Theses Doctoral

Drivers of Functional and non-Functional Drug Use: A Latent Class Analysis

Roberts, Eric Thomas

Drug prohibition has dramatically affected countries worldwide. It fuels violence and corruption in Latin America, and Central and Southeast Asia, and is a major contributing factor behind the United States having the highest rate of incarceration in the world. Yet there is scant evidence that prohibition reduces drug use. Despite this lack of evidence, prohibition is the preferred policy stance of governments worldwide. One of the primary justifications of prohibition is that drug use causes individuals’ harm. While there is evidence of individual harms associated with drug use there is also a literature suggesting it is possible to use drugs functionally – defined here as use with minimal impairment to mental and physical health, and social roles and expectations.

However, drug use is a politically charged topic and as such little research on functional drug use has come to prominence. The existence of persons who use drugs functionally would allow us to consider alternative approaches to drug control that address the harms that stem from both prohibition and individual use.In this dissertation I conducted three independent but related studies to explore the existence and drivers of functional drug use.

In Chapter 1 I systematically reviewed peer-reviewed literature from Ovid MEDLINE, PsycINFO, Scopus, and Web of Knowledge databases regarding functional drug use and find robust evidence that all illegal drugs can be used functionally. Drawing on the narratives of participants across the studies the typical person who uses drugs functionally is marked by three characteristics. First, they actively avoid addiction and take steps to maintain overall good physical and mental health. Second, they are socially integrated with lives that do not revolve solely around procuring and consuming drugs; hallmarks of this included holding a job, attending school, and maintaining connections to non-drug using family and friends. Third, persons who use drugs functionally take pains to avoid negative stereotypes attached to persons who use drugs – paying for their drugs with excess income, avoiding other illegal activities and attending to important socially sanctioned activities.

In Chapter 2 I used data from the Inner-City Mental Health Study Predicting HIV/AIDS, Club and Other Drug Transitions (IMPACT) study, a cross-sectional dataset of former and current persons who use drugs in New York City selected via random street intercept between 2005 and 2008, to apply the findings of our review to find participants reflective of the phenomenon of functional drug use. Using exploratory latent class analysis on questions regarding drug use behaviors I report different patterns of drug use within the IMPACT sample and regress measures of social functioning on these classes as distal outcomes to assess the functionality of each class. My solution is a 6-class model consisting of the following use patterns: former non-persons who inject drugs (PWID); former PWID; marijuana use; cocaine, crack and marijuana use; low frequency polydrug use; high frequency polydrug use.

Among the classes containing persons who use drugs currently, there was a clear pattern of relative functionality based on the probability of drug related interference and having an illegal main source of income. From most functional to least functional these were: marijuana use (2% interfering use; 5% illegal main source of income), cocaine, crack and marijuana use (48%; 31%), low frequency polydrug use (58%; 38%), and high frequency polydrug use (80%; 57%); compared to 37% of the overall sample reporting interfering use and 24% reporting having an illegal main source of income. Comparing the classes to former non-PWID, marijuana use had a lower odds of drug use interference (OR = 0.07, p-value < 0.01) whereas all other classes had significantly increased odds of drug use interference with increasing odds from former PWID (OR = 1.80, p-value = 0.04), cocaine, crack and marijuana use (OR = 4.46, p-value < 0.01), low frequency polydrug use (OR = 6.48, p-value < 0.01), and high frequency polydrug use (OR = 18.66, p-value < 0.01). Regarding main source of income there was no significant difference between the marijuana use class with the former non-PWID (OR = 0.88, p-value = 0.81). The other classes however, followed a similar step-wise pattern as for drug use interference: former PWID (OR = 2.68, p-value = 0.04), cocaine, crack and marijuana use (OR = 7.21, p-value < 0.01), low frequency polydrug use (OR = 10.08, p-value < 0.01), and high frequency polydrug use (OR = 21.30, p-value < 0.01).

In Chapter 3 I built on the results from Chapter 2 to test whether childhood physical abuse, sexual abuse, and neglect (CAN) was associated with membership in less compared with more functional drug use classes using multinomial logistic regression. This analysis builds on literature summarized in Chapter 1 suggesting non-functional drug use is associated with feelings of negative affect and a large body of work documenting associations between CAN and psychiatric and behavioral disorders generally, and drug use specifically. I report that childhood neglect is not associated with different patterns of drug use behaviors but is positively associated with having an illegal main source of income (OR = 1.40, 95%CI 1.02, 1.92). Participants experiencing physical abuse were 1.65 (95% CI 1.06-2.59) times more likely to engage in high frequency polydrug use compared to marijuana use but had no association with drug use interference after adjustment for drug use class. There were positive associations between all measures of sexual abuse with drug use interference and having an illegal main source of income. Adjustment for drug use class accounted for the association with drug use interference but not having an illegal main source of income.

This exploration of functional drug use found a strong evidence base of qualitative work supporting its existence; however, there are few extant quantitative investigations. Applying the results of our review to an epidemiologic sample I found a hierarchy of functionality related to different patterns of drug use. Moving this body of work forward requires the development of new scales to measure functional drug use to more fully characterize the phenomenon. Replication across samples will generate much needed estimates of the prevalence of functional and non-functional drug use, key data for drug policy debates.

These scales should take into account the three key dimensions outlined by participants across the studies reviewed and be applicable across various kinds of drugs and meaningful cross-culturally. I report evidence supporting an association between CAN with different patterns of drug use and reduced social functioning. Future analyses should measure other sources of childhood trauma if they are interested in the direct effect of CAN on drug use, as well as modifiers of the CAN-drug use relationship to fully characterize the phenomenon. However, it should also be noted that the model this analysis is based on, like most extant theories of use, is rooted in the moral panic over drugs that has engulfed the United States for the last 100 years.

These models treat drug use as unequivocally harmful, hence an irrational activity and therefore, implicitly, the result of some trauma. Functional drug use subverts this paradigm and considers multiple reasons for and patterns of use. While there are likely negative inducements towards less functional patterns of drug use we would do well to update our models by considering pleasure and incorporating both positive and negative inducements. New models should then be tested systematically across samples.


  • thumnail for Roberts_columbia_0054D_17398.pdf Roberts_columbia_0054D_17398.pdf application/pdf 1.12 MB Download File

More About This Work

Academic Units
Thesis Advisors
Galea, Sandro
Ph.D., Columbia University
Published Here
August 3, 2022