Indoor off-body wireless MIMO links between a mobile user equipped with wearable textile patch antennas and a fixed base station exhibit specific channel behavior due to the near presence and movements of the human body. Therefore, they require a dedicated channel model that captures the effects of correlated small-scale Rayleigh fading and correlated lognormal shadowing. A methodology is presented to construct such a model, allowing to predict the bit error characteristics and channel capacity curves based on the shadowing and fading correlation matrices that are extracted from channel measurements. It is shown that by separating shadowing, including effects caused by movement and reorientation of the human body, from small-scale fading, the main mechanisms of the off-body communication link are accurately captured by the model. A clear dependence of the shadowing correlation values on the physical layout of the antenna system is found. In our measurements, shadowing is not significantly decorrelated by polarization diversity or front-to-back diversity whereas the small-scale fading is clearly decorrelated. From the model, MIMO channel realizations with identical bit error rate and channel capacity characteristics as the measured channel can be quickly generated for link emulation purposes.