Implementing an evidence-based prize contingency management protocol for stimulant use

Abstract

Introduction:

Contingency management (CM) is an efficacious treatment for stimulant use disorders. Research has developed support materials for the clinical delivery of prize-based CM and they are widely accessible, but few resources are available to support design and preparation for CM implementation. This guide aims to fill that gap.

Method/Results:

The article outlines a suggested prize CM protocol and discusses the best practices most aligned with evidence-based practices and acceptable-if-necessary modifications. The article also highlights modifications that are not evidence-based and not recommended. In addition, I discuss practical and clinical aspects of preparing for CM implementation.

Conclusions:

Deviations from evidence-based practices are common, and poorly designed CM is unlikely to impact patient outcomes. This article provides planning stage guidance to support programs’ adoption of evidence-based prize CM for the treatment of stimulant use disorders.

1. Introduction

Contingency management (CM) interventions are built on the behavioral principle of positive reinforcement. Tangible incentives are delivered when objectively verifiable target behaviors are demonstrated, with the intent to increase the probability of these behaviors occurring again in the future. In the context of substance use disorder treatment, the target behavior is often negative toxicology samples (i.e., urine, breath) and reinforcement is delivered to increase the chances of future negative samples (i.e., abstinence). CM has a decades-long evidence base (Benishek et al., 2014; Bolivar et al., 2021; Griffith et al., 2000, Lussier et al., 2006; Prendergast et al., 2006), which demonstrates that CM engenders better abstinence outcomes compared to treatment-as-usual and other robust interventions such as cognitive behavioral treatment (De Crescenzo et al., 2018; Dutra et al., 2008). Not only is CM efficacious in the short term when contingencies are in place, but it also produces more robust responses in the long term compared to other active therapies when objective long-term outcomes are used (Ginley et al., 2021).

After many years of limited CM implementation in clinical settings (Benishek et al., 2010), interest in CM adoption in increasing. In part, this interest is driven by a need to find effective treatment options for patients with stimulant use disorders as stimulant use surges nationally alongside the opioid crisis (Ellis et al., 2018). With no FDA-approved medications available for patients with stimulant use disorder, CM has gained attention as one of the few effective treatment options available from both the National Institutes of Health (Volkow, 2020) and the Substances Abuse and Mental Health Administration (2020). Reviews and meta-analyses of treatments for stimulant use disorders (Ciccarone & Shoptaw, 2022; De Crescenzo et al., 2018; Farrell et al., 2019; Ronsley et al., 2020) are uniform in recommending CM as the treatment of choice for this population. Although clinical competence rating scales for delivering CM are available (Ledgerwood & Petry, 2010; Petry & Ledgerwood, 2010; links to these freely available resources provided in the reference section), fewer resources are available for the protocol design stage, which is a critically important step toward implementing an intervention that will produce benefits in patient outcomes.

This guide aims to fill that gap, with a specific focus on designing prize-based CM protocols for stimulant abstinence. Stimulant-focused protocols are the concentration given the clear need for evidence-based strategies for these disorders; however, these recommendations also apply to other substances with similar drug monitoring detection windows. Substances with substantially different monitoring detection windows (e.g., breathalyzers with detection windows of hours rather than days) would require major modifications to the proposed protocol, and these modifications are not discussed in the current article. Readers are referred to Petry (2012) for a comprehensive introduction to CM protocol development for a wide range of substances, as well as nonabstinence behavioral targets, such as attendance, treatment-related activities, and medication adherence.

First, we describe a model protocol, which is based on the Clinical Trials Network studies (Peirce et al., 2006; Petry et al., 2005) and the preferred model suggested in the CM trainings for the national Veterans Affairs implementation effort in their intensive outpatient programs (DePhilippis et al., 2018; Petry et al., 2014; Rash & DePhilippis, 2019; Rash et al., 2013). In recognition that providers may need to modify standard protocols, the subsequent sections systematically present CM design considerations that impact CM’s efficacy, highlighting those areas that can be modified and the potential impact of those modifications. We also highlight changes that would significantly decrease the impact of CM interventions, as deviations from evidence-based CM parameters are common (Rash et al., 2012; 2020) and produce ineffective programs. The last section focuses on practical issues important to the planning stage, including the importance of reinforcer selection, supplies and auditing needs, and integration into clinical workflows.

2. Two systems of CM delivery

The two major systems for CM delivery are voucher and prize CM; the latter is also referred to as the Fishbowl technique. Both systems target a specific, objectively verifiable behavior using tangible reinforcement. Both systems are efficacious and have decades of science to support them (Benishek et al., 2010; Dutra et al., 2008; Griffith et al., 2000; Lussier et al., 2006, Prendergast et al., 2006). Possible earnings in voucher CM are predictable, with both the patient and clinician knowing the amount of possible reinforcement in advance of any meeting—this is a major difference between the two approaches. In a voucher schedule, the first demonstration of the target behavior earns vouchers of a specific dollar amount, and the voucher amount typically escalates with consecutive performance (Higgins et al.,1993;1994; 2019). For example, for the first stimulant negative sample, a patient earns $2.50; for the second consecutively negative sample, the patient earns $3.75; for the third, they earn $5, and so on. Earned vouchers can be exchanged for goods or services.

In contrast, prize CM introduces probability (i.e., sometimes you win a prize and sometimes you do not) as well as variability in prize magnitude (small, large, jumbo). Given these features, neither the patient nor the clinician can predict actual expected reinforcement value in advance of a session. These prize-based features were originally designed to lower the costs of voucher CM protocols without sacrificing efficacy. Notably, both prize and voucher CM have effective magnitudes below which CM is unlikely to impact behavior, as discussed below. When utilizing evidence-based parameters, both systems are effective in changing behavior and the selection of which to use is often based on preferences of one system versus the other. However, for simplicity, this article focuses solely on the prize-based approach. Readers interested in a model voucher CM protocol are referred to Higgins et al. (2019) and Petry (2012).

3. A model prize CM protocol for stimulant abstinence

Table 1 presents two prize CM protocols, the second of which (far right column) is based on the suggested schedule used in the VA regional CM trainings (Petry et al., 2014; Rash et al., 2013; Rash & DePhilippis, 2019). Nancy Petry designed the VA schedule with real-world implementation in mind, as such it already uses some “acceptable” modifications. For example, best practice would be thrice weekly urine testing and reinforcement; however, in recognition that thrice weekly schedules are difficult for many clinics, the suggested protocol uses twice weekly monitoring and reinforcement. Similarly, the average expected maximum magnitude of prizes per patient is in the middle of the acceptable, evidence-based range for prize-based protocols (recommended: $385 to $533; this recommendation is explained in detail in section 4.6.2 below). The original value of the suggested VA model (in 2011–2012) was $364 average expected maximum per patient, but has been adjusted here to account for inflation using the CPI index. The adjusted protocol produces an average maximum expected value of $432 (far right column of Table 1) and was accomplished by changing the value of the small prize (previously $1) to $2 (as shown in Table 4). We discuss best practices and acceptable-if-necessary modifications below. Because this protocol already includes some modifications away from ideal, we do not recommend additional modifications or they should be limited in scope.

Table 1.

Prize CM Protocols for Stimulant Abstinence – Simple Escalation and Capped

Weeks	Twice-weekly Sessions	Number of Prize Draws Possible
		Simple Escalation (for comparison to the capped schedule at right; illustrates the impact of capped schedule versus no cap on magnitude)	Draws Capped at 8 (Suggested Schedule)
Week 1	Visit 1	1	1
	Visit 2	2	2
Week 2	Visit 1	3	3
	Visit 2	4	4
Week 3	Visit 1	5	5
	Visit 2	6	6
Week 4	Visit 1	7	7
	Visit 2	8	8 (cap)
Week 5	Visit 1	9	8
	Visit 2	10	8
Week 6	Visit 1	11	8
	Visit 2	12	8
Week 7	Visit 1	13	8
	Visit 2	14	8
Week 8	Visit 1	15	8
	Visit 2	16	8
Week 9	Visit 1	17	8
	Visit 2	18	8
Week 10	Visit 1	19	8
	Visit 2	20	8
Week 11	Visit 1	21	8
	Visit 2	22	8
Week 12	Visit 1	23	8
	Visit 2	24	8
Summary:
12-weeks Duration	24 Visits	300 draws possible, with an expected maximum prize value of ~$791	164 draws possible, with an expected maximum prize value of ~$432

Notes. A client who attends all sessions and submits all stimulant negative urines with no disruptions would follow the schedule as shown. Clients with disruptions, whether it be absences or stimulant positive urines, would not earn the full amount of draws possible. Average expected maximum prize value is based on the use of a “standard” fishbowl involving 500 slips, with 250 non-winning, 209 smalls (~$2 in value), 40 larges (~$20 in value), and 1 jumbo (~$100 in value).

Table 4.

Fishbowl Composition

Slip Categories	Quantity of Slips in Fishbowl	Probability	Prize Value
Positive Affirmations (non-winning)	250	.500	$0
Small prize	209	.418	$2
Large prize	40	.080	$20
Jumbo prize	1	.002	$100
	Total: 500 slips

Note. Positive affirmations can use all the same phrase (e.g., “good job!”) or have varied positive messages (“keep up the good work!”, “fantastic progress”). Be sure that the writing cannot be viewed through the paper or alternately use an opaque fishbowl, otherwise clients will favor slips likely to produce prizes (e.g., slips with long sentences would tend to be non-winning affirmation and might be a signal to avoid selection), which could greatly increase the costs of the program.

The suggested protocol in Table 1 (far right column) is a 12-week protocol targeting stimulant abstinence, involves twice-weekly visits, and uses a draw schedule that starts at 1 draw for the first stimulant negative urine and escalates by 1 draw for each consecutive stimulant negative urine. Once the patient earns 8 draws per negative urine, the draws are capped. The schedules displayed in Table 1 reflect “perfect” performance. That is, this patient did not miss any sessions and submitted 24 stimulant negative urines, thereby earning the full number of draws possible. Clinicians should clearly and explicitly link the chances to earn draws to the patient’s efforts toward abstinence at each session (see Ledgerwood & Petry 2010 for specific examples of clinical delivery).

3.1. Resets

As noted above, draws escalate with consecutive negative samples. When that consecutive string of negative urines is broken either by a stimulant positive urine or an unexcused absence (described below), no (zero) draws are earned for that visit, and the draw schedule resets back to its initial value of 1 draw per stimulant negative urine starting with the next negative urine submitted. If urine samples again return to consecutive negative submissions, the draws resume escalating by 1 draw up to the cap of 8 draws. Each instance of a stimulant positive urine, refused sample, or unexcused absence will reset the draw schedule. Multiple resets during the 12-week protocol are possible.

Table 2 displays the schedule of a patient who had interruptions in the progression of their CM intervention (using the “suggested protocol” from Table 1 with draws capped at 8). This patient scenario includes an unexcused absence (a “no show”) in week 3, a stimulant positive urine in week 8, and an excused absence in week 11. The patient earned zero draws on those visits (i.e., “withholding reinforcement when behavior is not demonstrated”). In addition, following the unexcused absence and the positive urine, the draw schedule was reset back to 1 draw for the next negative sample submitted (referred to as a “reset”) and resumed escalating by 1 draw for consecutive negative samples. In contrast to these events, the excused absence did not trigger a reset when the patient returned with a negative sample, and the patient’s schedule continued escalating uninterrupted. Because the patient did not attend the visit and submit a stimulant negative urine, they did not earn any draws even for that excused visit; however, because the visit was excused, the reset contingency was not applied. If the patient fails to show or the urine was stimulant positive for the return visit, then they would earn 0 draws for that return visit, and the schedule would reset back to 1 draw for the next stimulant negative urine submitted.

Table 2.

Hypothetical Schedule of a Client with Excused and Unexcused Absences and a Stimulant Positive Urine

Weeks	Twice-weekly Sessions	Number of Prize Draws Possible Using Suggested Capped Model
Week 1	Visit 1	1
	Visit 2	2
Week 2	Visit 1	3
	Visit 2	4
Week 3	Visit 1	No Show/Unexcused Absence (0 draws)
	Visit 2	1 (Reset)
Week 4	Visit 1	2
	Visit 2	3
Week 5	Visit 1	4
	Visit 2	5
Week 6	Visit 1	6
	Visit 2	7
Week 7	Visit 1	8 (Reaches cap)
	Visit 2	8
Week 8	Visit 1	8
	Visit 2	Stimulant Positive (0 draws)
Week 9	Visit 1	1 (Reset)
	Visit 2	2
Week 10	Visit 1	3
	Visit 2	4
Week 11	Visit 1	Excused Absence (0 draws)
	Visit 2	5 (No reset following excused absence)
Week 12	Visit 1	6
	Visit 2	7
Summary:
12-weeks Duration – Duration does not extend because of absences	Attended 22 out of 24 scheduled visits	Client earned 90 draws out of 164 draws possible in this hypothetical scenario.

Patients who experience absences or submit positive samples will not earn the full amount of draws. Note the impact of these events on the total number of draws earned. In this hypothetical scenario, the patient earned only 90 out of the 164 draws possible during the 12-week intervention period with only three events. Also note that patient performance (i.e., absences, resets) did not extend the duration of the CM intervention. The duration of the CM protocol is always 12-weeks regardless of engagement or performance. As an example, consider a client who met with a clinician who described the CM protocol and invited the client to participate; however, this client fails to show in weeks 1 through 7 and resurfaces in week 8. The client would be welcomed back to the CM program and informed they have 5 weeks remaining (weeks 8 through 12) to earn draws for stimulant negative urines. The duration did not extend because of the patient’s lack of engagement with the program.

3.2. Absences

Unexcused absences (and stimulant positive urines) reset the draw schedule. The most common unexcused absence circumstance is the patient who fails to show for a scheduled appointment (a “no show”). The reset feature of the schedule is applied to these unexcused absences. Unexcused absences are penalized with resets because these absences may be attempts to avoid detection of substance use (e.g., a no show to a scheduled CM appointment because of recent stimulant use). Another possibility is an excused absence, which would not disrupt the draw schedule. In contrast to unexcused absences, excused absences do not trigger resets in recognition that these events are often out of the control of the patient (e.g., clinician cancels due to illness, patient experiences a medical emergency).

See Table 3 for common examples of excused and unexcused absences. Determination of excused and unexcused absences would typically follow or expand on existing clinic policy. If no such policy is in place (or if clinicians are variable in applying the absence policy), then the clinic should develop explicit definitions specific to the CM protocol and prior to initiating the CM program to align with best practices. As a rule of thumb, if the absence is therapist- or clinic-driven (e.g., clinician is ill and patient cannot be seen by another clinician; entire clinic is closed for holiday), then the absence is excused. Among patient-driven absences, clinics can decide what reasons they will accept (if any) for excused absences and what type of verification will be required. Examples might be a medical emergency or a court day in which the patient could not be seen the day before or the day after to accommodate the absence. However, the rules related to absences—what is accepted and what is not accepted, expectations (e.g., related to advance notification), and types of acceptable verification should all be outlined from the start of the CM program. This information would be a routine aspect of explaining how the CM program works to each patient in the first CM visit. Similarly, staff should fully explain what constitutes an unexcused absence, such as a no show, and the impact of such an absence on the CM program.

Table 3.

Suggestions for Defining Excused and Unexcused Absences

Excused Absences	Unexcused Absences
Medical emergencies	No shows
Non-emergency appointments in which it is not possible to reschedule the patient’s CM appointment	Failure to cancel in advance per explicitly defined clinic policy
Non-emergency illness (per clinic policies, e.g., medical provider documentation might be required)	Failure to provide documentation as explicitly defined by clinic policy
Holidays in which timely rescheduling was not possible
Clinic-causes (e.g., entire clinic closed, and timely rescheduling was not possible due to compressed work week)
Clinician-causes (e.g., clinician was ill and coverage by another clinician or timely rescheduling was not possible)
Court date in which timely rescheduling was not possible
Short-term inpatient stay or jail
Quarantine periods (e.g., COVID exposures)
Other legitimate reasons for non-attendance (e.g., out of town, funeral)

Notes. This table lists suggestions, but each clinic/organization should determine the accepted excused and unexcused reasons for absences. These policies should be explicitly defined and applied uniformly across patients, and are often based on existing clinic policies related to absences. Whether verification and the types of acceptable verification should also be explicitly defined as part of the CM policies. Whenever possible, rescheduling a session is preferred, and we routinely offer the opportunity to come at a different time of the day (whenever possible) or the day before/after their scheduled session. However, it may not always be possible to reschedule in a timely manner and excused absences can be used in these cases. Absences, even excused or extended absences, do not extend the duration of the CM program.

A final point on absences—clinicians should apply the rules with consistency across patients, and all clinicians should apply the policy in the same manner. A situation where some clinicians are perceived as flexible and others as rigid in applying absence policies should be avoided. For example, if the clinic policy is that verification of absences is required, then all CM clinicians must adhere to requiring verification with the same rigor. The CM supervisor should monitor for consistent application of these policies across patients and therapists.

3.3. Fishbowl

Figure 1a shows a clinic fishbowl (panel a) and a small prize slip (panel b). Table 4 displays the composition of the prize slips within the bowl. A standard fishbowl composition is 500 slips, with half winning and half containing nonwinning positive affirmations. Among the winning slips, the probabilities are arranged such that smaller prizes are more common, and the large and jumbo prizes are less common (probabilities listed in Table 4). Slips are replaced after each patient, so that the probabilities remain constant over time and across patients, meaning the same patient could win the jumbo more than once and the jumbo could be won by multiple patients over time. Changing the probabilities has a direct impact on the magnitude of the protocol and thereby costs of the CM program to the organization; we recommend adherence to empirically validated fishbowl compositions.

Figure 1:

Panel 1a shows a fishbowl containing 500 draw slips. Panel 1b shows a small prize slip drawn from the fishbowl; draws are replaced between patients. Panel 1c (bottom) shows an example of an electronic/virtual fishbowl from Project MIMIC in which the clinician inputs the correct number of draws for a given patient (i.e., 3 in this example) and the results (2 positive affirmations and 1 small prize) are revealed on the left by dragging down the grey box.

As noted, this schedule is similar to the one provided as a model during the VA regional CM trainings (Petry et al., 2014; Rash et al., 2013). However, that model used small prizes with an average value of $1, and its average expected maximum magnitude per patient was about $364. I have included one change from that original schedule in acknowledgement that small prizes that are desirable for around $1 are often challenging to find. In response, the small prize category has been adjusted to include prizes worth ~$2 in value. With this adjustment, the revised average maximum expected magnitude per patient is $432 (and is within the recommended range of $385 to $533 for magnitude). This higher value accounts for inflationary changes over time and is likely to yield more desirable small prizes (and make shopping easier).

In the fishbowl picture (Figure 1a), note that the patient cannot see the writing on the slips when folded (use heavier weight paper and ensure that slips are folded). Larger fishbowls are preferred, providing plenty of room to stir the slips without spillage. I recommend auditing the fishbowl periodically to ensure that the correct distribution is maintained over time. Clinics could elect to have one shared fishbowl that all CM therapists use or one fishbowl for each CM therapist.

Prize earnings across many patients should reflect the underlying composition of the fishbowl slips/categories. If a clinic is experiencing a higher than expected (based on probabilities) rate of “larges”, for example, it may be time to audit the fishbowl (a patient could have palmed a large slip, slips are not getting replaced, etc.). The person tasked with supervising the CM protocol can check these probabilities by clinician, if necessary, to narrow down problem areas and re-training needs.

4. Designing the CM protocol

I recommend that clinics do not start from scratch in developing a CM protocol. Rather, use a protocol from the scientific literature that demonstrated a positive impact on patient outcomes. Be cautious about adopting a protocol that has not yet been scientifically vetted (e.g., a protocol paper published in advance of outcomes) or an article describing an implementation of CM that is presented without solid evidence of efficacy (e.g., uses self-report rather than objectively measured abstinence). Start with a protocol with demonstrated success and account for inflation by translating the magnitude to current dollars for older protocols. Readers might scan meta-analyses (e.g., Benishek et al., 2014; Bolivar et al., 2022; Griffith et al., 2000; Lussier et al., 2006; Prendergast et al., 2006) for protocols with demonstrated efficacy and then refer to the primary article(s) for details of the protocol(s). As a starting place, the meta-analyses provide an average estimate of effect size, often with an indication of the population (i.e., patients with stimulant use disorders). Readers can identify studies with similar populations, settings, and target behaviors, and then review where an individual study’s effect size falls relative to the average (i.e., above the average, below the average) to evaluate whether a protocol is suitable and likely to maximize impact on patient outcomes. In reviewing studies, consider population factors (severity), number of target behaviors (only stimulants or multiple drugs or multiple behaviors), duration, and so on. Protocols targeting stimulants and/or opioids have similar schedules because of the windows of detection in the urine tests; however, studies based on substances with very different windows of detection (e.g., breathalyzer-based reinforcement of alcohol or tobacco/smoking abstinence because of their short detection windows, or marijuana abstinence because of its long detection window) are unlikely to translate to a stimulant protocol without substantial modifications.

If modifications are necessary, they should be done with an understanding of how the changes could impact patient outcomes. I strongly recommend the involvement of a CM scientist who understands the underlying reinforcement principles and their impact on outcomes. Prior research (Rash et al., 2012; Rash et al., 2020) suggests that rewards used in clinical practice diverge from evidence-based practices at high frequencies. For example, in Rash et al. (2020), roughly half of substance use treatment providers using rewards as part of clinical practice failed to reinforce in a timely manner (at least same day), used low frequency reinforcement opportunities, and did not utilize an escalating reinforcer schedule—all of which are features of research-based CM protocols. Reinforcer magnitude (Rash et al., 2012, 2020) is another parameter rarely at an effective “dose”, an issue compounded by the SAMHSA-imposed incentive cap of $75 per patient per year for organizations attempting to use those fund sources. Given the impact of these schedule parameters on efficacy, great caution should be used in making changes to research-based CM protocols and, again, most organizations would likely benefit from expert consultation before considering deviations from empirically supported protocols. Importantly, poorly designed CM is unlikely to impact patient outcomes, which is the overarching goal of treatment. Using a well-tested CM protocol that has demonstrated efficacy is far more likely to achieve this goal than a stripped-down version of CM that meets clinic needs or budgetary constraints rather than prioritizing patients’ benefit. Although a thorough discussion of clinical competence is beyond the scope of this article, poor clinical delivery or clinician nonadherence to the established protocol also impacts the efficacy of CM and benefit to patients (Hartzler et al., 2017; Petry et al., 2012a). Thus, both CM design and CM delivery are important to implementation fidelity. See Ledgerwood and Petry (2010) or Petry (2012) for additional discussion of clinical skills in delivering CM with patients and plan for supervision of these issues.

This article assumes a focus on stimulant abstinence as the behavioral target for the CM program. Below, I outline some important CM design considerations and highlight possible modifications that have the potential to impact outcomes, including those modifications that have already been incorporated into the protocol described above. Consider that the impact of changes on efficacy is cumulative—while one change may be acceptable, multiple changes from research-based protocols, even if small in nature, are likely to decrease the effectiveness of the CM protocol and the potential benefits to patients. Table 5 outlines the best practices most aligned with the evidence base, acceptable-if-necessary modifications (which should be limited in number in a given protocol), and practices that are not acceptable when implementing an evidence-based CM protocol.

Table 5.

Recommended Best Practices, Acceptable Compromises, and Unacceptable Practices in CM protocols for Stimulant Abstinence

CM Schedule Parameters	Best Practices	Acceptable Modifications (limit the number used)*	Not Acceptable
Target Behavior	Objective, verifiable, and therapeutic	None	Self-report of Behavior
Number of Behaviors Targeted	Single drug or single drug class	1) Multiple drugs (e.g., stimulants plus other substance targets), consider increasing magnitude. 2) Multiple behavior targets (e.g., stimulant abstinence AND attendance), consider increasing magnitude.	--
Frequency of Monitoring/Reinforcement	Should have complete coverage to detect if stimulant use occurred. Optimal is thrice weekly.	Twice-weekly, well-spaced (e.g., M&Th or T&Fr)	-Poorly spaced twice-weekly -Less than twice-weekly
Immediacy	Results provided immediately, i.e., Rapid screens	Delayed but at least same day	Extended delayed (not same day), e.g., sending urine sample to laboratory for testing
Escalation/Resets	Yes	---	---
Magnitude (Prize CM, 12-week duration, single behavior target, in 2022 dollars)	Typical population: ~$385-$533 Severe/Special pop.: Increase magnitude. Multiple Behaviors Targeted: Increase magnitude. Longer duration: Increase magnitude.	Must use an effective magnitude	Must use an effective magnitude
Reinforcer Selection	1) Stocked in-house prizes that are desirable and with sufficient variety 2) A selection of various store gift cards. 3) Loadable gift cards.	--	Failure to use desirable reinforcers

Note.

^*Acceptable modifications should be limited in total number. In other words, adopting all the acceptable modifications in the same protocol is not recommended. Judicious selection of a limited number of modifications should be used with recognition that the optimal choice is to adhere to the best practices recommendations for the majority of CM protocol features. “Not acceptable” practices should not be employed in CM protocol design.

4.1. Single versus multiple drug targets

I recommend targeting a single behavior and avoiding complex multi-behavior protocols. One of the more robust findings from the CM literature is that single drug targets (e.g., cocaine) or single drug class targets (e.g., stimulants) generate more robust impact on outcomes compared to protocols that reinforce multi-drug abstinence (e.g., requiring abstinence from more than one drug, such as stimulants, marijuana, alcohol, and opioids concurrently) (Ainscough et al., 2017; Bolivar et al., 2021; Griffiths et al., 2000; Lussier et al.., 2006; Prendergast et al., 2006). See Table 6. From various experiences providing CM trainings, my general observation is that an initial or default perspective by clinics is that they want to focus on many behaviors (abstinence, attendance, therapy behaviors, etc.), in essence, they are aiming for “perfect” behavior or the “perfect” patient.

Table 6.

Comparison of Meta-analysis estimates of Effect Sizes for Abstinence Outcomes in Single Drug Target CM versus Multiple Drug Target CM Protocols

Meta-analysis	Effect Size (d) for Single Drug Target CM	Effect Size (d) for Multi-Drug Target CM
Griffiths 2000	Single drug = 1.32	Polydrug = 0.45
Lussier 2006	Cocaine Only = 0.75 Opiates Only = 0.85	Dual Cocaine & Opiate = 0.43 Polydrug = 0.41
Prendergast 2006	Cocaine Only = 0.66 Opiates Only = 0.65	Polydrug = 0.42
Ainscough 2017	Cocaine = 0.75	Dual Cocaine & Opiates = 0.48 Polydrug = 0.62
Bolivar 2021	Stimulants = 0.70	Polydrug = 0.46

Notes. When two or more effect sizes for abstinence were available within a given meta-analysis report, preference was given to outcomes reflecting sustained abstinence (Ainscough = longest duration of abstinence).

This expectation of perfect performance also carries to protocols that focus on abstinence, with clinics often stating a preference to require abstinence from multiple drugs rather than a single drug target. As the meta-analytic results in Table 6 demonstrate, multi-drug target CM is still efficacious, but far less so than a streamlined and focused CM protocol targeting a single drug or single drug class. Single-drug target protocols allow more patients to gain access to the reinforcers by focusing on the core issue and offer an achievable goal (because we want the patient to experience successes in the CM program). If choosing a multi-target CM protocol, be aware that some patients may be excluded because they will never achieve the goal we have set. Also be aware that the protocol magnitude should likely be adjusted to a higher amount per patient to compensate for multi-target CM designs.

4.2. Frequency: Monitoring and reinforcement

CM protocols that offer higher frequency reinforcement opportunities yield larger effect sizes (Griffith et al., 2000; Pfund et al., 2022). In an abstinence CM protocol, the frequency of monitoring and reinforcement is largely determined by our available testing technologies, with the goal being near complete ability to detect stimulant use if it occurs. With rapid urine screens, testing about 3 times per week would likely detect most instances of stimulant use, and this frequency is associated with larger effect sizes than lower frequency schedules (Griffith et al., 2000). An acceptable modification would be well-spaced, twice-weekly urine screens (Petry, 2012; Rash, 2017). Examples of well-spaced screens would be Monday/Thursday, Tuesday/Friday, or Monday/Friday. Poorly spaced screens risk missing stimulant use when it is occurring, and thus reinforcing a patient for stimulant abstinence when they are not actually abstinent. Once-weekly urine testing is not an acceptable practice, because such a practice even if well-intentioned to accommodate clinic schedules, tends to result in ineffective CM programs (Metrebian et al., 2021). Clinics that cannot schedule at least twice weekly urine screens and reinforcement sessions should not implement an abstinence-based CM program.

Note that the model protocol provided in Table 1 has already adopted the modification of twice-weekly frequency rather than thrice weekly. Caution should be used in adding any additional changes that weaken the effect of CM (e.g., opting to target multiple behaviors simultaneously). Each change should be viewed as cumulative in its impact on the ability of the CM protocol to successfully impact patient outcomes.

4.3. Immediacy

Reinforcers that are delivered immediately following the behavior are more effective than delayed reinforcers (Griffith et al., 2006; Lussier et al., 2006; Packer et al., 2012). Effect sizes in Lussier et al. (2006) were d = 0.85 for immediate versus d = 0.39 for delayed delivery; and in Griffith et al. (2000), the effect sizes were d = 1.35 and d = 0.39 for immediate versus delayed delivery. The practical application of this principle in abstinence-based CM protocols is that clinics should use rapid/instant screens that provide results immediately and within the same visit. Sending urine samples out to laboratories for testing (not a recommended practice) is a clear deviation from this principle, as it introduces delay; it should be avoided. Fortunately, point-of-care testing is increasingly reliable and is preferred over laboratory testing for routine clinical use by the American Society of Addiction Medicine consensus document (ASAM, 2017).

4.4. Duration

Twelve weeks seems to be the default duration for CM protocols focusing on abstinence from illicit substances. In Ginley et al. (2021), of 24 CM conditions included, only 3 conditions had an incentive duration of fewer than 12 weeks, 15 treatment conditions used a 12-week CM protocol, and 8 conditions involved 16 or more weeks of incentives (5 protocols used 16 weeks, 2 had 24-week durations, and 1 protocol was 52 weeks). In Ainscough et al.’s (2017) meta-analysis, studies using CM protocols of fewer than 12-weeks duration failed to demonstrate a significant impact on sustained abstinence. In Bolivar et al. (2021) and De Crescenzo et al. (2018), both of which focused on stimulants, the mean duration of CM was 17 weeks (SD = 13.8), and median duration was 12 weeks (range 6–36), respectively. Specific to prize CM studies, of 19 studies examined in Benishek et al. (2014), 15 used an incentive duration of 12 weeks, 1 study used a duration of 24 weeks, and 3 studies were fewer than 12 weeks.

Intervention duration (i.e., duration that the CM contingencies are in place) also appears to be related to longer term outcomes, with longer interventions conferring additional benefit (Ginley et al., 2021). However, the Ginley analysis included only 3 treatment conditions involving more than 16 weeks of CM. Additional research should explore long-term CM given this early signal of long-term impact.

In determining duration, 12 weeks would be the ideal starting point for protocol duration, based on the consensus of studies reviewed above. If resources allow, then providers can use longer incentive durations. Another consideration for CM protocol duration is the severity of the population targeted (e.g., Petry et al., 2012b); if severity is high (e.g., specifically targeting a high-risk for relapse group, high rates of comorbidities, significant barriers to recovery), then longer durations of incentives may be warranted to allow sufficient time to demonstrate a stable behavior pattern. We note that longer protocols, while possibly desirable for their long-term benefits, are often more costly, which may be a substantial factor in decision-making for programs with limited resources.

4.5. Escalation, resets, and caps

The vast majority of scientifically vetted CM protocols in randomized controlled trials have incorporated escalating reinforcers for successive performance and reset contingencies following a positive sample or unexcused absence (Ginley et al., 2021). When directly compared, escalating protocols with resets tend to produce better effects on abstinence outcomes compared to those with fixed (flat, non-escalating) schedules (Roll & Higgins, 2000; Roll & Shoptaw, 2006; Roll et al., 2006; Romanowich & Lamb, 2015; Stoops et al., 2011). Therefore, we recommend use of an escalating schedule with resets. However, we do note a study that found no effect of fixed versus escalating schedules when resets were not included in the schedule (Hutchison et al., 2012; Tuten et al., 2012) and a recent study (Regnier et al., 2022) that suggested a fixed schedule with higher initial value is more likely to produce a negative sample on the first visit when magnitude is high ($1980 over 12 weeks).

Escalation does increase overall costs of the CM program, and many schedules incorporate a cap that serves to leverage the benefit of escalation while controlling impacts on costs. For comparison purposes, Table 1 displays simple escalation (first schedule) and a capped schedule (second and the suggested model). In the simple escalation schedule, draws start at 1 draw for the first negative sample and escalate by 1 draw for each consecutive negative sample. In total, assuming the patient attends all sessions and submits all stimulant negative samples, the patient would earn 300 draws from the fishbowl with an average expected maximum magnitude of about $728. The second and suggested schedule (far right column) starts similarly with 1 draw for the first stimulant negative sample and escalates by 1 draw for consecutive negative samples. Once the patient reaches 8 draws, the draws are capped at 8 for the remainder of the protocol as long as the patient continues to submit negative urines. In this second schedule, if the patient attends all visits and submits all negative urines, then they can earn 164 chances to draw from the fishbowl and an average expected maximum magnitude of $432.

As this comparison shows, capping draws serves to decrease the total number of draws possible and subsequently the total magnitude available to the patient. Beyond serving as a cost-control feature, capping a schedule also serves to limits the amount of time completing the draws with the patient. Selecting, opening, and tallying 8 slips from the fishbowl is quicker than doing the same for 24 draws.

Table 2 provides a hypothetical patient scenario in which resets back to the starting value (1 draw) occur following positive samples and unexcused absences. An alternate approach to resets includes the same reset following a lapse but reinstitutes draws to the level achieved prior to the lapse when a patient has submitted a defined number of consecutive negative samples (e.g., after two weeks of demonstrated abstinence). An example of this feature would be a patient who has been earning 8 draws, but then submits a stimulant positive sample. The patient earns no draws the day of the positive samples and draws reset back to 1 draw for the next negative sample and escalate by 1 draw until the patient submits 4 consecutive stimulant negative samples (i.e., two weeks of abstinence). Once the patient achieves that goal, draws possible for the next visit return to the value prior to the lapse (8 draws in this case). Some clinicians find this arrangement to be less punitive for patients while maintaining the function of the reset. In uncapped schedules, this return to previous values may be more important than in capped schedules (because the set-back could be much greater in uncapped protocols).

4.6. Magnitude

Perhaps the most important feature of an evidence-based prize CM protocol is its average expected maximum magnitude. The Lussier et al. (2006) and Bolivar et al. (2021) meta-analyses found that larger possible incentive magnitudes produce larger impacts on patient outcomes. Lussier et al. (2006) indicate that protocols with >$16/day generated the largest effect size (d = 0.95) compared to protocols using <$5/day, which generated the smallest effect (d = 0.47). In Bolivar et al. (2021), CM studies (k = 22) focused on stimulant abstinence had mean maximum daily earnings of $14.51 (SD = $11.94). For a 12-week protocol, that value would translate to ~$1,000 per patient, which is typical of many voucher CM magnitudes.

Most prize CM studies have employed lower average expected maximum magnitudes per patient than the magnitudes used in voucher schedules, and prize CM compared to voucher CM can be similarly effective even at lower magnitudes (Petry et al., 2007; 2015). In Benishek et al. (2014), of 16 prize CM protocols with a 12-week duration, the median magnitude per patient was $420 (range $80–$1,391), which is about $533 in 2022-adjusted dollars. Remember that magnitude should be considered along with other issues such as the complexity of the target behavior (single versus multiple), the severity of the population (Petry et al., 2012b), and duration of the CM protocol. If these factors increase, then magnitude should likely also be increased.

Given the focus on making CM accessible for application in real-world clinical settings, studies have explored attempts to use lower cost prize CM. Petry et al. (2004) compared the “standard” magnitude at the time for prize CM of $240 (note that the 2022 consumer price index-adjusted value of this protocol is ~$385) to a lower cost prize CM condition of $80 ($129 in 2022 consumer price index-adjusted dollars). Participants in the $240 condition had statistically better abstinence outcomes compared to treatment-as-usual. However, participants in the $80 condition did not statistically improve their duration of abstinence beyond participants randomized to treatment-as-usual, which is notable given the additional expense, staff effort, and time involved in implementing additional interventions such as CM. The nonresponse rate in this study is of clinical interest. Rates of nonresponse (i.e., had 0 weeks of consecutive abstinence) were more than doubled in the treatment-as-usual (35%, 13 of 37 pts.) and $80 CM condition (31%, 14 of 45 pts.) compared to the $240 condition (13%, 5 of 38 pts). Choosing low magnitude CM translates to fewer patients benefiting.

4.6.1. Deviations from effective magnitudes and the SAMHA incentive cap

I have chosen to highlight Petry et al.’s (2004) study above because of the similarity of the lower cost condition ($80, which would translate to $129 in 2022 dollars) to the current $75 per patient per year SAMHSA cap on incentives. Many programs are attempting to implement CM protocols that stay within that $75 SAMHSA limit. This practice is counter to scientific evidence and we do not recommend it. A protocol that uses this $75 maximum for incentives is not employing an evidence-based magnitude, nor is it likely to adhere to best practices beyond magnitude, such as frequency, duration, etc. Until this SAMHSA limit is lifted, if a program is intending to use these monies for a prize CM protocol, they should supplement beyond the $75 from another allowable source up to an effective magnitude. Dr. Yngvild Olsen (2023), director of The Center for Substance Abuse Treatment at SAMHSA, noted a common misperception that supplementation of CM incentives above the currently $75 limit was not permissible and highlighted several states that are successfully braiding different allowable funding streams to bring the incentive dollars per patient to an evidence-based magnitude while remaining compliant with federal policies.

4.6.2. Recommended magnitude

I recommend prize-based CM magnitudes be in the range of $385 to $533 for a 12-week protocol (note that the range of effective magnitude for voucher schedules is different from this recommended range, which is specific to prize CM schedules). The lower bound of this recommendation ($385) is based on the “standard” prize CM magnitude supported by a number of Nancy Petry’s randomized clinical trials (adjusted to 2022 dollars). The upper bound is from the median value from Benishek et al. (2014), adjusted to 2022 dollars. The protocol in Table 1 has an average expected maximum per patient of $432. Magnitudes below the recommended range should not occur, though increases in magnitude might be warranted (and recommended) based on severity, number of target behavior, extended duration, etc.

In general, consider magnitude in the context of what you are asking of your patients. A rule of thumb is to offer a reinforcer magnitude that is commensurate with the difficulty of the goal. If, for example, you elect to use a CM protocol that requires multiple behaviors be met, then we would increase the expected maximum magnitude beyond the starting range of $385 to $533 to a higher prize magnitude value per patient. Note also that other features such as duration and frequency can impact magnitude. This articlec has focused on a 12-week protocol. If a longer duration is desired, the magnitude should also increase accordingly. For example, a 24-week protocol would be roughly twice the magnitude of the 12-week protocol proposed here, unless it introduces more complex fading schedules.

A last point—CM magnitude is important to sustained benefits for our patients. For example, in Higgins et al. (2007), compared to lower magnitude CM, a protocol using higher magnitude CM increased cocaine abstinence rates at all follow-ups through an 18-month follow-up period (well beyond the period that contingencies were delivered in the first 12 weeks). If the purpose of treatment is to create long-lasting cocaine abstinence in our patients, cutting the CM magnitude (e.g., to stay within the SAMHSA $75 cap) is not going to help us reach that goal.

5. Reinforcer selection

5.1. Desirability

The most critical aspect of reinforcer selection is that the items be desirable to your patients. Asking individual patients what reinforcers they would like to work toward (e.g., a gift card to a specific store or prizes related to a specific need) is a good way to ensure desirable prizes. In practical terms, the desirability of reinforcers will vary from population to population and even from patient to patient. Consider the needs and wants of your specific patient group. Are many of your patients in unstable housing situations or moving from program to program? If yes, then bulky items needing storage are unlikely to be appealing. Are a significant portion of your patients for the CM program experiencing homelessness? If yes, then items that are waterproof, lightweight, and address a need might be popular. Is your program geared toward parents with children? Then items that facilitate play (e.g., sidewalk chalk, games, coloring books) might be top items. Also consider seasonality (warm versus cold weather needs)— sunscreen and small personal portable fans may fly off the shelf in summer, but gloves/socks and handwarmers may be valued in winter.

5.2. Gift cards

Gift cards are a popular choice, but programs still must consider desirability and patient population factors. In our programs, we stock about 15 or so different brands of gift cards, so that patients have a wide selection of cards available to them. Patients will have varying transportation and store availability options, as well as different thoughts about what they would like to purchase, so having a wide selection of different store brands is a safer strategy to ensure broad appeal of the CM reinforcers across many patients. Alternately, programs might ask individual patients which stores they prefer at the start of the CM program (assuming the program has capacity to quickly acquire desired gift cards). Some organizations may wish to avoid purchasing gift cards for stores that sell alcohol, cigarettes, or weapons, but little evidence exists that patients in CM programs use cash or gift cards to purchase substances or other problematic items (e.g., Festinger et al., 2014; Vandrey et al., 2007).

We encourage patients to mix and match different cards up to their earned dollar amount. For example, if today’s patient earned a large prize worth $20, then they might select two different store cards worth $5 each and a $10 card to total their $20 prize. To facilitate this practice, we only buy cards in small denominations ($5, $10, and $20). We do not stock $100 cards for the jumbo prize because the odds that the one or two stores you selected for the jumbo denomination will appeal to all patients winning a jumbo prize is slim. Rather, patients can stack five $20 cards to total $100. This stacking of cards works for patients and benefits us such that we are not retaining undesirable items in large values. Be careful about large bulk purchases—it is best to order moderately and restock as needed. You do not want to be left with thousands of dollars of gifts cards to a store that closed or a store that is undesirable to patients. Similarly, if you live in a location in whih gift cards have expiration dates, be careful about the quantity purchased. I tend to avoid cards that require an activation fee whenever possible because this fee reduces the amount available to the patient. Loadable gift cards are another option but be mindful of activation and load fees when shopping vendors.

Be sure to ask how patients are planning to spend their gift cards—genuine interest often opens a discussion into a therapeutic topic. Some common examples include patients saving up their gift cards for a special event or holiday gifts. These events might tie into other treatment goals such as expanding their nonsubstance use social network or reconnecting with supportive family, providing a natural opportunity to link their progress with the CM program, and their recovery to the patient’s broader treatment goals. These discussions also offer natural opportunities to highlight the patient’s progress and efforts and to build self-efficacy, which is a core aspect of the clinical delivery of CM (Ledgerwood & Petry, 2010).

5.3. Variability in prize selection

Variability is often a critical factor in successful prize cabinets. As noted above, you are aiming to appeal to many different patients with different needs and desires. However, another important consideration is that you are also aiming to keep the interest of the same patient over time. The goal should be that Patient A is as excited about prize opportunities in week 12 as they were in week 1. To meet this goal, we routinely solicit suggestions from patients for prizes, particularly when we are restocking, and we restock the prize cabinets often.

When shopping, you want a large selection of different items in the small and large categories. Limit multiples only to items that are known to be popular. The small prize category (~$2) will be the most frequently earned prize but is often the most difficult to shop. Petry (2012) provides examples of popular small prizes and suggests starting with about 40 to 50 small prizes representing at least 25 different items (some multiples allowed, but still a large selection of different items). Though prize popularity varies across patient groups, fuzzy socks, toiletries, and food/drink items are often at the top of my list. For the large prize category (~$20), Petry (2012) suggests an initial stocking of 8 to 10 large prizes and to avoid multiples in this category until you have feedback on desirability. Again, depending on the population, I have found kitchen sets and body care sets (e.g., beard/hair trimmers, razor kits) to be popular, as well as the ever-popular gift cards. For the jumbo category (~$100), Petry recommends at least 2 options for your cabinet. In my experience, this category has been the most sensitive to changing technology. Flip through old photos of our prize cabinets, and you will see the passage of time through our jumbo prizes: boom boxes, then CD players, then portable DVD players, and more recently, blue-tooth headphones. Given the infrequency of jumbo wins, you might consider using only gift cards for this category (e.g., stacking five $20 cards for a total of $100 in gift cards) and avoid getting stuck with a prize that is outdated and no longer desirable.

6. Other considerations

Below, I list some of the practical items needed to purchase (e.g., lockable storage for the prizes) or to create (e.g., reminder slips) to start a prize CM program. A checklist is provided in Appendix A that may be useful in the planning phase of implementation.

6.1. Lockable storage

We prefer a lockable 5-shelf, 2-door office cabinet (or the half-size, 3-shelf version) to store the prizes. We also use a smaller lockbox for the gift cards. Consider location carefully—it should be easily accessible to CM clinicians at the times they will need access, but in a location that offers some level of security. Consider who will have keys (or where keys will be stored) in terms of accessibility.

Best practice would be to walk the patient who has earned a prize to the cabinet and allow them to review and select their desired item. Occasionally, we have encountered a situation with very limited space for the cabinet or a storage location that patients could not access. In those situations, we have used a picture menu of the prize items (be sure to update the menu to remove taken items and add new items following purchases). Patients select their prize from the menu, and the therapist then goes to the cabinet to pull the selected item and return it to the patient’s location.

6.2. Prize bowl

I prefer a large plastic, flat-sided fishbowl (shown in Figure 1ab) because it is lightweight and stores more readily than some other shapes. However, the actual container used is not critical to CM delivery as long as it is large enough that slips do not easily spill when drawn. Patients should be able to stir the slips comfortably and dig down to the bottom of the container if desired. The fishbowl should be secured when not monitored (e.g., in a locked office, or in the locked prize cabinet). If you have multiple clinicians involved in the CM program, you may want more than one fishbowl or even one fishbowl per clinician.

Traditionally, we have always had the patients do their own fishbowl draws. At the onset of COVID, we made a number of adaptations to limit potential exposures, one of which was to have the clinician draw the slips so that fewer hands were involved and potential contamination reduced. Another option would be to have patients wear gloves while drawing or disinfect their hands prior to drawing. At this point, we have resumed our typical approach of allowing the patient to draw, but these adaptations may be useful for some programs.

Electronic fishbowls replace the physical fishbowl with an electronic/virtual draw generator. For example, the clinician would input the correct number of draws for a given patient’s session (e.g., 3 draws), and the program generates the draw results (e.g., 2 good jobs, 1 small prize) following the same probability distribution that would be used to create the physical fishbowl. I still prefer and recommend using physical fishbowls, but this choice is driven by personal preference rather than research. In the rare circumstances where a patient questions whether the CM program is fair, I can offer a patient the physical fishbowl to examine the slips for themselves to see that the draws match the promised distribution (i.e., the patient can see the jumbo slip is in the fishbowl, see that the number of larges match the stated number). I also find patients enjoy the physical fishbowl draws and are more excited about the process. Nonetheless, clinicians may find the electronic fishbowl to be easier and quicker to administer, and utilizing this resource is unlikely to damage efficacy. The website for Project MIMIC (Becker & Garner, 2023) provides a downloadable electronic fishbowl (Figure 1c).

I will share one caution that applies to both the physical and electronic fishbowl—clinicians may be tempted to “re-do” draws when they are not desirable, often because a clinician wants their patient to win more or larger prizes. It is imperative that this practice not occur; doing so could substantially increase costs of the program and skew average earnings per patient well beyond expected magnitudes, as well as damage the integrity of the program and lead to patients believing the system is rigged and potentially unfair. If this practice is suspected, the person charged with oversight of the CM program can review the distribution of draws for each clinician’s CM caseload (i.e., all of Clinician A’s patients combined). Each clinician’s distribution should follow the expected probabilities once a sufficient density of patients/draws is achieved.

6.3. Prize slips

Our CM research group used mostly paper slips over the years because it is a cheap and a readily available option. We use either colored paper or a slightly heavier-than-typical weight paper to make sure the writing is not visible once the slip is folded. If you notice after the fact that your slips are not opaque, you can paint the fishbowl or cover the container so that the slips are not visible during the draw process. All slips should be the same color, equal in size, and folded in the same manner. Deviations in consistency (e.g., a patient notices that all the “good jobs” are a different color, or that the larges are a different shape/size) can lead to differential picking, which could drive up the costs of the CM program.

6.4. Reminder slips

Though not required for CM delivery, we recommend reminder slips. Patients often find them useful for clear communication of the CM draws and expectations. However, I find that the reminder slips are more useful for reminding the clinician to cover all the necessary steps of a skillful CM session. The reminder slips can be printed text with space for the clinician to write in key information, including the current session’s draw results and draws possible at the next session. Petry (2012) provides an example, but these slips can be tailored to your program and preferred verbiage.

6.5. Accounting/auditing procedures

Someone in your organization should be tasked with the CM auditing. Three aspects should be monitored over time:

1. The slips in the fishbowl. We audit our fishbowls monthly to ensure that the distribution of slips (probabilities) stay constant over time. Occasionally, slips might fall out and be lost, and the auditing provides an opportunity to correct this discrepancy. Less frequently, you might suspect cheating (e.g., a patient retained a large slip and is presenting it at subsequent sessions resulting in higher than typical prize earnings). If so, you can easily change the color of the slips or change the font color of the writing. Careful attention to the draw process by the clinician and monitoring the return of the slips to the bowl reduces the likelihood of these events. In addition to monitoring the frequencies of the slip categories, the counting process is also an opportunity to look for worn or marked slips (e.g., corner folds, writing).

2. The prize inventory. All items should be inventoried on purchase and tracked when distributed to patients. We maintain a prize release sheet for each patient that includes the date of prize distribution, a description of the prize (usually 1–2 words with a unique numeric tracking number), patient’s initials acknowledging receipt of item(s) on that day, and staff initials. Each time the patient earns a prize, this information should be entered into that specific patient’s prize release form. Multiple lines might be used on the same day if the patient earned multiple prizes that session. For the unique code, we have either used part of the item’s UPC code (e.g., last 4 digits) or a handwritten number with permanent marker. The goal is that every prize purchased and distributed by your organization can be linked to a patient, and these prizes should match the clinical record of prizes earned (discussed below). The goal with your auditing system would be to track all CM funds spent from purchase (via receipts) through distribution to individual patients. Any inventory not distributed should still be present in the prize cabinet. The person assigned to monitor inventory need not be a clinician and does not need to be the same person that provides clinical supervision of the CM delivery.

3. The clinician’s delivery of the correct number of draws. On the clinical side (often stored with or part of the patient records), providers should maintain a session-by-session tracking of the CM session (see Appendix B for an example or Petry 2012 for another format). At minimum, this tracking should include the date of sessions, the stimulant result (positive/negative), the total number of draws earned (6 picks today), and the breakdown of draws selected (4 “good jobs”, 1 small, 1 large). These draw breakdowns will match up with the prize release form for that patient. In this example, the prize release form should show that the patient picked one small and one large prize from the cabinet. Supervision should include a review of these documents to assess whether any deviations from the protocol are occurring (e.g., incorrect number of draws administered, draws awarded when urine was not collected or positive, excused or unexcused absences not applied correctly).

4. Tracking patient performance. This same clinical record also serves to track the patient’s progress over time. A skillful delivery of CM will leverage this information, for example, highlighting the patient’s efforts: “Today’s sample is the fourth session in a row that is stimulant negative!” I prefer a low-tech, paper version of clinical tracking that displays the 12 weeks of the CM program, in a week-by-week format. As much as possible, the process can be simplified so that the clinician need only circle the day or write in a date (or if the patient was absent, circle or write-in excused or unexcused), circle the stimulant result (+ or –), write in the number of draws and the draw slip breakdown.

If maintaining CM session results in the electronic clinical record, I suggest you concurrently use this printed version (or an electronic version) that allows the entire CM record to be seen without needing to click into and out of individual session records. The view-in-a-glance format helps the clinician to administer the correct number of draws because they can see the entire record in a quick glance, and it also serves to remind the clinician to deliver critical information (such as how many draws the patient earned today and how many they can earn at the next session, see Ledgerwood & Petry, 2010 for an indepth review of the clinical delivery of CM). Last, the ability to see the entire CM record in one glance facilitates supervision.

7. Integration with clinical care

Below, I review some of the common considerations for integrating CM programs into existing clinical services, including patient selection, workflow options, and supervision requirements.

7.1. Patient selection and eligibility

Given the costs of CM, it is highly unlikely that an organization will offer CM to all patients. Rather, a specific subset of patients or a specific program will be selected for CM. In this article, we have started with the assumption that the target behavior will be stimulant use. Therefore, patients with stimulant use disorder and/or recent evidence of stimulant use will be the target population. Some programs may have further eligibility considerations, such as a specific program. In general, you want to target patients who have room to benefit from the CM program. If your patients with stimulant use disorder are already stimulant abstinent or stimulant use is rare (e.g., as found in some residential programs), then using CM to target stimulant negative urines is likely to be expensive and provide little benefit to the patients. However, the same program that finds little stimulant use in their residential program might observe that many of those patients struggle with stimulant relapse when transitioning from residential to outpatient care. This pattern would suggest a clear area where CM could have a beneficial impact, and the CM program should be designed to target these “pinch points” where patients tend to struggle with relapse or attrition.

Some exclusionary criteria are worth considering. Prize CM has an element of chance, and though it is not a gambling activity, our research group has historically excluded patients who are in recovery from a gambling disorder. Note that those with ongoing gambling activity who are not in recovery have participated in our research studies with no noted escalations in gambling activity or problems (Petry & Alessi, 2010; Petry et al., 2006; Rash & Petry, 2011).

Another possible exclusion to consider is those patients involved in the legal system. Generally, participation in CM involves a higher than typical rate of urine screens. When we screen more often, we are more likely to detect use. Thus, some patients (e.g., those involved in child and family cases, custody concerns, criminal justice involvement) might suffer negative consequences beyond the clinical environment because of the increased chance of detecting ongoing stimulant use through CM participation. Great care should be taken to ensure that patients do not encounter unintended negative consequences because of their participation in CM. These patients can be excluded from the CM program, or, in some cases, organizations have worked with these external partners to devise a system in which CM participation does not have a negative impact (e.g., developing an agreement that typical frequency clinic laboratory-based urine testing results at usual clinic frequencies will be shared per normal procedures with these agencies, but CM rapid screens will not be shared).

7.2. Workflow

Several options exist for integrating CM into workflows, and the choice will likely differ across organizations depending on staffing, clinic schedules, and the patient’s level of care. My preferred approach is to tie CM to the extent possible to other important appointments that are already part of the patient’s schedule. This scheduling approach serves to reduce burden on the patient and often boost attendance to the other activity. For example, if the patient is attending groups, then the CM session for stimulant abstinence might be conducted before or after group. If the patient is scheduled for individual sessions, then CM could be conducted in the first 5 to 10 minutes of that session. A possible sequence (option 1) might be as follows: 1) collect urine sample, 2) read urine sample result for stimulant and conduct CM session (administer or withhold draws per protocol), 3) if prizes are earned, walk patient to cabinet and have them select item(s) from prize cabinet, 4) log the prizes selected and document with initials from patient and staff member, 5) provide reminder slip with draws possible at the next session and an indication (e.g., circle M/Th) or date of the next appointment, and 6) continue into therapy session material.

An alternate approach for clinics using a prize menu and/or a harder to reach cabinet would be: 1) collect urine sample, 2) read urine sample result for stimulant and conduct CM session (administer or withhold draws per protocol), 3) use prize menu to select item(s) or tell patient that you will walk to the cabinet together at the end of your session together, 4) provide reminder slip, 5) conduct therapy session, 6) deliver the selected prize or walk with patient to the prize cabinet together for the prize selection, and 7) log prizes. Option 1 minimizes delay between the urine collection/reading and the delivery of selected prizes, and from the clinical perspective, the CM session tends to open naturally to therapy topics (e.g., slips, relapse, coping skills, opportunities to build self-efficacy). Option 2 is an acceptable modification that moves the sometimes time-consuming process of prize selection to the end of the therapy session and may work better for some clinicians.

While options 1 and 2 are my preferred strategies for integration of CM, other options would include conducting the CM session after the therapy session or conducting the CM session as a stand-alone appointment. Both of these latter strategies may be necessary if the staff member delivering CM is not the same person as the patient’s therapist. This decision of whether to have the clinicians deliver CM versus another staff member varies across organizations and is worth discussion. The content of CM sessions integrates seamlessly into therapy session, but other members of the staff may have more time to meet with patients, which is an important factor in twice-weekly CM. Even if the therapist does not deliver, effort should be made to integrate patient progress in the CM program.

Workflow integration can be challenging, and many arrangements for how CM will fit into clinic schedules are possible. However, keep in mind that the spacing of the twice-weekly sessions is important. Sessions should not be back-to-back, and acceptably spaced visits would be Mondays/Thursdays or Tuesdays/Fridays.

Considerations of staff availability, as well as economic feasibility (billable versus nonbillable), may ultimately guide decisions related to who delivers the CM session to patients. However, CM does not require a minimum degree or license. At minimum, the person delivering CM should be 1) thoroughly trained in quality CM delivery; 2) supervised/supported by a clinician in case clinical issues arise; and 3) supervised for adherence to protocol fidelity.

7.3. Preparing detailed protocol and patient handout

Before starting a CM program with any patients, the program should write a thorough CM protocol that includes all relevant policies (e.g., how absences will be handled). The goal should be a protocol that is specific to your program and your program’s policies, includes responsible parties (who will be trained to deliver CM, who will provide coverage, who will supervise clinical aspects, who will monitor/audit, etc.). Be comprehensive, add clarifying language to the procedures when/if issues arise, and address all possible what-if scenarios.

In addition to the protocol, programs can prepare a patient handout. I use a 1-page, double sided format that includes a brief narrative description of what the CM program is about, and in addition to the narrative, I include a bulleted list of key information (e.g., duration, target behavior) and charts with the fishbowl composition (Table 4) and the draw schedule (“suggested” schedule from Table 1). The narrative description is useful for patients, the latter components (bulleted list and charts) are intended to serve as prompts for the CM clinician to cover all the necessary information and help to clearly communicate how the program works. Including these elements is not a requirement; for example, the fishbowl composition could be memorized, or a cheat sheet could be taped to the bottom of the fishbowl. However, in my experiences training new CM therapists, having all the information in one handout is helpful for improving the explanation of CM to the patient in a thorough and accurate manner.

7.4. Consistency: Initial training and monitoring for drift

Our laboratory’s approach to training new staff in CM includes a workshop involving didactic (e.g., behavioral principles, evidence of CM’s efficacy, best practices) and experiential (e.g., demonstration, group exercises, peer practice, learning to apply the rating scale for competent CM delivery) components (see Petry et al., 2014; Rash et al., 2013). We follow this training with 4 audio-recorded roleplay scenarios that include:

1. A first CM session—therapist explains the CM program to a new patient;

2. An “on-track” session—patient has tested stimulant negative and therapist delivers correct number of draws;

3. An “off-track” session—patient has tested stimulant positive, and therapist must respond according to the protocol;

4. A session following an excused or unexcused absence—patient who missed the last session returns today with a stimulant negative urine result. Therapist must apply the clinic CM policies related to excused/unexcused absences and administer the correct number of draws per protocol.

A CM-experienced clinician provides feedback on the roleplays using Petry et al.’s CM Competence Scale (2010) and, if necessary, new roleplay scenarios are completed until the staff member achieves competent delivery (e.g., administers the correct number of draws; links the opportunity to draw for prizes to the patient’s efforts toward abstinence; see Ledgerwood & Petry 2010 for more on clinical delivery of CM sessions). Once the staff member demonstrates an initial level of competence, subsequent supervision of audio-recorded sessions can be less frequent, but the clinical supervisor should still regularly review the CM tracking forms for errors in protocol administration. Particular attention should be paid to incomplete records and whether the number of draws were correctly administered, especially following positive urine results and excused and unexcused absences. Prompt feedback on errors allows for corrective actions for the affected patient and retraining for the therapist to prevent further errors.

For programs new to CM, I strongly suggest limiting the number of initial CM patients to only a few (one or two) patients at the start. These patients help to pilot the program and often gaps in the protocol and/or staff training can be addressed. Once the CM program seems to be solid and staff are competent and comfortable with CM delivery, then the CM program can be expanded to a larger number of patients.

8. Patient factors

8.1. Patient characteristics

CM is generally effective regardless of patient income level (Rash et al., 2009; Rash et al., 2013; Secades-Villa et al., 2013); sex (Burch et al., 2015; Rash & Petry, 2015); sexual orientation (Zajac et al., 2020); housing status (Rash et al., 2017); prior treatment history (Blanken et al., 2016; but also see Rash et al., 2008a who found a benefit of CM for patients with multiple prior treatment episodes and muted response from patients 0–1 prior treatment episodes); race/ethnicity (Barry et al., 2009; but also see Montgomery et al., 2015 who found CM results differed for White versus African American patients when intake urine toxicology status was considered); comorbid substance use disorders (Byrne & Petry, 2011; Rash et al., 2008b); and HIV status (Burch et al., 2017). Two studies find CM effective regardless of involvement in the criminal justice system (Ginley et al., 2017; Petry et al., 2011) and receiving CM treatment may decrease future illegal activity among those with criminal justice involvement at baseline (Ginley et al., 2017). However, results from DeFulio et al. (2013) suggested a stronger and statistically significant response among those initiating treatment without a criminal justice referral compared to those initiating treatment with a referral.

Some clinical indicators are associated with enhanced response in CM treatments (see Forster et al., 2019 for a review). For example, in Weinstock et al. (2007), CM was effective for a range of psychiatric severity compared to standard care in terms of abstinence outcomes, but CM relative to standard care is especially robust in preserving retention among those with moderate and high psychiatric severity compared to those with low psychiatric severity. Other clinical characteristics may signal a muted response to CM. Perhaps the most consistent of these markers of decreased response is positive baseline toxicology status, which is known to predict poor treatment response in non-CM psychosocial treatment modalities (Alterman et al., 1996, 1997; Kampman et al., 2001; Sanchez-Hervas et al., 2010) and in CM studies (Petry et al., 2004; Stitzer et al., 2007a,b). To examine whether higher “dose” CM would improve outcomes in these baseline positive patients, Petry et al. (2012) randomized patients with a drug-positive toxicology result to usual intensive outpatient services, standard CM ($250), or a higher magnitude CM ($560) condition. Although drug-positive patients benefited in both the standard and higher magnitude CM conditions, the higher magnitude condition produced more robust outcomes, suggesting that enhanced CM protocols may be beneficial for this group.

8.2. Cultural adaptations of CM

Input directly from targeted populations and from the clinicians providing their care can be valuable in the design stage. For example, the cultural adaptation of CM for American Indian/Native Americans (AI/NA) communities involved input from stakeholders who indicated that the acceptable modification of twice weekly visits would be more feasible than thrice weekly visits given practical barriers such as limited access to transportation (McDonell et al., 2016; 2021). The advisory board in the McDonell et al. study and focus groups in Hirchak et al. (2018; 2019) suggested desirable reinforcers, which are critical to the success of a CM program. Stakeholders also provided input on cultural adaptations and awareness for study materials (e.g., compatibility with cultural perspectives on gift-giving, Buduli et al., 2018; language and symbols, Hirchak et al., 2018), suggestions for integration into clinical versus community services (Hirchak et al., 2019), and use of local leaders to administer the CM program (Hirchak et al., 2018). Stakeholder involvement also serves the function to increased buy-in among community leaders. This leadership endorsement is important for implementation success (i.e., local trusted champions of CM). Other groups have developed culturally sensitive CM protocols for homeless gay and bisexual men (Nyamathi et al., 2017). Overall, efforts to include stakeholder input can be informative for practical considerations (e.g., highlighting areas of concern to address in training workshops), as well as for efficacy as in the selection of meaningful, desirable reinforcers.

9. Conclusion

CM is one of the few effective options currently available to treat stimulant use disorders. However, deviations from evidence-based practices are common and undermine clinical impact. I recommend organizations start with a research protocol that has proven its effectiveness in achieving desired patient outcomes (e.g., increasing abstinence). Modifications should be considered only when absolutely necessary, and organizations should be aware that modifications to CM protocols are cumulative in impact. In this article, I have presented design considerations categorized as best practice based on extant research, acceptable-if-necessary modifications, or practices that are not recommended. The aim, of course, should be for the best practice option for each parameter. One or two acceptable-if-necessary modifications might be selected if unavoidable, and none of the not recommended practices should be used. A model prize CM protocol is provided that notably already includes one acceptable modification away from optimal practice (twice weekly rather than thrice weekly urine sampling and reinforcement); further modifications should be avoided. This model protocol focuses on stimulant negative urines as the target behavior, is 12 weeks in duration, incorporates escalation and resets, and uses a magnitude likely to be effective in changing patient outcomes.

CM is the treatment of choice for stimulant use disorder because it produces better outcomes than comparator therapies. However, these robust impacts will only be reproduced in the clinical realm when programs use evidence-based CM protocols. Our patients deserve the full benefit of this treatment option.

Highlights.

• Few resources support the design of contingency management (CM) protocols.

• This guide outlines evidence-based practices for CM protocol design.

• Acceptable-if-necessary modifications are outlined.

• Practices that should be avoided and are not evidence-based are also described.

Appendix A

Checklist for CM Implementation:

1. Develop detailed CM protocol for your program

   1. Which patients are eligible?
   2. Which member of the team will enroll patients?
   3. Which member of the team will deliver CM with patients?
   4. How will absences be handled?
   5. How will CM be integrated into the clinic workflow?
   6. Who will provide coverage for vacation/illness/etc.?
2. Fill supervisory roles

   1. Auditor
   2. Clinical supervisor

3. Develop a CM description handout for patients

4. Obtain rapid urine screens

5. Prepare fishbowl and slips

6. Stock cabinet

7. Create Prize Inventory Forms (tracks all inventory purchased)

8. Create Prize Release Form (tracks prizes released per patient)

9. Create CM Session Tracking (tracks all sessions per patient)

10. Create Reminder Slips

Appendix B

Example of Clinical Tracking for CM Sessions

Week 1			Week 2
Date:			Date:
Day of week (circle)	M, T, W, Th, F	M, T, W, Th, F	Day of week (circle)	M, T, W, Th, F	M, T, W, Th, F
Circle:	Show Ex Unex	Show Ex Unex	Circle:	Show Ex Unex	Show Ex Unex
Cocaine Utox Result:	+ −	+ −	Cocaine Utox Result:	+ −	+ −
Draws	___1___draw	_____draw(s)	Draws	_____draw(s)	_____draw(s)
Drawings (if appl) small (S), large (L), ,jumbo (J) or affirmation (A)			Drawings (if appl) small (S), large (L), ,jumbo (J) or affirmation (A)
Week 3			Week 4
Date:			Date:
Day of week (circle)	M, T, W, Th, F	M, T, W, Th, F	Day of week (circle)	M, T, W, Th, F	M, T, W, Th, F
Circle:	Show Ex Unex	Show Ex Unex	Circle:	Show Ex Unex	Show Ex Unex
Cocaine Utox Result:	+ −	+ −	Cocaine Utox Result:	+ −	+ −
Draws	_____draw(s)	_____draw(s)	Draws	_____draw(s)	_____draw(s)
Drawings (if appl) small (S), large (L), ,jumbo (J) or affirmation (A)			Drawings (if appl) small (S), large (L), ,jumbo (J) or affirmation (A)