Evaluation of the World Health Organization 2009 classifi cation of dengue severity in autopsied individuals , during the epidemics of 2011 and 2012 in Brazil

Introduction: The dengue classifi cation proposed by the World Health Organization (WHO) in 2009 is considered more sensitive than the classifi cation proposed by the WHO in 1997. However, no study has assessed the ability of the WHO 2009 classifi cation to identify dengue deaths among autopsied individuals suspected of having dengue. In the present study, we evaluated the ability of the WHO 2009 classifi cation to identify dengue deaths among autopsied individuals suspected of having dengue in Northeast Brazil, where the disease is endemic. Methods: This retrospective study included 121 autopsied individuals suspected of having dengue in Northeast Brazil during the epidemics of 2011 and 2012. All the autopsied individuals included in this study were confi rmed to have dengue based on the fi ndings of laboratory examinations. Results: The median age of the autopsied individuals was 34 years (range, 1 month to 93 years), and 54.5% of the individuals were males. According to the WHO 1997 classifi cation, 9.1% (11/121) of the cases were classifi ed as dengue hemorrhagic fever (DHF) and 3.3% (4/121) as dengue shock syndrome. The remaining 87.6% (106/121) of the cases were classifi ed as dengue with complications. According to the 2009 classifi cation, 100% (121/121) of the cases were classifi ed as severe dengue. The absence of plasma leakage (58.5%) and platelet counts <100,000/mm3 (47.2%) were the most frequent reasons for the inability to classify cases as DHF. Conclusions: The WHO 2009 classifi cation is more sensitive than the WHO 1997 classifi cation for identifying dengue deaths among autopsied individuals suspected of having dengue.

Dengue is a major public health problem worldwide (1) (2) .The dengue cases in Brazil account for approximately 60% of all the dengue cases reported to the World Health Organization (WHO) from the Americas (3) .In the State of Ceará in Northeast Brazil, dengue was fi rst noted in 1986, and it has since been endemic to the region, with many documented epidemics (4) (5) .
There are diffi culties in the classifi cation of dengue cases using the WHO 1997 classifi cation; therefore, since 2001, Brazil has adopted the term dengue with complications (DWC) (4) (6) .However, the term DWC has been used only in Brazil, and its use has complicated comparisons of dengue rates between Brazil and other countries (7) (8) .
In this context, based on a multicenter study [Dengue Control (DENCO)], the WHO proposed a new classifi cation system in 2009 (WHO 2009 classifi cation) (9) .This classifi cation includes cases of dengue without warning signs, dengue with warning signs, and severe dengue (SD) (9) (10) .
Many studies across the world have shown a higher sensitivity and greater ability to include a large number of cases with the WHO 2009 classifi cation that with the WHO 1997 classifi cation (8) (10) (11) (12) (13) (14) (15) (16) (17) .However, limited evidence exists on the impact of the WHO 2009 classifi cation on the notifi cation of deaths and on the evaluation of dengue fatality indicators around the world (18) .
In Northeast Brazil, where dengue is endemic, autopsies have identified many deaths caused by dengue, although dengue was not suspected during the clinical courses of

METHODS
the patients (19) .No previous study has assessed the ability of the WHO 2009 classifi cation to identify dengue deaths among autopsied individuals suspected of having dengue.The objective of the present study was to evaluate the ability of the WHO 2009 classifi cation to identify dengue deaths among autopsied individuals suspected of having dengue in Northeast Brazil.

Study design
This retrospective study evaluated dengue deaths among autopsied individuals suspected of having dengue at the Coroner's Offi ce Dr. Rocha Furtado (CO-RF), Northeast Brazil, during the epidemics of 2011 and 2012.
All the individuals suspected of dengue were reported to the Center for Epidemiological Surveillance of the State Department of Health for epidemiological investigation.Following the report of a suspected case, the epidemiological surveillance team of the state investigates the case to confi rm or exclude dengue (4) (6) (18) (20) .
Data were collected from January 2011 to December 2012 using a structured questionnaire.The data included age, sex, symptoms and signs, warning signs, shock evidence (signs of poor perfusion and hypotension), effusion (ascites, pleural effusion, pericardial effusion, and respiratory distress), laboratory test results, and autopsy fi ndings.
We systematically retrieved data from medical records, notifi cation forms used by the Brazilian Ministry of Health, and laboratory test records for all reported cases of dengueassociated deaths during the study period.

Ethical considerations
This study followed the ethical principles of research involving human subjects prepared by the National Health Council of Brazil and was approved by the Ethics Review Boards of Centro Universitário Christus -UNICHRISTUS (protocol #011/2011) and Hospital São José de Doenças Infecciosas (protocol #078/2011).Identifying information in the medical records was codifi ed to preserve confi dentiality.

Patients and defi nitions
In this study, the autopsied individuals suspected of having dengue had features clinically consistent with dengue, and they underwent reverse transcription polymerase chain reaction, NS1 antigen enzyme-linked immunosorbent assay, immunoglobulin M antibody assay, virus isolation, or immunohistochemistry to confi rm the presence of dengue.All the autopsied individuals included in this study were confi rmed to have dengue based on the fi ndings of the laboratory examinations.
In this study, leukopenia was defi ned as a leukocyte count <1,000/mm 3 , hypoalbuminemia was defined as a serum albumin level <3g/dL, liver injury was defi ned as an aspartate aminotransferase or alanine aminotransferase level <1,000U/L, and severe thrombocytopenia was defi ned as a platelet count <20,000/mm 3 .

Statistical analysis
Data were stored in Microsoft Offi ce Access version 2010 (Microsoft Corp., Redmond, WA) and analyzed using Epi Info version 3.5.1 (Centers for Disease Control, Atlanta, GA).Data were classified in accordance with the classification systems used currently in Brazil adapted from the WHO 1997 classifi cation and the WHO 2009 classifi cation.The descriptive analyses included the frequencies of quantitative variables, number of deaths, and case-fatality rates.
The absence of plasma leakage (58.5%) and platelet counts <100,000/mm 3 (47.2%)were the most frequent reasons for the inability to classify cases as DHF (Table 2).
Of the 106 cases classifi ed as DWC, 9 (8.5%) did not fulfi ll any of the four criteria for DHF according to the WHO 1997 classifi cation, 38 (35.8%) did not fulfi ll three criteria, 28 (26.4%) did not fulfi ll two criteria, and 31 (29.3%) did not fulfi ll one criteria.
With respect to the signs and symptoms, fever was noted in 87% (100/115) of the cases, and respiratory distress syndrome, persistent vomiting, cough, prostration, dyspnea, and abdominal pain were noted in over 70% of the cases (Table 3).

DISCUSSION
In the present study, of the 121 cases, 106 (87.6%) were not classified as DHF/DSS because the WHO 1997 criteria were not fulfi lled, which indicates the low sensitivity of the WHO 1997 classifi cation for the detection of severe cases (8) (15) (16) (17) (22) and emphasizes the importance of adopting the WHO 2009 classifi cation (10) .Using the WHO 2009 classifi cation, we found that all autopsied individuals in this study could be considered as having SD, suggesting that the WHO 1997 classifi cation is unable to identify severe cases, including those with a high probability of death.Based on these fi ndings, from an epidemiological viewpoint, countries that use only the DHF criteria in a strict sense may tend to underreport dengue-associated deaths when considering only deaths due to DHF (23) .
In the present study, fi ve laboratory techniques were used to evaluate the presence of the dengue virus.The Brazilian Ministry of Health and the State Department of Health in Ceará recommend that all dead individuals suspected of having dengue undergo post-mortem examination (6) , and this protocol allows the confi rmation of dengue in suspected dengue deaths that would have been otherwise unconfi rmed (19) .Findings from previous studies reinforce the importance of using techniques to confi rm dengue in post-mortem examinations of individuals suspected of having dengue, especially children (15) (24) .
The post-mortem examinations performed by the Coroner's Offi ce Dr. Rocha Furtado identifi ed the presence of plasma leakage and bleeding, and evidence of comorbidities that were not reported during the clinical course of the disease.These fi ndings contributed to the classifi cation of the cases as DHF, although the cases did not completely fulfi ll the WHO 1997 criteria, and similar fi ndings have been reported in other countries (25) (26) .A recent study of 13 autopsied children with dengue reported thrombocytopenia in all the children (23) .However, in that study, only individuals who fulfi lled all four criteria for DHF according to the WHO 1997 classifi cation underwent post-mortem examination.Comparison of this previous study with the present study is diffi cult because in the previous study, autopsies were performed because a diagnosis of DHF had already been made, while in the present study, dengue was diagnosed only after the post-mortem examination.
Many studies have reported on autopsies of patients who died from dengue; however, most of the patients in those studies died from DHF.In the present study, the majority of dengue cases did not fulfi ll all the criteria for DHF/DSS and dengue was confi rmed only after autopsy completion.In the literature, such cases are usually described as atypical manifestations, published as separate reports, or described as exotic fi ndings after autopsies (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) .
We found that patients aged >65 years had a high possibility of death, which may be related in part to the high frequency of comorbidities reported in the elderly (38) (39) (40) (41) (42) (43) (44) .Information on comorbidities is not included in the Brazilian Information System for Notifi able Diseases, and this information is obtained only through investigation of fatal dengue cases (40) (42) .
An important issue with the WHO 1997 classifi cation is that countries usually adopt different methods to calculate fatality from dengue, and comparison of dengue case-fatality rates among different countries is diffi cult because countries interpret the DHF defi nition differently (45) (46) .We believe that the WHO 2009 classifi cation facilitates comparisons of results and data from different countries.
In Brazil, fatality from dengue is calculated by dividing the number of DHF deaths by the number of confi rmed cases of DHF.However, this indicator is infl uenced by the surveillance capacity to capture all suspected cases of DHF.Additionally, some regions divide the number of DHF deaths by population size.The WHO 2009 classifi cation proposes that countries calculate the fatality from SD by dividing the number of deaths from SD by the population of each region, and then multiplying the obtained value by 1,000.The adoption of the WHO 2009 classifi cation should improve the comparison of dengue fatality rates among different countries because it captures severe cases (8) (15) (22) .It is believed that if this indicator is used, the fatality rates from dengue can be easily compared among different regions, considering population size and the impact of the disease in each region.
One important limitation of this study is that it included a small number of cases.The study could not obtain data from many patients because some were not hospitalized and epidemiological investigations were performed after postmortem examinations.
The WHO 2009 classifi cation is more sensitive than the WHO 1997 classifi cation for identifying dengue deaths among autopsied individuals suspected of having dengue.

FIGURE 1 -
FIGURE 1 -Characteristics of the 106 cases classifi ed as dengue with complications according to World Organization 1997 classifi cation adapted by Brazil and as severe dengue according the 2009 WHO classifi cation.WHO: World Health Organization.

TABLE 1 -Comparison of the World Health Organization 1997 classifi cation adapted by Brazil and the WHO 2009 classifi cation for the 121 autopsied individuals in Northeast Brazil in the 2011 and 2012 dengue epidemics.
WHO: World Health Organization; DF: dengue fever; DWC: dengue with complications; DHF: dengue hemorrhagic fever; DSS: dengue shock syndrome.

TABLE 2 -Criteria not met in the 106 cases for classifi cation as DHF.
DHF: dengue hemorrhagic fever.