Cannabis Use Directly Linked to Psychosis Relapse

Batya Swift Yasgur, MA, LSW

July 24, 2017

Cannabis use by patients with psychosis is tied to a significantly increased risk for relapse. That increase in risk for relapse may be directly due to nonadherence to antipsychotic medication, new research shows.

A team of researchers led by Sagnik Bhattacharyya MD, PhD, of the Department of Psychosis Studies, Institute of Psychiatry, Kings College London, United Kingdom, found that continued use of cannabis was predictive of poor outcomes, including increased risk for relapse, number of relapses, length of relapse, and care intensity at follow-up. Further statistical analysis found that medication adherence partly mediated the effect of continued cannabis use on these outcomes.

"Our current study presents clear evidence that, depending on the type of outcome measure one looks at, there seems to be fairly consistent effect of cannabis use on these outcome measures, and it seems to be mediated by nonadherence. Cannabis users seem to be less adherent [to their medication], therefore, they have worse outcomes," Dr Bhattacharyya told Medscape Medical News.

The study was published online July 10 in Lancet Psychiatry.

Nonresponsive or Nonadherent?

Previous research has shown that cannabis use may affect psychosis outcomes. Its effect on relapse risk was reduced when researchers controlled for adherence, which suggests that "cannabis use could adversely affect psychotic outcome partially by influencing adherence to antipsychotic medication," the authors write.

However, previous studies have not systematically investigated the extent to which the association between cannabis use and psychosis relapse is mediated by nonadherence with psychotropic medications.

"In our previous work, we demonstrated that cannabis use was associated with antipsychotic treatment failure, but we did not know whether the mechanism was increased risk of true nonresponsiveness to the medication or increased nonadherence to medication. This study set out to investigate that question," said Dr Bhattacharyya.

The researchers investigated patients aged 18 to 65 years who had been clinically diagnosed as having first-episode nonorganic (nonaffective or affective) psychosis.

The main outcome variable was risk for relapse, defined as "admission to a psychiatric inpatient unit owing to exacerbation of psychotic symptoms within 2 years following the first presentation to psychiatric services."

Other outcome measures related to relapse included the number of relapses, the length of relapse, the time to first relapse, and the intensity of care at follow-up.

Cannabis use was assessed using a modified version of the Cannabis Experience Questionnaire. The researchers categorized cannabis users on the basis of their pattern of continuance of use after onset. Participants were categorized as being either not a cannabis user (no use or use only once or twice after onset), an intermittent cannabis user (the participant used cannabis more than twice but not every month), or a continued cannabis user (the participant used cannabis every month) throughout the 24-month follow-up period.

Medication adherence was assessed as a mediator variable, using the Life Chart Schedule. Classification was determined on the basis of information about prescription and ratings of adherence: nonadherence (67% to 100% of the time noncompliant); irregular adherence (34% to 66% of the time noncompliant); good adherence (0% to 33% of the time noncompliant); or medication not prescribed within the 2 years following illness onset.

The researchers used structural equation modeling analyses to measure the mediating effect of medication adherence on the association between cannabis use and relapse.

Of the 397 patients who presented with first-episode of psychosis, 245 patients were followed for 2 years from onset. Of these, 91 (37%) experienced relapse during the 2 years after onset of psychosis. Most patients reported regular or irregular adherence with their medication (45% and 42%, respectively); a small subset of patients (14%) reported nonadherence.

Following the onset of psychosis, 146 patients (60%) were classified as not cannabis users. These included 98 patients (40%) who were never regular users and 48 (20%) who had formerly been regular users.

The remaining patients were classified either as intermittent cannabis users (36 patients, 15%) or continued cannabis users (63 patients, 26%).

Potential Mechanisms

When the researchers compared patients who experienced relapse to those who did not, they found that patients in the relapse group were more likely to be continued cannabis users (P = .0018; 95% confidence interval [CI], 0.09 - 0.43) and as nonadherent (P < .0001; CI, 0.18 - 0.58) or irregularly adherent (P = .0001; CI, 0.13 - 0.39) for the prescribed medication.

After adjusting for ethnic origin, other illicit drug use, and intensity of care at origin, structural equation modeling analyses revealed that the association between continued cannabis use and risk for relapse was mediated (26.4% as the proportion of total effect mediated) by medication adherence (direct effect, β = 0.22; 95% CI, 0.03-0.42; P = .027; indirect effect: β = .08, 95% CI, .004-0.16; P = .040), "suggesting a partial but not full mediation by medication adherence of the effect of cannabis use on risk of relapse," the authors write.

They found a similar effect with respect to intensity of care at follow-up. Medication adherence mediated 19.7% of the effect of continued cannabis use, "again implicating partial mediation based on significant indirect and direct effects."

The effect of continued cannabis use on the number of psychosis relapses and time until relapse was likewise mediated by medication adherence. However, no significant mediation effect was found for length of relapse.

The researchers tested an alternative model in which cannabis use rather than medication adherence was the proposed mediator. They found that continued cannabis use did not mediate the association between medication nonadherence and the various relapsing outcomes, after controlling for covariates.

In addition, they tested a mediation model that assessed whether medication adherence mediated the effect of cannabis use on outcome and a reverse arrow model that tested whether cannabis use mediated the effect of medication adherence on outcome.

In those models, medication adherence had a significant direct effect on risk for relapse (β = 0.45; 95% CI, 0.20 - 0.70; P = .0004), number of relapses (β = 0·42; 95% CI, 0.18 - 0.65; P = .0005), care intensity at follow-up (β = 0.42, 95% CI, 0.19 - 0.65; P = 0.0003), and time until relapse (β = -2.00, 95% CI, -3.18 to -0.81; P = .0010), "indicating that cannabis use did not fully confound the effects of medication adherence on outcome."

Medication nonadherence "mediated the effect of continued cannabis use on risk of relapse (26%), number of relapses (36%), time until a relapse occurred (28%), and care intensity index at follow-up (20%)," the authors write.

"In our previous research, we found that people who respond to one antipsychotic will remain on it, but a person who shows no response, for whatever reason, will be treated by the clinician with a second or third antipsychotic. We found that people who use cannabis have a history of more treatments with antipsychotics and more failed treatments," said Dr Bhattacharyya.

"The question is whether patients with a 'failed antipsychotic' are truly nonresponsive, or whether they may be nonadherent and this is why they are not showing response," he said. "In the case of cannabis users, we showed that cannabis was associated with treatment failure but did not know if it was as a result of increased nonadherence or increased true nonresponsiveness."

He acknowledged that the mechanism whereby cannabis leads to nonadherence has yet to be elucidated.

"There are a few potential mechanisms. Cannabis use may lead people to feel more unwell and more psychotic, and one of the hallmarks of psychosis is that patients do not realize they are unwell. They do not have insight into their illness, so they may stop taking their medications because they do not have insight that they need medication."

He added that his research group is carrying out studies to test that hypothesis.

"Another potential mechanism is that those who use cannabis regularly might experience more medication side effects and therefore be less adherent to treatment, or that there may be a biological mechanism whereby cannabis reduces responsiveness to antipsychotics, but again, these hypotheses need to be tested," he added.

Definitive Finding

Commenting on the study for Medscape Medical News, Deepak D'Sousa, MBBS, MD, professor of psychiatry, Yale University School of Medicine, New Haven, Connecticut, said the findings of the study were not surprising.

"Most clinicians who treat this population are somewhat aware that cannabis presents greater risk of relapse and that this might be mediated by medication nonadherence, but it is good to see this studied in a systematic way, and the finding seems to be definitive."

He noted that the researchers relied on self-report to assess cannabis use and medication adherence, a limitation that the researchers acknowledge.

"Nevertheless, one of the issues of self-report is that people may underreport nonadherence and underreport the use of cannabis, and despite the possibility that patients may have been underreporting, the researchers still found significant relationship and association between cannabis use, medication adherence, and relapse."

The study has "obvious clinical implications," he said.

"The take-home message for clinicians is that they should really consider measures to address nonadherence in patients who are known to be using cannabis on some regular basis.

"The good news is that we do have a number of approaches to managing nonadherence. Obviously, the simplest of those would be educating patients, having a discussion with patients, and asking them on a regular basis if they are adherent with their medication regimen," Dr D'Sousa added.

He also noted that there are newer injectable formulations of antipsychotic medications whereby patients can receive their medications once every 2 weeks, once a month, or once every 3 months. "This provides a guaranteed delivery of the medication,'" he said.

Dr Bhattacharyya agreed.

"In an ideal world, we would want to stop patients from using cannabis, but current treatments, which are typically psychological, do not seem to work very well in that direction. In the absence of effective means to stop cannabis use, clinicians should focus on ways of improving adherence to antipsychotics."

These include cognitive-behavioral therapy to "help with insight so that people recognize they are unwell and therefore adhere to treatment." Long-acting injectable medications "may be another helpful approach."

The study was funded by a National Institute of Health Research Clinician Scientist Award. Dr Bhattacharyya and Dr D'Sousa have disclosed no relevant financial relationships.

Lancet Psychiatry. Published online July 10, 2017. Full text


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.