This study is among the first to examine the association between neurocognitive impairment and HAART medication adherence in HIV patients with and without cocaine dependence. Overall, participants demonstrated substantial neurocognitive impairment, with mean T-scores under 40 on most measures. This is consistent with prior research findings that neurocognitive deficits are common in a majority of HIV patients, particularly as the disease progresses (Antinori et al., 2007
; Tozzi et al., 2007
). In our sample, all participants had a history of CD4 cell count < 500, reflecting disease progression; they had been diagnosed with HIV for a mean of 13 years, and 78% had a diagnosis of AIDS. Across the whole sample, neurocognitive impairment was associated with poorer HAART adherence, even after accounting for current cocaine dependence. However, active cocaine users had significantly greater neurocognitive impairment and poorer medication adherence, and neurocognitive impairment was found to partially mediate the relationship between cocaine dependence and medication non-adherence. That is, the higher rate of non-adherence in cocaine users was partially explained by the negative correlation between cocaine dependence and neurocognitive functioning. The degree of neurocognitive impairment and poor medication adherence among active cocaine users suggests that this is a population in need of more intensive interventions to optimize treatment outcomes.
Active cocaine users had greater overall neurocognitive impairments, with specific deficits in verbal memory, processing speed, and visuospatial construction. This is generally consistent with studies conducted among HIV-negative samples, which also found specific rather than global deficits (Beatty et al., 1995
; Hoff et al., 1996
; Medina et al., 2006
; Rippeth et al., 2004
; Rosselli et al., 2001
; Simon et al., 2000
; Verdejo-Garcia & Perez-Garcia, 2007
). The effects of cocaine dependence on neurocognitive functioning may be particularly profound in HIV patients, given that HIV infection itself also causes impairments in many of the same neurocognitive domains, notably information processing, memory, and executive function (Heaton et al., 1995
; Moore et al., 2006
; Reger et al., 2002
). Our results lend further support to the independent effect of current cocaine dependence on neurocognitive impairment in HIV patients, but further research is necessary to explore whether or not HIV infection and cocaine dependence have additive and/or synergistic effects on neurocognition.
Few studies have examined the association between cocaine use and neurocognitive functioning among HIV patients, and results have been variable, likely due to methodological differences. Durvasula and colleagues found that cocaine users had poorer performance on information processing but not in other domains (Durvasula et al., 2000
). However, the use of a 12-month recall period meant that some of the "active" users could have been abstinent for several months, and the quantity of cocaine use in the sample was modest. Chang and colleagues also found no differences between HIV patients with and without cocaine dependence on verbal memory, psychomotor speed, and executive function, but the cocaine users in that sample had been abstinent for an average of 5 years (Chang et al., 2008
). This is consistent with our finding that recovered cocaine users demonstrated no neurocognitive impairments relative to drug naïve participants. In contrast, Levine and colleagues found that participants who had used psychostimulants (cocaine or methamphetamine) in the past month, as verified by toxicology screen, had impairments in sustained attention relative to those who had not used in the past month (Levine et al., 2004
). A major strength of our study was that participants were carefully assessed using several validated clinical interviews and urine toxicology screens to differentiate groups of active, recovered, and naïve cocaine users, enabling us to tease apart the effects of current versus past cocaine dependence. Furthermore, the active cocaine users in our sample were chronic users who met diagnostic criteria for cocaine dependence and used cocaine at least weekly, strengthening our ability to identify effects due to cocaine use. Cumulatively, findings from these studies suggest that neurocognitive deficits associated with cocaine dependence are due to the impermanent effects of active use, and that cocaine use may have to exceed a certain threshold before any measurable impact on neurocognitive impairment is observed.
Our results lay the groundwork to examine whether prolonged abstinence may reverse the neurocognitive impairments associated with cocaine dependence. The recovered group in our study met full diagnostic criteria for past cocaine dependence in sustained full remission, with a mean of nearly 14 years of regular cocaine use. They demonstrated no neurocognitive impairments relative to drug naïve participants and were equally likely to adhere to their HAART regimens. Evidence from HIV-negative samples supports the idea that drug- induced neurocognitive impairments can be reversed. In a study of twins with and without a history of heavy psychostimulant use, after 1 year of abstinence, deficits were evident only on motor skills; there were no differences in verbal learning, visuospatial construction, or executive function (Toomey et al., 2003
). Another study found no difference between cocaine users who were abstinent for 1 month and controls on verbal memory, visuospatial construction, processing speed, verbal fluency, and executive function (Bolla, Rothman, & Cadet, 1999
). A third study of veterans entering a 3-week substance abuse treatment program found significant improvements in memory, attention, motor skills, and executive function, but not in visuospatial construction or language (Schrimsher & Parker, 2008
). Longitudinal studies that measure changes in neurocognitive functioning over the course of treatment and longer-term recovery (> 1 year) in HIV patients are needed to establish causality.
As expected, active cocaine users reported poorer HAART adherence compared to non users. Specifically, 64% of active cocaine users missed at least one dose in the past month compared to only 26% of non drug users. Our results support the hypothesis that neurocognitive impairments associated with cocaine dependence partially explain the higher rates of non-adherence in cocaine users. Cognitive remediation strategies may be helpful for improving medication adherence among HIV-positive cocaine users. However, given that neurocognitive functioning only partially mediated the relationship between cocaine dependence and medication adherence, further research is needed to identify additional mediators that may be amenable to intervention. For example, disruptions in sleep and eating patterns, increased environmental instability, and a generally chaotic lifestyle may lead to poor adherence among psychostimulant users (Reback, Larkins, & Shoptaw, 2003
; Tucker et al., 2004
). Furthermore, many of the active cocaine users also abuse alcohol and marijuana, which may further impair neurocognitive functioning and ability to adhere to HIV medications. Ultimately, treatment for cocaine dependence and additional co-occurring substance use disorders may be critical for improving HIV clinical outcomes.
Results of this study should be interpreted in light of the following limitations. First, the cross-sectional design precludes any conclusions regarding causality. Differences in neurocognition may have preceded the onset of cocaine dependence and/or interfere with substance abuse treatment (Aharonovich et al., 2006
; Aharonovich, Nunes, & Hasin, 2003
; Fals-Stewart & Lucente, 1994
; Fox, Jackson, & Sinha, 2009
; Teichner, Horner, & Harvey, 2001
). It is also likely that the relationship between neurocognition and medication adherence is bidirectional: neurocognitive impairment may limit a patient's ability to adhere perfectly to HAART, and poor adherence may lead to further declines in neurocognitive functioning (Lovejoy & Suhr, 2009
). Clearly, longitudinal studies are needed to elucidate the temporal relationship between these variables. Second, this study relied on self- report of medication adherence. Self-report is the most practical means of collecting adherence data in cross-sectional studies, and it has been found to be a valid measure that is predictive of HIV viral load (Fletcher et al., 2005
; Giordano et al., 2004
; Lu et al., 2008
; Reynolds et al., 2007
). A notable strength of this study was the use of multiple adherence measures, with consistent results across measures. Nevertheless, results may have been biased by the poor neurocognitive functioning evidenced by the sample. Future studies may wish to combine self-report with objective measures of adherence, such as pill counts and pharmacy refills, given the high rates of memory impairment observed in our study (Lovejoy & Suhr, 2009
). Third, global neurocognition variables have the potential to underestimate performance in specific domains (Lovejoy & Suhr, 2009
). However, these are commonly used in neuroAIDS research (Avants et al., 2001
; Barclay et al., 2007
; Hinkin et al., 2002
), and use of a global neurocognition variable allowed us to test a single regression model, minimizing the risk of a type I error. Fourth, the proportion of African-Americans was higher in the active versus non cocaine group. Whenever possible, we employed demographic connections that accounted for race when converting raw scores to T-scores, and results did not change when accounting for race in the regression models. A matched control study that ensures equal proportions of African-Americans across groups would help to eliminate potential confounding effects of race on neurocognitive functioning. Finally, this study utilized a convenience sample of patients receiving HIV treatment in the United States. Replication studies with larger sample sizes and in other regions of the world are needed to determine if results generalize to the broader population of HIV patients.
In conclusion, cocaine dependence among HIV patients is associated with suboptimal clinical outcomes, including higher viral loads, greater immune suppression, faster disease progression, and death (Baum et al., 2009
; Chander et al., 2008
; Webber, Schoenbaum, Gourevitch, Buono, & Klein, 1999
), and poor medication adherence may be an important contributing factor. This study found that HIV patients with current cocaine dependence had greater neurocognitive impairment and poorer HAART adherence compared to non drug users, and that neurocognitive impairment associated with cocaine dependence partially explained the lower rates of medication adherence. Results underscore the importance of assessing and treating both neurocognitive impairment and cocaine dependence among HIV patients, including the development of empirically supported treatments. Further research is needed to identify potential additive and/or synergistic effects of HIV infection and cocaine dependence and their long-term effects on clinical outcomes.