Sampling and data collection methods
The UK National Recovery Survey was modelled on a similar process conducted in the USA in 2017 [16, 17]. The target population was the general population in the United Kingdom (England, Scotland, Wales and Northern Ireland) aged 18 or over who perceived that they had overcome a problem with drugs or alcohol. The survey was conducted by the market research and data analytics company YouGov, and ethical approval was obtained from the University of Birmingham Science, Technology, Engineering and Mathematics Ethical Review Committee (ERN_21_0565).
In stage 1 the screening question ‘Did you use to have a problem with drugs or alcohol, but no longer do?’ was administered on a UK nationally representative telephone omnibus survey in December 2021. The question was run twice to generate 2,000 responses. This provided (a) an estimate of the prevalence of AOD problem resolution and (b) the demographic profile (such as age, gender, social grade, region) of everyone who reported problem resolution. These data were used to create representative sample frames of those who have resolved a problem with AOD, which were then used to sample and weight the data.
Stage 2 involved the administration of the screening question on YouGov’s online panel of 400,000 active panellists in the UK in January 2022, allowing them to send the survey to those who qualified. YouGov employ an active sampling method, drawing a sub-sample from its panel that is representative of the group in question in terms of socio-demographics. YouGov has a proprietary, automated sampling system that invites respondents based on their profile information and how that aligns with targets for surveys that are currently active. Respondents are automatically, randomly selected based on survey availability and how that matches their profile information. Respondents are contacted by email and invited to take part in an online survey without knowing the subject at this stage. A brief, generic email invitation was used which informed the respondent only that they were invited to a survey. This helped to minimise bias from those opting in/out based on level of interest in the survey topic. Following this, the full survey was administered online, and the final sample consisted of n = 1,373 UK adults. All participants gave informed consent via the YouGov webpage prior to completing the survey.
Weighting adjusted the contribution of individual respondents to aggregated figures and is used to make surveyed populations more representative of a larger, project-relevant population by forcing it to mimic the distribution of that larger population’s significant characteristics. The weighting tasks happened at the tail end of the data processing phase on cleaned data. YouGov used RIM (Random Iterative Method) weighting as its standard approach, as there were a number of different standard weights that all needed to be applied together. This weighting method calculated weights for each individual respondent from the target and achieved sample sizes for all of the quota variables. RIM weighting is an iterative process, whereby the weights are recalculated several times until the required degree of accuracy is reached. The samples were weighted to be representative of all UK adults who had overcome an AOD problem by age, gender, region and social grade (ABC1 C2DE ) based on the initial nationally representative telephone survey in stage 1.
Participants were grouped on their responses to questions about being ‘in recovery’. Firstly, they were asked “Do you consider yourself to be in recovery?” and given the option of responding ‘yes’ or ‘no’. Participants answering ‘yes’ were categorized as “currently in recovery.” Participants answering “no” were asked the follow-up question “Did you ever consider yourself to be in recovery?”, also with a yes/no response option. Participants responding “yes” were categorized as ‘used to be in recovery.’ Participants responding “no” were categorized as ‘never in recovery.’ No definition of recovery was given in the survey, and so the definition used in each case was self-determined.
Qualitative questions about being in recovery
Participants who indicated having never been ‘in recovery’ were asked, “You indicated that you once had a problem with alcohol or drugs but you no longer do, and you have never considered yourself to be ‘in recovery’. What is the main reason why you have never considered yourself to be ‘in recovery’?” Participants who indicated no longer being in recovery were asked: “You indicated that you once considered yourself to be in recovery but no longer do. Why is that?” Both were given an unlimited word count text box in which to type their response.
Alcohol or drug use and recovery-related characteristics
Participants responded to items from the Form-90  about a list of substances (alcohol, cannabis, cocaine, heroin, other opioids, amphetamines, benzodiazepines, hallucinogens, synthetic drugs, and ‘others’). They were asked 1) whether they considered each substance had ever been a problem, 2) age of first use (from which we dichotomized as < 15 vs. ≥15 years) and 3) to select a primary problem substance . Participants were also asked how long it had been since resolving their problem (split into three groups: 0–5 years; 5–15 years; 15 + years). The survey included items about history of 18 psychiatric disorders including alcohol use disorder and other drug use disorder (“Which of the following substance use and/or mental health conditions have you ever been diagnosed with?“) . Criminal justice history was assessed with an item adapted from the Form-90 , ‘Have you ever been arrested?’. Possible responses included ‘no’, ‘yes – in the past year’ and ‘yes – but not in the last year’.
Sex, age, and ethnicity were all captured as part of the YouGov panel process.
Use of recovery support services or treatment services
Participants were asked “Which of the following recovery support services or treatment programs have you ever participated in?” We grouped the nine response options into (a) used formal treatment (i.e., primary care, specialist outpatient addiction treatment, inpatient alcohol/drug detoxification services or residential rehabilitation), and (b) used recovery support services (i.e., sober living environment, recovery school, university recovery programs/communities, faith-based recovery services such as those provided by a church, synagogue, or mosque, or local peer-led recovery organization (LERO)). Participants were also asked “Which of the following self-help groups have you ever attended to help you with your alcohol or drug problem?.” We coded endorsement of any such group (e.g., AA, SMART Recovery, ‘other’) as ‘used mutual help group’.
Indices of psychological well-being and functioning
Quality of life was assessed using the EUROHIS-QOL , a widely used eight-item measure of quality of life adapted from the World Health Organization measure on quality of life. Items are rated on a 5-point Likert scale ranging from 1 (very dissatisfied) to 5 (very satisfied), with larger values indicating greater quality of life. In addition, we assessed happiness and self-esteem using single-item, 5-point Likert measures, with larger values indicating greater happiness/self-esteem, respectively [23, 24], and psychological distress using the Kessler-6 , a six-item scale where participants rate how often they experienced mental health difficulties (e.g., nervousness and depression) during the previous 30 days on a 5-point Likert scale ranging from 0 (none of the time) to 4 (all of the time).
The 10-item Brief Assessment of Recovery Capital (B-ARC)  is a brief version of the Assessment of Recovery Capital (ARC) scale . Participants reported level of agreement (1 = strongly disagree to 6 = strongly agree) with statements on their recovery, environmental support, and well-being (e.g., “I regard my life as challenging and fulfilling without the need for using drugs or alcohol”). The total score is between 10 and 60, with higher scores representing more overall Recovery Capital. This measure has demonstrated excellent concurrent validity with the longer recovery capital measure (r = .92) as well as excellent internal consistency (a = 0.95) .
We calculated weighted frequencies and cross-tabulations by recovery identity group to provide a descriptive comparison of participants who consider themselves to be in recovery versus not. To identify factors associated with identifying as being ‘in recovery’ we compared the three recovery status groups (i.e., 1 – currently in recovery, 2 – used to be in recovery, and 3 – never in recovery) in univariate multinomial regression models. The univariate predictor variables included demographic, substance use, mental health, criminal justice, recovery support system use variables, quality of life and recovery capital indices. Analyses were exploratory and we did not control for multiple testing. To provide an indication of the strength of association between each tested univariate predictor and identifying as being in one of the three recovery status groups we calculated pseudo r-squared values of the overall model, where larger values represent stronger associations. In addition, we also provided odds ratios and 95% confidence intervals (CI) for each pairwise comparison of the three groups (i.e., currently vs. never, used to be vs. never, and currently vs. used to be). All analyses were conducted using SPSS version 29.
To provide insight into why participants self-identified in the way they did regarding recovery status, we coded the responses to the open-ended recovery questions. Two authors (ED and IM) reviewed the open-ended responses and applied the coding structure created by Kelly et al. for summarizing responses . Discrepancies between coders (5.0% for “no longer;” 4.8% for “never”) were resolved by consensus in a meeting between the two coders. Results were summarized by computing weighted frequencies.