Feasibility of a telehealth-based contingency management intervention for alcohol use disorders using the phosphatidylethanol (PEth) 16:0/18:1 alcohol biomarker: a pilot randomized trial

Abstract

Background:

Phosphatidylethanol (PEth) is a blood-based biomarker for alcohol consumption that can be self-collected and has high sensitivity, specificity, and a longer detection window compared to other alcohol biomarkers.

Objectives:

We evaluated the feasibility and acceptability of a telehealth-based contingency management (CM) intervention for alcohol use disorder (AUD) using the blood-based biomarker PEth to assess alcohol consumption.

Methods:

Sixteen adults (7 female, 9 male) with AUD were randomized to Control or CM conditions. Control participants received reinforcers regardless of their PEth levels. CM participants received reinforcers for week-to-week decreases in PEth (Phase 1) or maintenance of PEth consistent with abstinence (<20 ng/mL, Phase 2). Blood samples were self-collected using the TASSO-M20 device. Acceptability was assessed by retention in weeks. Satisfaction was assessed with the Client Satisfaction Questionnaire (CSQ-8) and qualitative interviews. The primary efficacy outcome was PEth-defined abstinence. Secondary outcomes included the proportion of visits with PEth-defined heavy alcohol consumption, negative urine ethyl glucuronide results, and self-reported alcohol use.

Results:

Retention averaged 18.6 ± 8.8 weeks for CM participants. CM participants reported high levels of satisfaction (CSQ-8, Mean = 30.3 ± 1.5). Interview themes included intervention positives, such as staff support, quality of life improvement, and accountability. 72% of PEth samples from CM participants were consistent with abstinence versus 34% for Control participants (OR = 5.0, p = 0.007). PEth-defined heavy alcohol consumption was detected in 28% of CM samples and 52% of Control samples (OR = 0.36, p = 0.159). CM participants averaged 1.9 ± 1.7 drinks/day versus 4.2 ± 6.3 for Control participants (p = 0.304).

Conclusion:

Results support the acceptability and satisfaction of a telehealth PEth-based CM intervention, though a larger study is needed to assess its efficacy [NCT04038021].

Introduction

In 2020, 28 million Americans experienced alcohol use disorder (AUD), yet only 25% of those with AUD received treatment (1). The gap between the need and provision of treatment suggests there are major barriers to accessing AUD care. Barriers to in-person treatment include stigma, transportation costs, time off work, and the need for childcare. Due to the COVID-19 pandemic, substance use disorders treatment shifted to telehealth delivery. Nearly 60% of patients who received substance use disorder treatment did so via telehealth-based care in 2020 (2). While telehealth delivery has limitations (e.g., access to and comfort with technology), it provides an opportunity to improve AUD treatment access for the more than 20 million Americans with an AUD who currently do not seek in-person care.

Contingency management (CM) is an AUD intervention that was adapted for telehealth (3–6). CM is based on the principles of operant conditioning, in that tangible reinforcers are provided when a participant submits evidence of alcohol or drug abstinence (7,8). CM requires the collection of urine, blood, breath, sweat, or saliva biospecimens to verify abstinence. Previously, we developed and tested a CM model for AUD using the alcohol biomarker ethyl glucuronide (uEtG). uEtG detects alcohol use over a two- to five-day window, thus requiring twice-weekly in-person visits to monitor abstinence. In three previous studies, we demonstrated that CM is an effective intervention for AUD. However, attrition in this 12-week model was as high as 57%, suggesting that many people are not able to attend frequent in-person visits (9–11). There are commercially available telehealth-based CM products designed to treat AUD, and initial data suggests these CM models have high patient acceptability (5). At the same time, these telehealth models have limitations, such as the requirement of wearing a costly and potentially stigmatizing transdermal monitor or submitting multiple breath samples daily to accurately confirm alcohol abstinence (12).

Phosphatidylethanol (PEth) is a blood-based biomarker for alcohol consumption that has high sensitivity, specificity, and a longer detection window compared to other alcohol biomarkers (13). The PEth homologue 16:0/18:1 has a half-life of 7.8 ± 3.3 days (14). This detection window requires less frequent sample collection to verify abstinence, thus reducing burden on providers and individuals receiving AUD interventions. In a within-subjects pilot study of an in-person PEth-based CM intervention, participants were 2.3 times more likely to submit a PEth-negative sample during the CM phase than the non-contingent control phase (15). These findings demonstrate that PEth is a feasible biomarker for monitoring alcohol use during CM.

Based on these promising results, we designed a larger feasibility trial of the in-person PEth-based CM intervention (16). We focused this study on formerly homeless, currently housed, individuals because in a previous clinic-based study we found that those who were unhoused were eight times more likely to withdraw from CM (17). We hypothesized that removing treatment barriers, by bringing CM directly to housing programs, would reduce drinking in addition to increasing housing tenure in this high-risk population. Therefore, the goal of our study was to test the feasibility of implementing the intervention in supported housing.

We utilized the lengthy detection period of PEth to create a 26-week CM intervention, where the frequency of PEth monitoring and the delivery of reinforcement decreased once participants achieved a PEth level consistent with two to four weeks of abstinence (i.e., <20 ng/mL) (18,19). Extending the duration of CM in a feasible manner is important, as a longer CM intervention period is associated with improved post-treatment outcomes (20). Due to COVID-19 stay-at-home orders, we modified our in-person CM protocol to a telehealth format with interviews and sample collection conducted with staff via Zoom. To facilitate remote self-collection of capillary blood during this study, we used the TASSO-M20 device to collect samples from the upper arm rather than the typical finger stick method for dried blood spot cards.

The current pilot study reports a telehealth PEth-based CM intervention for adults with AUD who were formerly homeless. Quantitative and qualitative measures were used to assess levels of acceptability and satisfaction for the intervention. We hypothesized that the CM intervention would meet our a priori thresholds for acceptability and satisfaction as defined below, and that qualitative data would support satisfaction with the CM intervention. Additionally, we hypothesized that CM efficacy, as assessed by PEth-defined abstinence and housing tenure, would be higher in the CM group than in the Control group.

Materials and methods

Participant recruitment and eligibility

Participants were recruited from housing programs in Spokane, WA, and digital advertising on social media outlets (i.e., Facebook, Instagram) from the Mountain and Pacific Time Zones. Participants were recruited from January 2020–February 2022 and were eligible for baseline if they 1) were at least 18 years of age, 2) consumed at least three standard drinks (assigned female at birth) or four standard drinks (assigned male at birth) on two or more occasions or had ≥14 drinks in the last 2 weeks; 3) self-reported stable housing (i.e., housed for ≥2 weeks at the time of study enrollment); 4) had unstable housing for ≥30 days; and 5) had access to WiFi for video conferencing. Participants were ineligible if they had 1) an inability to provide informed consent or failed tests for competency using the MacArthur Competency Assessment Tool for Clinical Research for 18–64 years old (21) or the Montreal Cognitive Assessment 5-minute protocol for 65+ years old (22); 2) history of alcohol withdrawal-related seizures; 3) plans to become pregnant or were currently pregnant; 4) severe substance use disorder other than AUD or cannabis use disorder, as determined by SCID-5, and used that substance in the past 30 days or tested positive for that substance at baseline (i.e., UScreen 5-Panel Drug Test Cup: cannabis, cocaine, opioids, amphetamines, methamphetamine); or were 5) currently enrolled in another CM study.

Eligible individuals signed informed consent electronically, then completed a baseline interview over HIPAA compliant Zoom using their personal phone or tablet. Study procedures were approved by the Washington State University’s Institutional Review Board and registered on clinicaltrials.gov [NCT04038021].

Randomization and treatment conditions

After the baseline visit, those who 1) submitted blood samples consistent with alcohol use in the past 28 days (PEth ≥20 ng/mL) (19) and 2) had a diagnosis of a current AUD, as assessed by the Structured Clinical Interview for DSM-5 (SCID-5, mild, moderate and severe, score 2–12), were randomized to CM or the Control (non-contingent) condition using permuted block randomization. Notably, a randomized allocation table was uploaded to REDCap which allowed participants to be randomized by study staff in real-time. Participants were stratified for randomization based on sex assigned at birth and baseline uEtG (i.e., positive/negative, cutoff ≥300 ng/mL). We used uEtG results to randomize participants instead of PEth 16:0/18:1 because there was not a verified excessive drinking cutoff for PEth when study procedures were established (19). However, our previous studies established that a uEtG cutoff of ≥300 ng/mL represents recent heavy drinking and that heavy drinking at baseline predicts CM treatment outcomes (23).

Previously we determined that 50 randomized participants would be needed to detect differences in alcohol-related outcomes using the original in-person protocol (16). However, due to pandemic-related delays and additional costs associated transitioning to telehealth format, we randomized 16 participants. For both conditions, we administered gift cards following a desired behavior being achieved (i.e., positive reinforcement), be that attending visits and submitting samples (i.e., CM and Control) or reducing alcohol consumptions (i.e., CM). Both conditions received a $10 compensation for submitting blood samples in Phase 1 and a $20 compensation for submitting blood samples and completing monthly surveys in Phase 2. Compensation for attending the study visit alone was not provided.

Control condition (non-contingent)

During Phase 1, participants attended visits once a week for 4 weeks then transitioned to Phase 2, during which they attended a visit every other week for a 4-week duration followed by one visit every 4 weeks for the remainder of the 26-week intervention period (see Figure 1). Following the analysis of samples, Control participants received additional reinforcers, regardless of their PEth levels, equal to the average amount earned by the CM group over the prior 4 weeks.

Figure 1.

Study flow for both conditions. in Phase 1, CM participants attended weekly visits and received reinforcement for week-over-week decreases in their PEth levels. Once abstinence was established in Phase 1 (PEth <20 ng/mL), CM participants transitioned to Phase 2 where they met every-other-week for the first 4 weeks and then once every 4 weeks for the remainder of the intervention, as long as abstinence was maintained. Should CM participants’ PEth results indicate alcohol consumption (≥20 ng/mL), they returned to Phase 1 with weekly visits until abstinence was once again achieved (see dashed line). Control participants met weekly for 4 weeks in Phase 1, then in Phase 2 transitioned to every-other-week for the first 4 weeks, and then once every 4 weeks for the remainder of the 26-week intervention, regardless of their PEth results. Reinforcers were not contingent on PEth levels for the Control condition. Both conditions received 26 weeks of treatment in total.

CM condition

CM participants received reinforcers for week-over-week reductions in PEth levels during Phase 1. If after 4 weeks a participant attained a PEth level consistent with abstinence (PEth <20 ng/mL), they then transitioned to Phase 2. During Phase 2, they submitted blood samples every other week for 4 weeks followed by visits once every 4 weeks for the remainder of the 26-week intervention, as long as PEth levels remained below 20 ng/mL. In addition to reinforcers for submitting samples, CM participants received $20 at each visit if there was a week-over-week decrease in their PEth levels (Phase 1) or PEth indicated sustained abstinence (<20 ng/mL, Phase 2). An additional $5 was earned for each week of consecutive abstinence as confirmed by PEth. If PEth levels were ≥20 ng/mL during Phase 2, then no reinforcement was given and participants reentered Phase 1 (i.e., weekly visits and sampling). Once PEth levels were again consistent with abstinence (<20 ng/mL), participants reentered Phase 2 and their reinforcer magnitude was reset to $20 for the next sample that was consistent with abstinence (see Figure 1). Whether a CM participant returned to Phase 1 or not, the CM treatment duration remained 26 weeks in total. Accordingly, the total CM earnings for a participant who submitted all samples and was continuously abstinent over 26 weeks was $2,045.

Blood collection and PEth analysis

Before each on-line visit, blood and urine sample collection supplies, with the visit number and participant deidentified number specified, were sent by FedEx to the participant’s residence. A prelabeled and prepaid FedEx Clinical Pak was included for shipping samples post-collection. Blood sample collection was conducted on camera so research staff could verify authenticity of the sample and provide technical support.

Blood samples were self-collected by participants using the TASSO-M20 device (Tasso, Inc. Seattle, WA, US). Prior to blood collection, participants vigorously rubbed their upper arm for approximately 30 seconds to increase blood flow. Following, they used an isopropanol alcohol pad to clean the area. The TASSO-M20 device was then adhered to the bicep (i.e., pods facing down) using the device’s adhesive surface. Capillary blood flow was initiated by firmly pressing the TASSO-M20 button to lance the skin with a microneedle. The device remained in place for 5 minutes after the skin was lanced or until all four sample plugs were adequately saturated with blood (20 μL ± 5%/plug). The device was then removed from the arm, film cover over the plugs was removed to allow drying, and the device was placed into a sealable bag with a silica packet. The participant then mailed the TASSO-M20 device to the University of Texas Health San Antonio for PEth analysis using the prepaid FedEx Clinical Pak.

PEth 16:0/18:1 was quantified using high-performance liquid chromatography with tandem mass spectrometry detection, as previously described in (24). All solutions were prepared with Milli-Q Plus water and reagents were LC/MS grade (Fisher Scientific, Waltham, MA). PEth 16:0/18:1 was acquired from Cerilliant (Round Rock, TX) and deuterated PEth 16:0/18:1 (PEth 16:0/18:1-d₅) was used as the internal standard (Echelon Biosciences, Salt Lake, UT). Calibrator and control samples had a volume of 40 μL. Two Tasso plugs from each experimental sample were analyzed. Each sample, excluding the blank control, was spiked with 10 μL of a 0.1 μg/mL 16:0/18:1-d₅ internal standard solution to achieve a 25 ng/mL concentration. 1 mL of isopropanol was also added to each sample (calibration, control, and two TASSO plugs). Samples were vortexed (1 minute), shaken (45 minutes) using an Eberbach Shaker set at high speed, and sonicated (single 5-second burst; Q-Sonica sonicator at 10% max). Samples were then centrifuged for 30 minutes (3,200 g) at 4°C. The clear supernatants were transferred to new tubes and evaporated to residue with a gentle stream of nitrogen at 30°C. The residues were dissolved in 100 μL of a 1:1 resuspension solution of mobile phase A (40:60, 2 mM ammonium acetate:acetonitrile) and mobile phase B (isopropanol) and then transferred to 1.5 mL snap-cap polypropylene microcentrifuge tubes and centrifuged for 10 minutes (3,200 g) at 4°C. The eluted samples were transferred to 300 μL polypropylene auto-sampler vials and then 10 μL were injected into the HPLC/MS/MS system. PEth levels were expressed as ng/mL.

Acceptability and satisfaction outcomes

We assessed acceptability by the number of weeks retained. Satisfaction was measured using the Client Satisfaction Questionnaire (CSQ-8). CSQ-8 scores range from 8 to 32 with higher scores indicating greater satisfaction. Our a priori threshold for intervention satisfaction was a summed score of ≥24 (25). CM participants also completed up to three qualitative interviews focused on assessing the feasibility of the TASSO-M20 device and CM intervention at weeks 4, 26 (completion of CM), and 38 (follow-up interview). Interviews were recorded and later transcribed using the Amazon Web Transcribe Service.

Interview questions included: How was the blood collection process? How would you change blood collection to make it better? The intervention was on zoom; tell me what you liked/didn’t like about it? What do you think about waiting a week to get your rewards? What about the process (like blood draws and online visits) do you think might encourage others to participate in the intervention? What about the process do you think might discourage others from participating in the intervention?

Efficacy outcomes

The primary alcohol-related efficacy measure was PEth-defined abstinence at each study visit. All participants had PEth levels positive for alcohol use (non-abstinence) at randomization (PEth ≥20 ng/mL). Abstinence was defined as a decrease in PEth level from the previous week in Phase 1 and a PEth level <20 ng/mL in Phase 2. We also assessed the longest duration of PEth-defined abstinence (i.e., consecutive weeks) and total duration of abstinence over the course of treatment (i.e., total weeks). Secondary alcohol-related outcomes included: PEth-defined heavy alcohol consumption (PEth ≥200 ng/mL) as a binary outcome (19), uEtG-defined evidence of drinking (uEtG ≥300 ng/mL) as a binary outcome, and self-reported number of drinks per day assessed at baseline and over the 181-day intervention period using the Timeline Follow Back (TLFB; Sobell & Sobell, 2000). Urine collection took place off camera while the dip card (Healgen Scientific Ltd, Confirm Biosciences) procedure and interpretation of results was conducted on camera with study staff. Once the uEtG results were recorded, participants discarded the sample cup and dip card at their residence.

The primary non-alcohol related outcome assessed was housing tenure. During the baseline interview, housing history was assessed via the Housing History Timeline (26) and during monthly interviews, housing was assessed via the Residential Timeline Followback (27). The Residential Timeline Followback allowed us to assess the days each participant was housed or unhoused across the study period.

Analytic plan

Demographic variable analyses

Groups were compared for differences in age, sex, race/ethnicity, and years of education with independent group t-tests for continuous variables and exact tests for categorical variables. Demographic variables were evaluated for use as fixed covariates or categorical fixed-effect independent variables when differences among groups were at p < .05 and as auxiliary variables to address missing data in multiple imputation procedures (see below).

Qualitative interview analysis

Two independent coders developed a codebook to assess interview themes. Codes were templated on the qualitative interview question guide and included intervention positives, intervention weaknesses, telehealth delivery, TASSO experiences, TASSO improvements, reinforcer delay, and intervention improvements. Once codes were established, the coders met with the principal investigator to discuss the themes with textual evidence for each. Minor revisions were made, resulting in the final themes of intervention positives, telehealth delivery, TASSO experiences, intervention weaknesses, research procedure improvements, reinforcer delay, and intervention improvements. The frequency of themes and subthemes were assessed using template analytic methods to evaluate implementation outcomes (i.e., satisfaction with the intervention and related procedures) (Crabtree and Miller, 1992).

Acceptability and satisfaction outcome analyses

Acceptability was measured as weeks of intervention retention. For satisfaction, the proportion of CM participants who reported a CSQ-8 score that exceeds the acceptability cutoff score of 24 was calculated.

Primary and secondary efficacy outcome analyses

All study participants retained housing, likely due to COVID-19 related eviction prohibitions, thus we were unable to compare group differences in housing outcomes. Generalized estimating equations (GEE) with the appropriate link function (logit for binary outcomes or identity for continuous outcomes) were used to evaluate between-group differences in longitudinal change in the primary outcome, PEth-defined abstinence, and each secondary alcohol-related outcome. Models were fit with and without a group-by-time interaction to assess improvement in goodness of model fit; in all models, the interaction term did not improve model fit, and was, therefore, removed to obtain a more parsimonious model. Goodness of fit was also assessed with various working correlation matrix structures; in all models, goodness of model fit was optimized with an autoregressive correlation matrix. Prior to employing GEE, multiple imputation was used to account for missing data per outcome. Imputation utilized existing instances of data to be imputed and auxiliary variables age, sex, race/ethnicity, years of education, and group assignment to help satisfy missing-at-random assumptions and improve the precision of imputation. Reported GEE results are pooled analyses based on five imputed data sets with odds ratios (ORs) and proportions reported for binary outcomes and estimated marginal means reported for continuous outcomes. Groups were also compared for differences in the longest duration of abstinence and total duration of abstinence during treatment as primary outcomes with independent group t-tests.

Means and standard deviations were obtained for continuous variables, with frequencies and percentages obtained for categorical variables. Quantitative variables were analyzed on an intention-to-treat basis to reduce the potential for bias in group comparisons. We established a type I error rate of p < .05. Reported p-values are two-sided type I error rates and all quantitative analyses were conducted with IBM SPSS Statistics, v. 28.0 (IBM, Armonk, New York, US).

Results

Participant demographics

All study recruitment and procedures took place after transitioning the study to a telehealth format. A total of 106 people were screened. Thirty-nine people were eligible for the baseline assessment, but 30 consented to study procedures and completed baseline measures. Of these, 16 individuals had a PEth level ≥20 ng/mL and were randomized to the Control condition (n = 9) or CM condition (n = 7). The primary causes for study ineligibility at screening included not meeting housing requirements and not meeting alcohol use requirements based on self-report. The primary cause for ineligibility following baseline was having a PEth level below 20 ng/mL cutoff (i.e., no recent drinking). As seen in Table 1, groups were statistically similar on all demographic and baseline alcohol variables (p values ≥ .05).

Table 1.

Demographics and baseline alcohol measures.

Variable	Control (n = 9)	CM (n = 7)	p-value
Female	4 (44%)	3 (43%)	1.0
Race/Ethnicity			.585
Hispanic or Latino	2 (22%)	0 (0%)
Black or African American	1 (11%)	2 (29%)
White	6 (67%)	5 (71%)
Age	47.4 (14.9)	44.0 (15.6)	.659
Education (Years)	12.9 (2.2)	15.3 (2.2)	.055
Lifetime duration unhoused (Months)	79.1 (144.7)	57.9 (58.0)	.813
Baseline PEth level (ng/mL)	279 (255)	1259 (1588)	.131
Positive uEtG at Baseline (≥300 ng/mL)	6 (67%)	5 (71%)	.714
Drinks/day	4.3 (4.4)	4.6 (3.4)	.889

Mean (SD); n (%).

CM: Contingency management.

PEth: phosphatidylethanol.

uEtG: urine ethyl glucuronide.

Acceptability and satisfaction outcomes

Quantitative

CM participants averaged 18.6 weeks of retention (SD = 8.8) and had a 43% retention rate (i.e., 3 of 7 participants completed the entire 26-week intervention). Reasons for CM attrition included disagreement with PEth results (n = 1), unrelated health problems (n = 2), and time commitment conflict (n = 1). All CM participants that completed the CSQ-8 had a total score that exceeded our a priori threshold of 24 points. The mean score CSQ-8 score for CM participants was 30.3 (SD = 1.5).

Qualitative

Five CM participants completed a total of 12 qualitative interviews (see Table 2). The most discussed theme was intervention positives (45 mentions) with subthemes of staff support, abstinence-associated quality of life improvement, accountability, and gift card reinforcers. The second most mentioned theme was feedback on the telehealth-based intervention (26 mentions), convenience, using and learning to use Zoom, and the pros/cons of virtual versus in-person interaction. The third most reported theme was coded as TASSO-M20 device experiences (19 mentions) including subthemes related to discomfort, device usability, initial anxiety about using the device, and suggested TASSO-M20 device improvements. Additionally, participants discussed intervention weaknesses (14 mentions), including subthemes of blood collection, the impact of drinking culture on the willingness to participate (e.g., social drinking, gaming drinking patterns to still receive incentives, etc.), and intervention procedures. Participants also offered suggestions to improve the intervention (7 mentions), such as expanding eligibility criteria beyond formerly homeless people, providing education about substance use, utilizing more time-sensitive alcohol measures, such as breath alcohol content. Lastly, participants gave feedback about ways to improve the research procedures (14 mentions). When asked about the week delay between providing samples and receiving reinforcers, no participants were bothered by waiting (9 mentions).

Table 2.

Qualitative interview theme frequencies and samples.

Theme	Example Quote
Intervention Positives (45)
Staff Support System	I’ve got, you know, kind of a support system with [staff].
Quality of Life/Stopped Drinking	I feel a lot better. I don’t feel sick. Drinking that much every weekend obviously puts a strain on your body. I have a two-year-old and … I want to be able to be the best I can be for him … This [gave] me the push that I probably was gonna keep making excuses not to.
Accountability	It gave me accountability having someone, you know, track my progress. It wasn’t really hard at all. I liked it. I liked the program, I would do it again
Incentives	I came into the study sort of wanting to stop drinking, but sort of wanting to continue. And so I wasn’t … I wasn’t desperate to become sober. But I ended up doing it for the money and you know it was the result. I got six great months of…Waking up in the morning feeling good.
Telehealth Methods (26)
Convenience	The testing part is easy, I mean, I don’t think it could be any easier … everything shows up and then you just mail it off.
Zoom	I’ve done telehealth but I’ve never done zoom, so it’s been pretty neat learning how it all works … I feel comfortable with it finally, after I don’t know how long we’ve been doing it.
Telehealth-based intervention	I would say the drawback is not being able to come in, face-to-face talk to everyone. You can hear each other clearly talk. So the positive thing about it is yes, you do get to stay home and it takes away that from that time of having to drive, walk or take the bus to meet up.
Tasso Experience (19)
Discomfort/Pain	I’m a diabetic and I stick myself every day … The finger prick hurts worse and it makes you irritated. I did [the TASSO] in the same place every time … other than a little bit of blood that comes out afterward it’s nothing.
Device Usability	When it works? It’s not bad at all. We did have some difficulties in … getting any blood out of me. Probably my biochemistry … multiple sticks on a day, way less fun than just one stick.
Initial Anxiety	At first I was kind of nervous. I didn’t know what the prick would be like. It’s not bad at all.
TASSO Improvements	Once I figured out things … needed to be warmed up [it worked]. I would make sure that I had it in the bathroom with heat turned up, I open the package, take it out, so that it had more chance to get warm. When delivered and it’s zero [outside] … They need to be, they need to be warm [to work].
Intervention Weaknesses (14)
Drinking culture	It’s a huge social part of our lives in this country, it’s about being able to socially adapt. [Not drinking] it’s not for everyone.
Blood Collection	You definitely can see it [blood]. Some people aren’t going to want to be sticking that [needle] into their arm.
Intervention Procedures	The [gift card] site. You can’t combine each week. So that kind of sucks … I have to go and use different cards or can’t get some of the cards sometimes.
Research Procedure Improvements (14)
Data Collection	If you could do [monthly questions], like online … that gives you more time to think. I’m very visual, more than audio. We don’t all receive information in the same way … maybe if they could have a choice of how [to do questions] … that would make it easier for some people.
Other Procedures	How long the monthly meeting is … I wouldn’t necessarily say discourage it, but … it’s annoying to sit in front of a camera for an hour to an hour and a half. If it was in person and took 30 minutes to an hour, it would be a little bit easier.
Other (16)
Intervention Improvements	I don’t know why some people joined because they wanted help, rewards or what. I don’t know how you guys market it to participants, my friend told me she didn’t qualify, I know she wanted to participate but she didn’t qualify, but the ability to accept people – her boyfriend is in the restaurant industry and it’s a big problem there, always trying out new drinks.
Incentive Delay	I knew what I was gonna use it for. It’s a pretty big list of places you can choose from … and [incentives] happened you know a week later or less than that. So I didn’t care about [waiting].

Efficacy-related outcomes

Primary outcome

CM participants were five times more likely than Control participants to submit PEth samples indicative of alcohol abstinence (OR = 5.0 [CI = 1.54–16.67], p = .007). The CM condition was abstinent on approximately 72% of study visits compared to 34% of study visits for the Control condition.

Secondary outcomes

The longest duration of abstinence (i.e., consecutive weeks) was more than 3 times longer for the CM group (12.3 ± 12.2 weeks) compared to the Control group (3.4 ± 3.1 weeks, p = .106). Likewise, the total number of weeks abstinent during treatment was more than 2 times longer for the CM group (13.9 ± 10.9 weeks) than for the Control group (4.9 ± 4.7 weeks, p = .078). PEth samples from CM participants were 64% less likely to indicate heavy alcohol consumption than those from Control participants (OR = 0.36 [CI = 0.09–1.49], p = .159). Additionally, CM participants were 57% more likely to submit negative uEtG samples (OR = 2.33 [CI = 0.55–10.00], p = .252). The CM condition also self-reported fewer drinks per day averaged across the intervention period (1.9 ± 1.7 drinks per day) compared to those self-reported by the Control group (4.2 ± 6.3 drinks per day, p = .304). No participants lost their housing during the study. Therefore, we did not analyze the impact of the intervention on housing status.

Discussion

Here we report data that supports the feasibility and initial efficacy of our pilot telehealth PEth-based CM intervention. We developed and tested this telehealth intervention during the COVID-19 pandemic due to stay-at-home orders restricting our originally proposed in-person study (16). Procedural changes included moving our in-person data collection to a Zoom format, replacing finger stick sample collection with the TASSO-M20 device for blood self-collection, using social media to recruit participants, and providing e-gift cards instead of physical gift cards. Additionally, we were not able to assess the effects of our intervention on housing tenure because none of our participants lost housing during the study, likely due to COVID-related eviction moratoriums. Though our sample size was small, our initial results support the acceptability, satisfaction, and efficacy of our telehealth PEth-based CM intervention.

Our CM intervention was 26 weeks versus the typical 12-week CM model used in other studies (28). This is an important distinction when considering that 71% of the CM participants (n = 5) in the present study completed the first 12 weeks of the intervention compared to 47% in a 12 week in-person CM for AUD model (10). In contrast, one person in the Control group withdrew from the study due to difficulties with study technology. The higher retention rate in the Control group versus CM is comparable to our previous studies that utilize a non-contingent control group (10). Furthermore, three of the four individuals who withdrew from our CM intervention did so for reasons unrelated to alcohol use or CM procedures (e.g., medical issues, schedule conflicts, etc.). Satisfaction with CM was high on quantitative and qualitative measures.

Though not part of our original pilot research design, our pivot to self-collection of blood was also feasible regarding the timely shipment of samples. A total of 135 blood samples were collected and shipped to the lab for assessment of PEth levels. The median days between participants collecting blood and shipping samples was 0. We considered a blood sample to be “late” if the PEth result was unavailable before the subsequent blood collection (i.e., next study visit). Of the 24 (15%) samples that were late, 21 (87%) occurred during Phase 1, where blood collection occurred weekly. For CM, 18 samples in total were late and all occurred during Phase 1. Additionally, there was a median of 6 days between sample collection and administration of electronic gift card reinforcers. Despite delays in CM reinforcers associated with “late” sample submissions, qualitative data indicated that CM participants did not find this delay problematic. Nevertheless, modifications to our approach, such as providing a bonus incentive to participants who immediately ship their sample or utilizing overnight shipping, might be necessary during Phase 1 to assure that CM reinforcers are provided prior to a participant’s next study appointment.

Group differences for our primary alcohol outcome, PEth-defined abstinence, were consistent with previous in-person CM studies for AUD (4,9,11,15). In our previous in-person PEth pilot study, we found that CM participants were 2.3 times more likely than Control participants to submit PEth-negative samples (15), whereas the current telehealth intervention found that CM participants were five times more likely to submit PEth-negative samples. Additionally, our current results indicate that CM participants attained clinically meaningful periods of alcohol abstinence, with the number of weeks abstinent being approximately three times longer for the CM condition than for the Control. Though our small sample size precluded significant findings across our secondary alcohol measures, we did find a similar trend with PEth-defined heavy alcohol consumption, uEtG positive results, and self-reported number of drinks per day all being reduced in the CM condition compared to the Control condition.

A major strength of our telehealth CM intervention is the use of the PEth homologue 16:0/18:1 to monitor alcohol consumption. The long half-life of this biomarker enabled us to have a flexible sampling and reinforcement schedule, with collection occurring as infrequently as once a month. Though decreasing sample and reinforcement frequency is not unique for CM interventions, utilizing a biomarker that can accurately assess alcohol consumption over a period of a month is novel. Using the PEth biomarker also allowed us to assess heavy alcohol consumption (PEth level ≥200 ng/mL). Additionally, our PEth-based CM had an intervention period of 26-weeks compared to that of a typical CM model (12–16 weeks). Longer intervention periods have been shown to be the primary predictor of improved post-treatment outcomes (20). Lastly, our telehealth format allowed us to reach individuals who would otherwise not access treatment due to burdens related to location (e.g., rural areas), schedule restraints (e.g., work, family) or stigma (e.g., wearable devices or breathalyzers). In fact, 64% of our total sample was treatment naïve.

The primary limitation of this feasibility study is its small sample size. Stay-at-home measures implemented during COVID-19 compromised study recruitment, to include the time recruitment was suspended to transition our in-person procedures to telehealth. A larger study is needed to test the efficacy of our PEth-based telehealth CM intervention; however, our results are consistent with previous large in-person CM studies that found medium to large effects on drug and alcohol use (10,29). The relatively high costs of PEth testing ($100/sample) and the TASSO-M20 device ($25/device), as well as the added costs of sample shipping also challenged the feasibility of implementing our telehealth CM intervention. The implementation of a TASSO-M20 device in the current study decreased the rate of sample collection failure (i.e., fewer than two saturated plugs out 4) to 3% compared to a failure rate of 16% using dried blood spots in our previous study (30).

Savings associated with reductions in alcohol use and related health benefits also have the potential to offset costs associated with the procuring TASSO-M2 devices and PEth analysis. Another potential limitation is the delay between sample collection and the administration of reinforcers. Though CM participants reported the delay was not problematic, and we observed group differences in our primary study outcome. Other biomarkers, such as smartphone synced breathalyzers allow for more timely delivery of reinforcers. However, in our studies we found that breathalyzer measures have to be collected multiple times a day to ensure accurate detection of drinking. Some participants find this to be inconvenient and burdensome. Breathalyzers also require calibration and direct observation or video recording technology to record collections and verify the identity of the person submitting the sample.

The current study supports the feasibility of our telehealth PEth-based CM intervention. The associated acceptability, satisfaction, and efficacy of CM procedures observed in this study are consistent with other in-person CM interventions for AUD. Though our primary and secondary alcohol-related outcomes are promising, a larger study is needed to definitively assess the efficacy of our intervention. The ability of our telehealth approach to reach participants that cannot attend in-person appointments due to their location or schedules (e.g., rural areas, work, childcare), and lower participant burden associated with reduced monitoring, makes our PEth-based telehealth CM intervention a potential strategy to treat the 20 million Americans with AUDs that currently do not seek in-person care.