We found support for our hypothesis that depression treatment patterns are associated with ART adherence. For patients with a depression diagnosis, the odds of being adherent with ART are greater for treated than untreated patients, and the odds of being adherent with ART are significantly greater for drug therapy alone than psychotherapy alone.
In Models 3, 4 and 5, we examined, simultaneously, adherence to antidepressant treatment and number of psychotherapy visits attended, with consistent findings. In the model limited to antidepressant users (Model 3), the model limited to psychotherapy users (Model 4), and the model that included those receiving either one or both treatments (Model 5), antidepressant adherence of ≥ 80% was significantly and strongly associated with greater odds of ART adherence. Number of psychotherapy visits attended was not associated with ART adherence in any of the models.
Our findings concord with those of previous research; among persons with HIV and depression, treating depression symptoms with pharmacotherapy is associated with improvements in ART adherence (24
). However, the current literature, including this study, is limited to retrospective observational studies establishing associations. The current study is not a controlled study comparing antidepressant treatment to a placebo in relation to ART adherence. Difficulties in establishing the nature of the association (i.e., whether treating depression symptoms would increase ART adherence) with observational data have been noted (56
). Future research should be designed to both investigate whether better antidepressant adherence would improve ART adherence, and to quantify the gains from antidepressant treatment.
We did not observe an association between psychotherapy utilization and ART adherence. Yet, there are empirically supported therapies that have been shown to be effective in other studies (for example, specially adapted protocols that combine cognitive behavioral therapy with adherence promoting components --“Cognitive Behavioral Therapy for Adherence and Depression” or CBT-AD) (39
). Many factors (that we were not able to measure) may be responsible for the lack of an association. First, patients who are not inclined to take medications (regardless of their class) may be more likely to seek psychotherapy as the treatment of choice for depression. Second, our claims data did not include information on the types of psychological treatment provided (e.g., cognitive-behavior therapy, nondirective supportive treatment, behavioral activation treatment, or interpersonal psychotherapy), the problems it targets, or the specific interventions used. It may be that adherence benefits require therapy adaptations of the interventions included in CBT-AD. Perhaps, some therapists emphasize adherence to ART as an important goal of psychotherapy, and engage patients in a discussion of barriers to and problem solving around adherence while other therapists may emphasize more traditional psychotherapy goals (e.g. improved mood, improved interpersonal relationships, or emotional resilience) that may have a less direct impact on ART adherence. Future research should include more detailed data on the psychotherapy provided (e.g., type, frequency, targets, emphasis, and interventions) and its association with ART adherence. PLWHA would be better served by mental health providers who try to motivate them to adhere to appropriate treatments for their conditions. Mental health practitioners serving the needs of people living with HIV should be trained with workforce development efforts, such as post graduate certificate programs and continuing education courses that emphasize the skills (e.g., motivational interviewing, behavior change) and knowledge (e.g., disease progression, medication side effects) needed to increase ART adherence.
We found that more than three-fourths of those with depression filled a prescription for an antidepressant during the observation period, a rate higher than we had anticipated. Analyses of HCSUS data from 1996 indicated that 43% of those with a mood disorder diagnosis, based on a structured interview, reported antidepressant use in the prior 6 months (57
). This may reflect the fact that we studied a commercially insured population, with better access to both mental health providers and medications than patients studied in the HCSUS data. It may also reflect increased treatment rates over the decade (58
The contributions of the current study include: a geographically diverse sample residing in all 50 states, the inclusion of both antidepressant treatment and psychotherapy, and a large sample size that enables us to investigate dose-response relationships, and provides us with the ability to quantify the association between various depression treatments and ART adherence. ART adherence was measured during an ART treatment episode, with discontinuations longer that 60 days viewed as conceptually different from other lapses in adherence.
There are several study limitations. First, because this is an observational study, we cannot attribute causation to the relationships found. That is, we cannot conclude that better antidepressant adherence was responsible for or caused improved ART adherence. Demonstration of causation would ideally require an experimental study design, but can be time consuming and relatively difficult to implement. It is possible that there are third variables that are responsible for the association we found. For example, certain patients may just be more willing to take pills, irrespective of pill type (antidepressants or ART medications). Such variables cannot be captured with the current design. If experimental designs are not possible, future research should try to measure potential confounders, such as willingness to take pills, irrespective of the type of pill (antidepressant or ART).
Second, although the dataset is large, it is nonetheless a convenience sample of privately insured beneficiaries. Information on race, education level, income, depression severity and previous depression treatment outcomes are not available in these claims data; their inclusion would provide further insights. Third, our adherence measures are based on pharmacy refills and do not guarantee ingestion of the medications. High concordance between claims based pharmacy measures and pills counts have been found (59
), and among HIV patients, pharmacy-based adherence measures have predicted survival (60
). However, a recent study suggests poor correlation between pharmacy-based claims and electronic drug monitoring (61
). Future research should investigate other definitions of adherence.