia; TT, tourniquet test ª 2006 Blackwell Publishing Ltd

10 Table 4 Fulfilment of WHO case classification criteria for DHF (additional to fever) as evidenced in the published literature. Evaluation studies of the usefulness of the WHO classification Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Comments 35 Rigau-Pérez (1997), Puerto rico Rigau-Pérez et al. 36 (1999), Puerto Rico 217 case records for DHF screened; 102 DHF 88 DHF I/II; 14 DSS; 115 DF Evaluation of modified case definition 1075 suspected DI: 915 LP, 160 LN A total of 118 cases satisfied the WHO criteria. 97/915 (11%) LP and 21/160 (13%) LN cases satisfied the WHO criteria for DHF 57/102 (56%) DHF had PL.11/53 (21%) of LP DF had discharge diagnosis of DHF without evidence of PL ALL DHF had HC 11% non-dhf (108/957) had PL 57/102 DHF had HM; 5/16 was TT positive 28/54 (52%) had microscopic hematuria All DHF had HM 40% non-dhf (378/957) had HM All 57 LP DHF had TP Only 57/102 (55.8%)of clinical LP DHF satisfied the WHO case definition, Similar CFR in DF cases 5.2% (6/115) and DHF cases 4.9% (5/102) suggest that DI could be fatal without capillary leakage. Observed incidence was nearly three times more than the reported incidence 57 vs. 20 cases. DHF is under diagnosed, ST NM All DHF had TP 10 deaths in DHF and 17 deaths in non-dhf cases Different PL criteria used to find out the maximum severe compatible LP DHF cases. Options were: (1) HcT level >50% at any time or, (2) HC P10% or, (3) Either HcT level >50% or HC P10%, ST NM ª 2006 Blackwell Publishing Ltd 1247

11 Table 4 (Continued) Authors, year, study type Patients studied Evidence of plasma leakage (PL) Haemorrhagic manifestations (HM) Thrombocytopenia (TP) Comments Murgue et al. (1999), Haiti Phuong et al. (2004), Vietnam Prospective A severity score based on basic clinical and biological criteria assigned to 403 LP cases. WHO case criteria were applied to 50 severe cases 17 cases (34%) were DF and 33 (66%) were DHF 712 Paediatric LPDI cases enrolled 310 DSS cases satisfying WHO case criteria analysed HcT was >45% in all DHF 11/17 (65%) DF had PL 20 did not reach the cut off value HC >44% (mean baseline HcT 37%) HM present in all DHF cases Haemorrhage was one of the parameters in the severity score of dengue manifestation 22 had no HM/TT positively throughout the observation period TP present in all DHF cases TP was also included in severity score 17/50 (34%) severe cases did not satisfy the WHO case criteria of DHF, eight severe DHF cases were without signs of shock, ST NM 15 did not have TP 57/310 (18%) DSS cases did not meet all four WHO criteria of DHF. Several models of modified classification to capture all DSS presented, (1) Fever &HMwithPLor TP identifying 93% of DSS cases or, (2) Fever and HC with either HM or TP identifying 94% of DSS cases, Serial HcT done DI, dengue infection; DHF, dengue hemorrhagic fever; DF, dengue fever; GIH, gastrointestinal hemorrhage; HM, hemorrhagic manifestation; HC, hemoconcentration; HcT, hematocrit; NM, not mentioned; PE, pleural effusion; PL, plasma leakage; PC, platelet count; LP, laboratory positive; LN, laboratory negative; ST, serial tests of HcT and PC; TP, thrombocytopenia; TT, tourniquet test ª 2006 Blackwell Publishing Ltd

12 1999). Difficulty in applying the TT may also result from faulty blood pressure sets or inappropriate sized cuffs. In Dhaka during the 2000 outbreak (Pervin et al. 2004) 19% of cases were TT positive when TT was done only once. Unless the TT is repeated after restoring the peripheral circulation, the TT might be negative in a poorly perfused patient with DSS. Some clinicians felt that the test should be judged qualitatively with the development of petechiae anywhere on the entire limb being considered 8 positive (Rohde 1978) or by the appearance of just one purpuric spot after performing a standard TT in a dark skinned individual (M. Narayanan, personal communication). Further modifications to the TT included the use 9 of an elastic cuff (Phuong et al. 2002) instead of the blood pressure cuff in areas where such resources were scarce. Thrombocytopenia in DHF In dengue the platelet counts show a trend towards lownormal levels as the disease approaches defervescence and become less than /l during the critical phase of plasma leakage. In a number of studies only a reduced proportion of DHF cases had such low platelet counts. Thrombocytopenia in DHF has been reported with a range of 8.6% in Indonesian (Samsi et al. 1990) and 48% in Sri Lankan (Lucas et al. 2000) prospective studies, 54% in Bangladesh (Ahmed et al. 2001), 70% in Indian (Srivastava et al. 1990) and 78% in Cuban (Guzman et al. 1987) outbreak studies. In mild forms of DHF, the duration of thrombocytopenia may be brief, with a brisk recovery of platelet counts as soon as the increased vascular permeability resolves. Although a daily platelet count may suffice to detect thrombocytopenia in moderate to severe DHF/DSS, more frequent and timely repeats may be necessary to detect thrombocytopenia in the milder cases, in which physicians might feel the redundancy to pursue frequent platelet evaluations. In Thailand however, where routine serial platelet count is undertaken, 94% of 2475 DHF cases had less than /l platelet count (Kalayanarooj et al. 2002). In a previous prospective study by Kalayanarooj et al. in Bangkok, Thailand (1997), 5 of the 28 virologically proven DHF cases (34%) did not have a platelet count of lower than /l even when serial counts showed significant decreases in platelet counts. Routine examination on admission and perhaps even subsequent regular 6 h examinations may miss this important clinical finding. Many severe DSS cases do not have platelet counts less than /l. Although they do not fulfil the WHO criteria, they are reported as DSS due to the presence of capillary leakage (B. Wills, personal communication). In a prospective clinical study by Chairulfatah et al. (1995) in Indonesia, admission thrombocytopenia was found in 5.3% of DHF cases increasing to 15.3% during hospitalisation. In another prospective clinical study of 46 laboratory-positive DHF cases by Lucas et al. (2001), 68% of DHF patients, had thrombocytopenia. Most studies did not state the frequency of repeating the platelet count or the method of the platelet count if done by direct method or indirect method. It might be possible that the recommended lowest value of platelet counts were missed. Plasma leakage in DHF Haemorrhagic manifestations and thrombocytopenia occur in a significant minority of DF as well as non-dengue febrile illness but increased vascular permeability occurs only in DHF. This particular feature is most noticeable during defervescence, about 3 5 days after the onset of fever. Several studies had difficulties to document the evidence of increased permeability. Although these studies were undertaken in tertiary health facilities, it is difficult for the reviewers to assess if the efforts to document the plasma leakage were sufficiently exhaustive or if the patient never had this clinical sign at all. In the WHO guidelines, the evidence of increased capillary permeability are 20% increase in Hct above the baseline value or clinical evidence of third space losses such as pleural effusion and/or ascites. The convalescent Hct may be a substitute for the baseline level but the decision on whether a patient has evidence of plasma leakage, in the absence of any other, will have to be deferred to the convalescent phase when it is largely an academic question if a patient had DF or DHF. Some physicians (Lucas et al. 2000; Phuong et al. 2004) extrapolated the baseline Hct from the respective healthy population in order to identify the 20% increase early. This value would obviously differ amongst the various communities; Lucas used an Hct of >40%, assuming the baseline Hct of about 33%, perhaps reflecting the prevalence of anaemia in the population studied. Using this limit, all DHF patients would fulfil the criteria for increased vascular permeability, but 26% (9/34) of DF and 30% (53/177) of non-di would also meet these criteria, thus reducing its specificity significantly. Intravenous fluid therapy for dehydration, before the onset of increased vascular permeability would mask the expected rise in Hct (Guzman et al. 1987, 1999). Alternatively, it may be impossible to measure the Hct in poorly perfused patients until the circulation is restored with intravenous fluids. Other circumstances where haemoconcentration may be difficult to document are concurrence of or predominance of bleeding over shock (Sumarmo et al. 1983; Pushpa et al. 1998), high prevalence of anaemia (Agarwal ª 2006 Blackwell Publishing Ltd 1249

13 et al. 1998; Shah et al. 2004), difficulty to ascertain haemoconcentration with a single Hct (Van Peenen et al. 1978; Kabra et al. 1999), excessive laboratory workload during epidemic outbreaks (Sharma et al. 1998), requirement for blood transfusion and sudden death of patients (Reynes et al. 1994) where repeat tests could not be performed. Hard evidence of pleural effusion/ascites is the most unequivocal documentation of increased vascular permeability. A right pleural effusion/presence of ascites is readily detected in a typical case of DSS. However, plain radiography of the chest, especially in the supine position, may not demonstrate small sized pleural effusion. A right lateral decubitus film or an ultrasound may be required in 10 such cases (Chumdermpadetsuk 1978; Nimmannitya 1996; Rigau-Perez 1999). Again, the timing of this investigation is crucial. Free fluid in the third spaces may become detectable only after intravenous fluid therapy in mild cases of DHF. Small pleural effusion could be detected also in DF or other non-dengue illnesses (Kalayanarooj et al. 1997) but thrombocytopenia or haemoconcentrtion will be absent. One has to look for other indirect evidence of increased vascular permeability, which are hypoalbuminaemia, excessive positive fluid balance (Nimmannitya 1993), and a reduction in urine output despite adequate 11 intravenous fluid therapy (Wills et al. 2004). In the absence of any bleeding, the rate of IV fluid infusion and the rate of leakage of plasma determine the Hct level. Rapidly declining Hct in a hemodynamically unstable patient is considered a warning sign for bleeding. The interpretation of Hct values, thus is not straightforward, especially when bleeding is significant but occult. It should always be done in conjunction with the clinical picture and haemodynamic status of the patient. A normal Hct value in a hypotensive patient is an indication of severe bleeding (Anh & Tram 1995; Lum et al. 2002). On the other hand, a declining Hct value in a well-perfused patient is a common occurrence during the reabsorption phase of DHF/DSS and can be mistaken as haemorrhage when pulse and urine output are not monitored. The establishment of a clinical algorithm that specifies the action to be taken at each step of evaluating a particular complaint would help clinicians manage cases better. Martinez (1995) has proposed one such classification system for dengue epidemics and an algorithm for important clinical signs of diagnosis and therapy. In an active and passive surveillance study, focused on the criteria for diagnosis of DHF (Rigau-Perez & The Puerto Rico Association of Epidemiologists 1997) it was observed that evidence of increased vascular permeability is not commonly considered by physicians to diagnose DHF. Instead, the diagnosis hinged more on the presence of clinical haemorrhage (excluding positive TT) rather than that of plasma leakage resulting in under-diagnosis and underreporting of DHF. A high case fatality among DF patients, as reported by Rigau-Perez and The Puerto Rico Association of Epidemiologists (1997) suggests that the authors might have included actual DHF cases in the DF category because they did not fulfil the DHF case definition for lack of information, or, that dengue infections with fatal outcomes at times develop without increased capillary permeability. Severe dengue not fulfilling all WHO criteria for DHF Dengue with haemorrhage without plasma leakage The paradox with DHF/DSS is that the haemorrhagic manifestation is minor in most cases and that the increased vascular permeability can be readily corrected with crystalloids or a colloid infusion. However, sometimes shock may be accompanied by or associated with severe bleeding, which would require blood transfusion. Severe haemorrhage may be a prominent manifestation as Kalayanarooj et al. (2002) observed in about 13% of DSS cases. Dengue with severe haemorrhage, not accompanied by increased vascular permeability are being reported from various parts of the world: Indonesia (Sumarmo et al ), China (Qiu et al. 1991), India (Pushpa et al. 1998, 13 Tripathi et al. 1998; Kabra et al. 1999) Philippines (Chan ; Hayes et al. 1988), Thailand (Kalayanarooj et al. 2002; Wichmann et al. 2004), South Pacific (Murgue et al. 1999) and Latin America (Reynes et al. 1994; Rigau-Perez & The Puerto Rico Association of Epidemiologists 1997; Harris et al. 2000; Corwin et al. 2001). Diaz et al. (1988) reported that gastrointestinal haemorrhage occurred before or on the same day as the onset of shock and of the 23 nonsurvivors only 2 had autopsy findings of effusions. Kabra et al. (1999) used the term DF with unusual bleeding (DFB) to describe the 21 cases (9%) who had more mucosal bleeding than those with DHF/DSS but required the same amount of fluids as those with DF. The mortality rate during the 1996 outbreak in Delhi was high (Broor et al ), indicating the severity of DFB. The difficulty to manage DF with bleeding was also experienced by other clinicians (Pushpa et al. 1998; D. G. Rubio 2004, personal communication). For a similar set of cases, Narayanan et al. (2002) used the term DF with haemorrhage. In the PAHO guidelines (1994) this has been called DF with unusual haemorrhage. Dengue with Shock without fulfilling the four DHF criteria In the study by Phuong et al. (2004) in Vietnam, 310 cases of severe dengue with shock were analysed in whom 1250 ª 2006 Blackwell Publishing Ltd

14 57 (18%) did not fulfil all the four criteria for DHF, and by a strict application of WHO criteria, would be classified as DF. Others (Wallace et al. 1980; Reynes et al. 1994; Murgue et al. 1999; Rigau-Perez & The Puerto Rico Association of Epidemiologists 1999; Harris et al. 2000; Narayanan et al. 2002) have reported similar experiences where dengue cases had signs of shock but no clear evidence of increased vascular permeability and/or thrombocytopenia and/or haemorrhage. Likewise in a recent prospective study in Nicaragua the strict application of the WHO criteria failed to detect a significant number of patients with shock and other severe manifestations of dengue, especially in adults (Balmaseda et al. 2005). Organ dysfunctions as prominent manifestations of severe dengue The majority of cases of severe manifestations and organ impairment in dengue are the consequences of plasma leakage and shock; however, increasingly, organ dysfunction, out of proportion to the degree of plasma leakage, has been reported. These disorders of the (Nimmannitya et al. 1987; George et al. 1988; Lum et al. 1993; Wichmann et al. 2004), brain (Hendarto & Hadinegoro 1992; Patey et al. 1993; Lum et al. 1996; Witayathawornwong 2004), and respiratory (Lum et al. 1995) are not accounted for in the present WHO classification. Murgue et al. in Tahiti (1999) proposed a scoring system based on haemorrhage, thrombocytopenia and organ dysfunction to define the severity of illness in DF and DHF cases. Applying these scores, 50 cases identified as the most severe were DHF in 33 cases (66%) and DF in 17 (34%) cases. The main drawback with a scoring system based on laboratory abnormalities is their clinical relevance, for example, increased liver transaminases may not be associated with any form of liver failure. Discussion Methodological and other issues in the present review Although some articles in Spanish and other languages could not be included and an exhaustive assessment of scientific quality was not the objective of the review, the countries with the highest disease burden have been included and relevant methodological issues have been identified. Possible differences in clinical presentation, case management, poor planning and constraints associated with conducting research in endemic or outbreak situations could have resulted in varied findings. Lack of sound statistical methods in some studies due to retrospective study design and inadequately described case selection, non-representative sample or small sample size in some studies remain major weaknesses of the reviewed papers. With the exception of a few articles, the authors have not described the use of standard operating procedures or standard equipment to measure the Hct or thrombocytopenia, which weakens the comparison among different studies. The presence of plasma leakage in all DHF cases was documented in only 38% (8/21) of outbreak studies, 29% (2/7) of record review studies and 33% (2/6) planned prospective clinical studies. The presence of thrombocytopenia in all DHF cases could not be documented in 12 (57%) outbreak studies, six record reviews (86%) and in any of the prospective studies. This highlights the difficulties in using the WHO case classification criteria even in the setting of a large health care facility with excellent planning, adequate resources and laboratory services. The diagnosis and documentation of thrombocytopenia and evidence of plasma leakage depends upon the timely clinical examination and serial testing of blood samples as the disease evolves by the days and hours. In reality, clinicians face a number of challenges in trying to meet these criteria. Conditions are seldom ideal in routine clinical care. Early admission and management reduce the number of deaths due to shock (Chairulfatah et al. 1995). But the risk of death due to the severe and unusual manifestations of dengue massive haemorrhage, pulmonary overload and congestive heart failure- has 17 increased (George & Duraisamy 1981; Sumarmo et al. 1983; Chaimahaprug et al. 1989; Kalayanarooj et al. 1989, 2002; Pimchaipong et al. 1990; Rigau-Perez et al ; Cam et al. 2001; Patumanond et al. 2003). On the other hand, if DF is interpreted as a mild disease and DHF as the severe form of dengue (Panhuis et al. 2005) the magnitude of life-threatening dengue manifestations and their geographical extent cannot be estimated. This was particularly observed by Tripathi et al. (1998) in India during the 1996 Delhi outbreak when only 14 cases (2.5%) among 560 hospitalised adult dengue cases could be classified as DHF. It underestimates the true extent of the problem, as more than 423 deaths have been officially recorded among suspected dengue infections. The current case classification system does not capture cases of severe and life threatening dengue, which do not fulfil the criteria for DHF and lead to under-diagnosis and underreporting of severe disease. Conclusion After more than 30 years of applying the WHO classification in many parts of the world in different health facilities, several challenges have been identified and it is timely to ª 2006 Blackwell Publishing Ltd 1251

15 re-visit the classification. A prospective multi-centre study across dengue endemic regions and age groups and across the whole health care system is required which describes the clinical presentation of dengue disease including simple laboratory parameters in order to review and if necessary modify the current WHO case classification system. Acknowledgements The study was supported by the UNICEF/UNDP/ WORLD BANK/WHO Special Programme for Research and Training in Tropical Diseases (TDR/WHO). The authors wish to thank Drs. Jose G. Rigau Perez (Puerto Rico), Eva Harris (Nicaragua), Maria Guzman (Cuba), DG Rubio (Cuba), Eric Martinez (Cuba), Jacqueline L. Deen (Paediatric Dengue Vaccine Initiative, Korea), Manjith Narayanan (India), Bridget Wills (Vietnam), Gerardo Gonzalez (Colombia), Xavier Bosch (Liverpool) and Neal Alexander (Colombia), for their valuable help with their published articles and/or comments on earlier drafts. We also acknowledge the valuable support of George Peterson and Steven S Bjorge (WHO office, Indonesia), Dr. Nyoman Kandun, Director General of Disease Control and Environmental Health, Indonesia, Anchalee Chamchuklin (Library, SEARO) and A. Thomas (Library, WHO, Geneva) to procure important documents and articles. References Agarwal A, Chandra J, Aneja S, Patwari AK & Dutta AK (1998) An epidemic of dengue hemorrhagic fever and dengue shock syndrome in children in Delhi. Indian Pediatrics 35, Ahmed FA, Mahmood CB, Das Sharma J, Hoque SM, Zaman R & Hasan MS (2001) Dengue and DHF in children during the 2000 outbreak in Chittagong, Bangladesh. Dengue Bulletin 25, Anh NN & Tram TT (1995) Integration of primary health care concepts in a children s hospital with limited resources. Lancet 346, Balmaseda A, Hammond SN, Pèrez MA et al. (2005) Assessment of the World Health Organization scheme for classification of dengue severity in Nicaragua. American Journal of Tropical Medicine and Hygiene 73, Broor S, Dan L, Sengupta S, Chakraborty M & Seth P (1997) Recent dengue epidemic in Delhi, India. In: Factors in the Emergence of Arbovirus Disease (eds JF Saluzzo & B Dodet) Elsevier, Paris, pp Cam BV, Fonsmark L, Hue NB, Phuong NT, Poulsen A & Heegaard ED (2001) Prospective case control study of encephalopathy in children with dengue haemorrhagic fever. American Journal of Tropical Medicine and Hygiene 65, Chan VF (1987) Virological and epidemiological studies of dengue haemorrhagic fever in the Philippines. Southeast Asian Journal of Tropical Medicine and Public Health 18, Chairulfatah A, Setiabudi D, Ridad A & Colebunders R (1995) Clinical manifestations of dengue haemorrhagic fever in children in Bandung, Indonesia. Annales de la Societe Belge de Medecine Troicale 5, Chaimahaprug P, Chanwong N & Megarapong P (1989) Causes of death in dengue haemorrhagic fever at Phichit Hospital in Bulletin of the Department of Medical Services, Thailand 14, Chumdermpadetsuk S (1978) Early recognition of severe dengue haemorrhagic fever. Journal of Medical Association, Thailand 61, Corwin AL, Larasati RP, Bangs MJ et al. (2001) Epidemic dengue transmission in Southern Sumatra, Indonesia. Transactions of the Royal Society of Tropical Medicine and Hygiene 95, Diaz A, Kouri G, Guzman MG et al. (1988) Description of the clinical picture of DHF/DSS in adults. Bulletin of the Pan American Health Organization 22, George R & Duraisamy G (1981) Bleeding manifestation of DHF in Malayasia. Acta Tropica 38, George R, Lam CK, Chua CT et al. (1988) Unusual clinical manifestation of dengue virus infection. Southeast Asian Journal of Tropical Medicine and Public Health 19, Gubler DJ (1998) Dengue and dengue haemorrhagic fever. Clinical Microbiology Reviews 11, Guha-Sapir D & Schimmer B (2005) Dengue fever: new paradigms for a changing epidemiology. Emerging Themes in Epidemiology. Open access journal. 1/1. Guzman MG, Kouri G, Martinez E et al. (1987) Clinical and serologic study of Cuban children with Dengue Haemorrhagic Fever/Dengue shock syndrome. Pan American Health Organization Bulletin 21, Guzman MG, Alvarez M, Rodriguez R et al. (1999) Fatal dengue haemorrhagic fever in Cuba, International Journal of Infectious Diseases 3, Guzman MG, Kouri G, Diaz M et al. (2004) Dengue, one of the great emerging health challenges of the 21st Century, Expert Review. Vaccine 3, Harris E, Videa E, Perez L et al. (2000) Clinical, epidemiologic, and virologic features of dengue in the 1998 epidemic in Nicaragua. American Journal of Tropical Medicine and Hygiene 63, Hayes CG, Manaloto CR, Gonzales A et al. (1988) Dengue infections in the Philippines: clinical and virological findings in 517 hospitalised patients. American Journal of Tropical Medicine and Hygiene 39, Hendarto SK & Hadinegoro SR (1992) Dengue encephalopathy. Acta Paediatrica Japonica 34, Kabra SK, Verma IC, Arora NK, Jain Y & Kalra V (1992) Dengue haemorrhagic fever in children in Delhi. Bulletin of the World Health Organization 70, ª 2006 Blackwell Publishing Ltd