Skip to main content
. Author manuscript; available in PMC: 2017 Feb 1.
Published in final edited form as: Drug Alcohol Depend. 2015 Dec 24;159:174–180. doi: 10.1016/j.drugalcdep.2015.12.013

Table 2.

Unadjusted estimates for the predictors of the difference between the percentage of doses taken on weekdays and weekends using Generalized Estimating Equations.

Estimate 95% CI p value
Sex Male vs. female 1.42 0.04 2.80 0.04
Age Per 10 year increase −0.58 −1.30 0.15 0.12
Treatment arm mDOT vs. standard of care 0.98 −0.41 2.37 0.17
Site Haiti vs. other −1.14 −2.47 0.20 0.10
Years on cART before entry Per one year increase −0.26 −0.54 0.01 0.06
CD4 count (cells/mm3) Per 100 cell increase 0.36 −0.10 0.83 0.13
Viral load (log 10 copies/ml) Per 1 log increase −0.35 −1.18 0.48 0.41
Binge drinking Yes vs. No 4.23 1.95 6.51 <0.01
*Any substance use last 30 days Yes vs. No −2.22 −5.20 0.75 0.14
Self-perception of health Fair/poor vs. Excellent /Very good/Good −2.09 −3.29 −0.89 <0.01
Quarter on the study Per quarter increase 0.44 0.13 0.76 0.01

Positive numbers represent an increase in the difference between weekday and weekend adherence.

*

Other than alcohol;

cART=combination antiretroviral therapy; mDOT=modified directly observed therapy; (Q1–Q3)=Interquartile range; s.d.=standard deviation; vs.=versus; Quarter on study= 4 successive 12-week period