Primary progression model, and estimated effects of total time in previous stages when added to the primary model.

aOdds ratios indicate the estimated effects of the predictors on the risk of progression to the next stage at any given time step. Odds ratios below 1.0 indicate lowered risk; ratios above 1.0 indicate increased risk.Based on 4 imputed data sets; the fifth had estimated odds ratio (OR) = 0 with likelihood ratio (LR) p-value bBased on 4 imputed data sets; the fifth had estimated odds ratio (OR) = 0 with LR p-value cAll imputed data sets had estimated OR = 0; the largest of the 5 LR p-values and the corresponding profile likelihood confidence bound are shown.dBased on 3 imputed data sets; the other two estimated odds ratios were 0.13 and 0.15 with LR p-values of 0.028 and 0.021, but these were close to degenerate, with much larger estimated standard errors and deviance nearly identical at OR = 0.eAlthough the estimated OR was finite, the estimates in all 5 imputed data sets appeared to be nearly degenerate, with large standard errors and deviance at OR = +∞ nearly as good as at the finite estimated values.fEstimates in all 5 imputed data sets were degenerate. As these are not parameters of interest, we did not pursue LR p-values or profile likelihood confidence bounds.gBased on 4 imputed data sets; the fifth had estimated OR = 0 with likelihood ratio p-value hBased on 3 imputed data sets; the other two had estimated OR = +∞ with LR p-values 0.021 and 0.082.Abbreviations: ALIVE: AIDS Link to Intravenous Experience study [13], [14], [15]; HALS: Hepatitis C and Alcohol Study [16]; SFVA: San Francisco Veterans Affairs Medical Center liver studies cohort [17].