Share this article on:

CD4 Natural History and Informative Censoring in Sub-Saharan Africa

Duvignac, Julien; Thiébaut, Rodolphe

JAIDS Journal of Acquired Immune Deficiency Syndromes: November 1st, 2006 - Volume 43 - Issue 3 - p 380-381
doi: 10.1097/
Letters to the Editor

INSERM U593 Epidemiology, Université Victor Segalen Bordeaux 2, Bordeaux, France

To the Editor:

In most Sub-Saharan African settings, the challenge to standardize HIV-decision criteria implies a better understanding of the CD4 count evolution.

In their cohort study of HIV-infected untreated adults living in Cape Town, South Africa, Holmes and colleagues reported a steeper decline of CD4 count with higher baseline CD4 count.1 The CD4 count annual decline was estimated at 47, 31, and 20 cells/mm3 in patients with >500, 351-500, and 50-200 CD4/mm3 at baseline, respectively. These estimates could be biased by informative dropout. The authors stated that they did not find evidence of informative dropout, although concerns could be raised on the method used to check its existence.

Among the 957 patients who had at least two CD4 cell count measurements and who were therefore included in their analyses, 50% were lost-to-follow up and 31% died. Both loss to follow-up (depending on the reason for loss-to-follow-up) and death can lead to informative censoring, as they may be associated with the marker value at the time of the event.2,3 In their discussion of this potential bias, Holmes and colleagues argue that they did not find any association between the time to dropout and the CD4 cell count, using a Cox proportional Hazards model with time to dropout as the outcome and CD4 count as a time dependent variable. However, a dropout process can be classified as informative-or non-ignorable-only if it is associated with the unobserved value of the marker at the time of the event. Analyzing the association between time to dropout and observed values of CD4 count during follow-up is thus not a valuable method to diagnose an informative censoring issue. This would more likely diagnose a missing-at-random (MAR) dropout process that depends on the observed trajectory of the marker.4

A method that ignores informative dropout is likely to lead to overoptimistic statements about the marker trends because subjects with worse CD4 count evolutions tend to have shorter follow-up times and hence are weighted less in the estimations of the group rate of the average marker decline.3 In their CD4 study, Holmes and colleagues' estimates for the CD4 decline are thus likely to be underestimated.

Julien Duvignac, MSc

Rodolphe Thiébaut, MD, PhD

INSERM E0338 Biostatistics, Université Victor Segalen Bordeaux 2, Bordeaux, France

Back to Top | Article Outline


1. Holmes CB, Wood R, Badri M, et al. CD4 decline and incidence of opportunistic infections in Cape Town, South Africa: implications for prophylaxis and treatment. J Acquir Immune Defic Syndr. 2006;42:464-469.
2. Thiebaut R, Jacqmin-Gadda H, Babiker A, et al. Joint modelling of bivariate longitudinal data with informative dropout and left-censoring, with application to the evolution of CD4+ cell count and HIV RNA viral load in response to treatment of HIV infection. Stat Med. 2005;24:65-82.
3. Touloumi G, Pocock SJ, Babiker AG, et al. Estimation and comparison of rates of change in longitudinal studies with informative drop-outs. Stat Med. 1999;18:1215-1233.
4. Little JW, Rubin DB. Statistical Analysis With Missing Data. New York, NY: John Wiley and Sons; 1987.
© 2006 Lippincott Williams & Wilkins, Inc.