Improved plausibility bounds about the 2005 HIV and AIDS estimates

Area of Science:

• Epidemiology
• Public Health
• Biostatistics

Background:

• Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO) provide annual HIV estimates.
• Since 2003, plausibility bounds have been reported to quantify uncertainty in HIV estimates.
• Bounds guide understanding of plausible ranges for HIV incidence, prevalence, and mortality.

Purpose of the Study:

• To develop and report updated plausibility bounds for HIV estimates.
• To base bounds on individual country surveillance and survey data.
• To refine the assessment of uncertainty in HIV epidemic magnitude.

Main Methods:

• Plausibility bounds developed for generalized epidemics, low-level/concentrated epidemics (LLC), and regions.
• Techniques built upon previous methods but utilized country-specific data.
• Data sources included surveillance and survey data from individual countries.

Main Results:

• Uncertainty bounds vary based on surveillance system quality and data availability.
• Countries with seroprevalence surveys have the tightest bounds (e.g., adult HIV prevalence ARR -18% to +19%).
• Lower quality data and fewer surveys in LLC countries lead to wider bounds (e.g., ARR -53% to +183% for poor quality data).

Conclusions:

• Plausibility bounds are scientific tools, not formal statistical confidence intervals.
• Bounds are derived using reasonable statistical approaches and assumptions.
• Understanding uncertainty aids policymakers in making informed decisions for HIV epidemic control.