CD4 cell counts of 200 x 10(6)/1 or below in natural history studies and surveillance: is one enough or are two better?
Easterbrook PJ., Yu LM., McLean K., Hawkins D., Gazzard B.
A retrospective cohort study was performed to examine the extent and clinical significance of misclassification associated with using the current United States AIDS case defining category of an initial CD4 count < or = 200 cells x 10(6)/l (< or = 200) compared with a definition requiring two consecutive counts below this level. The main outcomes examined were the probability of subsequent CD4 counts > 200 x 10(6)/l (> 200) and progression times to AIDS and death. Of the 2025 predominantly male homosexual HIV-positive patients attending two hospital based HIV clinics with initial CD4 cell counts < or = 200, 1524 (75%) subsequently had consecutive counts < or = 200, but only half did so at the next CD4 count. Ten per cent had either no further or only non-consecutive counts < or = 200, and 15% had only one CD4 count available for analysis. The cumulative proportion of patients with a CD4 count > 200 at one year after a first count of < or = 200 was about twice (39%) that observed among the subgroup with at least two consecutive counts < or = 200 (19%). The times from the initial counts of < or = 200 to AIDS and death were also shorter by six months and two months, respectively, in the subgroup with two or more consecutive counts < or = 200. A significant proportion of patients will be prematurely classified as having a CD4 cell count < or = 200 if a single CD4 count below this level is accepted. A definition of two consecutive counts < or = 200 should be adopted in preference to a single count < or = 200 for natural history studies and clinical trials, in which precise estimates of the time to or from a defined CD4 threshold are important. In surveillance programmes, however, such an approach may be impractical because of missing or infrequent serial CD4 counts, although adjustments can be made based on these estimates of premature misclassification.