Dynamic transitions between marijuana use and cigarette smoking among US adolescents and emerging adults

ABSTRACT

Background: Marijuana and tobacco are considered two closely related substances. It is of great significance to understand the mutual impact of marijuana and cigarette use when more states in the US have legalized marijuana use. Objective: This study aims to investigate the transitions between marijuana and cigarette use among adolescents and emerging adults. Method: Guided by the probabilistic discrete events systems (PDES) theory, a five-stage model with 21 transition paths was constructed to quantify dynamic transitions between marijuana and cigarette use. The five stages were NU: Never-user, MU: Current marijuana user, CU: Current cigarette user, MCU: Current marijuana–cigarette user, and FU: Former-user. The proposed five-stage PDES model was tested using the 2013 National Survey on Drug Use and Health data (N = 26,665, 50.45% male). Transition probabilities were estimated using the Moore–Penrose generalized inverse matrix method. Result: Among the adolescents, 51.14% of the CUs transited to use marijuana (MCU/MU), higher than the proportion of those who first used marijuana and then transferred to cigarettes (MCU/CU) (41.66%). The quitting rates for MUs, CUs and MCUs were 29.38%, 25.93% and 27.76%, respectively. Of the total FUs, 31.90% transited to MUs, 17.06% to CUs, and 17.39% to MCUs. Among the young adults, more people progressed from MUs to CUs. Transition probabilities by single year of age were also estimated. Conclusion: This is the first study to quantify marijuana–cigarette transitions. Study findings indicate more cigarette-to-marijuana transitions for adolescents and more marijuana-to-cigarette transitions for emerging adults. Future intervention programs should consider this age-related difference in marijuana–cigarette use transitions.

KEYWORDS:

• PDES
• marijuana
• cigarette
• transition probability
• state probability

Acknowledgments

We thank Dr. Xingdi Hu for his assistance in programing R codes to solve the proposed PDES models with M–P generalized inverse matrix method, and we also thank Ms. Amy Elliott to help us revise the language and improve the manuscript.

Additional information

Funding

This work is funded by the National Institutes of Health (1RO1DA022730, U01-AA020797 and U24AA02202). The funding source has no role in study design, data collection, analysis and results interpretation, preparation and final approval of the manuscript, and decision to submit the manuscript for publication.