The data were analyzed using a 3 × 2 analysis of variance (disease status × marijuana use). As noted above, for age and education there were no significant main or interaction effects. However, on the Hamilton Depression Rating Scale there was a significant effect of marijuana use (F = 8.98, df = 1,276, p < 0.003). Subjects with frequent marijuana use reported more symptoms of depression. On the Hamilton Anxiety Rating Scale, there also was a significant effect of marijuana use (F= 8.58, df = 1,276, p < 0.004). Once again, subjects with frequent marijuana use reported more symptoms of anxiety. It was also considered that subjects with frequent marijuana use might also be prone to frequent use of other drugs such as alcohol. As shown in t1, for weekly consumption of alcohol, there was a significant effect of disease stage (F= 4.14, df=2, 276, p < 0.013), for marijuana use (F=12.41, df=1,276, p < 0.001), and a trend toward a significant interaction (F=2.67, df=2,276, p <0.071). Subjects with symptomatic HIV infection and subjects with infrequent marijuana use used significantly less alcohol. There was no difference in current or past use, abuse, or dependence of other drugs. Therefore, depression, anxiety, and alcohol use were included in the analyses as covariates. Comparison of CD4 levels revealed the expected group differences. Among the symptomatic subjects, there were no significant effects of marijuana use on CD4 levels and no interaction. Thus, any interactions of disease stage and marijuana use cannot be attributed to differences in level of immunosuppression.
The initial analysis was based on the overall impairment score, which reflected the total number of measures on which an individual achieved a score that was one standard deviation below the mean of the control group. As shown in t2, after controlling for the effects of depression, anxiety, and alcohol consumption, there was a significant effect for disease stage (F=7.18, df=2,276, p <0.001), a trend toward a marijuana effect (F=3.49, df=1,276, p < 0.063), and a significant interaction effect (F=3.53, df=2,276, p < 0.031). Inspection of the data indicates that the difference between the no/minimal and frequent marijuana use groups was greatest among the subjects with symptomatic HIV infection.
The data were further analyzed to determine which areas of neuropsychological performance were affected by marijuana use. This was accomplished by a series of analysis of covariance (ANCOVAs) controlling for the effects of depression, anxiety, and alcohol consumption. The results of these analyses are also presented in t2, which demonstrates that the effect of marijuana use was reflected only on a measure of delayed memory, on which there was also a significant interaction effect. Analysis of simple main effects revealed that the differential impact of marijuana was progressively greater in relation to increasingly severe HIV disease. There were no differences on other measures of attention, learning, or memory related to marijuana use.