The WHO recently revised criteria for initiating antiretroviral therapy (ART) among adults in low-resource settings . The new guidelines recommend ART for adults with clinical stage III or IV (regardless of CD4 cell count) or with CD4 cell count below 350 cells/μl (regardless of clinical stage) . These guidelines simplify and liberalize prior guidelines that required stage III if CD4 cell count was 200–350 cells/μl. Programs to prevent mother-to-child HIV transmission (MTCT) are now one of the major sites in which HIV-infected adults are identified and treated. This population is of high priority for ART, as treatment is life saving for women, a major factor protecting the health of their children, and ART has a direct role in reducing HIV transmission to the child .
Here, we describe the potential impact of these guideline revisions using data collected from a trial in Lusaka, Zambia, conducted prior to the widespread availability of ART . As part of this trial, 1025 HIV-infected women were recruited during pregnancy and followed after delivery for 24 months. At enrollment, CD4 cell counts and plasma viral load were measured and clinical stage determined using a standardized history. No woman with stage IV disease was identified. PCR tests were performed on child blood samples collected at birth, 1 week, 1 month, monthly up to 6 months and 3-monthly up to 24 months. Children were breastfed to 4 months, thereafter practices were heterogeneous: some women weaned as per random assignment and others continued to breastfeed . The median duration of breastfeeding in the whole cohort was 12 months.
We examined associations between maternal characteristics measured during pregnancy used to define eligibility for ART, including CD4 cell count, viral load, clinical stage and old and new WHO criteria. We examined the capacity of these characteristics to predict: mortality among women between delivery and 24 months; perinatal transmission, defined as HIV infection in the child detected before 6 weeks of age; and postnatal transmission, defined as infection detected only after 6 weeks. Kaplan–Meier methods were used to calculate the risk of these outcomes, Cox proportional hazards models to determine relative risks (RRs), and, among those with the endpoint, the proportion of patients who would have met criteria for ART, were calculated to estimate the preventive fraction. The latter parameter can be used to infer the highest possible percentage of endpoint that could be averted if ARTs were fully effective. ART only became available towards the end of the study and all person-time was censored once ART was initiated.
In our cohort, 54% of women had CD4 cell count below 350 cells/μl but most deaths (88%) occurred in this group (Table 1). Sixty-eight percent of women met the new WHO treatment criteria, which include clinical stage III in addition to CD4 cell count below 350 cells/μl, and 92% of deaths occurred in this group. Applying the new criteria or using only CD4 cell count below 350 cell/μl would result in treating 10.1 and 8.4 women, respectively, per death averted. A viral load of 48 428 copies/ml classified the same proportion of women in need of treatment as a CD4 cell count below 350 cells/μl but identified a smaller proportion of deaths (76%) consistent with prior data . Adding viral load to CD4 cell count resulted in performance similar to the new WHO criteria. Adding viral load to the new WHO criteria would identify 96% of deaths but would require treating 76% of women. To account for hemodilution during pregnancy, some have suggested that lower CD4 thresholds should be used in pregnant women to avoid overtreatment. Our data suggest that making this adjustment is unwise as CD4 cell count thresholds of 200, 250 and 300 cells/μl would identify only 59, 72 and 79% of deaths among women, respectively.
The old WHO criteria and clinical stage perform poorly to detect perinatal and postnatal HIV transmission. The new criteria, however, would detect 88% of perinatal and postnatal HIV infections. CD4 cell count below 350 cells/μl alone would detect almost the same number of postnatal infections (83%) but a lower percentage (76%) of perinatal infections (Table 1). Both viral load and CD4 cell count are independent predictors of transmission. For perinatal transmission, viral load [RR = 3.1, 95% confidence interval (CI) 2.0–4.6] is the stronger predictor in multivariate analysis controlling for CD4 cell count (RR = 2.0, 95% CI 1.3–3.0). For postnatal transmission, their contributions are similar: viral load (RR = 3.8, 95% CI 2.2–6.3) and CD4 cell count (RR = 3.8, 95% CI 2.1–6.8). Combining viral load and CD4 cell count as either/or criteria for initiating therapy would lead to better results than the new WHO criteria while treating slightly fewer women.
If we adjust for the expected efficacy of ART to reduce perinatal and postnatal transmission , applying the new criteria among pregnant women could prevent 82% of all infections even if no extended postnatal interventions are applied among the other women. It is desirable to implement these extended interventions to reduce transmission among women not needing ART  but these interventions will be more costly per infection averted, as the transmission rate in this subset is lower.
Clinical staging is useful in settings in which laboratory testing cannot be done, is unreliable or delayed but its value when laboratory results are available is limited. Our data indicate that its inclusion increases the number of women treated with only marginal increase in coverage of women and infants at risk.
Our estimates of the proportion of deaths and infections averted are directly proportional to the proportion of the population meeting each criterion. The distribution of the severity of disease is likely to vary across settings and over time depending on stage of the epidemic, referral patterns and service coverage and so on. Other studies have reported less advanced disease than our cohort  but the distribution of CD4 cell counts and clinical stage that we observed is similar to a large multisite aggregation of data on pregnant women participating in the MTCT-Plus program in eight African countries .
Our data provide evidence-based support for the treatment thresholds in the revised WHO treatment guidelines. Our analysis also provides estimates of the large positive impact these guidelines could have if widely implemented on reducing mortality among women and HIV transmission to children.
The study was supported in part by grants from the National Institute of Child Health and Human Development, National Institutes of Health (R01 HD 39611 and R01 HD 40777). G.M.A. is a recipient of the Elizabeth Glaser Pediatric AIDS Foundation Scientist Award.
2. Becquet R, Ekouevi DK, Arrive E, Stringer JS, Meda N, Chaix ML, et al
. Universal antiretroviral therapy for pregnant and breast-feeding HIV-1-infected women: towards the elimination of mother-to-child transmission of HIV-1 in resource-limited settings. Clin Infect Dis 2009; 49:1936–1945.
3. Kuhn L, Aldrovandi GM, Sinkala M, Kankasa C, Semrau K, Mwiya M, et al
. Effects of early, abrupt cessation of breastfeeding on HIV-free survival of children in Zambia. N Engl J Med 2008; 359:130–141.
4. Brown ER, Otieno P, Mbori-Ngacha DA, Farquhar C, Obimbo EM, Nduati R, et al
. Comparison of CD4 cell count, viral load, and other markers for the prediction of mortality among HIV-1-infected Kenyan pregnant women. J Infect Dis 2009; 199:1292–1300.
5. Taha TE, Kumwenda J, Cole SR, Hoover DR, Kafulafula G, Fowler MG, et al
. Postnatal HIV-1 transmission after cessation of infant extended antiretroviral prophylaxis and effect of maternal highly active antiretroviral therapy. J Infect Dis 2009; 200:1490–1497.
6. Dugan K, Carter R, Toro P, El-Sadr WM, Abrams EJ. Comparing clinical and immunologic criteria for determining HAART eligibility among HIV-infected women in low resource settings
. First International Workshop on HIV Pediatrics
; 17 and 18 July 2009; Cape Town, South Africa; 2009.