Assessing the precision of a time-sampling-based study among GPs: balancing sample size and measurement frequency