Panel data analysis in medical research

A longitudinal study involves repeated observations of the same items over long periods of time. Panel studies are longitudinal studies of batch which combine cross-sectional and time- series data in observations on a number of over time that play special role in medical research or clinical trials. In this paper, we primarily discuss about the panel data and various modeling of panel data. It is not possible to use ordinary regression methods due to inter- correlation of observation related to one unit. Generalized estimation equation (GEE) for estimation of panel regression coefficient by considering inter correlation was used among observations.

We considerd the importance of panel data modeling with GEE in real panel data samples as applications in the effect of estrogen patches on the postnatal depression.

In the estimation of panel regression coefficients on real data, starting treatment effect (b(pre)=0.428, p<0.001), treatment group effect (b(group)=4.025, p<0.001) and treatment period effect (b(visit)=1.218,p<0.001 with use of robust method to misspecification of the correlation structure were significant.

According to the important role of panel data inter- correlation structure in analyzing in medical research, we propose to estimate the coefficients of panel via GEE method.