Progressreportonthegrant``StatisticalMethodsfor.docVIP

Progress report on the grant ``Statistical Methods for Partially Controlled Studies; RO1 EY 014314-01 for 02/2003-02/2004 PI*: Constantine E. Frangakis, PhD Co-PI: Donald B. Rubin, PhD *Dr. Frangakis is on leave of absence in the Greek Navy, until April 04. Until then, Dr. Ronald S. Brookmeyer is the substitute PI. Description We are continuing good progress in meeting the objectives of the grant. The PI (Frangakis) and Co-PI (Rubin) have published or have in press with the support of the grant the work given in the reference list below. The contribution of the work is also briefly described below. Research by Frangakis et al. (2004) listed in the references develops new methodology for partially controlled studies that can be used to evaluate the effect that using needle exchange programs (NEP) has in reducing HIV incidence among drug users. In the application, using data from the Baltimore NEP, we use location of the NEP sites from drug users as the factor that was controlled by the study staff, and we use the actual exchange of needles and the measurement of HIV tests as the partially controlled factors. The research addresses two major methodologic problems in NEP evaluation. First, NEPs attract people who are at higher risk for HIV in measured factors, and thus, quite likely, in unmeasured factors, than those who do not use the NEPs. Second, as in other partially controlled studies with multiple factors, existing methods, including instrumental variables, is not appropriate to estimate effects of the NEP. The research develops a method that can estimate effects of using the NEP based on the framework of “principal stratification” (Frangakis and Rubin, Biometrics 2002, 21—29, see 2002-3 progress report), addressing also the longitudinal nature of the study, and the multiple levels of the distance of the NEP sites from the participants. Using the new method, we have found that utilization of the NEP can reduce HIV incidence among drug users by up to 90