Longitudinal Relationships Between Cannabis and Tobacco Use and Symptom Severity in Individuals at Clinical High Risk for Psychosis

In a longitudinal study of individuals at clinical high risk for psychosis, current tobacco and cannabis use predicted increased future anxiety, while heavier baseline cannabis use was associated with slower improvement in negative symptoms and greater role function deterioration.

The Gist

Imagine a group of young people who are standing on the edge of a cliff, looking down into a foggy valley. In the medical world, these are people at "Clinical High Risk" (CHR-P) for developing psychosis (a condition where reality gets distorted, often leading to schizophrenia). They haven't fallen yet, but they are wobbling, and their symptoms are getting a bit shaky.

Now, imagine that many of these young people are trying to steady themselves by smoking cigarettes or using marijuana. The big question doctors have been arguing about for years is: Are they smoking to calm their nerves (self-medication), or is the smoking actually pushing them closer to the edge?

This study, which followed 764 of these young people over two years, decided to stop guessing and start tracking the timeline. They asked three simple questions:

1. If they smoke now, will they feel worse later?
2. If they feel worse now, will they smoke more later?
3. Did how much they smoked at the very beginning predict how their whole journey would go?

Here is what they found, explained with some everyday analogies:

1. The "Anxiety Loop" (Smoking makes anxiety worse)

Think of anxiety like a leaky bucket. When you are anxious, you feel like you need to plug the hole. Many people think smoking or vaping is the "plug."

The study found that for these young people, smoking isn't a plug; it's actually pouring more water into the bucket.

• The Finding: Whether it was tobacco or cannabis, the more they used at one visit, the more anxious they were at the next visit.
• The Metaphor: It's like trying to put out a fire with gasoline. They might feel a momentary sense of relief, but the next time they check the fire, it's actually bigger. The study suggests that for these at-risk youths, substance use is a short-term fix that creates a long-term problem with anxiety.

2. The "Social Paradox" (Doing well leads to smoking)

This was the most surprising part. Usually, we think people smoke because they are lonely or depressed. But this study found the opposite was true for social life.

• The Finding: Young people who had better social skills and were doing well with friends were actually more likely to smoke more at the next visit.
• The Metaphor: Imagine a party. If you are the life of the party, feeling confident and connected, you are more likely to be offered a drink or a joint, and you are more likely to accept it. The study suggests that social success opens the door to substance use.
• The Twist: Interestingly, using cannabis also seemed to predict better social functioning the next time around. It's a two-way street: Being social leads to smoking, and smoking (in the short term) might make them feel more social. It's a complex dance where "fitting in" and "using substances" are tangled together.

3. The "Head Start" Problem (Starting heavy means a harder road)

The researchers looked at who started the study with the heaviest smoking habits and tracked their progress over two years.

• The Finding: Those who started with heavy cannabis use didn't just stay the same; they got stuck.
  • Their negative symptoms (like lack of motivation, not feeling emotions, or withdrawing) got better much slower than non-smokers.
  • Their ability to function in school or work got worse.
  • Their visual learning (how they process what they see) got worse.
• The Metaphor: Imagine two runners starting a race. One runner is wearing heavy boots (heavy cannabis use), and the other is barefoot. Even if both runners try to get faster, the one in the heavy boots is dragging their feet. They don't necessarily fall down immediately, but they can't improve their speed, and their legs (their daily life and motivation) start to ache and fail faster than the others.

What Does This Mean for Real Life?

The old idea was: "These kids are smoking because they are scared or depressed, so if we treat the depression, the smoking will stop."

This study says: "Not quite."

1. Anxiety is the real driver: Smoking makes anxiety worse, creating a vicious cycle. We need to treat the anxiety directly, not just tell them to stop smoking.
2. Socializing is a double-edged sword: Helping these kids make friends is great, but we also need to teach them how to say "no" when their new friends offer them substances. Being popular shouldn't mean being high.
3. Early warning sign: If a kid at high risk is already smoking heavily, doctors should know that this kid is on a harder path to recovery. They need extra support to keep their motivation and daily life on track.

In a nutshell: For young people at risk of psychosis, smoking isn't a magic cure for their problems. It's more like a heavy backpack that makes the climb up the mountain harder, especially if they start the journey already carrying a heavy load. The goal is to help them put the backpack down so they can run faster toward a healthy future.

Technical Summary

1. Problem Statement

Individuals at Clinical High Risk for Psychosis (CHR-P) exhibit high rates of tobacco (up to 64%) and cannabis (up to 48.7%) use. A critical unresolved question in the field is the directionality of the relationship between substance use and clinical outcomes:

• Causal Model: Does substance use drive symptom exacerbation and the transition to psychosis?
• Self-Medication Model: Do emerging psychiatric symptoms (e.g., anxiety, psychosis) drive individuals to use substances to cope?

Previous research has yielded conflicting results, often due to reliance on cross-sectional designs (which cannot establish causality) or binary classifications of substance use (use vs. non-use), which fail to capture dose-response relationships. This study aims to disentangle these temporal associations using longitudinal data to determine if substance use predicts clinical deterioration or if clinical status predicts substance use escalation.

2. Methodology

Study Design & Cohort:

• Data Source: North American Prodrome Longitudinal Study 2 (NAPLS2).
• Participants: 764 individuals meeting CHR-P criteria (excluded healthy controls).
• Duration: Longitudinal follow-up over 2 years with assessments every 6 months.
• Exclusions: Data from visits where participants converted to psychosis were excluded to maintain consistent time intervals for modeling.

Measures:

• Substance Use: Assessed via the Alcohol Use Scale/Drug Use Scale (AUS/DUS). Frequency was measured on an ordinal scale (e.g., cigarettes per day for tobacco; frequency per month/week for cannabis) rather than a binary variable.
• Clinical Symptoms:
  • Psychosis: Scale of Psychosis-Risk Symptoms (SOPS) covering positive, negative, disorganized, and general symptoms.
  • Affective: Self-Rating Anxiety Scale (SAS), Social Interaction Anxiety Scale (SIAS), and Calgary Depression Scale for Schizophrenia (CDSS).
• Functioning: Global Functioning: Social (GFS) and Role (GFR) scales.
• Cognition: Six domains from the MATRICS Consensus Cognitive Battery (MCCB).

Statistical Approach:
The authors employed Linear Mixed Effect (LME) models with subject-level random intercepts to account for repeated measures. Three specific analytical frameworks were used:

1. Lead-Lag Analysis (Current Use $\to$ Future Symptoms): Predicting symptoms at Visit $T+1$ based on substance use at Visit $T$, controlling for current symptoms, age, sex, and antipsychotic dosage.
2. Lead-Lag Analysis (Current Symptoms $\to$ Future Use): Predicting substance use at Visit $T+1$ based on symptoms at Visit $T$.
3. Baseline Trajectory Analysis: Modeling the interaction between baseline substance use and time to determine if initial use frequency altered the slope of symptom recovery or deterioration over the 2-year period.

• Correction: False Discovery Rate (FDR) adjustments were applied to p-values for multiple comparisons.

3. Key Contributions

• First Longitudinal Lead-Lag Analysis in CHR-P: This is the first study to use longitudinal modeling to explicitly test temporal precedence between substance use and clinical presentation in a CHR-P cohort.
• Dose-Response Modeling: Unlike prior studies treating use as binary, this study utilized frequency data to model dose-response relationships.
• Differentiation of Anxiety vs. Psychosis: The study specifically isolates the relationship between substances and anxiety, finding it distinct from the relationship with core psychotic symptoms.
• Bidirectional Social Dynamics: It identifies a complex, bidirectional loop between social functioning and substance use, challenging the notion that social improvement is purely protective against substance use.

4. Key Results

A. General Trajectories (2-Year Overview):

• Clinical Improvement: Psychiatric symptoms (total, anxiety, depression), social/role functioning, and global cognition all showed significant improvement over the 2-year period.
• Stable Substance Use: The frequency of tobacco and cannabis use did not change significantly over the 2 years for the cohort.

B. Current Use Predicts Future Symptoms (Short-Term):

• Anxiety: Heavier current tobacco use ($\beta=0.712, p=0.033$) and cannabis use ($\beta=0.511, p=0.018$) predicted worse anxiety at the next visit.
• Social Functioning: Heavier current cannabis use predicted better social functioning at the next visit ($\beta=0.060, p=0.009$).
• Null Findings: Current substance use did not predict future positive/negative psychotic symptoms, depression, role functioning, or cognitive performance in the short term.

C. Current Symptoms Predict Future Use:

• Social Functioning: Better current social functioning predicted heavier tobacco ($\beta=0.178, p<0.001$) and cannabis ($\beta=0.162, p<0.001$) use at the next visit.
• Null Findings: Current levels of anxiety, depression, or psychosis symptoms did not predict an increase in substance use at the next visit.

D. Baseline Use Predicts Long-Term Trajectory:

• Cannabis: A significant baseline cannabis-by-time interaction was found. Heavier baseline cannabis use predicted:
  • Slower recovery of negative symptoms ($\beta=0.159, p=0.0017$).
  • Deterioration of role functioning (work/school performance) ($\beta=-0.046, p=0.018$).
  • Deterioration of visual learning cognition.
• Post-Hoc on Positive Symptoms: While baseline cannabis did not alter the trajectory (slope) of positive symptoms, post-hoc analysis confirmed that heavier baseline users had persistently higher positive symptom severity throughout the study.
• Tobacco: Heavier baseline tobacco use predicted slower improvement in social functioning.

5. Significance and Clinical Implications

Rejection of Simple Self-Medication:
The study does not support the hypothesis that CHR-P individuals increase substance use to self-medicate worsening anxiety or psychosis. Instead, symptom severity did not predict future substance use escalation.

The "Social Paradox":
The findings reveal a counter-intuitive dynamic: better social functioning facilitates future substance use. This suggests that social interventions aimed at improving peer connections in CHR-P youths may inadvertently increase exposure to substance-using environments. Clinicians must equip patients with skills to navigate social settings without resorting to substances.

Anxiety as a Primary Target:
Since substance use predicts the exacerbation of anxiety (but not necessarily psychosis symptoms in the short term), early interventions should simultaneously target anxiety reduction and substance use cessation to break the cycle.

Prognostic Value of Baseline Cannabis:
Baseline cannabis frequency serves as a robust prognostic marker. Heavier use at enrollment predicts a poorer long-term trajectory for negative symptoms and functional recovery, even if the rate of symptom change remains similar to non-users. This suggests that baseline use should be incorporated into treatment planning and prognosis formulation.

Conclusion:
The study shifts the clinical focus toward a model where substance use in CHR-P is tightly linked to anxiety levels and social functioning dynamics, rather than being a direct driver of acute psychotic episodes. It advocates for biopsychosocial interventions that address anxiety, teach social navigation skills, and utilize baseline substance use as a key indicator for long-term recovery potential.