Fig 1. Estimates for proportion of patients without hospital encounter (i.e., 1 –rate of hospital encounter) over the years from the baseline visit.

The results showed that, compared to those without hospital encounter prior to the baseline visit, those with encounter were significantly more likely to experience hospital encounter in the following years (Adjusted hazard ratio of 1.67, p<0.0001 from Cox regression). Estimated rates of no hospital encounter (and number of subjects at risk) were displayed at bottom-left corner of the plot by group and year.