• Rate of development of TB (microbiologically confirmed TB or highly probable TB) during the 36 month risk period [ Time Frame: Patients are assessed for TB one two monthly at each ART re-fill appointment ] [ Designated as safety issue: Yes ]
  

• Rate of drug toxicity (specifically, peripheral neuropathy, hepatitis +/-raised ALT grade III or worse and allergic rashes grade III or worse [ Time Frame: during the intervention period (ALT determined at baseline, 1, 2 and 3 months and then 3-monthly. the last safety determination is at 12 months post initiation of the study drug) ] [ Designated as safety issue: Yes ]
  
• Proportions adhering to study drug and HAART at the end of each study year as measured by pharmacy refills [ Time Frame: 1 month to two monthly, depending on the individual patient's clinic appointment ] [ Designated as safety issue: Yes ]
  
• Rate of development of INH monoresistance during the 36 month risk period. [ Time Frame: 36 months ] [ Designated as safety issue: Yes ]
  
• Death [ Time Frame: 36 months ] [ Designated as safety issue: Yes ]
  
• Worsening ART outcomes (virological and immunological failure) [ Time Frame: CD4+count and viral load are assessed as per clinic protocol (6 monthly post ART initiation) ] [ Designated as safety issue: Yes ]
  

The incidence of Tuberculosis (TB) in poor settlements around Cape Town continues to rise despite highly-active-anti-retroviral therapy (HAART) roll-out and DOTS. In Khayelitsha district, where this project will be conducted, TB incidence is about 1600/100000. There is an equally high HIV prevalence, currently 33%. Over 50% of adults presenting with active TB are co-infected with HIV and a third of all patients starting HAART have active TB. Although HAART has been shown to reduce the overall risk of TB by 59-80%, this risk still far exceeds the general risk. In the Khayelitsha HAART cohort, the risk of developing TB whilst on HAART is ~12 per 100 p-y. In the nearby community of Gugulethu, there is a 14% risk of active TB with at least half of the cases occurring within the first 3months on HAART. In a region where RD1-detected prevalence of latent TB infection is at least 80%, there is a real concern that TB will likely undo the benefit of HAART in the long run. Additional measures are therefore required to reduce the risk of TB in those already receiving or starting HAART. Isoniazid preventive therapy (IPT) represents an option but there is insufficient evidence to determine whether IPT can further (and safely) reduce the risk of TB in the HAART era. In a RCT, we propose to evaluate whether IPT can reduce the risk of active TB in patients receiving HAART.

A total minimum sample size of 1204 is required for the study to detect a 35% reduction in the hazard rates for tuberculosis in the intervention group (h1= 0.052) compared to the control group (h0=0.085) at a power of 80% and a Type II error of 0.05. Our maximum targeted sample size when losses to follow-up and subgroup analyses are considered is 1445. Development of TB will be the primary endpoint.