This paper reports results of qualitative exit interviews conducted as part of the Alcohol Screening and Preoperative Intervention Research (ASPIRE) study, a randomized pilot trial of 2 preoperative alcohol interventions [22]. ASPIRE recruited 51 participants between July 2019 and February 2021 (with a pause from April to July 2020 due to COVID-19–related reductions in elective surgeries) from a large academic medical center in the Midwestern United States. To be eligible for the parent trial, participants had to meet the following inclusion criteria: (1) be scheduled for elective or semielective surgery in the next 35 to 120 days in select subspecialties (plastic, knee or hip arthroplasty, minimally invasive, endocrine, gynecology, urology, colorectal, or hepatobiliary); (2) receive regional or general anesthesia; (3) be aged 18 to 75 years; (4) have positive or unknown alcohol use in their social history in the electronic health record; and (5) meet criteria for high-risk alcohol use (AUDIT-C score ≥5). Selected surgical types included orthopedic, plastic, urology, gynecological, hepatobiliary, and general surgery. Individuals were excluded if they were undergoing surgeries requiring only local anesthesia, could not read or understand English, or had substantial cognitive impairment or evidence of psychotic symptoms (ie, delusions or hallucinations). Research staff contacted potential participants by phone (call or text). Eligible participants provided written informed consent for all trial procedures including qualitative exit interviews. Randomization into the brief advice or health coaching intervention conditions was stratified based on sex and alcohol screening scores.

The University of Michigan’s Institutional Review Board approved all study protocols (HUM00156743). This is a registered clinical trial (ClinicalTrials.gov NCT03929562).

The ASPIRE pilot trial involved 3 quantitative assessment time points: at baseline and 1- and 4-months postbaseline. After the final (4-month) follow-up, treatment response and low response were assessed by calculating percent change in alcohol use at follow-up time points relative to baseline.

Following completion of the final follow-up assessments at 4 months for the ASPIRE pilot trial, we invited participants to take part in qualitative exit interviews based on study condition (brief advice and health coaching) and treatment response. Our goal was to enroll approximately at least 25% of the trial sample of N=51. Of the 26 participants invited based on these criteria, 19 agreed and completed exit interviews.

A trained graduate-level interviewer conducted qualitative exit interviews via the internet (through Health Insurance Portability and Accountability Act–compliant teleconferencing) using a semistructured interview guide containing open-ended questions and detailed probes designed to explore how the interventions (or low-intervention factors) may have influenced alcohol and other substance use during the perioperative period (see Multimedia Appendix 1 for the guide). Questions also explored participants’ perspectives on the acceptability (ie, whether the sessions and content were appropriate, appealing, and impactful) and feasibility (ie, whether participation was convenient, clear, and achievable), as well as how comfortable they felt sharing accurate and honest information about substance use with study staff and medical providers. Study staff regularly reviewed transcripts against audio recordings to confirm accuracy and deidentification.

We undertook thematic analysis of transcribed data that began with a collaborative codebook development process involving 4 research team members (the principal investigator, study coordinator, research assistant, and a lead qualitative investigator) [28-30]. First, key topics of interest were independently reviewed from the interview guide and several selected transcripts to develop a preliminary list of potential codes and definitions [31]. We met to discuss and refine this preliminary list, which we compiled into a draft codebook. Interviews were audio-recorded and transcribed verbatim by a professional transcription company that omitted potentially identifiable information. The draft codebook was then independently tested using another set of transcripts and the team members met again to discuss and refine the codebook. Through 3 additional rounds of this iterative codebook development and testing process, discrepancies in code application were identified and resolved; codes were merged, added, and removed; and code definitions were revised as necessary until consensus was reached on a final codebook. To validate the final codebook and assess consistency in code application, 2 core coders (the study coordinator and research assistant) double-coded 2 transcripts using MAXQDA (VERBI Software). After assessing and determining that there was a high degree of consistency, we divided up the remaining transcripts between the 2 coders, who then independently coded the remaining transcripts under the supervision of the principal and lead qualitative investigators. We continued holding weekly meetings to review coding progress, discuss emergent themes, and identify and clarify preliminary findings [32]. To delve deeper into emergent themes that were identified during an initial round of coding (described below), midway through data collection, new questions were added on COVID-19 experiences, and intervention changes that emerged in earlier interviews were suggested. For our thematic analysis for this paper, we focused on understanding intervention acceptability and feasibility and potential reasons for intervention response. The transcripts across these domains were then reviewed to identify key themes, which are described below and exemplified using anonymized quotes.

Among 19 participants, 9 (47%) were female, 9 (47%) were from the health coaching intervention condition, 10 (53%) were from the brief advice condition, 9 (47%) were considered high responders, and 10 (53%) were considered low responders (Table 1). From our analysis of interview data, we identified the following three themes: (1) the intervention content was novel and impactful, (2) the choice of intervention modality enhanced participant engagement in the intervention, and (3) factors external to the interventions also influenced alcohol use. These themes are detailed below and summarized in Table 2.

Overall, participants found both intervention conditions to be acceptable, of appropriate length, and impactful. Several participants stated that the intervention directly influenced their perioperative alcohol consumption: “It was very enjoyable. It was easy...there hasn’t been a thing that’s made me uncomfortable or that I haven’t agreed with [in] the study.” In general, participants reacted positively to the information presented in both intervention conditions. Participants considered the information regarding the risks of preoperative alcohol use on surgical outcomes to be novel (yet believable) and motivating. By receiving this information, participants described feeling more prepared for surgery and “in control” of their surgical outcomes. As one health coaching intervention high responder explained:

Several participants commented on what they learned about the connections between alcohol use and the body’s ability to heal, including alcohol’s impact on the immune system. In considering what “stood out” from the brief advice condition, one high responder explained:

Other high responders generally perceived the information shared as part of the intervention as impactful, identifying it as one of the main reasons they reduced their preoperative alcohol use. One brief advice high responder (48 years old, male) stated “If I was not a part of the study or aware of the study, I probably would not have reduced [my drinking].” Health coaching participants described how conversations with a nonjudgmental health coach provided them an opportunity to carefully consider changing their alcohol use.

Both interventions presented information about alcohol use in the preoperative period that some participants found helped them make a “connection” between alcohol use and health, ultimately changing their personal risk perceptions. Participants found the concept of prehabilitation, or optimizing health prior to surgery, to be particularly impactful because the health coach encouraged them to view themselves as important members of their surgical teams who could enhance their own health, as opposed to passively receiving surgery. As such, they felt an improved sense of control over their surgical outcomes. As one health coaching high responder explained:

Among low responders in both intervention conditions, there was a sense that the information presented was true but did not apply to them personally. This was particularly true for participants on the lower end of the AUDIT-C eligibility assessment, who did not view their drinking habits as risky. These individuals felt they had been included in the study erroneously, with one health coaching low responder (71 years old, female) saying, “it just didn’t feel like a good fit for me.” A brief advice low responder (47 years old, male) further elaborated that he thought the content was “focused on people that drink a lot all the time...while I don’t have a problem having a beer or two on the weekends, it’s really not an issue for me.”

Some participants who had undergone major surgeries in the past described never receiving information about alcohol use and related surgical risks. Accordingly, some participants mentioned feeling surprised that their doctors had not communicated this kind of information to them in the past. One brief advice low responder (22 years old, male) explained that he would have appreciated receiving better information from his surgical team given the significance of the topic, saying, “I’ve had two surgeries in the past on my knee...these issues literally never came up. I was never briefed on what complications might arise around alcohol or tobacco use, so that was new.”

By design, intervention modality differed between conditions. While the brief advice condition was only delivered by phone, health coaching participants were given a choice between phone or face-to-face sessions (by videoconferencing or, prior to COVID-19, in-person). Overall, it appeared that having some choice in intervention modality increased participants’ satisfaction; while nearly all health coaching participants were satisfied with their modality of choice, brief advice participants, who had no choice, indicated mixed experiences and dissatisfaction with phone participation.

Some health coaching participants who chose face-to-face intervention delivery expressed high satisfaction with this modality, explaining that they felt more engaged with the information presented, took intervention recommendations more seriously, and were more open and honest because it felt more “personal.” As one health coaching high responder (58 years old, male) explained, “You’re talking to somebody, a human, and it just makes you want to participate and follow the rules, so [there’s] more accountability.” Most face-to-face sessions involved videoconferencing, which participants could schedule around their work and personal lives; several noted that this flexibility enabled them to save time and energy compared with traveling to in-person sessions. They were also able to participate in a setting of their choice, which in addition to being convenient, was more comfortable. Instances of dissatisfaction included one participant stating the sessions were “a bit too long” and another participant who would have preferred in-person sessions, but this was impossible due to COVID-19 pandemic restrictions at that time.

In contrast, brief advice participants’ experiences with phone sessions were more mixed, possibly because they lacked any choice in intervention modality. While some participants discussed specific advantages of phone sessions (eg, greater ease of use and convenience over videoconferencing or traveling to an in-person appointment), others felt they would have been more comfortable face-to-face, with one participant (48 years old, male) saying he would be “a lot more engaged [having] a video in front of me and [knowing] I’m being watched at the same time.” Furthermore, they perceived a poorer connection between themselves and the interventionist, viewing the phone modality as “un-intimate.” As a brief advice low responder explained:

Most felt that the length of health coaching sessions (~45 minutes) was appropriate and afforded them sufficient time for discussion without feeling rushed while also showing respect for their time. In contrast, while most participants in the brief advice condition felt that the session length was acceptable, some felt the length was too short to feel comfortable opening up about the sensitive topic of alcohol use.

There were several factors external to the intervention content that influenced participants’ alcohol consumption, including (1) self-report and tracking of alcohol use for the ASPIRE trial assessments, (2) “intervention” language used in the study, and (3) surgical experiences.

This study describes the experiences of participants in a preoperative pilot trial of alcohol-focused interventions of varying intensities and modalities. Participants in our qualitative exit interviews were selected to represent high responders and low responders to help better understand potential effective and ineffective elements of the interventions to inform a future sequential multiple assignment randomized controlled trial, assigning participants to various treatment options based on initial treatment response. Overall, high responders in our pilot trial described intervention content as novel and relevant. The health coaching condition, which involved motivational interviewing and goal setting, provided an opportunity to reflect on alcohol use in an open, nonjudgmental conversation with supporting materials that framed the patients’ roles as active, thereby empowering them to take actions to improve their surgical outcomes. Surgery preparation and recovery appeared to motivate change in alcohol use for participants, with the intervention content helping some participants link their alcohol use to surgical health for the first time. While most participants found both interventions to be novel and motivating, a few believed the link between alcohol use and surgical health was intuitive and that they would have stopped or reduced alcohol use before and after surgery regardless of the intervention. Likewise, longer surgical recovery and prolonged medication use led to longer alcohol cessation for some participants.

Though all participants met “risky” alcohol use eligibility requirements for the study using a validated alcohol use screening and risk identification tool, many did not perceive themselves as drinking enough to put them at risk, especially in the low-responder group. Instead, some of these participants described dismissing the intervention content about alcohol use and surgical risk because they felt it did not apply to them. A modified approach that uses different terminology or alternative discussion points could help address this barrier to change. This could include acknowledging that some participants may be surprised by feedback suggesting their level of alcohol use could be linked to surgical risks and discussing their reactions to this information.

There were also aspects of study participation and surgery that influenced participants’ alcohol use separate from the intervention content. There was evidence of assessment reactivity, particularly for participants in the brief advice condition for whom the baseline and follow-up study assessments represented a much larger proportion of their participation time than the actual intervention itself. Being able to visually see patterns of alcohol use on our assessment calendar was illuminating and motivating for some, which is consistent with research finding that the simple act of measuring behaviors and comparing them to external standards or goals can result in lasting behavior change [33], with larger effects on goal attainment when outcomes are physically tracked and recorded [34].

Importantly, several participants reacted negatively to the term intervention, which was used to describe study treatment conditions in several study-related documents and contact scripts. For these individuals, the term intervention triggered negative associations with confrontational Johnson-style interventions [35] during which a person with an addiction is confronted by family and friends and sent to long-term rehabilitation, sometimes against their will. After learning of this unintended association, and the negative reaction across several participants, in July 2020, we removed this term from study documents. For example, in the consent form, we changed references from study interventions to study conditions or treatments. In addition, using prehabilitation program language could also reduce stigma by indicating the program’s goals are to optimize a patient’s health for surgery and subsequent recovery. In our next study phase, we will engage community members in reviewing all our study documents prior to recruitment for clarity and interpretation.

This study has several limitations. Recruitment was paused due to the COVID-19 pandemic and the resulting cancellation of elective surgeries. When surgeries and recruitment resumed at a reduced capacity, it is possible that surgeries and participants were different in significant ways from those scheduled prior to the pandemic. Interviews were conducted by a researcher on the study team, which may have increased the risk of social desirability bias; however, it was a team member who had not been in contact with participants prior to the interview (ie, not the health coach or recruiter), and we reminded participants that the purpose of the interview was to inform intervention improvement, and they were encouraged to provide honest and critical feedback. Finally, while qualitative data can help deepen understanding of participants’ intervention experience, results from this study should not be considered generalizable to broader populations. The next step in research includes a fully powered sequential multiple assignment randomized trial that will randomly assign participants to treatment at baseline and then reassign them to different treatments and or treatment intensities based on initial treatment response and low response.

In conclusion, through in-depth interviews with a select sample of participants in the ASPIRE pilot trial, we found that both health coaching and brief advice conditions were acceptable pending some minor improvements and modifications. Further research is needed to test the effectiveness of these interventions in reducing drinking before and after surgery, both in the short and long term. Future research and programmatic work on these topics should carefully consider the use of lay versus technical (ie, psychiatric) terminology (eg, intervention) to avoid unintended negative reactions. Future research should also investigate whether assessments and feedback alone, without being coupled with brief interventions or health coaching, may be sufficient in changing participant alcohol use behaviors before or after surgery.