HIV transmission through health care in sub-Saharan Africa

Abstract

Listen

Translate article

3 Gaist D, Jeppesen U, Andersen M, Garcia Rodriguez LA, Hallas J, Sindrup SH. Statins and risk of polyneuropathy: a casecontrol study. Neurology 2002; 58: 1333–37. 4 Novella SP, Inzucchi SE, Goldstein JM. The frequency of undiagnosed diabetes and impaired glucose tolerance in patients with idiopathic sensory neuropathy. Muscle Nerve 2001; 24: 1229–31. 5 McManis PG, Windebank AJ, Kiziltan M. Neuropathy associated with hyperlipidemia. Neurology 1994; 44: 2185–86. Rakai, Uganda, that more individuals HIV-positive at baseline had injections during follow-up than did individuals HIV-negative at baseline (44·5% vs 31·8%). From this result, one can estimate that reverse causality (individuals with HIV-related symptoms seeking injections) accounts for a rate ratio of 1·4 for prevalent HIV associated with injections. Even so, reverse causality only partly explains observed associations between prevalent HIV and injections (relative risks in 19 studies range from 1·16 to 2·96), and says nothing about observed associations between incident HIV and injections. Thoma and colleagues report that “other analyses” from Rakai show that injections are “not associated with HIV1 acquisition”. However, if these other analyses report data from the same Rakai cohort and observation interval as that of Thoma and colleagues, there are some curious inconsistencies. The proportion of HIV-negative individuals at baseline who report injections during follow-up is lower in the Lancet letter than the AIDS letter (31·8% vs 41·7%), and the analysis of injections as a risk for incident HIV in the AIDS letter seems to exclude 1500 individuals HIV-negative at baseline and with few injections during follow-up who are identified in the Lancet letter. If these letters report data from different cohorts or observations intervals, then even more information on injections remains unreported. At least some of the analyses seem to be based on an incomplete account of available data, which undermines the conclusions. Designing efficient programmes to prevent HIV infections requires good information about risks. Basing programmes on either underestimates or overestimates of risks in health-care settings weakens prevention.