The "START" (a Streamlined ART Initiation Strategy) Study

Overview

START is clinic-level (not individual-patient) intervention to catalyze the process of ART initiation among HIV-infected adults who meet CD4-based criteria for combination ART (e.g., adults with a CD4 T cell level \< 350/ul). The three START components are: (1) real-time point-of-care CD4 testing using the Alere PIMA platform to ascertain treatment eligibility in real time at first presentation to care, (2) targeted knowledge transfer (i.e., dissemination) of recent scientific evidence regarding effects rapid ART initiation on survival to front line health care workers; and 3) feedback and reporting to clinic and providers. There three components represent empirically validated steps in the PRECEED implementation model.

RATIONALE: In Africa, up to a 50-75% of HIV-infected, adults fail to initiate (FTI) ART or encounter delays of months even after eligibility through CD4+ T-cell testing or clinical criteria are established. Recent randomized trials including ACTG Protocol 5164, ACTG 5221, CAMELIA and SAPIT have demonstrated that delays of even weeks in ART initiation increases mortality among patients presenting with active opportunistic infections. Even among those without active infections, mortality among untreated persons with a CD4 \< 350 approaches 15/100 person-years - a large proportion of which should be avoidable in patients who have presented to care. New dissemination and implementation strategies that are generalizable across resource-limited settings are needed to address FTI and make the public health approach more effective. The overall goal of the multi-component START strategy is to initiate the greatest number of eligible patients in the shortest amount of time possible, while maintaining safety, efficacy and cost effectiveness. HYPOTHESIS: The START intervention, a combined intervention using provider education, novel technology and reinforcing feedback, will increase the rapidity and completeness of ART initiation. INTERVENTION: START is clinic-level (not individual-patient) intervention to catalyze the process of ART initiation among HIV-infected adults who meet CD4-based criteria for combination ART (e.g., adults with a CD4 T cell level \< 350/ul). The three START components are: (1) real-time point-of-care CD4 testing using the Alere PIMA platform to ascertain treatment eligibility in real time at first presentation to care, (2) targeted knowledge transfer (i.e., dissemination) of recent scientific evidence regarding effects rapid ART initiation on survival to front line health care workers; and 3) feedback and reporting to clinic and providers. There three components represent empirically validated steps in the PRECEED implementation model. STUDY DESIGN: A cluster-randomized, step-wedge trail of 20 clinics operated by the Makerere Joint AIDS Program. Four clinic sites will be randomly assigned to the START at 6-month intervals between month six and 24 during the three-year trial. PRIMARY OBJECTIVE 1\. Evaluate the programmatic change of START on the cumulative incidence of ART initiation 14 days after first clinical eligibility for ART in treatment eligible HIV-infected patients receiving care under routine program conditions in a large, multi-site, HIV care program in Uganda. SECONDARY OBJECTIVES 1. Evaluate the effect of START on the incidence of mortality in treatment-eligible, HIV-infected patients. 2. Evaluate the effect of START on plasma HIV RNA levels one year after treatment eligibility in HIV-infected patients presenting to care in a multi-site, HIV care program. 3. Evaluate the effect of START on the incidence of vertical transmission in all treatment-eligible, HIV-infected women. 4. Evaluate the effect of START on retention among ART-eligible, HIV-infected patients in the study sites. 5. To assess the cost-effectiveness of the START intervention. 6. To evaluate adaptation of START intervention into clinical care through qualitative analysis of counseling interactions and limited interviews with counselors and patients regarding their experience of counseling and identification of content of current counseling activities and how this changes over time. OUTCOMES: The primary outcome of the study is ART initiation. Secondary outcomes are all-cause mortality, plasma HIV RNA at one year after clinic presentation, vertical transmission events and cost-effectiveness measures. ANALYTIC PLAN: Primary analysis for primary outcome: We will treat the outcome as binary, in which we consider patients as either initiating ARV by 14 days or not. In other words, the outcome of ART start within 14 days or less is met if the start is by 14 days inclusive (\<=14 days), otherwise the outcome is not met. Patients with less than 14 days of observation time before cross-over or end of study database closure will not be analyzed. This means that even if a patient initiates ART but has less than 14 days of possible observation time, that event will not be counted. Also, deaths prior to ARV initiation are considered non-initiations. The analysis will use a mixed effect logistic regression model (meglm in Stata 14.0) with a Normal random effect for site and a fixed effect for intervention. Marginal proportions will under intervention and control be calculated using the "margins" command in Stata. We will represent the effect as a risk ratio of marginal proportions. Secondary sensitivity analyses for primary outcome 1. We will adjust the primary analysis for calendar time specified as a restricted cubic spline. 2. We will restrict analyses to the incidently eligible population. 3. We will perform an analysis including all observation time censoring those with less than 14 days of observation due to database closure or cross-over. 4. We will carry out an analysis in which we estimate the effect of crossing into the intervention condition among patients who are eligible but who had not started ART at the time clinics changed from control to intervention through introduction of a time-varying covariate representing cross over date in a multivariable regression model.