Article Text

PDF
Improved plausibility bounds about the 2005 HIV and AIDS estimates
  1. M Morgan1,
  2. N Walker2,
  3. E Gouws3,
  4. K A Stanecki3,
  5. J Stover4
  1. 1Centers for Disease control and Prevention, Atlanta, GA, USA
  2. 2UNICEF, New York, NY, USA
  3. 3Joint United Nations Programme on HIV/AIDS, Geneva, Switzerland
  4. 4Futures Group/Constella, Glastonbury, CT USA
  1. Correspondence to:
 Dr M Morgan
 Centers for Disease Control and Prevention, Atlanta, GA, USA; wmm1{at}cdc.gov

Abstract

Background: Since 1998 the Joint United Nations Programme on HIV/AIDS and the World Health Organization has provided estimates on the magnitude of the HIV epidemic for individual countries. Starting with the 2003 estimates, plausibility bounds about the estimates were also reported. The bounds are intended to serve as a guide as to what reasonable or plausible ranges are for the uncertainty in HIV incidence, prevalence, and mortality.

Methods: Plausibility bounds were developed for three situations: for countries with generalised epidemics, for countries with low level or concentrated epidemics (LLC), and for regions. The techniques used build on those developed for the previous reporting round. However the current bounds are based on the available surveillance and survey data from each individual country rather than on data from a few prototypical countries.

Results: The uncertainty around the HIV estimates depends on the quality of the surveillance system in the country. Countries with population based HIV seroprevalence surveys have the tightest plausibility bounds (average relative range about the adult HIV prevalence (ARR) of −18% to +19%.) Generalised epidemic countries without a survey have the next tightest ranges (average ARR of −46% to +59%). Those LLC countries which have conducted multiple surveys over time for HIV among the populations most at risk have the bounds similar to those in generalised epidemic countries (ARR −40% to +67%). As the number and quality of the studies in LLC countries goes down, the plausibility bounds increase (ARR of −38% to +102% for countries with medium quality data and ARR of −53% to +183% for countries with poor quality data). The plausibility bounds for regions directly reflect the bounds for the countries in those regions.

Conclusions: Although scientific, the plausibility bounds do not represent and should not be interpreted as formal statistical confidence intervals. However in order to make the bounds as meaningful as possible the authors have tried to apply reasonable statistical approaches and assumptions to their derivation. An understanding of the uncertainty in the HIV estimates may help policy makers take better informed decisions to address the epidemic in their respective countries.

  • ANC, antenatal clinic
  • ARR, average relative range
  • ART, antiretroviral treatment
  • EPP, Estimation and Projection Package
  • LLC, low level or concentrated epidemics
  • MSM, men who have sex with men
  • UNAIDS, Joint United Nations Programme on HIV/AIDS
  • HIV
  • estimates
  • plausibility
  • uncertainty

Statistics from Altmetric.com

Footnotes

  • The findings and conclusions in this report are those of the authors and do not necessarily represent the views of their respective organisations.

  • Edited by Peter Ghys, Neff Walker, Helen Ward and Rob Miller

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.