Comparison of Methods for Correction of Mortality Estimates for Loss to Follow-Up after ART Initiation: A Case of the Infectious Diseases Institute, Uganda

Abstract

BackgroundIn sub-Saharan Africa, a large proportion of HIV positive patients on antiretroviral therapy (ART) are lost to follow-up, some of whom are dead. The objective of this study was to validate methods used to correct mortality estimates for loss-to-follow-up using a cohort with complete death ascertainment.MethodsRoutinely collected data from HIV patients initiating first line antiretroviral therapy (ART) at the Infectious Diseases Institute (IDI) (Routine Cohort) was used. Three methods to estimate mortality after initiation were: 1) standard Kaplan-Meier estimation (uncorrected method) that uses passively observed data; 2) double-sampling methods by Frangakis and Rubin (F&R) where deaths obtained from patient tracing studies are given a higher weight than those passively ascertained; 3) Nomogram proposed by Egger et al. Corrected mortality estimates in the Routine Cohort, were compared with the estimates from the IDI research observational cohort (Research Cohort), which was used as the “gold-standard”.ResultsWe included 5,633 patients from the Routine Cohort and 559 from the Research Cohort. Uncorrected mortality estimates (95% confidence interval [1]) in the Routine Cohort at 1, 2 and 3 years were 5.5% (4.9%–6.3%), 6.6% (5.9%–7.5%) and 7.4% (6.5%–8.5%), respectively. The F&R corrected estimates at 1, 2 and 3 years were 11.2% (5.8%–21.2%), 15.8% (9.9%–24.8%) and 18.5% (12.3% –27.2%) respectively. The estimates obtained from the Research Cohort were 15.6% (12.8%–18.9%), 17.5% (14.6%–21.0%) and 19.0% (15.3%–21.9%) at 1, 2 and 3 years respectively. Using the nomogram method in the Routine Cohort, the corrected programme-level mortality estimate in year 1 was 11.9% (8.0%–15.7%).ConclusionMortality adjustments provided by the F&R and nomogram methods are adequate and should be employed to correct mortality for loss-to-follow-up in large HIV care centres in Sub-Saharan Africa.

Outcomes of HIV‐positive patients lost to follow‐up in African treatment programmes

“…We excluded 1589 items for the reasons detailed in Figure . Thirty articles met inclusion criteria . Two studies reported results for children and adults separately.…”

“…Our regression analyses can inform such assumptions, taking into account the proportion of patients LTFU in the treatment programme and calendar year. A nomogram or more sophisticated methods can then be used to obtain estimates of overall mortality that are corrected for LTFU .…”

Growth and Mortality Outcomes for Different Antiretroviral Therapy Initiation Criteria in Children aged 1-5 Years

“…While censoring by drop out can be handled using g-computation, and we conducted sensitivity analyses using multiple imputation, it may be possible that those defined to be lost have a particular high risk of being dead. In some settings, this is known to be true for adults; 27–29 however, in children less is known about those lost to follow-up because reasons for missing an appointment may relate to the caregiver’s work responsibilities, family situations, economic opportunities, and own health status 30 . Studies that trace lost children or link them to vital registries are needed to gain more knowledge about this group.…”

Impact of and Correction for Outcome Misclassification in Cumulative Incidence Estimation

“…Our estimator can also be used when misclassification probabilities are not known in advance such as, for example, in sensitivity analyses, when ranges of plausible misclassification rates can be considered. In the problem of death under-reporting in HIV care and treatment programs in low or middle-income settings, a number of papers [ 14 , 15 , 24 , 28 ] can provide plausible estimates of death under-reporting in similar settings. In the study by Hinchliffe and colleagues [ 4 ] on the other hand, consideration of various levels of misclassification showed that cancer mortality estimation is more sensitive to over-recording of cancer-related causes of death.…”