In this paper, we develop a theoretical model that links the demand for telecare to the length of stay in hospital and formulate three models that can be used to derive the treatment effect by making various assumptions about the probability distribution of the outcome measure. We then fit the models to data and estimate them using a strategy that controls for the effects of confounding variables and unobservable factors, and compare the treatment effects with that of the Propensity Score Matching (PSM) technique which adopts a quasi-experimental study design. To ensure comparability, the covariates are kept identical in all cases. An important finding that emerges from our analysis is that the treatment effects derived from our econometric models of interest are better than that obtained from an experimental study design as the latter does not account for all the relevant unobservable factors. In particular, the results show that estimating the treatment effect of telecare in the way that an experimental study design entails fails to account for the systematic variations in individuals' health production functions within each experimental arm.