RESEARCH IN PROGRESS 629

that non-invasively detects the presence of schistosome circulating cathodic antigen (CAA) in urine. The field microscope had a sensitivity of 85% and a specificity of 96%, using standard faecal smears, in detecting & mansoni. The CCA test had a sensitivity of 86% and specificity of 87%, although this is likely to be an underestimate. The principal defects were a lack of sensitivity at low intensities of infection in the field microscope and the prohibitive cost of the CCA test. Despite these deficiencies, it is considered that both tools are able to assess the prevalence of schistosomiasis with sufficient accuracy to implement the appropriate mass control programme as defined by the WHO and therefore have a future as a tools for diagnosis of schistosomiasis under the appropriate circumstances.

Burkit t ' s l y m p h o m a in C a m e r o o n

Peter M c C o r m i c k and Franc ine T c h i n t s e m e Chil- dren's Department, Banso Baptist Hospital, North West Province, Cameroon

The profound effect of monotherapy using cyclo- phosphamide for Burkitt's lymphoma (BL) has been recognized since original observations by Denis Burkitt et al. (1965: Cancer, 18, 399-410). The need for appro- priate, affordable, effective and equitable treatment for BL in the resource-poor setting of its area of endemicity is now equally well recognized (Agenda of 1st SIOP Meeting, 1994: personal communication). We report the use of monotherapy with cyclophosphamide in high dose at frequent intervals, for the first 26 clinically suspected, Murphy-staged cases of BL in a pilot trial at Banso Baptist Hospital, Kumbo, Cameroon between February and September 2003. There was no control group, since no alternative therapy was available and placebo/decision not-to-treat would have been unethi- cal. Institutional Review Board approval was granted; informed consent was obtained from all carers. Treat- ment success was recorded for total or near total disappearance of the lesion(s) at day 15 (third chemo- therapy). Otherwise treatment failure was recorded. Of 26 clinically suspected cases, the specimens of 2 were lost, and 2 subsequently proved negative on cyto- logical analysis. Because of the certainty in clinical diagnosis we calculated treatment failure and success rates based on all 26 cases. Treatment failures with proven cytology were 7 (26.9%). Treatment successes (potential cures) at 7 months were 19 (73.1%). The case numbers are small and further follow-up is neces- sary but these preliminary findings are encouraging.

The aet io logy o f n o n - r e s p o n s i v e paediatr ic pneu- m o n i a in Du rb an , South Africa

L. M. McNal ly l'z, P. M. Jeena 2, S. A. Thula 2, K. Gajee 2, A. W. S t u r n l 2, S. Cassol 2, A. S m i t h 29 H. M. Coovadia 2, D . Goldblatt ~ and A. M. T o m k i n s ~ 1Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK; 2Nelson R Mandela School of Medicine, University of Natal, Durban, South Africa

HIV-infected children hav~ increased morbidity and mortality from pneumonia but the organisms responsi- ble have not been fully defined. The Durban Paediatric Pneumonia Pathogen Study is the first to determine the response rates of very severe pneumonia in an HIV- prevalent area. It also aimed to determine the aetiology in non-responders in order to aid the choice of empiric second-line therapy. Children aged 1-59 months with (very) severe pneumonia were included. Respiratory history and examination were recorded. Unlinked anonymous HIV testing was performed on all children. Non-responders were investigated further by non- bronchoscopic bronchoalveolar lavage or a lung/pleural aspirate. A total of 362 children was recruited and the

majority (70%) was aged < 1 year. Fifteen percent died. Sixty-one percent responded to WHO therapy. Seventy percent of children were HIV-infected (78% of non-responders). A probable pathogen was isolated in 85% of the children who had further investigations. The most common pathogens were Pneumocystisjiroveci pneumonia, cytomegalovirus, tuberculosis, and gram- negative bacteria. Multiple pathogens were present in 61%. The majority of children in an HIV-endemic area still respond to standard WHO therapy. However, non- responders are more likely to have HIV. The majority of non-responders had more than one organism iso- lated.

[This oral presentation was awarded first prize at the meeting.]

Quant i fy ing d ivers ion o f m a l a r i a - v e c t o r m o s q u i - toes f r o m repe l l ent l t rea ted 'bait' to unprotec ted 'col lectors ' in f ield tests in the Bol iv ian A m a z o n

S. J. Moore , C. R. Dav ie s and M. M. C a m e r o n London School of Hygiene and Tropical Medicine, Keppel Street, London WCIE 7HT, UK

Insect repellents provide useful protection from malaria for high-risk groups including forest-workers, especially in the Amazon where vectors are difficult to control through conventional means due to their exophily, early biting, and forest-breeding. The WHO recommends that repellents are field-tested using human-landing catches, with individual collectors po- sitioned > 10 m apart. However, it is still common- place to perform tests using repellent-treated 'bait ' and an unprotected 'collector'. This could overesti- mate repellent protection and longevity since mosqui- toes will feed preferentially from the unprotected 'collector', thus potentially exposing subsequent users to disease. Mosquito diversion from 15% DEET (repellent) to 15% baby oil (control) was quantified utilizing pairs of testers with each team situated > 10 m apart. Permutations of repellent repellent; control control; and repellent control were replicated over 18 nights with 6 pairs. Control individuals received 22.82% more bites when paired with repellent wear- ers than untreated individuals (likelihood ratio test, Z z, P - 0 . 0 4 8 5 ) , and repellent wearers received 11% more bites when paired with other repellent wearers than controls (not significant). In a further test with 25% lemongrass (Cymbopogon citratus) repellent, con- trols received 23.53% more bites when paired with repellent wearers, than when with unprotected part- ners (not significant). Thus it is advisable to field-test repellents using individuals and not pairs of collec- tors.

Rehabi l i tat ion o u t c o m e s in people wi th m o v i n g diff iculty d ischarged f r o m hospita l in L i m p o p o Prov ince , South Africa

Michae l M. S. M w a m b o l a Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK

The study assessed the rehabilitation outcomes in people with moving difficulty discharged from hospital in Limpopo Province, South Africa, and aimed to com- pare the outcome between people referred to commun- ity rehabilitation workers (CRW) and those discharged to their homes in villages where there are no CRWs. The study employed both quantitative and qualitative methods to describe demographic characteristics and disabilities, and to explore the participants' percep- tions. The study recruited 46 adults with difficult mov- ing who were discharged from hospital 1-3 years before the study started. There were no striking differ- ences in types and causes of disability between the 2 settings. At hospital discharge the villages with CRWs