Background:  CD4 counts and viral loads are integral components of HIV disease monitoring, allowing decision-making regarding treatment initiation and changes. However, these technologies are costly and require lab equipment and expertise that is not available in many resource-poor settings. There is a paucity of data regarding the correlation between CD4 and total lymphocyte count (TLC) in HIV-infected children. However, based on limited data, the World Health Organization (WHO) has made recommendations regarding TLC cut-offs for initiation of ART for children. Using our database of children with HIV, we attempted to further evaluate the relationship between CD4 and TLC in order to inform current practice in HIV disease management in resource-limited settings.

Methods:  We retrospectively reviewed clinical and laboratory data of all HIV-infected children seen at The Hospital for Sick Children from 1995 to 2005. Using paired CD4/TLC data for each patient, we determined the degree of relatedness between these 2 measures, taking into account potential confounders such as age, ART, and race. We evaluated the reliability of the TLC cut-offs suggested in the WHO 2006 guidelines on pediatric ART in resource limited settings.

Results:  Data for 126 patients aged 2 weeks to 18 years were collected, including information from 3650 visits. There was no significant difference in TLC count between children of black race and children of other races. In regression analysis adjusted for age, race, and ART, TLC and total CD4 count were highly correlated (p <0.0001) as were change in TLC and change in CD4 count (p <0.0001). A mean change in TLC of 2.0 x10⁶ cells/L (SE 0.038) was observed for every 1 cell/mm³ change in CD4 count. In evaluating the cut-offs proposed by WHO, the TLC cut-off for each age group (<1, 1 to 3, 3 to 5, 5 to 8, and >8 years) predicted the appropriate CD4 cut-off in 80%, 88%, 83%, 84%, and 89% of the time, respectively.  

Conclusions:  TLC is a good predictor of CD4 count, though the cut-offs for treatment proposed by the WHO may be suboptimal.