ReadMe file for GP outcome data GP data: n=126 rows of data; 63 GPs recruited, at baseline and two follow-up time points (one row per follow-up time, see note below) (n.b. GP# 16 withdrew after completing baseline measures and before training, and hence all follow-up data columns are blank on both rows.) The data dictionaries explain the naming of each variable in each dataset (naming convention follows above format, all in CSV format). For further information on variables and analysis, including the timing of data collection for each outcome, please refer to the protocol paper (Open access at https://doi.org/10.1186/s13063-017-2225-8) and/or the main results paper (DOI as per this Dryad entry). Data for the main GP outcomes are provided in long format – that is, one row for each follow-up time point after baseline. Baseline data is replicated across each row (to allow adjustment in outcome models). To examine profile at baseline, you will need to first filter the dataset to just one particular week (e.g. week==4) to keep just one row per participant. Baseline versions of “outcome” variables all include “_base” at or near the end of their variable name (see GP file Data Dictionary for more information). To preserve anonymity in line with ethical principles and Dryad guidelines, some variables have been redacted (GP gender, age, ethnicity, and length of service). Researchers wishing to access these data should contact the study team (ben.darlow@otago.ac.nz, principal investigator; james.stanley@otago.ac.nz, biostatistician).