Estimating the number of people in a country or region with HIV who are undiagnosed and in need of ART
Polly Clayden, HIV i-Base
Estimating the number of people in a country or region with undiagnosed HIV in need of ART is essential for testing and treatment programmes.
Rebecca Lodwick and colleagues from the University College London showed a simple method for making this estimate, which uses data on simultaneous HIV/AIDS diagnoses at low CD4 count.
The HIV surveillance data needed for this method are the number of previously undiagnosed people presenting with AIDS (simultaneous HIV/AIDS) in a year with CD4 count at diagnosis. The number of people with simultaneous HIV/AIDS in a CD4 stratum represents a proportion of the total undiagnosed people with CD4 count in that stratum. The proportion is equivalent to the annual incidence of AIDS in people in that stratum, which can be estimated from cohort studies.
For each stratum the number of people with undiagnosed HIV is estimated by dividing the number of people with simultaneous HIV/AIDS diagnoses (with CD4 count in the stratum) by the CD4 specific AIDS rate.
The investigators obtained an uncertainty range associated with this estimate by assuming the AIDS rate varies according to a normal distribution and the observed number of diagnoses according to a Poisson distribution. They accounted for these two sources of uncertainty simultaneously over 10,000 runs.
They illustrated this method for people with undiagnosed HIV and CD4 count below 200 cells/mm3. Using data from CASCADE the incidence of AIDS in this CD4 stratum among ART naive patients has been estimated to be 0.25 per year (95% CI, 0.21-0.28). They used an example that supposed during the past year there were 50 simultaneous HIV/AIDS diagnoses with CD4 counts below 200 cells/mm3. Then the estimated number of undiagnosed people in this CD4 stratum would be 50/0.25=200. The estimated 95% uncertainty range would be 144-268.
They acknowledged that a potential source of bias is the possible under-diagnosis and under-reporting of simultaneous HIV/AIDS diagnoses.
They also note that the method depends on high levels of ascertainment of people presenting with simultaneous HIV/AIDS and on availability of CD4 count at diagnosis.
The investigators intend to make a SAS programme available that will perform this calculation.
Lodwick RK et al. A method to estimate the number of people in a country or region with HIV who are undiagnosed and in need of ART. 10th International Congress on Drug Therapy in HIV Infection, November 7-11. Glasgow. Poster abstract P165. Published in Journal of the International AIDS Society 2010 13(Suppl 4):P165. http://www.jiasociety.org/content/13/S4/P165