Bayesian reanalysis of null results reported in medicine: Strong yet variable evidence for the absence of treatment effects