The chippers, the quitters, and the highly symptomatic: A 12-month longitudinal study of DSM-5 opioid- and cocaine-use problems in a community sample

Abstract

Background: Individual trajectories of drug use and drug-related problems are highly heterogeneous. There is no standard taxonomy of these trajectories, but one could be developed by defining natural categories based on changes in symptoms of substance-use disorders over time.

Methods: Our study was conducted in a community sample in Baltimore, Maryland. At baseline, all participants were using opioids and/or cocaine, but none were in treatment. Drug use and symptomatology were assessed again at 12 months (N = 115).

Results: We defined Quitters as participants who had not used for at least 30 days at follow-up (17%). For the remaining participants, we performed longitudinal cluster analysis on DSM symptom-counts, identifying three trajectory clusters: newly or persistently Symptomatic (40%) participants, Chippers (21.5%) with few symptoms, and Converted Chippers (21.5%) with improved symptom counts. Logistic regression showed that profiles of Quitters did not resemble Chippers, but instead resembled Symptomatic participants, having high probability of disorderly home neighborhood, nonwhite race, and negative mood. Quitters tended to have two protective factors: initiating opioid-agonist treatment during the study (reffect = 0.25, CL95 0.02-0.48) and lack of polydrug use (reffect = 0.25, CL95 0.004-0.49). Converted Chippers tended to be white, with orderly home neighborhoods and less negative mood (reffects 0.24 to 0.31, CL95 0.01-0.54).

Conclusions: Changes in DSM symptomology provided a meaningful measure of individual trajectories. Quitters shared psychosocial characteristics with Symptomatic participants, but not with participants who continued to use with few symptoms. This suggests that Quitters abstained out of necessity, not because their problems were less severe.

Trial registration: ClinicalTrials.gov NCT01571752.

Keywords: Cocaine; DSM-5; Opioid; Substance use disorder; Trajectory.

Conflict of interest statement

The authors report no conflicts of interest.

Copyright © 2019 Elsevier Ltd. All rights reserved.

Figures

Figure 1.

Spaghetti plots showing symptom-count trajectories for the 115 individual participants assessed at two visits 12 months apart. Trajectories represent the number of DSM-5 SUD symptoms (out of a possible 11) endorsed at each visit. Each thin line represents one participant, thicker lines are in proportion to the number of participants showing the same trajectory. Note that: 1) participants who became abstinent over the course of the study were classified as Quitters; and 2) participants who had symptoms for both opioids and cocaine (dual use) were included in the Symptomatic Cluster if they had persistent or emergent symptoms for either drug, even if they had few or no symptoms related to the other. Rows show baseline drug-use categories: 1: opioids only (n = 45, 39%). 2: cocaine only (n = 28, 24%). 3 and 4: dual use (n = 42, 37%). Columns show clusters of 12-month trajectories: 1: Converted Chippers—high or moderate symptomatology that resolved by the 12-month visit (n = 25, 21.5%). 2: Chippers—low or no symptomatology at both visits (n = 25, 21.5%). 3: Symptomatic Use—high or moderate symptomatology at both visits; dual use was assigned to this group if their trajectory for either drug class met criteria (n = 46, 40%). 4: Quitters—defined outside of cluster analysis; the criterion was 30-day abstinence at the 12-month visit, regardless of symptom trajectory (n = 19, 17%).

Figure 2.

Heatmap showing changes in specific DSM-5 symptoms related to opioid use and/or cocaine use for each participant. Each column of cells shows data for one participant. Cell color indicates the trajectory for a single symptom, based on its presence vs. absence at the baseline visit and 12-month visit: (1) improved if endorsed only at baseline; (2) not endorsed if not endorsed at either visit; (3) persistent if endorsed at both visits; or (4) emergent if endorsed only at the 12-month visit. Green “Q” indicates a participant who reported being abstinent at Visit 2 and was therefore treated as a member of the Quitters “cluster” in regression analysis. Blocks of participants separated by white space within the Converted Chippers, Chippers and Symptomatic clusters represent subclusters.

Figure 3.

Summary values of variables included in the multinomial logistic regression model, showing percentage in each cluster (including the “cluster” of Quitters, defined prior to clustering). (a) number of participants in each cluster; (b) percentage of participants self-described as being white; (c) percentage of participants who initiated opioid agonist therapy between the baseline and 12-month visits;* (d, e, f) percentage of participants who enrolled in the study with use of opioids, cocaine, or both, respectively; (g) summary rating on the Profile of Mood States (POMS) questionnaire, referring to the 30-day period prior to joining the study, with higher values indicating a “better” state; (h) average value of residences in the participant’s home neighborhood; and (i) Neighborhood Inventory for Environmental Typology (NIfETy) ratings, with higher values indicating a more orderly neighborhood environment. *77.3% of the overall sample expressed interest in treatment at Visit 1, 87% of Symptomatic, 78.9% of Quitters, 92% of Converted Chippers and 44% of Chippers.

Figure 4.

Results of multinomial logistic regression on external criterion variables (the same ones described for Fig. 4) as predictors of membership in each cluster, with Symptomatic Use as the reference group. Each variable is plotted so that its ostensibly protective levels (more stable moods, more orderly neighborhood, etc.) are on the top half of the y axis. The left panel shows odds ratios (points) and 95% confidence intervals (lines). Because the magnitude of the odds ratios is affected by the units of measure of the predictor, the right panel shows effect sizes (reffect) and 95% confidence intervals for the same analysis. Confidence intervals are shown in only one direction to emphasize inclusion or exclusion of the null: predictors whose confidence intervals do not cross the central gray line had statistically significant associations with cluster membership. All tests were two-tailed. In general, Quitters were similar to ongoing Symptomatic Use group (the reference category) except for having used agonist treatment between the baseline and 12-month visits. In contrast, Chippers and Converted Chippers had significantly higher likelihoods of several person-level and environmental protective factors than ongoing Symptomatic Use.