Send to

Choose Destination
AIDS. 2017 Apr;31 Suppl 1:S31-S40. doi: 10.1097/QAD.0000000000001321.

All-cause mortality in HIV-positive adults starting combination antiretroviral therapy: correcting for loss to follow-up.

Author information

aInstitute of Social and Preventive Medicine (ISPM), University of Bern, Bern, Switzerland bCentre for Infectious Disease Epidemiology and Research (CIDER), University of Cape Town, Cape Town, South Africa cJohn Hopkins University, Baltimore, Maryland, USA dISPED, Centre INSERM U897-Epidemiologie-Biostatistique, Université Bordeaux, Bordeaux Cedex, France eKirby Institute, UNSW, Sydney, New South Wales, Australia fInstitute for Implementation Science in Population Health, City University of New York, New York, USA gDepartment of Epidemiology and Biostatistics, City University of New York School of Public Health, New York, New York, USA hDepartment of Biostatistics, Vanderbilt University, Nashville, Tennessee, USA iDepartment of Biostatistics, Indiana University School of Public Health, Indianapolis, Indiana, USA.



To estimate mortality in HIV-positive patients starting combination antiretroviral therapy (ART) and to discuss different approaches to calculating correction factors to account for loss to follow-up.


A total of 222 096 adult HIV-positive patients who started ART 2009-2014 in clinics participating in the International epidemiology Databases to Evaluate AIDS collaboration in 43 countries in sub-Saharan Africa, Asia Pacific, Latin America, and North America were included. To allow for underascertainment of deaths due to loss to follow-up, two correction factors (one for the period 0-6 months on ART and one for later periods) or 168 correction factors (combinations of two sexes, three time periods after ART initiation, four age groups, and seven CD4 groups) based on tracing patients lost in Kenya and data linkages in South Africa were applied. Corrected mortality rates were compared with a worst case scenario assuming all patients lost to follow-up had died.


Loss to follow-up differed between regions; rates were lowest in central Africa and highest in east Africa. Compared with using two correction factors (1.64 for the initial ART period and 2.19 for later), applying 168 correction factors (range 1.03-4.75) more often resulted in implausible mortality rates that exceeded the worst case scenario. Corrected mortality rates varied widely, ranging from 0.2 per 100 person-years to 54 per 100 person-years depending on region and covariates.


Implausible rates were less common with the simpler approach based on two correction factors. The corrected mortality rates will be useful to international agencies, national programmes, and modellers.

[Indexed for MEDLINE]
Free PMC Article

Publication types, MeSH terms, Substance, Grant support

Publication types

MeSH terms


Grant support

Supplemental Content

Full text links

Icon for Wolters Kluwer Icon for PubMed Central
Loading ...
Support Center