Assessing NIAAA’s Definition of Recovery from Alcohol Use Disorder: A Latent Class Analysis of a Heterogeneous Online Sample

Abstract

NIAAA’s 2022 definition of recovery from alcohol use disorder (AUD) includes two core components, remission of DSM-5 AUD criteria and cessation of heavy drinking. This study’s purpose was to assess patterns of AUD symptoms and heavy drinking in a heterogeneous national sample, in order to clarify the utility of the definition. Participants who self-reported having resolved an alcohol problem for at least six months were recruited through Amazon Mechanical Turk (N=386) and surveyed about their problem severity, current drinking, and AUD symptomology. We used latent class analysis to discern meaningful clusters of AUD symptoms and heavy drinking, as well as factors associated with class membership. A two-class model was the best fit for the data. The first class, which we termed Less Symptomatic, included 83.4% of the sample. Individuals in this class were unlikely to endorse of any of the 10 AUD criteria (<2.5% of the time) and 24.3% reported heavy drinking. In the second class (16.6% of the sample), termed Symptomatic, 45% of respondents endorsed at least one AUD criterion and 88.2% reported heavy drinking. These findings suggest that some individuals in recovery may continue to drink heavily with minimal problems, while others continue to experience AUD symptoms.

Introduction

Approximately one in ten adults in the United States has resolved an alcohol or other drug (AOD) problem (Kelly et al., 2017). Although numerous conceptions of what it means to recover from an AOD problem have long existed, recent research has highlighted the importance of person-centered recovery elements that extend beyond abstinence alone (Best & Hennessy, 2022; Kelly et al., 2018; Witkiewitz & Tucker, 2020). Researchers have found, for example, that self-care, spirituality, and outlook on life are salient components of resolving AOD problems (Kaskutas et al., 2014; Laudet, 2007; Neale et al., 2015). Consequently, many stakeholders have shifted from abstinence-centered definitions of recovery to more multifaceted conceptions that include holistic improvements in quality of life, as well as inclusion of reduced drinking or drug use as a recovery component. For example, the U.S. Substance Abuse and Mental Health Services Administration (SAMHSA) defines recovery as “…a process of change through which individuals improve their health and wellness, live a self-directed life, and strive to reach their full potential” (2012, p. 3).

Within this context, the National Institute on Alcohol Abuse and Alcoholism (NIAAA) recently released an updated definition of recovery from alcohol use disorder (AUD), incorporating the views of stakeholders including research experts, clinicians, and persons in recovery (Hagman et al., 2022). NIAAA’s AUD recovery definition includes two primary components: (1) remission from 10 DSM-5 AUD criteria, such as stopping doing enjoyable activities in order to drink or drinking despite it causing problems with family or friends (the DSM-5 criterion for craving alcohol is excluded) and (2) reduction of alcohol consumption below heavy drinking thresholds, i.e., three drinks or more per day and seven per week for females and four drinks or more per day and 14 per week for males, per NIAAA guidelines. In addition, the definition states that recovery is associated with improved functioning across multiple domains of well-being, such as relationships, work, mental health, and spirituality (Hagman et al., 2022).

NIAAA’s AUD recovery definition aligns with research conveying that resolving AUD can include low-risk drinking among subsets of individuals (Witkiewitz et al., 2017). Although the definition specifies that drinking must be below NIAAA’s heavy-drinking thresholds, there remains a debate in the literature regarding whether heavy drinking may be a component of AUD recovery processes if accompanied by quality-of-life improvements (e.g., reductions in depression symptomology). One study, for instance, found that among a sample of individuals who completed AUD treatment three years prior, one group evidenced improvements in global functioning alongside sustained heavy drinking (Witkiewitz et al., 2019). Another longitudinal study replicated these results among a sample who had completed AUD treatment ten years prior (Witkiewitz et al., 2020). In response to these findings, other researchers have rebutted that heavy drinking should not be considered a component of AUD recovery due to its associated biological health risks and the implication that those who sustain abstinence but are otherwise deemed low-functioning may feel stigmatized (Kelly & Bergman, 2021).

Due to the recency of NIAAA’s recovery definition, little research to date has assessed the definition’s utility and validity in community-based samples. Importantly, it is unknown how many people who self-identify as having resolved an alcohol problem meet NIAAA’s recovery definition criteria. Further, although prior studies have identified a range of sociodemographic covariates associated with other recovery definitions (Kelly et al., 2018), it is unknown if characteristics of personal identity (e.g., gender, race/ethnicity, age, sexual orientation), geography (e.g., rural, suburban, or urban residence), indicators of social position (e.g., income or education), and alcohol problem severity are associated with NIAAA recovery status. There is a particular need to explore these associations in community samples that include people with a range of treatment experiences. Individuals utilizing inpatient or outpatient treatment programs and/or self-help groups are over-represented in research, even though the majority of people in the United States who resolve an AOD problem are in unassisted recovery, e.g., resolving the problem without formal treatment or use of self-help resources (Kelly et al., 2017; Subbaraman et al., 2015; Tucker et al., 2020).

The current study aimed to assess patterns and covariates of NIAAA recovery definition criteria endorsement among a sample of individuals self-identifying as having resolved a prior problem with alcohol. Although the NIAAA definition acknowledges the importance of holistic changes in well-being, our analysis focused on the definition’s core criteria of no longer meeting any DSM-5 AUD criteria (excluding craving) and no recent heavy drinking. Considering the absence of research in this area, we conducted a latent class analysis to explore clustering based on NIAAA criteria and, secondly, examined auxiliary variables, which may vary as a function of class membership.

Methods

Data were collected as part of a larger project to develop and evaluate a new recovery-related measure (Authors, 2023a). Amazon’s Mechanical Turk (MTurk) was used for recruitment and data collection. MTurk is an online crowdsourcing platform increasingly used in survey-based research, including prior studies on alcohol use problems, which have shown comparability between MTurk samples and samples recruited through in-person methods (Jones et al., 2022; Kim & Hodgins, 2017; Strickland & Stoops, 2019;). The Institutional Review Board at the principal investigator’s university approved the study.

Participants and Screening

We used a multi-step process to identify and screen potential participants into the study. To view the study’s eligibility screening within the MTurk platform, participants had to live in the United States and have an approval rate of 95% or greater on prior MTurk tasks, a criterion commonly used in MTurk-based research to identify participants with greater reliability (Cobanoglu et al., 2021; Strickland et al., 2019). The eligibility screening contained decoy questions (e.g., “Do you use tobacco products at least once a month?”) to mask the research’s intended aim, to decrease the likelihood of participants trying to falsely self-select into the study. It also included questions assessing the following criteria: (1) age 18 or older; (2) ability to communicate in English; and (3) answering affirmatively to the question “Did you used to have a problem with alcohol, but no longer do?” and report having resolved the problem for at least 30 days. Other studies have used similar criteria (Kaskutas et al., 2014; Kelly et al., 2018). For this analysis, we limited the dataset to participants who identified as having resolved their problems for six months or longer, given the prototypical instability of early recovery (Hagman et al., 2022).

Procedures and Measures

Data collection took place between September 2021 and January 2022. We manually reviewed responses to the eligibility screening survey to confirm eligibility and invited these participants to complete the study survey. The survey began with information about the study and an affirmation of informed consent. Consenting participants proceeded to take the survey, which included questions about demographics, substance use problem severity, treatment history (e.g., engagement with inpatient or outpatient treatment and self-help groups), and fulfillment of the NIAAA recovery definition criteria. We measured problem severity using the Short Inventory of Problems-Lifetime Version, with questions phrased to ask separately about problems related to alcohol use and problems related to drug use (SIP-Alcohol and SIP-Drug Use; Blanchard et al., 2003; Morse & Roberson, 2017). To assess if participants fulfilled the NIAAA recovery definition, we asked a series of questions based on the 10 DSM-5 AUD criteria and about heavy drinking patterns. Participants were asked separately about each of the AUD criteria and indicated if they had experienced the criterion (i.e., “spent a lot of time drinking or hung over”) in the past three months, three months to one year ago, one to five years ago, more than five years ago, or if they never experienced it (American Psychiatric Association, 2013). To assess heavy drinking, participants indicated the last time they had consumed more than three drinks in one day or more than seven drinks in one week (for participants identifying their sex as female) and more than four drinks in one day or more than 14 drinks in one week (for participants identifying their sex as male). Upon completion of the survey (approximately 10 minutes), participants were compensated $4 through MTurk.

Following best practices for using crowd-sourced data from MTurk and similar platforms in academic research (Belliveau & Yakovenko, 2022), we employed several quality control procedures prior to using data for analysis. This included the use of multiple attention check questions, such as asking participants about age in two places within the survey to ensure that they answered consistently and questions designed directly for attention checking, e.g., “Please select 5 as the response to this question.” We also analyzed response times to identify respondents with implausibly brief completion times and assessed responses to individual measures within the survey for potential straight-lining. Responses that were potentially questionable were excluded from the final sample (n=386 for this analysis).

Data Analytic Plan

Latent class analysis (LCA) was used to identify potential classes using the 10 DSM-5 diagnostic items and an additional item indicating recent heavy drinking. LCA is a special case of finite mixture modeling that models categorical latent variables that represent subpopulations where membership is not known a priori, but is modeled as a function of the data (Lanza & Rhoades, 2013). LCA permits both classifications of individuals and the identification of auxiliary variables that may be associated with class membership. In this study, auxiliary variables included: demographics (e.g., gender, race/ethnicity, age, sexual orientation); geography (e.g., rural, suburban, or urban residence); socioeconomic status indicators (e.g., income, education level, employment); criminal justice system involvement (e.g., self-reporting prior incarceration or prior felony); problem severity (SIP-Alcohol and SIP-Drug Use scores); and treatment history variables (e.g., unassisted recovery vs. using treatment or mutual aid group services and resources). Since the DSM-5 AUD criterion assessing craving is not part of the NIAAA recovery definition and was not used to establish the classes, we included it as an auxiliary variable to determine if classes differed in terms of craving.

Starting with a one-class solution, classes were added sequentially until model fit did not improve. Model fit (see Table 2) was compared using the Akaike Information Criterion (AIC; Akaike, 1973), Bayesian Information Criterion (BIC; Schwarz, 1978), and classification entropy. Smaller AIC and BIC values suggest a better-fitting model, and entropy values closer to 1 suggest adequate separation of classes (Geiser, 2013). In addition, a significant p-value on the Bootstrap Likelihood Ratio Test (BLRT) indicated that a solution with k classes fit better than a k – 1 class solution. The selection of a final model balanced fit statistics with the substantive interpretation of model results.

Table 2.

Model fit statistics for 1- through 4-class solutions

	1	2	3	4
LL	−996.208	−823.76	−804.929	−790.931
AIC	2014.416	1693.519	1679.859	1675.861
BIC	2057.931	1784.503	1818.313	1861.786
aBIC	2023.029	1711.527	1707.262	1712.66
Entropy		0.89	0.94	0.911
LRT		340.14	37.14	27.61
p value of LRT		< .000	0.27	0.17

All analyses were conducted with MPlus software (Muthén & Muthén, 2007). Following identification of the best-fitting solution, the BCH command was used to examine auxiliary variables that may differ as a function of class membership. The BCH command examines differences in auxiliary variables by class using an inverse of the weighted class assignment error probabilities and a chi-square distribution. This is considered the most conservative method for estimating differences in auxiliary variables in latent class models (Asparouhov & Muthén, 2015).

Results

The sample (N = 386) included 43.6% women, 28% participants of color (8% Black, 7.8% other races, and 16.7% Latinx ethnicity), and 21% people in unassisted recovery (see Table 1). Each of the 10 AUD criteria was endorsed by 2.9–8.4% of participants in the past three months, and 34.8% reported heavy drinking during the past three months (see Table 2). Nearly two-thirds (61.9%) of the sample met full NIAAA recovery criteria, meaning they endorsed none of the DSM-5 AUD criteria (aside from craving) and reported no heavy drinking within the last three months.

Table 1.

Sample description and comparison of auxiliary variables in the Less Symptomatic and Symptomatic classes

	Full Sample %	Less Symptomatic Class %	Symptomatic Class %
Auxiliary Variables	(N=386)	(n=322)	(n=64)	p
Gender
Cisgender male	55.8	56.5	52.4	.60
Cisgender female	43.6	42.8	47.6	.54
Other gender identities	0.5	0.6	0.0	.16
Biological sex
Male	55.7	56.3	52.5	.62
Female	44.3	43.7	47.5
Age
18–29 years	15.3	14.4	19.5	.40
30–59 years	77.7	78.6	73.5	.45
60 years or older	7.5	7.6	7.0	.88
Race
Black	8.0	9.0	3.2	.07
White	84.2	82.7	91.6	.05
Other race	7.8	8.3	5.1	.38
Hispanic or Latinx ethnicity	16.7	16.0	19.8	.54
LGBQ identifying	22.8	21.4	21.0	.73
Criminal justice background
Previously incarcerated	26.6	29.2	13.3	.005
Previous felony	15.0	15.3	14.0	.81
Employment
Employed, retired, or student	92.0	90.2	100	<.001
Unemployed	8.0	9.8	.00
Education
High school diploma or less	6.5	7.5	1.5	.09
Some college but no degree	11.1	12.4	4.9	.04
Associate’s degree or higher	82.3	80.1	93.6	.002
Low-income	66.6	65.1	74.1	.19
Urbanicity
Urban	48.2	48.0	49.2	.87
Rural	20.5	20.6	20.1	.93
Suburban	31.2	31.3	30.7	.93
Time since problem resolution
Six months to 1 year	37.7	33.3	60.6	<.001
One to 2 years	26.4	27.5	20.0	.25
Two to 5 years	19.1	20.3	12.6	.16
More than 5 years	17.0	18.9	6.8	.007
Unassisted recovery	21.0	21.5	15.5	.30
SIP-Alcohol Score M (SD)	9.0 (4.0)	9.4	9.1	.80
SIP-Drug Use Score M (SD)	7.1 (5.0)	7.3	7.8	.52
Wanted a drink so badly you couldn’t think of anything else (craving)	13.5	8.9	37.0	<.001

Model fit statistics are reported in Table 3. The AIC and BIC suggested that the optimal number of classes is two. The LRT test was significant for the two-class solution (8.36; p < .001), suggesting that the two-class solution is a better fit than a one-class solution. The three-class solution was not statistically significant, indicating it did not fit better than the two-class solution. Based on fit statistics and the interpretability of the solutions, the two-class solution was retained for further analysis.

Table 3.

Probability of endorsing DSM-5 AUD criteria and heavy drinking (last 3 months) in full sample and by class¹

AUD criteria (paraphrased)	Full Sample % (N=386)	Less Symptomatic Class % (n=322)	Symptomatic Class % (n=64)
1. Drank more or spent more time drinking than intended	8.4	2.5	38.2
2. Tried to limit or stop drinking but couldn’t	7.6	2.0	35.2
3. Spent a lot of time drinking or hung over	6.2	0.0	37.4
4. Wanted a drink so badly you couldn’t think of anything else (craving)2	--	--	--
5. Had difficulty doing household work, parenting, job duties, or school work because of drinking or being hung over	4.7	0.6	25.0
6. Drinking despite it causing problems with family or loved ones	3.4	0.6	17.6
7. Stopped doing other enjoyable activities in order to drink	3.6	1.5	14.5
8. Repeatedly drank in situations that put you at risk of injury (e.g., driving, using machinery)	2.9	0.4	15.2
9. Continued to drink despite it causing you stress, sadness, health problems, or periods of memory loss	4.5	0.0	26.5
10. Had to drink more to feel the same level of intoxication (feeling drunk)	4.2	0.0	25.3
11. Had problems when alcohol wears off, such as trouble sleeping; being irritated, sad, or nervous; sweating; or having an upset stomach	4.5	0.0	27.8
Heavy drinking3	34.8	24.3	88.2

Notes:

¹The first column reports percentages for the full sample and the second and third columns report probabilities for the two classes.

²The craving item from the DSM-5 AUD diagnostic criteria was excluded from the latent class analysis, since it is not part of the NIAAA definition of AUD recovery.

³Heavy drinking criteria were determined per NIAAA’s 2021 guidelines.

The first class (see Figure 1 and Table 3) included 83.4% of the participants. Individuals in this class were unlikely to endorse of any of the 10 AUD criteria (<2.5% of the time) and about one quarter (24.3%) reported heavy drinking. We termed this the Less Symptomatic class.

Figure 1:

Probability of AUD criteria endorsement and heavy drinking (past 3 months) by class membership

The second class consisted of 16.6% of the sample. This class was more likely than the Less Symptomatic class to endorse each of the 10 AUD items. We designated this as the Symptomatic class. This class was also much more likely to report heavy drinking (88.2%). Forty-five percent of the Symptomatic class endorsed at least one AUD criterion. The items most likely to be endorsed were drinking more than intended (38.2%), unsuccessfully trying to stop drinking (35.2%), and spending a lot of time drinking or hungover (37.4%). The least likely to be endorsed items were related to social functioning and family relationships (e.g., stopping doing other enjoyable activities in order to drink; 14.5%) or risky behaviors (e.g., repeatedly drinking in situations that put one at risk of injury; 15.2%).

The proportions and p values of the auxiliary variables are listed in Table 1. The Less Symptomatic class had a higher percentage of members with some college and a lower percentage of members with an Associate’s degree or higher. More members of the Less Symptomatic class reported having been previously incarcerated and being unemployed, compared to the Symptomatic class. More members of the Symptomatic class reported a time since problem resolution of 6 months to 1 year; in contrast, there were more members of the Less Symptomatic class who reported 5 or more years since problem resolution. Craving varied significantly between the two classes, with members in the Symptomatic class being much more likely to report it (37% versus 8.9% in the Less Symptomatic class). Other auxiliary variables including gender, biological sex, age, income, unassisted recovery, and problem severity did not significantly differ by class.

Discussion

In this study, we aimed to identify to what extent individuals in a heterogenous online sample of people who self-identified as having resolved alcohol problems met the NIAAA recovery definition and which auxiliary variables were associated with symptomatic and less-symptomatic class membership based on AUD criteria and heavy drinking. Overall, LCA results indicated that most of the sample had a low likelihood of endorsing AUD items or heavy drinking. Thus, meeting the NIAAA recovery definition may be a realistic personal goal or method of assessment in treatment and recovery (Authors, 2023b; Hagman et al., 2022).

However, there was a substantial subset of individuals who endorsed recent heavy drinking and/or one or more of the 10 AUD criteria. There is debate in the literature regarding what can be considered “recovery” as well as the permissibility and risk of heavy drinking episodes in recovery (Kelly & Bergman, 2021; Witkiewitz et al., 2020; 2021). It may be true that the portion of the sample in the Symptomatic class is at greater risk of returning to previous alcohol problems or may require ongoing recovery supports. Notably, the proportion of participants who reported heavy drinking (34.8%) was much higher than the endorsement rate for any of the AUD criteria (2.9–8.4%), suggesting that at least some and perhaps a large proportion of individuals in recovery who continue to engage in heavy drinking do not report experiencing any of the problems that the AUD criteria assess. This is further evidenced by the fact that nearly one quarter of the Less Symptomatic class reported heavy drinking with no to minimal (2.5% or under) endorsement of AUD criteria. However, it is critical to recognize that heavy drinking is never risk-free, regardless of if individuals report AUD symptoms. Moderate to heavy alcohol consumption increases the risk and severity of many illnesses, including liver disease and several forms of cancer (Kelly & Bergman, 2021).

Study results are generally consistent with Witkiewitz et al.’s (2019) finding that a small subset of people was considered high-functioning while engaging in heavy drinking following AUD treatment. Future investigations should examine in greater depth the characteristics of people who report resolving alcohol problems but who do not meet NIAAA AUD recovery criteria, in order to clarify the nature and relationship of heavy drinking and functional improvements in recovery. Such improvements may be measured by the presence or absence of AUD criteria as well as other indicators of biopsychosocial wellness, such as recovery capital, quality of life, and post-traumatic growth (Best & Hennessy, 2022; Elison et al., 2017; Krentzman et al., 2022).

Although we did not anticipate some of the relationships found between auxiliary variables and class membership, these analyses were exploratory, and may also be partially explained by previous recovery research findings. First, we found that individuals who reported less time since resolving their alcohol problems (six months to one year) were more likely to be in the Symptomatic class, while those with more than five years since problem resolution were more likely to be in the Less Symptomatic class. This finding is supported by literature documenting that those in earlier stages of recovery are less stable with respect to their drinking (Maisto et al., 2020). Second, having a higher education and being employed were associated with membership in the Symptomatic class. These variables have been associated with alcohol use in recovery in other national research samples (Fan et al., 2019). Theoretically, this finding could also align with stress and coping-based models of recovery that posit relationships between stress (such as entailed in employment and education) and recovery outcomes (Goshorn et al., 2023; Kelly & Hoeppner, 2015).

Finally, those with incarceration histories may report lower symptom endorsement or less heavy drinking due to the idea of reaching “rock bottom,” perceiving that abstinence is integral to achieving their lifestyle goals following incarceration (Kirouac et al., 2015). Those with incarceration histories may also have had access to treatment and recovery services as part of their incarceration or re-entry. As the design of this study does not allow for confirmation of the mechanisms underlying associations between auxiliary variables and class membership, future studies should explore these possible explanations, as well as alternatives. Further, given that our use of the BCH command in MPlus uses a conservative chi-square distribution to detect differences between classes, we acknowledge the possibility that other auxiliary variables could emerge as significant if one applied a less conservative analytic approach.

Our findings carry several implications for the treatment and recovery field. While meeting NIAAA recovery criteria is a sound overarching treatment goal, service providers should keep in mind that there are multiple pathways to this outcome (Authors, 2024). Therefore, supporting people in recovery from alcohol problem should begin with conversations about individuals’ specific recovery goals, which may include abstinence, moderate alcohol use, resolution of specific AUD symptoms (e.g., resuming enjoyable activities previously stopped because of drinking), and/or other health and wellbeing gains (Gallagher et al., 2019). For individuals not focused on abstinence, nonjudgmental psychoeducation about the definition of and risks associated with heavy drinking may be instructive (Ray et al., 2019). People seeking to resolve alcohol problems outside of formal treatment systems may also benefit from education about AUD symptoms and heavy drinking, which they can use to define their own recovery goals. This could be accomplished through the use of digital recovery support services, including recovery-centered smartphone apps, social media groups, and mobile messaging interventions, to reach, support, and educate people beyond the traditional treatment infrastructure (Bergman & Kelly, 2021; Chavez & Palfai, 2020). Lastly, it is notable that the occurrence of craving differed substantially, with people in the Symptomatic group being much more likely to report it. This suggests that it may be informative for clinicians to check in regularly with clients about craving levels, or for people in unassisted recovery to self-assess this, as craving may be associated with greater likelihood of experiencing other AUD criteria and/or engaging in heavy drinking.

There are several limitations of this study to note. First, although we made efforts to recruit a sample that was diverse with regard to various characteristics, the final sample is not fully representative of the U.S. adult population of people in recovery from alcohol problems. For instance, while sexual minorities (i.e., people identifying as gay, lesbian, queer, or bisexual) are an estimated 11% of the U.S. population in recovery (Haik et al., 2022), this group comprised 22% of the study sample. Therefore, when interpreting findings, caution must be used to consider the over and under-representation of specific groups.

Second, participants completed the survey at one time point, and therefore it is unknown to what extent NIAAA recovery criteria endorsement might fluctuate over time. Going forward, researchers should continue to assess endorsement of these recovery criteria at multiple time points. One recent study found that meeting the NIAAA recovery definition predicted three-month, but not six-month, drinking outcomes (Authors, 2023b), implying that endorsement of these criteria may fluctuate across different recovery stages. Furthermore, research indicates that individuals often cycle between heavy drinking and abstinence in recovery (Maisto et al., 2020). Thus, further research is needed to examine longitudinal trajectories of NIAAA-defined recovery. Ideally, this research would include both quantitative and qualitative explorations of how people with lived experience of AUD experience and self-define their recoveries, including but not limited to the criteria specified in the NIAAA AUD recovery definition. Finally, because our sample consisted of individuals who had resolved their problems with alcohol for at least six months, findings from this study cannot be applied to people in the earliest stages of recovery.

Although we used best practices to minimize inaccuracy in the MTurk-delivered survey, the self-reported nature of the data may have impacted results. Specifically, determining if one meets AUD criteria requires considerable self-awareness. Thus, it is possible that some respondents might have denied some criteria, whereas a clinician administering a diagnostic interview could determine otherwise (though clinicians’ assessments are also subject to inaccuracies; Hasin et al., 2020). Relatedly, we paraphrased the criteria in plain language to facilitate participants’ self-assessments, creating the possibility that nuances contained in the exact phrasing of the criteria were lost. Modifying a structured interviewing tool for self-assessment, such as the Diagnostic Assessment Research Tool (Schneider et al., 2022), would provide a more robust way of assessing the criteria in future research. Lastly, the COVID-19 pandemic was in its early phases during data collection for this study. Therefore, given recent evidence suggesting an impact of pandemic stressors on those in recovery from alcohol problems (Scarfe et al., 2022), results may have been skewed toward a higher than usual prevalence of AUD criteria and heavy drinking in response to these stressors.

Conclusion

This study confirms that most individuals in our sample who self-identified as having resolved an alcohol problem also met the current NIAAA recovery criteria (Hagman et al., 2022). However, a significant portion did not meet the full criteria, with heavy drinking being particularly prevalent. The prevalence of heavy drinking in the Less Symptomatic (24.3%) versus Symptomatic (88.2%) classes suggests that some individuals in recovery continue to drink heavily with minimal problems, whereas for others, heavy drinking is accompanied by the reoccurrence of AUD symptoms. Moving forward, research can continue to assess the utility of the NIAAA recovery definition in clinical and community-based settings and consider how various factors, including heavy drinking episodes, influence people’s ability to meet and sustain recovery goals.