Associations between presence of detectable CMV DNA and subsequent non-AIDS event at baseline (case: N=85, control: N=147), year 1 (case: N=128, control: N=276) and pre-event (case: N=110, control: N=237). Analysis adjusted for HIV RNA (log10) at baseline, adjusted for CD4 count at year 1 and pre-event (where CD4 count adjustment includes quadratic and cubic terms). Sensitivity analyses exclude participants with low cell yield. *NE – Not estimable. Using a post-hoc Cochran-Mantel-Haenszel test, there was an increased risk of death for those with detectable CMV DNA at baseline [odds ratio (95% CI) = 9.0 (1.9, 42.6), p=0.002, not shown in the figure].