Research Article Review: Mathematically Modeling Substance Use Disorder Symptoms During Addiction Recovery

Here’s the link for the research paper I’ll be discussing in this post.

I’ll start by saying that in my estimation this paper is important in at least 3 ways:

One: In this study, “recovery” is not merely defined as “abstinence”, nor is abstinence their only marker of recovery.

Two: The study examines…

markers of SUD problems,
and markers of recovery,
while using relatively novel statistical techniques,
to look at the effect of key variables upon each other, in clusters, over time.

Three: This entire approach to research in addictions is relatively rare.

The reader probably knows that all too commonly, research examining clinical care of SUDs has merely aimed at comparing pre- and post-intervention differences in substance use, or abstinence, and perhaps a few other variables.

But in this paper, statistical techniques that are seldom seen were used to examine the possible inter-relation between and among the presence, absence, and management of:

craving,
self-efficacy,
positive emotion,
negative emotion, and
substance use

in real-time, and over-time, during a given day – and across the duration of the study.

A brief look at some sections of the paper

The paper is an open access paper; you can read it with no paywall.

I’ll highlight a few sections here.

The opening explains the background of this kind of statistical inquiry, the types of questions these statistical techniques can ask and answer, and their relevance to substance use problems.

Next comes a section on the statistical techniques themselves. Both the statistical methods and the accompanying research design are explained. Some readers might be especially interested in this information because of how technical it is. Such interest on the part of some would not surprise me, given what this paper does and achieves – it is very different.

The sections covering the results and their discussion were rewarding for me personally. They are examples of the kinds of data, data collection, data compiling, and topographical mapping of improvement over time that I envisioned and wrote about in my series titled “Addiction and the Stages of Healing”. The authors of this paper also hope to one day include GPS locations gathered passively, biometric data, and fMRI in real-time – as I had mentioned in my Stages of Healing series.

Exactly what did this study look at?

(In this paper the authors abbreviate “positive affect” as “PA” and negative affect as “NA”).

The authors note that:

“The PA items included ‘calm’, ‘content’, ‘relaxed’, ‘happy’, ‘elated’ and/or ‘excited’; and the NA items included ‘nervous’, ‘stressed’, ‘irritated’, ‘worried’, ‘upset’, ‘tense’, ‘angry’, ‘depressed’ and/or ‘sad or blue’.”

What kinds of comparisons were examined in real-time and over time?

Morning craving and its association with same-day use and same-day self-efficacy.
Morning negative affect (NA) or morning positive affect (PA) and their associations with same-day use and same-day self-efficacy.
And “…the moderating effect of morning negative affect on the associations between morning craving and same-day recovery outcomes and then examined morning positive affect as a moderator of those associations.”

Thus, in their study they were able…

“…to calculate the estimated time-varying prevalence of substance use and mean self-efficacy for four types of mornings: those with both morning craving and NA (or, alternatively, PA), those with morning craving but no NA (PA), those with no morning craving but morning NA (PA) and those with neither morning craving nor NA (PA).”

One example of their many findings was that:

“…experiencing morning PA attenuated the link between greater craving and lower self-efficacy during a brief time window…”

Conclusion

I’ll be delighted if (for complex, chronic, and severe SUDs) dozens of high-quality partial replications of this work are published, examining many more such variables, and doing so over at least the first five years after initiating care. The kind of statistical modeling done in this paper, applied to examining information like that, would be interesting and important.

Some longer quotations for those that might be interested:

“Only a few very recent investigations have used time-varying effect modeling to study addiction recovery processes. The present empirical study builds upon prior research by studying time-varying self-reported experiences of affect and craving and their associations with recovery outcomes (i.e. substance use and self-efficacy), as each of these constructs has important and complex roles in recovery. For example, craving reliably predicts substance use in many addiction treatment studies, but less understood is whether the association between craving and substance use remains constant or decouples with time in recovery. This association may be non-linear and could inform the timing and intensity of psychological interventions throughout recovery. The link between craving and other non-substance-related recovery outcomes such as self-efficacy could also vary dynamically – understanding this link could clarify how persistent cravings impact one’s belief in their ability to execute important recovery behaviors…”

(In their report “Time Varying Effect Modeling” is abbreviated as TVEM and “Intensive Longitudinal Data” is abbreviated as ILD. Understanding those abbreviations is important as you read on.)

“The data presented here and much of the research in the health sciences applying TVEM to ILD use behavioral or self-report data. However, the types of ILD that are collected but not yet analyzed using TVEM are vast, including passively collected ambulatory assessment data such as physiological arousal or global positioning system (GPS) data coordinates, which could be used for deriving continuous time metrics of ‘time spent’ in different socio-spatial contexts. TVEM could also be useful for other intensively repeated data collection situations, including laboratory-based studies more amenable to experimental manipulations and for assessing biological processes. TVEM has been used to analyze fMRI data collected across developmental ages to model age-varying functional connectivity between two brain regions (the amygdala and ventromedial prefrontal cortex) to study their implication in major depressive and anxiety disorders. Dynamic functional connectivity analyses using fMRI data have seen a rapid increase in recent years in the study of SUDs, and TVEM may offer a comparatively more straightforward and model-based technique for some of these analyses.”

Reference

Stull, S. W., Linden-Carmichael, A. N., Scott, C. K., Dennis, M. L. & Lanza, S. T. (2023). Time-varying effect modeling with intensive longitudinal data: Examining dynamic links among craving, affect, self-efficacy and substance use during addiction recovery. Addiction. doi.org/10.1111/add.16284

Suggested Reading

Ashford, R. D., Brown, A., Brown, T., Callis, J., Cleveland, H. H., Eisenhart, E., Groover, H., Hayes, N., Johnston, T., Kimball, T., Manteuffel, B., McDaniel J., Montgomery, L., Phillips, S., Polacek, Statman, M. & Whitney, J. (2019). Defining and Operationalizing the Phenomena of Recovery: A working definition from the recovery science research collaborative. Addiction Research & Theory, 27:3, 179-188, DOI: 10.1080/16066359.2018.1515352

I include this paper as a suggested reading, given the importance of defining our concepts and deciding what is worth measuring.

Galanter, M. (2014). Alcoholics Anonymous and Twelve-Step Recovery: A model based on social and cognitive neuroscience. American Journal on Addictions, 23: 300-307. doi.org/10.1111/j.1521-0391.2014.12106.x

In my opinion this is a landmark paper. It examines the personal work completed via mutual aid, operationalizes each effort in standard psychological terms, and identifies the associated brain regions of each. In this way, the author outlines a research agenda related to getting better, rather than studying the problem and its mere diminishment.