Evaluation of the WHO Clinical Case Definition for Pediatric HIV Infection in Bloemfontein, South Africa

Transcription

1 Evaluation of the WHO Clinical Case Definition for Pediatric HIV Infection in Bloemfontein, South Africa by Christine L. van Gend, a Maaike L. Haadsma, a Pieter J. J. Sauer, a and Cornelius J. Schoeman b a Department of Pediatrics, University Hospital, Groningen, The Netherlands b Department of Pediatrics and Child Health, University of the Free State, Bloemfontein, Republic of South Africa Summary The WHO clinical case definition for pediatric HIV infection has been designed to be used in countries where diagnostic laboratory resources are limited. We evaluated the WHO case definition to determine whether it is a useful instrument to discriminate between HIV-positive and HIV-negative children. In addition, clinical features not included in this case definition were recorded. We recorded clinical data from 300 consecutively admitted children in a state hospital in Bloemfontein, South Africa, and tested these children for HIV infection. A total of 222 children were included in the study; 69 children (31.1 per cent) were HIV positive. The sensitivity of the WHO case definition in this study was 14.5 per cent, the specificity was 98.6 per cent. Apart from weight loss and generalized dermatitis, the signs of the WHO case definition were significantly more often seen in HIVpositive than in HIV-negative children. Of the clinical signs not included in the WHO case definition, marasmus and hepatosplenomegaly especially occurred more frequently in HIV-positive children. Based on these findings we composed a new case definition consisting of four signs: marasmus, hepatosplenomegaly, oropharyngeal candidiasis, and generalized lymphadenopathy. HIV infection is suspected in a child presenting with at least two of these four signs. The sensitivity of this case definition was 63.2 per cent, the specificity was 96.0 per cent. We conclude that in this study the WHO case definition was not a useful instrument to discriminate between HIV-positive and HIV-negative children, mainly because its sensitivity was strikingly low. The simplified case definition we propose, proved to be more sensitive than the WHO case definition (63.2 vs per cent), whilst its specificity remained high. Introduction In 1987 the World Health Organization (WHO) published a clinical case definition for HIV infection in children. 1 It was developed primarily to be used in countries where other diagnostic resources are limited. The case definition consists of three major and six minor signs (Table 1); HIV infection is suspected in a child presenting with at least two of the major and two of the minor signs. The WHO stresses the need for an evaluation of the case definition in different settings. 1 However, since its publication the case definition has only been reviewed in a few studies. 2 5 Acknowledgements We thank the J. K. de Cock Stichting, the Marco Polo Fonds, the Stichting Dr H. Muller s Vaderlandsch Fonds, the Stichting Groninger Universiteitsfonds, the Faculty of Medical Sciences of the University of Groningen and the Department of Pediatrics of the University Hospital Groningen for their financial support and Dr Johannes Viljoen for his valuable advice on p24 antigen tests. Correspondence: P. J. J. Sauer, Department of Pediatrics, University Hospital, P.O. Box , 9700 RG Groningen, The Netherlands. Tel ; Fax <p.j.j.sauer@bkk.azg.nl>. TABLE 1 WHO clinical case definition for pediatric HIV infection HIV infection is suspected in an infant or child presenting with at least two of the following major signs associated with at least two of the following minor signs in the absence of known cases of immunosuppression, such as cancer or severe malnutrition or other recognized etiologies. Major signs weight loss or abnormally slow growth chronic diarrhea > 1 month prolonged fever > 1 month Minor signs generalized lymphadenopathy oropharyngeal candidiasis repeated common infections (otitis, pharyngitis, etc.) persistent cough generalized dermatitis confirmed maternal HIV infection To evaluate the WHO case definition, we compared the symptoms and signs, including those mentioned in the case definition, of pediatric patients Journal of Tropical Pediatrics, Vol. 49, No. 3 Oxford University Press 2003; all rights reserved 143

2 with and without HIV, in a state hospital in Bloemfontein, South Africa. The purpose of this study was to determine whether the WHO case definition is, in this setting, a useful diagnostic instrument to discriminate between HIV-positive and HIV-negative children. Materials and Methods From 13 May to 23 July 2000 we recorded data from 300 consecutively admitted children in Pelonomi Hospital, a state hospital in Bloemfontein. Pelonomi Hospital has a capacity of 720 beds and is the largest hospital in the health district Southern Freestate. All pediatric patients under 13 years of age admitted to the medical wards and the medical intensive care unit were included in the study. Those admitted to the surgical, oncological and neonatal wards, and the surgical intensive care unit were excluded. Approval for the study was obtained from the local Medical Ethical Committee. With the help of a checklist of over a hundred items we recorded data from every child included in the study. The data concerned the medical history, physical examination and outcome (length of stay or death of the child), and included the signs mentioned in the WHO case definition. Furthermore we obtained information on the child s social situation, including age and marital status of the mother, education level and employment of the parents, and living conditions (type of house, presence of water, electricity and telephone). We took the medical and social history from the person who accompanied the child to the hospital; this was usually the mother. From unaccompanied children we were able to obtain some of these data from the medical report. Six of the nine signs of the case definition had to be obtained from the history, three from physical examination. We asked the accompanying person about the maternal HIV status, but did not confirm it by laboratory testing as requested in the WHO case definition, since the case definition was developed to be used in countries where laboratory tests are not widely available. If parental consent was obtained, the child was tested for the presence of antibodies against HIV using both Vironostika HIV Uni-Form II Ag/Ab (Organon Teknika, Boxtel, Netherlands) and Enzygnost HIV Integral (Dade Beringh, Marburg, Germany). In children under 18 months of age and in children being breastfed, maternal antibodies against HIV can be found, which may cause an incorrectly positive result of the antibody tests. Therefore, if their antibody test results were positive, these children were also tested for the presence of the p24- antigen with Innotest HIV Antigen mab (Innogenetics, Ghent, Belgium). A positive p24-antigen test result can be used to diagnose HIV infection, since the specificity of this test approaches 100 per cent. However, since the sensitivity of the test is limited, a negative p24-antigen test result cannot be used to exclude HIV infection. 6 Laboratory tests with a high sensitivity, like those based on the principle of PCR, would have been more suitable but were too expensive and therefore not available for this study. We divided the 300 children into three groups: HIV negative (153), HIV positive (69), and HIV indeterminable (78). The HIV-negative group consisted of the children with two negative antibody test results (153). The HIV-positive group consisted of: (i) 16 children over 18 months of age and not being breastfed, with two positive antibody test results; (ii) 52 children under 18 months of age and/or being breastfed, with two positive antibody test results and a positive p24-antigen test result; and (iii) one child with two non-corresponding antibody test results and a positive p24-antigen test result. The HIV-indeterminable group consisted of: (i) children under 18 months of age and/or being breastfed, with two positive antibody test results and a negative (36 children) or indeterminable (one child) p24-antigen test result; (ii) two children with two non-corresponding antibody test results and a negative p24- antigen test result; and (iii) 39 children of whom we did not get any test results, mainly because the parents did not give consent or due to technical problems. The HIV-indeterminable group showed no marked differences in demographic or other data with the HIV-positive and the HIV-negative groups and was left out of further analysis. We applied the WHO case definition to the HIVpositive and the HIV-negative groups, and compared both groups on all data recorded. For this comparison we used, depending on the type of variable, the chi-squared test, the Mann Whitney U-test and Student s t-test. Statistical significance was defined as a p value less than We used logistic regression on the data involved in the composition of a simplified case definition to determine the contribution of each sign. We analysed the data with SPSS, version 9.0. Results Of the 222 children included in the final analysis 69 were HIV positive (31.1 per cent) and 153 HIV negative (68.9 per cent). The mean age in the HIV-positive group was 21.8 months and in the HIV negative-group, 26.6 months (p = 0.560). The HIVpositive group consisted of 37 male and 32 female patients, the HIV-negative group consisted of 74 male and 79 female patients. In 30 cases, 10 HIVpositive and 20-HIV negative children (13.5 per cent), we were unable to take a medical and social history, but could obtain part of the missing data from the medical report. We were able to apply the WHO case definition to 144 Journal of Tropical Pediatrics Vol. 49 June 2003

3 TABLE 2 Prevalence, p-value, sensitivity, specificity and positive predictive value of the complete WHO case definition and its separate signs in hospitalized children in Bloemfontein, South Africa HIV-positive HIV-negative p-value Sensitivity Specificity PPV children with sign children with sign n % n % WHO case definition 9/ / Weight loss or abnormally 31/ / slow growth Chronic diarrhea > 1 month 6/ / Prolonged fever > 1 month 9/ / Generalized lymphadenopathy 43/ / < Oropharyngeal candidiasis 28/ / < Repeated common infections 12/ / (otitis, pharyngitis, etc.) Persistent cough 19/ / < Generalized dermatitis 2/ / Confirmed maternal HIV infection 8/ / < of the 222 children (90.5 per cent). From the remaining 21 children we did not have the required data, since the history was missing and the data from the medical report were insufficient. Eleven children met the criteria of the WHO case definition; nine of them were HIV positive and two were HIV negative. The sensitivity of the case definition in this research population was 14.5 per cent, the specificity was 98.6 per cent, and the positive predictive value (PPV) 81.8 per cent. With the exception of weight loss and generalized dermatitis, the signs mentioned in the case definition were significantly more often seen in HIV-positive than in HIV-negative children (Table 2). Analysis of data obtained from the medical report, the medical and social history, and physical examination, showed the following (Table 3). HIV-positive children were admitted for longer periods, died more often during admission, and were more likely to have been previously admitted to hospital than their HIVnegative counterparts. HIV-positive children did not suffer more frequently from night sweats, vomiting, loss of appetite and convulsions than HIV-negative children. They did suffer more frequently from chronic dyspnea and abdominal pain. The social situation of HIV-positive children did not differ significantly from that of HIV-negative children (data not displayed in the Table). Physical examination showed that length, head circumference, and weight of HIV-positive children were more often below the third percentile on the growth charts. Likewise, marasmus defined by the Welcome Classification 7 as a weight below 60 per cent of the expected weightfor-age was more frequently seen in the HIVpositive than the HIV-negative group. Furthermore, finger clubbing, delayed developmental milestones (also after correction for prematurity), and hepatosplenomegaly were more often found in HIVpositive children. Hepatosplenomegaly proved a better predictive variable for HIV infection than hepatomegaly or splenomegaly alone (Table 3). The occurrence of respiratory distress did differ, but not significantly, between HIV-positive and HIVnegative children. Based on these findings we composed a new, simplified case definition for pediatric HIV infection (Table 4). For this new case definition we selected all signs observed significantly more often in the HIVpositive group. Data from the medical and social history were left out, since these data can be difficult to obtain and cannot be verified. After further analysis, one combination of four signs (marasmus, hepatosplenomegaly, oropharyngeal candidiasis, and generalized lymphadenopathy) proved the most useful for the new, simplified case definition. HIV infection is suspected in a child presenting with at least two of these four signs. The sensitivity of this simplified case definition was 63.2 per cent, the specificity was 96.0 per cent, and the PPV was 87.8 per cent. Discussion In this study the prevalence of HIV infection in hospitalized children was 31.1 per cent. Other South African studies, in KwaZulu Natal 4 and Soweto, 8 showed prevalences of 25.6 and 29.2 per cent, respectively. In this research population the WHO case definition for pediatric HIV infection was not a useful instrument to discriminate between HIV-positive and HIV-negative children. Its sensitivity of 14.5 per cent Journal of Tropical Pediatrics Vol. 49 June

4 TABLE 3 Prevalence, p-value, sensitivity, specificity and PPV of data concerning outcome, medical history and physical examination of hospitalized children in Bloemfontein, South Africa HIV-positive HIV-negative p-value Sensitivity Specificity PPV children with sign children with sign n % n % Outcome Length of stay (in days) (n = 57) 7.34 (n = 148) <0.001 Died during admission 12/ / < Previously admitted 31/ / < Medical history Night sweats 28/ / Vomiting 36/ / Loss of appetite 43/ / Convulsions 21/ / Chronic dyspnea 8/ / Abdominal pain 35/ / < Physical examination Length < p3 45/ / Head circumference < p3 29/ / Weight < p3 51/ / < Marasmus a 23/ / < Finger clubbing 10/ / < Delayed milestones 20/ / Delayed milestones after correction for prematurity 13/ / < Hepatosplenomegaly 28/ / < Hepatomegaly 48/ / < Splenomegaly 29/ / < Respiratory distress 40/ / a Marasmus is defined as a weight below 60 per cent of the expected weight-for-age. TABLE 4 Proposed simplified case definition for pediatric HIV infection HIV infection is suspected in a child presenting with at least two of the four following signs : marasmus hepatosplenomegaly oropharyngeal candidiasis generalized lymphadenopathy was strikingly low. In four other studies evaluating the WHO case definition, its sensitivity is higher (Congo 35 per cent, 2 Rwanda 47 per cent, 3 KwaZulu Natal 22 per cent, 4 and Ivory Coast 19 per cent 5 ), but is also limited. Furthermore, with its combination of nine items, divided into major and minor signs, the WHO case definition was unnecessarily difficult to use in a clinical setting. Six of the nine signs had to be obtained from the medical history and could not be objectively verified. In addition, not every child was accompanied by a person that knew the medical history of the child and even caretakers who did, were often unable to answer the questions on the duration of certain conditions (i.e. diarrhea, fever, common infections, and cough) and the maternal HIV status. In some situations a language barrier complicated taking the history. Apart from weight loss and generalized dermatitis, each sign mentioned in the WHO case definition was seen significantly more often in HIV-positive than in HIV-negative children. Compared to our results the occurrence in other studies of chronic diarrhea, prolonged fever, and generalized dermatitis is remarkably high. 2,3,9 On the other hand, generalized lymphadenopathy and oropharyngeal candidiasis are seen remarkably less frequently in these studies. The sign weight loss or abnormally slow growth is not clearly defined by the WHO. We regarded it as a subjective observation of a caretaker, that has to be obtained from the history, but it can be regarded as a more objective observation of a clinician as well. It is better to use a strict definition, for example a weight below the third percentile or below 60 per cent of 146 Journal of Tropical Pediatrics Vol. 49 June 2003

5 expected weight (marasmus) on a local growth chart. The minor sign confirmed maternal HIV infection does not belong in a case definition, that has been developed to be used in settings where laboratory tests are not available and where it is therefore not possible to confirm the maternal HIV status. In both this and the Soweto study 8 HIV-positive children were admitted longer than their HIVnegative counterparts, while in the KwaZulu Natal study 4 mean length of stay is similar in both groups. In all three studies HIV positive children had more often been admitted previously and died more frequently during admission. The Rwanda study 3 suggested a relation between socio-demographic data and HIV infection in children. We did not find a significant difference in the social situation of HIV-positive and HIV-negative children, but have to remark that this study was carried out in a state hospital with a homogenous mix of patients from the lower social classes. In this study marasmus was seen in approximately one-third of the HIV-positive children and much less in HIV-negative children (7.3 per cent). These findings are comparable to those of the other South-African studies. 4,8 Over 40 per cent of the HIV-positive children in this research population showed hepatosplenomegaly, much more than the HIV-positive children in the Nigeria study 9 (19 per cent). The difference in the occurrence of finger clubbing between the HIV-positive and HIV-negative children of the research population (14.5 vs. 0.7 per cent) was remarkable and suggests that HIV-positive children suffer often from conditions like chronic lung disease. Respiratory distress was seen in approximately half of the children in the research population and did not occur significantly more often in the HIV-positive than the HIV-negative group. In the Rwanda study the occurrence of respiratory distress is remarkably lower and does differ significantly between HIV-positive and HIV-negative children (26 vs. 5 per cent). 3 Because the WHO case definition did not prove useful in this research population we composed a new case definition. This new case definition consists of four equivalent items that can be obtained from simple physical examination. It does not require the medical history of a child, which means it is less likely that data are incorrect or missing. Compared with the WHO case definition, the new case definition is easier to use and less subjective. In this study it proved to be much more sensitive than the WHO case definition (63.2 vs per cent). However, even in the research population on which we based this case definition, one out of three HIV-positive children was missed. On the other hand, since the specificity (96.0 per cent) and positive predictive value (87.8 per cent) of the new case definition were high, children that did meet the criteria of this case definition were very likely to be HIV positive. We cannot assume that the children in this study are representative of children admitted to state hospitals in other developing countries, because of the differences in the occurrence of symptoms and signs between the children in this study and in studies in Congo, 2 Rwanda, 3 and Nigeria. 9 Whether the new, simplified case definition is useful outside Pelonomi Hospital remains to be tested. In developing countries appropriate diagnostic laboratory tests are often not available, especially for children under 18 months of age and/or those being breastfed. In the absence of diagnostic facilities, the clinical symptoms and signs are the only information one has to predict the HIV status of a child. In such a situation, the new clinical case definition might be a useful instrument to determine whether a child is likely to be HIV positive or not. However, a critical attitude and knowledge of the general clinical presentation of HIV infection in children are required when using the case definition. References 1. World Health Organization. WHO/CDC case definition for AIDS. Wkly Epidemiol Rec 1986; 61: Colebunders RI, Greenberg A, Nguyen-Dinh P, et al. Evaluation of a clinical case definition of AIDS in African children. AIDS 1987; 1: Lepage P, Perre P van de, Dabis F. Evaluation and simplification of the World Health Organization clinical case definition for paediatric AIDS. AIDS 1989; 3: Yeung S, Wilkinson D, Escott S, Gilks CF. Paediatric HIV infection in a rural South African district hospital. J Trop Pediatr 2000; 46: Vetter KM, Djomand G, Zadi F, et al. Clinical spectrum of human immunodeficiency virus infection disease in children in a West African city. Pediatr Infect Dis J 1996; 15: Fransen K, Mertens G, Stynen D, et al. Evaluation of a newly developed HIV antigen test. J Med Virol 1997; 53: Welcome Trust International Working Party. Classification of infantile malnutrition. Lancet 1970; ii: Meyers TM, Pettifor JM, Gray GE, Crewe-Brown H, Galpin JS. Pediatric admissions with human immunodeficiency virus infection at a regional hospital in Soweto, South Africa. J Trop Pediatr 2000; 46: Emodi IJ, Okafor GO. Clinical manifestations of HIV infection in children at Enugu, Nigeria. J Trop Pediatr 1998; 44: Journal of Tropical Pediatrics Vol. 49 June