Chloe Burke

Background: Systematic reviews are essential for evidence-based decision making in health sciences but require substantial time and resource for manual processes, particularly title and abstract screening. Recent advances in machine learning and large language models (LLMs) have demonstrated promise in accelerating screening with high recall but are often limited by modest gains in efficiency, mostly due to the absence of a generalisable stopping criterion. Here, we introduce and report preliminary findings on the performance of a novel semi-automated active learning system, JARVIS, that integrates LLM-based reasoning using the PICOS framework, neural networks-based classification, and human decision-making to facilitate abstract screening. Methods: Datasets containing author-made inclusion and exclusion decisions from six published systematic reviews were used to pilot the semi-automated screening system. Model performance was evaluated across recall, specificity and area under the curve precision-recall (AUC-PR), using full-text inclusion as the ground truth. Estimated workload and financial savings were calculated by comparing total screening time and reviewer costs across manual and semi-automated scenarios. Results: Across the six review datasets, recall ranged between 98.2% and 100%, and specificity ranged between 97.9% and 99.2% at the defined stopping point. Across iterations, AUC-PR values ranged between 83.8% and 100%. Compared with human-only screening, JARVIS delivered workload savings between 71.0% and 93.6%. When a single reviewer read the excluded records, workload savings ranged between 35.6 % and 46.8%. Conclusion: The proposed semi-automated system substantially reduced reviewer workload while maintaining high recall, improving on previously reported approaches. Further validation in larger and more varied reviews, as well as prospective testing, is warranted.

Background and aims: Cannabis is the most widely used illicit drug worldwide and is often co‐used with tobacco, the leading cause of preventable death. Although cannabis and tobacco have distinct neurobiological actions, their associations with brain volumes are unclear. We aimed to review studies investigating cannabis use, tobacco use, their co‐use and brain volume and triangulate evidence across different study designs. Methods: A systematic review and meta‐analysis preregistered on PROSPERO (CRD42022356982) and reported according to PRISMA 2020 guidelines. We searched SCOPUS, PubMed and PsycINFO up to 5 September 2024 for studies investigating cannabis use, tobacco use, co‐use and brain volume. Cross‐sectional, longitudinal and Mendelian randomisation studies were included. The outcome was brain volume of global, cortical and subcortical regions. We extracted adjusted and unadjusted estimates. Random effects meta‐analyses were stratified by exposure and study design across 33 brain regions. Risk of bias was assessed using a modified version of the Newcastle‐Ottawa scale. Results: Searches yielded 103 studies: 57 investigated cannabis use, 45 investigated tobacco use and one investigated tobacco and cannabis co‐use. Seventy‐seven studies were included in meta‐analysis ( n = 72 798), 44 ( n = 18 247) in the cross‐sectional cannabis analysis, 30 ( n = 51 194) for tobacco cross‐sectional and four ( n = 3357) in the tobacco longitudinal analysis. Meta‐analysis of adjusted estimates from cross‐sectional studies (k denotes the number of independent studies) indicated smaller amygdala volumes [k = 17, g = 0.13, 95% confidence interval (CI) = 0.03, 0.23)] in people who use cannabis compared with controls. Relative to controls, people who smoked tobacco had smaller volumes in the amygdala (k = 5, g = 0.17, 95% CI = 0.04, 0.31), insula (k = 5, g = 0.17, 95% CI = 0.06, 0.27), pallidum (k = 5, g = 0.17, 95% CI = 0.13, 0.21) and total grey matter volume (TGMV) (k = 7, g = 0.17, 95% CI = 0.04, 0.30). Longitudinal studies indicated a larger decrease in TGMV in people who smoke tobacco (k = 5, g = 0.05, 95% CI = 0.01, 0.10) relative to controls. Conclusions: Cannabis use appears to be associated with smaller volume in the amygdala. Tobacco use appears to be associated with smaller amygdala, insula, pallidum and total grey matter volume.

Background and Aims: Cocaine use disorder (CUD) is an increasingly widespread concern worldwide. Various pharmacological treatments have been investigated for CUD, but their efficacy remains unclear, and none have been licensed for treatment. Therefore, we assessed the comparative effectiveness, safety, and acceptability of pharmacological interventions for the treatment of CUD and prevention of relapse. Design: Systematic review and network meta-analyses of double-blind randomized controlled trials (RCTs). PROSPERO (CRD42024596434). We assessed risk of bias using the Cochrane Risk of Bias 2 tool and certainty of evidence was assessed using the Confidence in Network Meta-analysis framework. Setting: Any in/outpatient setting, with no restriction on geographic location. Participants: Adults with diagnosed CUD. Participants with co-occurring substance use disorders and/or common mental health conditions were eligible. Interventions: Pharmacologicalinterventions specifically tested for the treatment of CUD with a minimum follow-up of 4 weeks. Measurements: We assessed 12 effectiveness, three acceptability, and two safety outcomes. Our primary outcomes were continuous abstinence, point abstinence, dropout for any reason, and serious adverse events. We also examined longest duration of continuous abstinence, extent of cocaine use, craving, severity of dependence, dropout due to adverse events, adherence, and mortality. Findings: We included 218 reports of 163 studies (14871 participants) assessing 89 unique medications, grouped into 8 categories. Most study results had some concerns or high risk of bias, and the results of syntheses were generally judged to be very low certainty evidence. As such, the findings ought to be interpreted with caution. No single treatment type showed consistent beneficial effects across all effectiveness outcomes and there was limited evidence that results varied by intervention type or presence of co-occurring comorbidity. Conclusions: Despite continued work in this area, effective pharmacological treatments for CUD remain elusive.

Background and aims: There is growing evidence that tobacco smoking causes depression, but it is unclear which constituents of tobacco smoke (e.g. nicotine, carbon monoxide) may be responsible. We used Mendelian randomisation (MR) to measure the independent effect of nicotine on depression, by adjusting the effect of circulating nicotine exposure [via nicotine metabolite ratio (NMR)] for the overall effect of smoking heaviness [via cigarettes per day (CPD)] to account for the non‐nicotine constituents of tobacco smoke. Design: Univariable MR and multivariable MR (MVMR) were used to measure the total and independent effects of genetic liability to NMR and CPD on major depressive disorder (MDD). Our primary method was inverse variance weighted (IVW) regression, with other methods as sensitivity analyses. Setting and participants: For the exposures, we used genome‐wide association study (GWAS) summary statistics among European ancestry individuals for CPD ( n = 143 210) and NMR ( n = 5185). For the outcome, a GWAS of MDD stratified by smoking status was conducted using individual‐level data from UK Biobank ( n = 35 871–194 881). Measurements: Genetic variants associated with NMR ( n = 6) and CPD ( n = 53). Findings: Univariable MR‐IVW indicated a causal effect of CPD on MDD [odds ratio (OR) = 1.13, 95% confidence interval (CI) = 1.04–1.23, P = 0.003] but no clear evidence for an effect of NMR on MDD (OR = 0.98, 95% CI = 0.97–1.00, P = 0.134). MVMR indicated a causal effect of CPD on MDD when accounting for NMR (IVW: OR = 1.19, 95% CI = 1.03–1.37, P = 0.017; Egger: OR = 1.13, 95% CI = 0.89–1.43, P = 0.300) and weak evidence of a small effect of NMR on MDD when accounting for CPD (IVW: OR = 0.98, 95% CI = 0.96–1.00, P = 0.057; Egger: OR = 0.98, 95% CI = 0.96–1.00, P = 0.038). Conclusions: The role of nicotine exposure in risk of depression cannot be entirely dismissed. However, the causal effect of tobacco smoking increasing depression risk appears to be largely independent of circulating nicotine exposure, which implies the role of alternative causal pathways.

Background: Observational studies consistently report associations between tobacco use, cannabis use and mental illness. However, the extent to which this association reflects an increased risk of new-onset mental illness is unclear and may be biased by unmeasured confounding.Methods: A systematic review and meta-analysis (CRD42021243903). Electronic databases were searched until November 2022. Longitudinal studies in general population samples assessing tobacco and/or cannabis use and reporting the association (e.g. risk ratio [RR]) with incident anxiety, mood, or psychotic disorders were included. Estimates were combined using random-effects meta-analyses. Bias was explored using a modified Newcastle–Ottawa Scale, confounder matrix, E-values, and Doi plots.Results: Seventy-five studies were included. Tobacco use was associated with mood disorders (K = 43; RR: 1.39, 95% confidence interval [CI] 1.30–1.47), but not anxiety disorders (K = 7; RR: 1.21, 95% CI 0.87–1.68) and evidence for psychotic disorders was influenced by treatment of outliers (K = 4, RR: 3.45, 95% CI 2.63–4.53; K = 5, RR: 2.06, 95% CI 0.98–4.29). Cannabis use was associated with psychotic disorders (K = 4; RR: 3.19, 95% CI 2.07–4.90), but not mood (K = 7; RR: 1.31, 95% CI 0.92–1.86) or anxiety disorders (K = 7; RR: 1.10, 95% CI 0.99–1.22). Confounder matrices and E-values suggested potential overestimation of effects. Only 27% of studies were rated as high quality.Conclusions: Both substances were associated with psychotic disorders and tobacco use was associated with mood disorders. There was no clear evidence of an association between cannabis use and mood or anxiety disorders. Limited high-quality studies underscore the need for future research using robust causal inference approaches (e.g. evidence triangulation).

Background: Cannabis and tobacco use are consistently associated with major depressive disorder (MDD) in conventional observational studies. However, these substances are often co-used, and the independent causal role of cannabis use in risk of MDD remains unclear. Methods: Univariable and multivariable MR (MVMR) were used to explore the total and independent causal effects of genetic liability to tobacco use and cannabis use on MDD. Our primary estimator was the inverse-variance weighted (IVW) method, with other methods as sensitivity analyses. For the exposures, we used genome-wide association study (GWAS) summary statistics among European ancestry individuals for several tobacco use (i.e., smoking initiation, smoking continuation, smoking heaviness) and cannabis use (i.e., cannabis initiation, cannabis use disorder [CUD]) phenotypes. For the outcome, a GWAS of MDD was conducted using individual-level data from UK Biobank. Results: Univariable MR indicated a causal effect of smoking initiation on MDD (odds ratio [OR] ^IVW = 1.34, 95% confidence interval [CI] = 1.27 – 1.42), with consistent but weaker evidence for smoking continuation (OR ^IVW = 1.13, 95% CI = 0.93 – 1.37) and smoking heaviness (OR ^IVW = 1.15, 95% CI = 0.99 – 1.33). There was no clear evidence for a causal effect of cannabis initiation on MDD (OR ^IVW = 1.00, 95% CI = 0.91– 1.11). Univariable MR indicated some evidence for a causal effect of CUD on MDD (OR ^IVW = 1.14, 95% CI = 1.04 – 1.25), which attenuated to the null when adjusting for liability to smoking initiation (OR ^MVMR-IVW = 1.03, 95% CI = 0.97 – 1.08). Conclusions: This study provides limited evidence for an independent causal effect of cannabis use on MDD, and stronger evidence for an independent causal effect of tobacco use on MDD. Analyses were limited by low power, and future research should triangulate these findings with results from high-quality observational studies.

Background and aims: There is growing evidence that tobacco smoking causes depression, but it is unclear which constituents of tobacco smoke (e.g., nicotine, carbon monoxide) may be responsible. We used Mendelian randomisation (MR) to examine the independent effect of nicotine on depression, by adjusting the effect of nicotine exposure (via nicotine metabolite ratio [NMR]) for the overall effect of smoking heaviness (via cigarettes per day [CPD]) to account for the non-nicotine constituents of tobacco smoke. Design: Univariable MR and multivariable MR (MVMR) were used to explore the total and independent effects of genetic liability to NMR and CPD on major depressive disorder (MDD). Our primary method was inverse variance weighted (IVW) regression, with other methods as sensitivity analyses. Setting and participants: For the exposures, we used genome-wide association study (GWAS) summary statistics among European ancestry individuals for CPD (n=143,210) and NMR (n=5,185). For the outcome, a GWAS of MDD stratified by smoking status was conducted using individual-level data from UK Biobank (n=35,871-194,881). Measurements: Genetic variants robustly associated with NMR and CPD. Findings: Univariable MR indicated a causal effect of CPD on MDD (odds ratio [OR] ^IVW =1.13, 95% confidence interval [CI]=1.04-1.23, P =0.003) but no clear evidence for an effect of NMR on MDD (OR ^IVW =0.98, 95% CI=0.97-1.00, P =0.134). MVMR indicated a causal effect of CPD on MDD when accounting for NMR (OR ^MVMR-IVW= 1.19, 95% CI=1.03-1.37, P =0.017; OR ^MVMR-EGGER= 1.13, 95% CI=0.89-1.43, P =0.300) and weak evidence of a small effect of NMR on MDD when accounting for CPD (OR ^MVMR-IVW= 0.98, 95% CI=0.96-1.00, P =0.056; OR ^MVMR-EGGER= 0.98, 95% CI=0.96-1.00, P =0.038). Conclusions: The causal effect of tobacco smoking on depression appears to be largely independent of nicotine exposure, which implies the role of alternative causal pathways.

Importance: Traditional observational epidemiological studies have consistently found an association between tobacco use, cannabis use and subsequent mental ill-health. However, the extent to which this association reflects an increased risk of new-onset mental ill-health is unclear and may be biased by unmeasured confounding. Objective: To examine the association between cannabis use, tobacco use and risk of incident mood, anxiety, and psychotic disorders, and explore risk of bias. Data Sources: CINAHL, Embase, MEDLINE, PsycINFO and ProQuest Dissertation and Theses were searched from inception until November 2022, in addition to supplementary searches. Study Selection: Longitudinal studies assessing tobacco use and cannabis use and their association with incident mood, anxiety or psychotic disorders were included. Studies conducted in populations selected on health status (e.g., pregnancy) or other highly-selected characteristics (e.g., incarcerated persons) were excluded. Data Extraction and Synthesis: A modified Newcastle Ottawa Scale was used to assess study quality. The confounder matrix and E-Values were used to assess potential bias due to unmeasured confounding. Summary risk ratios (RR) were calculated in random-effects meta-analyses using the generic inverse variance method. Main Outcome(s) and Measure(s): Exposures were measured via self-report and defined through status (e.g., current use) or heaviness of use (e.g., cigarettes per day). Outcomes were measured through symptom-based scales, interviews, registry codes and self-reported diagnosis or treatment. Effect estimates extracted were risk of incident disorders by exposure status. Results: Seventy-five out of 27789 records were included. Random effects meta-analysis demonstrated a positive association between tobacco use and mood disorder (RR:1.39, 95%CI:1.30–1.47) and psychotic disorder (RR:3.45, 95%CI:2.63-4.53), but not anxiety disorder (RR:1.21, 95%CI:0.87–1.68). Cannabis use was positively associated with psychotic disorders (RR:3.19, 95%CI:2.07-4.90), but not mood disorders (RR:1.31, 95%CI:0.92-1.86) or anxiety disorders (RR:1.10, 95%CI:0.99-1.22). Confounder matrix and E-value assessment indicated estimates were moderately biased by unmeasured confounding. Conclusions and Relevance: This systematic review and meta-analysis presents evidence for a longitudinal, positive association between both substances and incident psychotic disorders and tobacco use and mood disorders. There was no evidence to support an association between cannabis use and common mental health conditions. Existing evidence across all outcomes was limited by inadequate adjustment for potential confounders. Future research should prioritise methods allowing for stronger causal inference, such as Mendelian randomization and evidence triangulation.