marginally significantp = 0.057
In the log-transformed model, the odds of not attending surgery were significantly predicted by loge(ED visits within 1 year) (β = 1.06, SE = 0.31, z = 3.43, p < 0.001), while logMAR showed marginally significant prediction of odds of missing surgery (β = 0.77, SE = 0.41, z = 1.91, p = 0.057).