OBJECTIVES: This study introduces a statistical model to study the longitudinal trends of prevalence rates in two hypothetical study populations. METHODS: A study population with a specific medical condition (cases) in 2003-2013 was extracted from Truven Health MarketScan Commercial Claims Database. This national representative database is longitudinal and patients can be followed over multiple years. In each calendar year of 2003-2013, patients with diagnosis of the given condition were included if they were continuously enrolled during that whole calendar year, and had at least 6-month of data prior to the earliest diagnosis (index date). They were matched to a control cohort based on age, gender, and region by calendar year. Controls’ index dates were randomly assigned based on the distribution of index dates of cases. In each calendar year, patients were flagged if there was at least one diagnosis of hypertension. Generalized estimating equations (GEE) models were employed to examine if the longitudinal trend of the prevalence rate of hypertension observed over time was statistically significant and if the prevalence rate between cases and controls over time was statistically different. RESULTS: A total of 21,180 cases (mean age: 47.7; male: 84.4%) and 66,027 of controls (mean age: 47.5; male: 83.9%) met the study criteria. Hypertension was recorded for 10.4% - 24.6% of cases and 12.8%-22.9% of controls in 2003-2013. The trend in hypertension prevalence was significantly increasing for both cases and controls over 2003-2013 (p < 0.001), but the prevalence rate was not consistently higher in cases than in controls. CONCLUSIONS: In circumstances where one is interested in studying comorbidity prevalence over time, a GEE model offers an flexible approach to compare cohorts, study time trends and examine the joint effect of both in one model while adequately addressing issues such as within patient variability.