Bayesian melding for estimating uncertainty in national HIV prevalence estimates
2008

Estimating Uncertainty in National HIV Prevalence Estimates

Sample size: 200000 publication Evidence: moderate

Author Information

Author(s): Leontine Alkema, Adrian E. Raftery, Tim Brown

Primary Institution: University of Washington

Hypothesis

To construct confidence intervals for HIV prevalence in countries with generalized epidemics.

Conclusion

The Bayesian melding approach provides a way to assess uncertainty in HIV prevalence estimates and produces formal statistical confidence intervals.

Supporting Evidence

  • Adult HIV prevalence in the Caribbean is estimated at 1.0% (95% CI 0.9% to 1.2%) in 2007.
  • Based on antenatal clinic data, the posterior median of prevalence is 5.3% for 2007 (95% CI 4.3% to 6.7%).
  • Median prevalence in Namibia is estimated at 16.9% (95% CI 13.2% to 22.1%) for 2007.

Takeaway

This study helps us understand how many people might have HIV in different countries by using a special method that combines data and expert opinions.

Methodology

The study used Bayesian melding to derive annual 95% confidence intervals for HIV prevalence based on antenatal clinic data and population-based surveys.

Potential Biases

There may be biases in antenatal clinic data compared to population estimates.

Limitations

The estimates are illustrative and may not include all available information.

Participant Demographics

The study focused on urban areas in Haiti and Namibia.

Statistical Information

Confidence Interval

95% CI

Digital Object Identifier (DOI)

10.1136/sti.2008.029991

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication