Adjusted odds ratios with 95% confidence intervals were calculated to approximate relative risks of adverse outcomes. A P value of less than .05 was considered statistically significant (ie, the 95% confidence intervals were calculated, and were considered significant if they did not include 1.0). However, due to the large sample size, statistical analysis was powerful enough to detect differences in risk between the ART groups that were statistically significant but where the actual size of the difference was small. In some cases the differences might be so small that they are not clinically meaningful. Therefore, since OR describes the magnitude of the effect between groups, an odds ratio cutoff of greater than 2.0 was selected to represent clinically meaningful risk to emphasize those outcomes with a marked association with invasive or noninvasive ART. All analyses were conducted using SAS 8.2 (SAS Institute Inc., Cary, NC).