Training healthcare workers on the use of electronic medical records in HIV clinics in Kenya:An evaluation of three training models The study is a quantitative program evaluation of the three training models comparing cost, geographic coverage, and quality of training, as measured by pre-and post-tests. Paired ttests were conducted to examine the changes in score from pre-test to post-test within training periods, and multiple linear regression was used to examine the associations between mean posttest scores variable by the training model and adjusted for pre-test score, age, sex, province, and cadre. The three-day, on-site model was the least expensive and had the largest geographic range, and quality of training did not differ substantially by the training model. Shorter trainings were associated with learning loss in computer literacy. The trainings were particularly effective for nurses, who had the highest mean change scores (p<0.001) in comparison to other cadres.