|| Checking for direct PDF access through Ovid
We investigated whether the onset of rapid declines in total lymphocyte counts and hemoglobin levels might be useful for staging HIV disease and initiating antiretroviral therapy. Using data from the Multicenter AIDS Cohort Study, we found that accelerated declines in both markers generally precede AIDS by 1.2 years and occur when the CD4 lymphocyte counts fall below 350 cells/mm3. These markers may thus be suitable for monitoring disease and timing therapy initiation in resource-limited settings.There is an emerging consensus that the HIV pandemic in the developing world requires treatment with antiretroviral drugs . However, because of the complexity, expense, and diversity of potentially effective treatments, there is not yet consensus on what treatments should be given and at what stage of HIV infection they should be initiated. In the developed world, guidelines for starting treatment are based on the CD4 lymphocyte counts and plasma HIV-RNA concentration , but these markers are considered impractical for widespread use in the developing world because of their expense and other factors.One marker that has been proposed for resource-limited settings is the total lymphocyte count (TLC). Current WHO guidelines recommend the use of TLC counts below 1200 cells/mm3 , although data supporting the use of this marker have been mixed [4–7]. One challenge to using TLC for predicting the disease stage is that it does not show a linear decline throughout HIV infection, but rather a period of stability followed by a more rapid decline preceding clinically defined AIDS . Furthermore, TLC can also be affected by a number of factors independent of disease progression.Therefore, using our understanding of the longitudinal patterns of TLC, we hypothesized that the rapid change in these markers could distinguish those with and without incipient AIDS. For this analysis, we used the extensive data from the Multicenter AIDS Cohort Study collected before the advent of potent antiretroviral therapy, which has supported the use of CD4 cell counts and HIV-RNA levels as markers of disease progression . In addition to TLC, we also evaluate hemoglobin, which has also been shown to correlate with the progression to AIDS under certain circumstances [10,11], declines around the time of AIDS , and is also relatively inexpensive and straightforward to measure.Of the 2770 HIV-infected men enrolled in the Multicenter AIDS Cohort Study, we restricted our analysis to 1455 men who provided marker data for at least four AIDS-free visits to October 1994 (before the use of protease inhibitors and the widespread use of stavudine and lamivudine), who either remained AIDS-free as of July 1997 or developed AIDS before October 1994, and contributed at least two visits within the 1.5 years preceding AIDS. Data obtained at the baseline visit of the seroprevalent men, and in the first year after seroconversion in the seroincident men, were excluded to avoid changes occurring around seroconversion. The median (interquartile range, IQR) number of marker measurements per individual was 12 (8–17).We modified methods previously applied to the retrospective identification of inflections in total T lymphocyte counts [13,14] and viral load measurements  to determine the occurrence and timing of the onset of rapid declines in TLC and hemoglobin. For each marker, we estimated the magnitude of decline that best distinguished men who did or did not develop AIDS over follow-up. To enhance the stability of our estimates, declines were calculated as the difference between the mean of the two most recent marker values from the mean of all previous marker values. The rates of decline that best (i.e.