Background:  The Kisumu Breastfeeding Study is a trial to evaluate safety, tolerance, and efficacy of HAART from 34 weeks’ gestation to 6 months post partum for prevention of mother-to-child transmission (PMTCT) of HIV among breastfeeding women. The initial regimen of nevirapine/zidovudine/lamivudine (NVP/ZDV/3TC—2 pills twice daily) was later revised: those with CD4 <250cells/mm³ got NVP/ZDV/3TC and those with CD4 ≥250 cells/mm³ got nelfinavir (NFV) and ZDV/3TC (6 pills twice daily). We evaluated the proportion of participants who completed either of the regimens initiated and reasons for switching or stopping.

Methods:  Participants, enrolled from July 2003 to November 2006, were seen weekly before delivery and 2, 6, and 14 weeks, and 5 and 6 months post partum. During study visits they were counseled on possible side effects, adherence, importance of following study requirements, and offered support in partner disclosure. Regular laboratory monitoring occurred. A database of drugs dispensed with reasons for change or discontinuation was maintained.

Results:  Overall, 522 participants started HAART, 82 (16%) stopped before 6 months, 440 (84%) continued to 6 months post partum, including 62 who switched to another regimen. Of 440, 263 (60%) started NVP/ZDV/3TC and 177(40%) NFV/ZDV/3TC. Of the 263 on NVP, 41 (16%) switched to NFV or efavirenz, 10% due to hepatotoxicity or rash and 6% due to initiation of TB therapy or warfarin; 21 of 440 (5%) switched from ZDV to stavudine because of anemia or neutropenia. None stopped NFV or 3TC. Overall, 402 (77%) had a CD4 ≥250 cells/mm³ of whom 194 (48%) received NVP and 208 (52%) received NFV. Of 28 regimen switches in this CD4 group, 22 (79%) were due to NVP, 6 (21%) to ZDV and none to NFV. Among the 82 who stopped HAART before 6 months, 32% stopped because of infant (n = 23) or maternal (n = 3) death (no deaths were attributed to HAART), 32% withdrew or relocated, 16% due to fear of disclosure, 11% due to early cessation of breastfeeding, 5% due to fear of side effects, and 4% due to pill burden. There was no difference in reasons for early drug cessation between those on NVP or NFV.

Conclusions:  Despite higher pill burden, the NFV-based regimen was better tolerated than the NVP-based regimen. Side effect rates due to NVP and ZDV are consistent with the literature. Some of those who stopped HAART early did so for reasons beyond direct control of study staff; however, many of the reasons were preventable and occurred despite counseling on adherence, possible side effects, and support in disclosure.