Risky decision‐making as an antecedent or consequence of adolescent cannabis use: findings from a 2‐year longitudinal study

Abstract

Background and aims Although poor decision‐making (DM) has been correlated with problematic cannabis use (CU), cross‐sectional designs make it difficult to determine whether poor DM represents an antecedent and/or consequence of CU. The current study measured bidirectional associations between CU and DM among adolescents over 2 years and compared these findings to those observed with episodic memory, which is consistently reported as a consequence of CU. We also measured the role of DM as a risk factor for cannabis use disorder (CUD) onset. Design Two‐year longitudinal study with five bi‐annual assessments. Participants Participants were 401 adolescents aged 14–17 years at baseline. Setting Miami, Florida, USA. Measurements CU frequency and CUDs were assessed at each time‐point through the Drug Use History Questionnaire and Structured Clinical Interview for DSM‐IV, respectively. Neurocognition was assessed at odd time‐points throughout the Iowa Gambling Task, Game of Dice Task and Cups Task [decision‐making (DM)] and the Wechsler Memory Scale IV and California Verbal Learning Test II (episodic memory). We used latent growth curve modeling to examine bidirectional influences between CU and neurocognition over time. We applied discrete time survival analyses to determine whether baseline DM predicted CUD onset. Findings Greater lifetime CU frequency was associated with poorer episodic memory at baseline (bs = −14.84, −16.44, Ps = 0.038, 0.021). Greater CU escalation predicted lesser gains in immediate episodic memory (b = −0.05, P = 0.020). Baseline DM did not predict CU escalation (b = 0.07, P = 0.421), nor did escalation in CU predict changes in DM (b = 0.02, P = 0.352). Baseline DM also did not predict CUD onset (adjusted OR = 1.01, 95% confidence interval = 0.98–1.06). Conclusions This study replicates findings that poorer episodic memory in adolescents appears to be a consequence of cannabis use, even among adolescents at earlier stages of use. Poor decision‐making does not appear to be either a consequence of or a risk factor for escalating cannabis use or onset of cannabis use disorder among adolescents.

An exploratory follow-up study of cannabis use and decision-making under various risk conditions within adolescence.

“…The unconditional linear growth models of CU frequency and CU-related problems using the MPS showed adequate to excellent fit, excluding the RMSEA for the CU frequency (RMSEA = 0.197; Table 3). As reported in our original study (Pacheco-Colón et al, 2022), CU frequency increased on average, with significant variability in the intercept and slope. In contrast, CU-related problems decreased on average, with significant variability in the intercept but not slope.…”

“…We also examined the bidirectional associations between DM performance across the four indices and CU-related problems as well as whether DM performance at baseline predicted CUD onset at some point in the study. Considering the aforementioned distinctions among these DM tasks, previous findings in this cohort that overall DM performance did not relate to predict CU outcomes (Pacheco-Colón et al, 2022) and known differences in DM performance between adolescents and adults (Almy et al, 2018; Hooper et al, 2004), we did not expect performance on all DM indices to predict changes in the CU-related outcomes. Rather, we considered the aims of the present study to be exploratory in nature, and thus did not preregister any hypotheses.…”

“…Additionally, our original intent was to replicate prior analyses (Pacheco-Colón et al, 2022) examining the bidirectional associations between a broad construct of DM and CU-related outcomes using the same covariates but with DM performance differentiated by task. However, it is possible that the omission of other covariates not included within the original study, including mental health concerns and positive toxicology other than THC (e.g., cocaine, amphetamines) could have confounded our current significant results.…”

“…The present study examined the bidirectional associations between separate indices of risky DM and CU among adolescents. A prior study using the same cohort of adolescents examined these associations using a composite score of DM across four indices and did not find longitudinal associations between DM and CU (Pacheco-Colón et al, 2022). However, given the nonsignificant-to-small correlations among the DM indices (Pacheco-Colón et al, 2022), we were interested in examining whether the association between risky DM and escalation in CU frequency varied across specific DM tasks.…”

“…A prior study using the same cohort of adolescents examined these associations using a composite score of DM across four indices and did not find longitudinal associations between DM and CU (Pacheco-Colón et al, 2022). However, given the nonsignificant-to-small correlations among the DM indices (Pacheco-Colón et al, 2022), we were interested in examining whether the association between risky DM and escalation in CU frequency varied across specific DM tasks. Multiple theories have supported the differentiation of DM tasks across explicit versus ambiguous risk conditions (Buelow & Suhr, 2009; Dunn et al, 2006; Huettel et al, 2006) and across explicit risk under gain versus loss domains (Kim-Spoon et al, 2017; Levin et al, 2007).…”