The purpose of this study is to examine potential implications of changes in the approach to adult growth hormone (GH) replacement (GHR) over the last 15 years. Therefore, we analysed the German KIMS database as one of the largest single country pharmacoepidemiological databases on adult GH deficiency (GHD). Based on the date of their first GH application patients were assigned to three intervals (1995-1999, 2000-2004, 2005-2009). A multivariate analysis of variance with interval and sex as independent variables was conducted. Differences were analysed with respect to IGF-I standard deviation score (SDS), quality of life, latency between GHD diagnosis and first GH dose, body mass index, waist-hip ratio, lipid profile, and GH dose. All analyses were conducted at baseline, 1 year, and 3 years of GHR. We detected significant associations between time interval and patient characteristics at baseline and with treatment effects. Recently, patients with less severe GHD (mean IGF-I SDS: -2.1, -1.6, -1.0 in the 1st, 2nd and 3rd interval; p = 0.000) are treated with lower GH starting doses (mean 0.30, 0.19, 0.21 mg/day in the 1st, 2nd and 3rd interval; p = 0.000). In the first time interval, IGF-I SDS was not normalized in females after 3 years of GHR. The results of our analysis demonstrate prominent changes in patient characteristics and handling of GHR. They highlight that approach to therapy and patient inclusion criteria change over time and may represent an important confounder for any analysis in epidemiological surveillance surveys.