Abstract:
Background: Self-motivated or intervention-based withdrawal of an addictive substance often
causes a conscious or unconscious search for a substitute, leading to substitute addiction. In
contrast, a new substance may lead to multiple addictions (i.e., polydrug use, concurrent
addiction) due to overlap between their neurochemical and behavioral factors.
Objectives: To identify multiple and substitute addictions among patients admitted to substance
abuse treatment programs in state agencies. The study also tested the suitability of the data mining
method Market Basket Analysis (MBA) to detect common drug use patterns in large-scale
datasets
Methods: Admissions data from 2019 and 2020 for patients who were aged 12 and older in the
Treatment Episode Data Set (TEDS) were analyzed. Primary, secondary, and tertiary drug use
information as self-reported by patients along with demographic data were analyzed, using
"Support%" and "confidence%" statistics of MBA to detect multiple and substitute addictions,
respectively.
Results: In 2020, of 1,416,357 patients, 31.2% used alcohol, 20.6% used heroin and 9.8% used
marijuana as a primary drug. Seven drugs (>1%) were used predominantly as either primary,
secondary, or tertiary. Alcohol users had 28% confidence in also using marijuana. Marijuana users
had 43% confidence in also using alcohol. For ages <20 years, support for alcohol >marijuana was
26% with 73% confidence in using marijuana. Marijuana and alcohol users had 18% confidence in
using cocaine or methamphetamine Pre-pandemic data had similar patterns.
Conclusion: MBA is useful for detecting common substance use patterns in large-scale datasets.
Identifying impacts of demographic characteristics on multiple and substitute addictions is
important for developing interventions that prevent these common patterns.
Keywords: Market basket analysis, Multiple addictions, Substance use, Substitute addictions
