ART treatments in HIV-infected patients

A study estimated the effect of 3-drug ART treatments versus 4-drug ART treatments for HIV-infected patients on the risk of virologic failure in the first 144 weeks. Virologic failure is a (binary) outcome of interest in HIV clinical trials, with treatments intended to prevent virologic failure. Treatment limiting adverse events and treatment limiting other events are also valid reasons to discontinue treatment.

a. In the ordinal regression model on the second page, interpret the coefficient for the predictor, log viral load. If you prefer to interpret exp(coefficient), that is fine as well.

b. In the logistic regression model on the third page, interpret the coefficient for the predictor, log viral load. If you prefer to interpret exp(coefficient), that is fine as well.

c. In the logistic regression model on the last page, interpret the coefficient for the predictor, log viral load. If you prefer to interpret exp(coefficient), that is fine as well.

d. Does the ordinal regression model on page 2 fits these data well? Motivate your answer.*

HIV example of Ordinal Regression Lok and Hughes (2016) analyzed efficacy and safety of antiretroviral treatment (ART) regimens. O Events (response): We create an ordinal response: (virologic failure (1), treatment discontinuation because of safety issues (2), treatment discontinuation for other reasons (3), treatment continued (4)) O Viral load (covariate 1, on the log_10 scale, as usual in HIV research: Ivl is log viral load) O Age (covariate 2)HIV example: Virologic failure 1.00 0.75 Parameter Estimates Term Estimate Std Error ChiSquare Prob> ChiSq vf_noworpast 0.50 P(YS1) Intercept[1] -4.1662021 0.2459544 286.93 <.0001 .o. lvl ivl whole model test df chisquare prob> ChiSq Difference 151.12609 302.2522 <.0001 full reduced rsquare aicc bic observations sum wgts example: virologic failure or treatment limiting adverse event parameter estimates erm estimate std error chisquare prob> ChiSq vitlae 0.50 P 2 Intercept[1] -4.460569 0.2824898 249.33 <.0001 lvl whole model test df chisquare prob> ChiSq Difference 207.85532 415.7106 <.0001 full reduced rsquare aic bic observations sum wgts example: virologic failure limiting adverse event other parameter estimates term estimate std error chisquare prob> ChiSq event_noworpast 0.50 YChiSq Difference 223.82262 1 447.6452 <.0001 full reduced rsquare aicc bic observations sum wgts example: ordinal regression parameter estimates term estimate std error chisquare prob> ChiSq Intercept[1] -4.6207484 0.2447052 356.56 <.0001 intercept lvl eventtype_le p. un . n. whole model test df chisquare prob> ChiSq Difference 204.44164 1 408.8833 <.0001 full reduced rsquare aicc bic ats>