A cross-sectional study from Ghana found that psychiatric patients using both cannabis and alcohol had the highest liver enzyme elevations, but a fundamental flaw in the study’s control group prevents researchers from distinguishing substance effects from the effects of psychiatric illness and its medications. These findings are hypothesis-generating, not practice-changing.

Psychiatric patients who use cannabis, alcohol, or both are a medically complex population in which the effects of substances, mental illness, and pharmacological treatment overlap in ways that are poorly understood. In Sub-Saharan Africa, where psychiatric populations are dramatically underrepresented in the biochemical literature on substance use, locally generated laboratory data carry special weight. This study from Accra Psychiatric Hospital attempts to fill that gap, and understanding both its contributions and its limitations is essential for clinicians working at the intersection of psychiatry and substance use medicine.

Substance use among people with psychiatric illness represents a common clinical challenge, yet surprisingly little laboratory data exist from African psychiatric populations to characterize its metabolic consequences. Researchers at Accra Psychiatric Hospital set out to compare hematological and biochemical markers across four groups: healthy non-users, alcohol-only users, cannabis-only users, and dual cannabis-alcohol users. The biological rationale was straightforward: both cannabis and alcohol have known hepatic and metabolic effects, and understanding their footprint in this population could inform clinical monitoring.

The study found that dual users had the highest AST (81.14 ± 72.26 U/L vs. 21.19 ± 8.83 U/L in non-users, p<0.001) and GGT levels (95.17 ± 175.47 U/L vs. 29.76 ± 10.99 U/L, p<0.001), suggesting hepatic stress. Cannabis-only users had significantly lower total cholesterol (4.53 vs. 5.30 mmol/L, p<0.001) and LDL-C compared to non-users. Both direct and indirect bilirubin were elevated across all substance user groups. However, the study’s use of healthy community volunteers as the control group, rather than non-substance-using psychiatric patients, means that observed differences may reflect psychiatric illness, psychotropic medications, sex imbalance, or dietary patterns rather than substance use itself. The authors recommend routine liver function monitoring for substance users, but they acknowledge the need for prospective, controlled studies to validate these findings.

Cannabis, Alcohol, and the Liver: What a Ghanaian Psychiatric Hospital Study Can, and Cannot, Tell Us

Here is a study asking exactly the right question for exactly the right population, and yet, a single design choice made it nearly impossible to answer that question cleanly. When researchers at Accra Psychiatric Hospital set out to measure whether cannabis and alcohol use change the blood chemistry of their patients, they compared those patients against healthy volunteers. That one decision put every result they found under a cloud. The paper appears to claim that cannabis and alcohol use cause liver enzyme elevations and lipid changes in psychiatric patients. What it actually tested is whether psychiatrically ill substance users differ biochemically from healthy people who use no substances and have no psychiatric diagnosis. Those are profoundly different questions. To its credit, the study does something genuinely valuable: it generates objective, machine-validated laboratory data from a population that is nearly invisible in the global biochemical literature. The liver enzyme findings for alcohol-involved groups, particularly the AST and GGT elevations in dual users, are directionally consistent with decades of established alcohol hepatotoxicity research. That directional consistency is worth noting. But the central methodological problem is inescapable. Comparing medicated psychiatric patients to healthy volunteers and then attributing the differences to substance use is like comparing cholesterol levels between people taking statins and people not taking them, then concluding the difference is caused by what they ate for breakfast. The medication is the elephant in the room, and in this study, it was never addressed.

This matters because antipsychotics and mood stabilizers independently alter lipid profiles, liver enzymes, and hematological markers. Without knowing which medications participants were taking, and without adjusting for them, we cannot parse which portion of the biochemical signal comes from cannabis, which from alcohol, which from psychiatric pharmacotherapy, and which from the illness itself. The sex imbalance compounds the problem: non-users were 57% female while substance user groups were 88 to 95% male, and sex independently influences hemoglobin, lipid levels, and liver enzyme activity. The wide standard deviations in the small subgroups further erode confidence. When the GGT of 21 dual users averages 95 U/L with a standard deviation of 175, that is the statistical equivalent of averaging the heights of 21 people when three of them are professional basketball players. The mean looks striking, but it does not represent the group. The paper’s recommendation for routine liver function screening in substance users is clinically intuitive and may well prove correct, but this dataset of 21 dual users compared against non-equivalent controls is not the evidence that should underpin such a recommendation.

To a patient, I would say: we are seeing patterns suggesting that using both cannabis and alcohol may put extra stress on your liver, and if you are using both, it makes sense for us to check your liver function periodically, but your medications also affect these tests and we are still learning. To a colleague, I would note that the signal here is interesting but uninterpretable as a true substance effect given the control group design. To a policymaker, I would caution against building screening mandates on this foundation while strongly encouraging investment in a properly controlled follow-up study. The most valuable thing this paper can do is motivate one. In substance use research within psychiatric populations, the comparison group is almost never “just healthy people.” Failing to match on psychiatric illness and treatment will always produce confounded results, regardless of how carefully the laboratory work is done.

This study sits very early in the research arc for biochemical characterization of substance use in African psychiatric populations. It is descriptive and observational, offering associations that may generate hypotheses for future investigation. Clinicians should recognize its contribution as a data point from a setting where almost no comparable data exist, while also acknowledging that none of its comparative findings can be interpreted as demonstrating substance-specific effects.

From a pharmacological perspective, the failure to account for psychotropic medications is particularly consequential. Antipsychotics such as olanzapine and clozapine are well-documented causes of dyslipidemia and hepatic enzyme elevation, and mood stabilizers like valproate are hepatotoxic in their own right. Any clinician interpreting liver enzymes or lipid panels in a psychiatric patient must first consider the medication profile before attributing abnormalities to substance use. The one concrete recommendation that does follow from these data, even if the study itself cannot prove it, is that clinicians caring for psychiatric patients who use both alcohol and cannabis should maintain heightened awareness of hepatic function and consider periodic liver enzyme monitoring as a matter of clinical prudence.

This is a cross-sectional observational study with original laboratory data collected from a single psychiatric hospital in Ghana. Cross-sectional designs occupy a lower tier in the evidence hierarchy because they capture associations at a single point in time and cannot establish temporal sequence or causation. The most important constraint on inference here is the non-equivalent comparator group: healthy volunteers rather than matched psychiatric controls, which makes it impossible to attribute observed biochemical differences to substance use alone.

The finding of elevated AST and GGT in alcohol-involved groups is directionally consistent with a large body of established literature on alcohol hepatotoxicity. The observation that cannabis-only users had lower total cholesterol and LDL-C aligns with some prior epidemiological studies, including analyses of NHANES data in the United States, which have reported similar associations, though the mechanisms remain unclear and confounders are difficult to control across studies. This study extends this line of inquiry into a previously unstudied Sub-Saharan African psychiatric population, though its design limitations prevent it from confirming or challenging prior findings with any certainty. A well-controlled prospective cohort study in this population would meaningfully advance the field.

The most consequential analytic choice was the decision to perform unadjusted group comparisons without multivariate regression controlling for sex, age, psychiatric diagnosis, and medication use. Had the researchers employed a regression model adjusting for these variables, the magnitude and possibly the direction of several findings could have changed substantially. In particular, adjusting for the dramatic sex imbalance (57% female non-users vs. 4 to 12% female substance users) might have attenuated or eliminated the hemoglobin and lipid differences. Similarly, using a psychiatric non-user control group, rather than healthy volunteers, and applying correction for multiple comparisons across the numerous biochemical outcomes tested, could have produced a markedly different pattern of statistical significance.

The most likely overinterpretation is that cannabis and alcohol use cause liver damage in psychiatric patients. This study found associations between substance use group membership and elevated liver enzymes, but its cross-sectional design cannot establish causation. More importantly, the comparison was between medicated psychiatric patients and healthy volunteers, meaning the differences could reflect psychiatric illness, pharmacological treatment, sex differences, or nutritional and lifestyle factors rather than substance use itself. Similarly, the finding of lower cholesterol in cannabis-only users should not be interpreted as evidence that cannabis lowers cholesterol; the confounders are too numerous and the comparison too uncontrolled to support that conclusion.

This study contributes rare, locally generated biochemical data from a psychiatric population in Ghana and generates biologically plausible hypotheses about hepatic stress in dual cannabis-alcohol users. It does not establish causation, cannot isolate substance effects from psychiatric illness or its pharmacological treatment, and should not be cited as evidence for screening guidelines. Its greatest value lies in motivating properly controlled prospective research in this underserved population.

Does this study prove that cannabis damages the liver?

No. The study found associations between substance use and elevated liver enzymes, but it cannot prove that cannabis caused those elevations. The participants using cannabis were also psychiatric patients likely taking medications that independently affect liver function, and the comparison group was made up of healthy people without psychiatric diagnoses. Multiple factors other than cannabis could explain the differences observed.

Should I get my liver tested if I use cannabis and alcohol together?

Periodic liver function monitoring is generally reasonable for anyone using alcohol regularly, regardless of cannabis use. This study suggests that using both substances together may be associated with higher liver enzyme levels, which is biologically plausible. Discussing this with your physician and including liver function in routine blood work is a sensible precaution, though this study alone does not establish a formal screening recommendation.

Does cannabis lower cholesterol?

This study found that cannabis-only users had lower total cholesterol and LDL-C than healthy non-users, and some prior epidemiological studies have reported similar associations. However, the difference could be explained by diet, medications, sex imbalance between the groups, or other unmeasured factors. It would be premature to view cannabis as a cholesterol-lowering agent based on this evidence.

Have thoughts on this? Share it: