Journal of Consulting and Clinical Psychology

Manuscript version of

Long-Term Efficacy of Contingency Management Treatment Based on Objective

Indicators of Abstinence From Illicit Substance Use up to 1 Year Following
Treatment: A Meta-Analysis
Meredith K. Ginley, Rory A. Pfund, Carla J. Rash, Kristyn Zajac

Funded by:
• National Institutes of Health

© 2021, American Psychological Association. This manuscript is not the copy of record and may not exactly
replicate the final, authoritative version of the article. Please do not copy or cite without authors’ permission.
The final version of record is available via its DOI: https://dx.doi.org/10.1037/ccp0000552

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
CONTINGENCY MANAGEMENT META-ANALYSIS 1

Long-term efficacy of contingency management treatment based on objective indicators of

abstinence from illicit substance use up to one-year following treatment: A meta-analysis

CONTINGENCY MANAGEMENT META-ANALYSIS 2

Abstract

Objective: Contingency management (CM) is often criticized for limited long-term impact. This

meta-analysis focused on objective indices of drug use (i.e., urine toxicology) to examine the

effects of CM on illicit substance use up to one-year following treatment. Method: Analyses

included k = 23 randomized trials of CM for stimulant, opioid, or polysubstance use disorders

that reported outcomes up to one year after the incentive delivery had ended. Using random

effects models, odds ratios (OR) were calculated for the likelihood of abstinence. Meta-

regressions and subgroup analyses explored how parameters of CM treatment, namely

escalation, frequency, immediacy, and magnitude of reinforcers, moderated outcomes. Results:

The overall likelihood of abstinence at the long-term follow-up among participants who received

CM versus a comparison treatment (nearly half of which were community-based comprehensive

therapies or protocol-based specific therapies) was OR = 1.22; 95% CI [1.01; 1.44], with low to

moderate heterogeneity (I2 = 36.68). Among 18 moderators, longer length of active treatment

was found to significantly improve long-term abstinence. Conclusions: CM showed long-term

benefit in reducing objective indices of drug use, above and beyond other active, evidence-based

treatments (e.g., CBT, twelve-step facilitation) and community-based intensive outpatient

treatment. These data suggest that policymakers and insurers should support and cover costs for

CM, which has hundreds of studies demonstrating its short-term efficacy, and now additional

data supporting its long-term efficacy.

CONTINGENCY MANAGEMENT META-ANALYSIS 3

What is the public health significance of this article?

This meta-analysis provides a summary of long-term outcomes of contingency management

treatment using objective indices of drug use. Contingency management was found to be more

efficacious than either standard care or other evidence based approaches up to one year following

the discontinuation of incentives.

Keywords: contingency management; urine toxicology; drug treatment; meta-analysis

CONTINGENCY MANAGEMENT META-ANALYSIS 4

Long-term efficacy of contingency management treatment based on objective indicators of

abstinence from illicit substance use up to one-year following treatment: A meta-analysis

Many patients with substance use disorders successfully reduce or cease drug use while

involved in treatment programs (Davis et al., 2016; Dutra et al., 2008; Irvin et al., 1999; Lee et

al., 2015; Lundahl et al., 2010; Magill et al., 2019; Magill & Ray, 2009; Roozen et al., 2004;

Sayegh et al., 2017). However, relapse rates are high (McLellan et al., 2000), and substance use

disorders are widely conceptualized as chronic relapsing conditions (Arria & McLellan, 2012;

McLellan et al., 2000). As such, treatment effects often diminish following the conclusion of

most active treatments (Benishek et al., 2014; Magill et al., 2009), and efforts to identify

treatments with lasting impact are essential for improving overall susbtance use outcomes.

Contingency management (CM) treatment is based on the theory of operant

conditioning. CM provides immediate, tangible reinforcers upon objective evidence of behavior

change (e.g., submission of drug negative urine samples). CM offers immediate positive

consequences for choosing not to use substances that compete with the positive aspects of drug

use, providing a bridge to the more substantial, but often very delayed benefits of recovery (i.e.,

employment, improved relationships; Petry, 2012). CM has the largest effect size of any

psychosocial treatment for reducing drug use during treatment (g = 0.54; Dutra et al., 2008), yet

remains one of the least likely evidence-based substance use disorder treatments to be offered in

clincal settings (Benishek, Kirby, Dugosh, & Padovano, 2010; Herbeck, Hser, & Teruya, 2008;

McGovern, Fox, Xie, & Drake, 2004; Willenbring et al., 2004).

Despite strong support for the efficacy of CM during the active treatment phase, many

professionals in the substance use treatment field question the durability of CM’s effect once

reinforcers are discontinued (Petry et al., 2017; Rash et al., 2012, 2014). Some meta-analyses
CONTINGENCY MANAGEMENT META-ANALYSIS 5

support this concern, finding that effect sizes decrease when reinforcers are discontinued

(Benishek et al., 2014, k = 19, 42% overlap [i.e., 42% of the studies in this meta-analysis are also

included in the current review]; Prendergast et al., 2006, k = 47, 4% overlap with current review;

Sayegh et al., 2017, k = 35 CM studies, 31% overlap with current review). However, a recent

systematic review reported nearly one-third of CM studies (across all drug and reinforcer types)

evidenced significant reductions in drug use after cessation of reinforcers (Davis et al., 2016, k =

69, 9% overlap with current review).

Our understanding of the long-term efficacy of substance use disorder treatments,

including CM, may be heavily impacted by variation in outcome measures. No single metric for

substance use treatment outcomes exists (Carroll et al., 2014; Donovan et al., 2012; Korte et al.,

2011; Tiffany et al., 2012). Objective indices (i.e., biological measures based on urine toxicology

screens) remove the risk of bias inherent in self-report (Del Boca & Noll, 2000; Hjorthøj et al.,

2012; Mangura & Kang, 1996; Schuler et al., 2009). However, most reviews of psychosocial

treatments for substance use disorders focus on self-reported drug use (Dutra et al., 2008;

Prendergast et al., 2002) or include objective biologically verified outcomes mixed in with

studies that only provide self-report (e.g., self-reported frequency of use) and/or randomly

biologically verified self-report outcomes (e.g., participants are told they could be drug tested to

verify self-report; Benishek et al., 2014; Sayegh, et al., 2017).

Past meta-analytic reviews have provided strong evidence of CM’s efficacy during

treatment and immediately post-treatment (Ainscough et al., 2017, k = 22, 9% overlap with

current review; Dutra et al., 2008, k = 14 CM studies, 21% overlap with current review; Griffith

et al., 2000, k = 30, 3% overlap with current review). Some meta-analyses have reported on

longer term outcomes (Benishek et al., 2014; Prendergast et al., 2006; Sayegh et al., 2017), but
CONTINGENCY MANAGEMENT META-ANALYSIS 6

these past reviews were limited because they either did not report biological outcomes

(Prendergast et al., 2006) or they combined biologically verified outcomes with self-report

and/or randomly biologically verified self-report outcomes (Benishek et al., 2014; Sayegh, et al.,

2017). Further, many reviews of CM focus on its efficacy when applied to only a single drug of

abuse (e.g., cocaine, Farronato et al., 2013, k = 8, 38% overlap with current review; nicotine,

Notley et al., 2019, k = 33, 0% overlap with current review) or do not comprehensively consider

samples recruited from outpatient settings as well as medication assisted treatment clinics

(Benishek et al., 2014). Additionally, no reviews that included long-term outcomes examined

how critical CM parameters found to influence efficacy during the active treatment phase (i.e.,

frequency, immediacy, escalation, fading, magnitude of reinforcement, and form of CM; e.g.,

prize-based CM, Beniskek et al., 2014 or voucher-based CM, Higgins et al., 2019) moderated

efficacy after the discontinuation of reinforcers.

The current systematic review and meta-analysis evaluated the efficacy of CM up to one

year after the reinforcer delivery ended using objectively verified substance use treatment

outcomes (i.e., urine drug screens). Only studies assessing outcomes following removal of the

incentives were included in the present analyses. The overall aim of the study was to determine

the relative efficacy of CM after reinforcers were discontinued compared to the long-term impact

of other psychosocial treatment approaches in reducing substance use. The secondary aim was to

understand critical moderators of the long-term efficacy of CM. This aim allows for an

integrated evaluation of several variables only examined in isolation or collapsed, and possibly

contributing to uncontrolled heterogeneity, in prior reviews (e.g., CM delivery method, single

illicit drug of abuse). This study seeks to directly address prior criticism of CM’s lack of durable
CONTINGENCY MANAGEMENT META-ANALYSIS 7

efficacy and includes comprehensive evaluation of factors that may enhance or detract from that

efficacy.

Method

Inclusion and Exclusion Criteria

Studies were included if they: 1) evaluated the efficacy of CM for treating illicit

stimulant, opioid, or polydrug use; 2) randomly assigned participants to two or more conditions;

3) enrolled at least 25 participants per condition (as recommended by Chambless & Hollon,

1998); 4) had at least one long-term follow-up (i.e., outcomes measured after the protocol

treatment was ended and the initial response to treatment had been determined); 5) presented

substance use outcomes based on urine toxicology; and 6) were published in English.

Studies were excluded if: 1) 25% or more of enrolled participants were under 18 years of

age; or 2) they presented secondary data from another trial included in the meta-analysis. When

two studies presented data from the same trial, we chose the study with the largest sample size.

Studies in which participants were recruited from a setting that provided medication assisted

treatment (e.g., methadone, buprenorphine) were included if they evaluated CM’s efficacy for

treating illicit substance use in that sample (e.g., CM to increase abstinence from stimulants in

persons also receiving methadone for an opioid use disorder) but were excluded if the CM

exclusively targeted a behavior other than abstinence, such as medication adherence.

Search Strategy

Studies published in any year through July 2020 were reviewed based on PRISMA

guidelines (Moher et al., 2009). Figure 1 provides a diagram of the implementation of these

guidelines. Searches were performed in PubMed and PsycInfo using the following combination

of search terms ["motivational interview*" OR "motivational enhancement" OR "motivation*

CONTINGENCY MANAGEMENT META-ANALYSIS 8

intervention" OR “contingency management" OR "voucher" OR "behavioral contracting" OR

"token economy" OR "community reinforcement" OR "matrix model" OR "aftercare" OR

"relapse prevention" OR "twelve step" OR "12 step" OR "twelve step facilitation" OR "12 step

facilitation" OR “family thera*” OR "family intervention*" OR “couples thera*” OR "couples

intervention" OR "seeking safety" OR "mindful*"] AND [“randomize*” OR "random" OR

"randomly" OR "clinical trial" OR “control* trial”] AND ["substance*" OR "drug*" OR

"cocaine" OR "crack" OR "stimulant" OR "amphetamine" OR "methamphetamine" OR "heroin"

OR "opiate*" OR "opioid" OR "marijuana" OR "cannabis"]. Search terms included the names of

evidence-based treatments besides CM to ensure inclusion of all studies with CM as the

comparator. Additionally, we searched: 1) reference lists of CM review articles identified using

the search terms (k = 17); 2) CM review articles identified in the Cochrane Database of

Systematic Reviews (k = 5), and 3) reference lists of studies that met our inclusion/exclusion

criteria (k = 23) to ensure we included all relevant studies.

Screening Abstracts

Titles and abstracts captured by the search strategy were screened in a multistep process.

One author imported all identified records into reference management software (Mendeley

Desktop, 2017). Confirmed duplicates were deleted, and the remaining records were screened

individually. Clearly non-relevant records (e.g., relapse prevention for a physical health

condition, smoking studies) were removed, as were records for which abstracts indicated the

paper would not meet inclusion criteria (e.g., small sample, non-random assignment). Articles

for which the abstract appeared to meet inclusion, or for which eligibility could not be

determined, were moved to full-text review. One author independently conducted full-text

review to determine inclusion of studies. Two additional authors completed a second full-text
CONTINGENCY MANAGEMENT META-ANALYSIS 9

review on a randomly selected sample of one-third of the 439 excluded articles as well as 100%

of studies classified by the first reviewer as meeting inclusion but not exclusion criteria.

Reviewers used a codebook for making inclusion/exclusion decisions. Inter-rater percentage

agreement overall for inclusion and exclusion of articles was 97%, k = 0.88. Discrepancies were

resolved through consensus, and when needed, discussion with a third reviewer.

Data Extraction

The primary outcome extracted from each study was urine toxicology results at the

longest available follow-up, which were either the percentage of negative urine toxicology

results (i.e., the number of negative urines samples provided as a ratio to the total number of

urine samples provided over the duration of the follow-up period with samples collected at

several time points) or the point prevalence rates of abstinence (i.e., whether a urine sample was

negative for a target drug at the single time point of the follow-up assessment). If both outcomes

were available, percentage of negative urine toxicology was prioritized for extraction, because

this outcome is more predictive of long-term outcomes (Preston et al., 1998; Stitzer et al., 2009).

If a study had multiple follow-up periods, results from the latest follow-up were reported. Results

from intent-to-treat samples were extracted whenever they were available. If the study did not

use intent-to-treat analyses, coders adjusted urine toxicology results with missing data assumed

positive (for k = 6 studies, 7 samples). If data were missing such that calculations using intent-to-

treat analyses were not possible, authors were contacted directly with a 100% response rate (k =

2; Chuzynski et al., 2015; Petry et al., 2015). Intent-to-treat calculations and author

correspondence allowed for inclusion of all studies identified by our search in the calculations of

overall odds ratios (ORs) described below.

CONTINGENCY MANAGEMENT META-ANALYSIS 10

In addition to the primary outcome variable, we extracted a number of study, participant,

and CM treatment variables. The study variables were publication year, comparison condition

(nonspecific therapy, community-based comprehensive therapy, protocol-focused specific

therapy; see next paragraph for further information about comparison condition categories),

targeted drug (stimulants only, opioids only, or polysubstance use), whether study recruitment

was conducted in a medication assisted treatment clinic (yes/no), outcome type (percentage of

negative urine samples or point prevalence of abstinence), and when long-term drug use

outcomes were evaluated since discontinuation of CM, measured in weeks. Participant variables

were demographic characteristics such as mean age, the percentage identifying as female, and

the percentage identifying as White. The CM treatment variables were: 1) escalation, i.e.,

participants could earn rewards of escalating value for consecutive negative urine toxicology

samples (yes/no), 2) fading, i.e., a design feature where reinforcers are reduced or become more

variable over time (faded versus not faded), 3) frequency (the number of times reinforcement

was earned per week) 4) immediacy (immediate reinforcement delivery versus delayed), 5)

maximum reinforcer magnitude available in average number of dollars available per participant,

6) CM delivery method (prize versus voucher), and 7) the duration of the CM protocol measured

in weeks.

Comparison conditions were grouped into nonspecific therapy, community-based

comprehensive therapy, or protocol-focused specific therapy. Nonspecific therapies included

“treatment as usual” or “standard care” where participants had assistance from providers related

to substance use treatment, mental health care, and other psychosocial needs but it was on an

infrequently/unstructured basis (~ < 1x per week). Community-based comprehensive therapy

included structured programs of care with more frequent contacts (e.g., intensive outpatient
CONTINGENCY MANAGEMENT META-ANALYSIS 11

substance use treatment). Comparison conditions were denoted as protocol-focused specific

therapies when the studies employed a specific recognized and/or manualized efficacious

substance use treatment (e.g., cognitive-behavioral therapy, twelve-step facilitation).

The Cochrane Risk of Bias tool was used to assess for possible bias in the included

studies (Higgins & Green, 2011). The following four criteria were assessed: random assignment,

allocation concealment, masking of outcome assessors, and incomplete outcome data. Selective

outcome reporting was not rated because most psychosocial treatment trials are still not

prospectively registered (Bradley et al., 2017). Each risk of bias criterion was designated as high,

low, or unclear risk of bias. We computed overall quality for each study; studies with three or

more indicators of low risk of bias were designated high quality, and studies with two or fewer

indicators of low risk were designated low quality.

For each study included in the final review, at least two authors extracted data using a

standardized form. A third author independently extracted data from a randomly selected third of

studies to ensure 3-way reliability. Inter-rater agreement for data extraction was 96%, and

differences were resolved through consensus or review by a third coder when necessary.

Analytic Plan

Descriptive data are provided for each study individually, with each study grouped by

whether or not participants were receiving medication assisted treatment (Table 1).

All statistical analyses were performed using Comprehensive Meta-Analysis v3.3070.

ORs, which represented the likelihood of a participant who received CM achieving abstinence

over a participant who received a comparison treatment, were calculated for each treatment

comparison group across the 23 studies with respect to urine toxicology outcomes. If studies

contained multiple CM groups compared to one control comparison, then the OR effect sizes
CONTINGENCY MANAGEMENT META-ANALYSIS 12

were combined and averaged into one OR effect size. Effect sizes were combined by aggregating

binary data of positive and negative urine toxicology outcomes from the treatment groups and

comparing the aggregated data to the positive and negative urine toxicology outcomes of the

comparison condition (Borenstein et al., 2009). A one-study removed analysis was conducted to

gauge the impact of each study on the overall ORs.

Calculations used natural log transformations of the OR with inverse variance weighting

to account for differences in sample sizes across studies. Results were then converted back to

ORs via inverse natural log transformations (Bland & Altman, 2000; Lipsey & Wilson, 2001).

Studies were expected to be at least moderately heterogeneous given the differing lengths of

follow-up periods and different target drugs, with variability not solely due to sampling error. As

such, overall ORs were calculated using random effects models (Neyeloff et al., 2012).

Cochran’s Q statistic estimated heterogeneity in effect sizes, and the inconsistency index (I2) was

reported as an estimate of variance due to heterogeneity (Higgins et al., 2003; Neyeloff et al.,

2012).

Multiple tests were conducted to test for possible publication bias, including the

examination of a funnel plot, the Egger regression test for asymmetry (Egger et al., 1997), and

calculation of the fail-safe N (Rosenthal, 1979). An asymmetrical funnel plot indicates

publication bias. The Egger test uses linear regression to assess the relation between error and

study effect sizes. A fail-safe N determines the number of non-significant studies not identified

by a review that would be needed for a significant OR to no longer be significant (Rosenthal,

1979).

We explored 18 possible moderators of CM efficacy (see Tables 1 and 2) at long-term

follow up. Meta-regressions were conducted to test for potential moderators with long-term
CONTINGENCY MANAGEMENT META-ANALYSIS 13

outcomes where the log OR effect sizes were regressed onto continuous variables detailed in the

data extraction section. Subgroup analyses tested for differences in long-term outcomes based on

the categorical variables detailed in the data extraction section. Subgroup analyses were

conducted using a mixed effects model, and significance was tested with the fixed effects model.

Variables were only tested as moderators if there were at least 10 studies reporting sufficient

statistical information (Higgins & Green, 2011). One-study removed analyses were conducted to

gauge the impact of each study moderator on the findings from meta-regressions and subgroup

analyses (Borenstein et al., 2013).

Results

The search yielded 5,510 records, which led to the identification of 23 independent

studies (see Figure 1 for the complete details of the search). Table 1 displays the 23 studies with

24 CM treatment-to comparison treatment contrasts at long-term follow-up along with the design

features of the CM treatment conditions.

Across the 23 studies, a total of 3,320 participants were allocated to study conditions. All

studies were conducted in the United States of America. Publication dates ranged from 1997 to

2015. Sample sizes across the 23 studies ranged from 52 to 388 (M = 138.3, SD = 74.2, median =

118.0). The mean age of participants was 39.1 (SD = 5.8, median = 37.6). Approximately 42.1%

(SD = 15.7, median = 44.1) of the sample identified as female, and 45.2% (SD = 20.5, median =

46.4) identified as White. Of the 23 studies, 38% included a sample of participants recruited

from a medication assisted treatment clinic (100% of which were methadone clinics, but

substances targeted by CM in these studies varied, see Table 2 for more details).

Of the 32 CM treatments, approximately 53% provided prize-based CM and 47% used

the voucher method. Almost all treatments utilized escalating reinforcers (91%) that were
CONTINGENCY MANAGEMENT META-ANALYSIS 14

delivered immediately (84%) upon submission of a negative urine screen. Few treatments (25%)

utilized a reinforcement schedule that faded over time. The frequency of reinforcement that

could be earned per week ranged from 1 and 7. The average maximum magnitude of

reinforcement available per participant per treatment episode was $914.46 (median = $466.0).

Table 2 displays the primary drug that participants used, the type of comparison group, how the

urinalysis outcome was reported (point prevalence vs percent negative), when the long-term

follow-up occurred relative to when CM was discontinued, and the risk of bias assessment for

each study criterion. Most studies examined the effect of CM on stimulant use only (67%),

followed by polysubstance use (29%) and opioid use only (4%). Slightly more than half the

studies (54%) utilized nonspecific therapy comparison groups, 33% used community-based

nonspecific comprehensive therapies, and 13% utilized protocol-focused specific treatments. At

long-term follow-up, about 58% of studies examined the point prevalence of abstinence and 42%

the percentage of negative urine samples submitted over the follow-up period. These outcomes

were assessed between 6 and 52 weeks (median = 24.0) since CM was discontinued.

Results of the Cochrane Risk of Bias tool indicated 54% of the studies were designated

high quality and about 46% were low quality. Approximately two-thirds (67%) of studies

provided an adequately detailed description of the method used to generate a random sequence

(e.g., used a computer program or a random table of numbers) and about 33% did not (e.g.,

simply stated participants were “randomized” with no further description of procedures). One

study (4%) described an adequate method to conceal the allocation of participants to conditions

from study investigators, while the remaining studies described no method to conceal allocation.

All (100%) studies were determined to have adequate masking of assessors, as they used the

objective assessment of urine toxicology to assess abstinence from drug use. Approximately 62%
CONTINGENCY MANAGEMENT META-ANALYSIS 15

of studies reported appropriate procedures to analyze outcome data (e.g., no missing outcome

data) and about 38% did not (e.g., completer analysis; imputed outcomes).

Long-Term Outcomes

Figure 2 displays a forest plot of the overall OR effect size of each study at long-term

follow-up and the weighted average OR effect size. At long-term follow-up, the weighted

average OR effect size was 1.22, 95% CI [1.03, 1.44], p = .02 (k = 23), with low to moderate

heterogeneity (I2 = 36.68). The one study removed analysis resulted in ORs ranging from 1.18 to

1.26, indicating that the results were not highly influenced by any single study. This weighted

effect size indicates that participants who received CM were 1.22 times more likely to be

abstinent based on urinalysis at follow-ups occurring after (median = 24 weeks) discontinuation

of the incentive delivery than participants who received a nonspecific therapy, a nonspecific

comprehensive therapy, or a specific therapy comparison condition. The funnel plot of the OR

effect sizes was symmetrical, and the Egger’s regression test did not indicate publication bias (p

> .05). The fail-safe N indicated that 37 unpublished studies with nonsignificant results would be

necessary to reduce this result to a nonsignificant level.

Moderators of Long-Term Outcome

We explored 18 possible moderators of CM’s long-term efficacy. Meta-regressions tested

the possible moderating effect of several participant demographics, publication year, reinforcer

magnitude, the number of weeks elapsed since the discontinuation of CM, and the duration of the

CM protocol (Table 3). There were significant moderating effects of publication year and

treatment duration on CM outcomes up to one year after the discontinuation of reinforcers. The

meta-regressions indicated a one-year increase in publication year was associated with a 0.04

decrease in the log OR (p = .04; k = 23). For treatment duration, a one week increase in the
CONTINGENCY MANAGEMENT META-ANALYSIS 16

duration of CM was associated with a 0.03 increase in the log OR (p = .04, k = 23). The log OR

effect size also increased as reinforcer magnitude increased (p = .05; k = 23). However, for

reinforcer magnitude, a one study removed analysis indicated the relation appeared mostly

driven by one study (i.e., Silverman et al., 2004) and became nonsignificant when that study was

removed from the meta-regression (p = .59; k = 22). Subgroup analyses tested the moderating

effects of several categorical variables (Table 4). No other moderators were significant, including

the CM delivery method (i.e., prize versus vouchers) and CM parameters (i.e., escalating, fading,

and immediacy) (all p’s > .05); however, this might be due to homogeneity in the types of CM

treatments examined in this meta-analysis, as many utilized similar reinforcement schedules.

Discussion

Objective assessment of long-term outcomes is relatively uncommon in clinical trials for

substance use treatments. This meta-analysis included 23 randomized trials of CM that had large

(>25/condition) samples of adult participants and reported urine toxicology results at long-term

follow-ups. This study focused specifically on CM, a psychosocial treatment model with strong

efficacy for reducing substance use during the active treatment period (Dutra et al., 2008) that

has faced skepticism about its durability (Petry et al., 2017). Despite this criticism, the overall

OR for CM at long-term follow up was significant, and participants who received CM evidenced

a 22% greater likelihood of abstinence at a median of 24 weeks after reinforcement ended than

participants receiving comparison treatments. These results provide support of lasting benefits of

CM after reinforcers have been discontinued using objective indices of drug use outcomes.

CM was found to be equally efficacious in the long-term regardless of participant age,

race, or gender. Length of follow-up, engagement of participants in medication assisted

treatment, type of comparison condition, drug(s) used, outcome type, and study quality did not
CONTINGENCY MANAGEMENT META-ANALYSIS 17

significantly moderate CM efficacy at long-term follow-up. Study publication year was found to

relate to a signficant decrease the efficacy of CM, with newer studies showing smaller effect

sizes. Additionally, longer treatment duration was associated with better long-term outcomes.

Longer treatment duration may allow for greater opportunity to establish durations of continuous

abstinence, a metric which has been consistently associated with better long-term outcomes

(Preston et al., 1998, Stitzer et al., 2009). Type of reinforcement (chance to win prizes versus

vouchers for each negative drug screen) did not impact long-term outcomes, providing further

support for the efficacy of both CM approaches (Petry et al., 2005; Petry et al., 2015).

Clinical Applications

CM’s costs are one of the foremost barriers to CM adoption (e.g., Benishek et al., 2010;

Kirby et al., 2006; Rash et al., 2012; Rash et al., 2014), and costs are often directly related to the

duration of CM treatment (i.e., longer CM protocols increase costs). Clinics will struggle to

implement CM with fidelity without external support to fund CM, which is especially important

in light of findings that CM is no more effective than standard care when reinforcement

magnitude drops below certain levels (Petry et al., 2004). To be effective, CM’s “dose” must be

in the effective range, and a significant proportion of addiction treatment providers in community

settings report maximum available reinforcement per patient well below effective “doses” (Rash

et al., 2013; 2020). The Veterans Affairs’ national implementation of CM in its IOP programs

provides a model of successful adoption and sustainment of CM programs that maintained

fidelity, including to parameters such as magnitude and duration (DePhilippis et al., 2018; Petry

et al., 2014; Rash & DePhilippies, 2019). Outside the Veterans Affairs, new pathways to fund

CM by payers (i.e., reimbursement) have yet to be made available despite CM’s clear evidence

of efficacy relative to other commonly used psychosocial treatments (e.g., CBT, MI, relapse
CONTINGENCY MANAGEMENT META-ANALYSIS 18

prevention; Dutra et al., 2008; Petry et al., 2017). These funding issues represent critical barriers

to the accessibility of CM that not only affect patients, but also limit potential societal benefits of

successful substance use disorder treatment in the form of improved employment and

productivity indices, reduced criminal activity, reduced risk behavior and spread of disease, and

improved family functioning (Petry et al., 2017). Despite these barriers, CM has some distinct

advantages beyond its clinical superiority. It can be integrated with wide variety of platform

therapies; it works with most client populations; and it can be readily adapted to clinic and client

needs. Further, both clinical and non-clinical staff can be trained to deliver CM, which may open

additional options for accessing treatment in non-traditional settings (e.g., housing programs,

employment programs) if funding barriers are resolved.

Long-Term Benefits

Long-term benefits were significant compared to comparison treatments even though

many of the trials contained design features that could have reduced the likelihood of uncovering

between-group differences. Notably, 13% of CM trials employed a treatment comparison that

was itself a specific protocol-based evidence based treatment (e.g., cognitive-behavioral

therapy), and 33% of “standard care” conditions were intensive outpatient treatment. Thus, many

participants, not only those in active CM, were engaged in robust, high intensity treatment and

likely received many of the indirect and often unmeasurable benefits of psychotherapy treatment

(Wampold, 2015). Effect sizes are larger when “passive” or no or minimal comparison

conditions are used relative to “active” and attention-matched comparison conditions

(Prendergast et al., 2002), yet this meta-analysis found benefits of CM even with rigorous

interventions as comparators. Further, effect sizes of CM were not significantly different for

participants who were or were not on medication assisted treatment. This finding suggests that
CONTINGENCY MANAGEMENT META-ANALYSIS 19

CM can be equally effective for those recruited from a medication assisted treatment treatment, a

result that may become even more salient as public health efforts seek to increase capacity for

individuals to access medication-based addiction treatments (Jones et al., 2015).

Despite the heterogeneity in effect sizes at long-term follow up, few clinical moderators

were significant. However, prior research indicates that several CM parameters are significantly

associated with enhanced efficacy of CM, including immediacy of reinforcement (Griffith et al.,

2000; Lussier et al., 2006), frequency of reinforcement (Griffith et al., 2000), and escalation of

reinforcement magnitude (Roll et al., 1996). We suspect that some of these null effects are driven

by the high quality designs used in the majority of included studies, resulting in little

heterogeneity in these particular CM parameters.

Limitations

Limitations include those common to all meta-analyses, such as publication bias.

However, we set rigorous inclusion and exclusion criteria, only included randomized trials

reporting objective indices of drug use, and conducted risk of bias assessment to increase

confidence in the results. No evidence of publication bias was found across multiple metrics (i.e.,

examination of funnel plots, the Egger’s regression test, and the calculation of the Fail-safe N).

Additionally, some statistical tests of moderators may have resulted in nonsignificant results due

to lack of power. For example, several CM parameters (e.g., immediate and escalating

reinforcers; Lussier et al., 2006; Roll et al., 1996) known to enhance abstinence were not

significantly associated with long-term outcomes in the present meta-analysis. This may be due

to few included studies utilizing CM protocols with delayed and non-escalating reinforcers.

Collectively, these measures constitute an improvement on past studies, especially those utilizing

participant self-report of drug abstinence.

CONTINGENCY MANAGEMENT META-ANALYSIS 20

Focusing on long-term outcomes led to the removal of studies without objective indices

of drug use at follow-ups. To create variables of meaningful moderators with groupings large

enough to analyze, some of the nuances of specific study variables may have been lost and may

contribute to unmeasured heterogeneity in the analyses. For example, polysubstance was used as

an overall label for studies assessing abstinence from more than one drug, but the number and

types of drugs covered by this classification ranged widely by study. We also could not examine

the influence of other key variables, such as comorbid mental health diagnoses, because few

studies reported sufficient details. Finally, the median length of follow-up was only about 6

months after treatment ended and, given the chronic nature of substance use disorders, it will be

important for future studies to utilize longer follow-up periods.

Urine toxicology provides an objective index of substance use, but it is not without

limitations. First, most urine toxicology tests only capture drug use in the few days preceding the

sample collection. Though CM protocols are designed with this limitation in mind, follow-ups

are often limited to a snapshot of abstinence. Second, sensitivity and specificity of toxicology

tests vary by drug of abuse (Peace et al., 2000). The studies primarily assessed abstinence from

stimulants or multiple substances concurrently. Persons with different drug use disorders may

respond differentially to CM and other treatments. For example, those with multiple drug use

disorders have more difficulties in achieving abstinence from all substances, resulting in lower

overall effect sizes of treatments for polysubstance users (Dutra et al., 2008).

Although no method for measuring drug use outcomes is standard across studies (Carroll

et al., 2014; Donovan et al., 2014; Tiffany et al., 2012), biological indices such as toxicology

testing should be prioritized. In assessing outcomes for chronic medical conditions such as

diabetes and heart disease, for example, most if not all studies include measurement of A1c
CONTINGENCY MANAGEMENT META-ANALYSIS 21

levels and blood pressure; reliance on self-reports when objective indices are available is

considered unacceptable in clinical trials targeting other medical conditions. The relatively small

number of trials that included toxicology testing in this review underscores that evaluation of

substance use disorders lags behind other chronic conditions. The lack of standardization of

outcome measures is not limited to the evaluation of CM and also may impact findings in other

reviews evaluating long-term effects of cognitive behavioral, relapse prevention, and other

addiction treatments (Burke et al., 2003; Magill & Ray, 2009; Ray et al., 2020; Sayegh et al.,

2017). To improve quality of research, and ultimately treatment, researchers and clinicians

should prioritize objective indices of drug use in evaluating both short- and long-term efficacy of

treatment approaches.

Conclusion

Past meta-analytic reviews of CM established its efficacy for improving substance use

outcomes during treatment and immediately post-treatment (Ainscough et al., 2017; Dutra et al.,

2008; Griffith et al., 2000). In addition, several meta-analyses focused on the long-term impact

of CM (Benishek et al., 2014; Prendergast et al., 2006; Sayegh et al., 2017), but were limited

because they did not report objective biological outcomes (Prendergast et al., 2006), or they

merged biologically verified outcomes with self-report and/or randomly biologically verified

self-report outcomes (Benishek et al., 2014; Sayegh, et al., 2017). Results of the current meta-

analysis provide new information to the field. Specifically, focusing on urine toxicology results,

this meta-analysis found a significant long-term effect for CM, directly addressing the common

concern that the effects of CM disappear once reinforcers are no longer provided. Of note, other

evidence-based psychosocial treatments for substance use disorders (e.g., cognitive behavioral

therapy; Magill et al., 2019) do not face this same level of criticism, yet none have demonstrated
CONTINGENCY MANAGEMENT META-ANALYSIS 22

significant long-term effects when only objective indicators of substance use are examined

(Burke et al., 2003; Magill & Ray, 2009; Magill et al., 2019). Further, the effect CM was robust

across a wide range of demographic and clinical moderators. Overall, CM increased odds of

abstinence across multiple investigative teams, participant demographics, and drugs of abuse.

Benefits of CM were present across rigorously designed trials, including those with comparison

groups using established, active treatment elements (Magill et al., 2019; Magill & Ray, 2009).

Patients with substance use disorders deserve access to treatments with the greatest evidence of

efficacy, and private and public insurers and society should support such treatments (Petry et al.,

2017; Rash et al., 2017; Roll et al., 2009). These results provide novel evidence that CM has

long-term efficacy in reducing drug use. However, no insurer or public payer, other than the

Veterans Administration (DePhilippis et al., 2018), presently covers costs of CM. It is time that

other healthcare systems and policy support this efficacious psychosocial intervention.
CONTINGENCY MANAGEMENT META-ANALYSIS 23

References
Ainscough, T. S., McNeill, A., Strang, J., Calder, R., & Brose, L. S. (2017). Contingency

management interventions for non-prescribed drug use during treatment for opiate

addiction: a systematic review and meta-analysis. Drug and Alcohol Dependence, 178,

318-339.

*Alessi, S. M., Hanson, T., Wieners, M., & Petry, N. M. (2007). Low-cost contingency

management in community clinics: Delivering incentives partially in group

therapy. Experimental and Clinical Psychopharmacology, 15(3), 293.

Arria, A. M., & McLellan, A. T. (2012). Evolution of concept, but not action, in addiction

treatment. Substance Use & Misuse, 47(8-9), 1041-1048.

Benishek, L. A., Dugosh, K. L., Kirby, K. C., Matejkowski, J., Clements, N. T., Seymour, B. L.,

& Festinger, D. S. (2014). Prize‐ based contingency management for the treatment of

substance abusers: A meta‐ analysis. Addiction, 109(9), 1426-1436.

Benishek, L. A., Kirby, K. C., Dugosh, K. L., & Padovano, A. (2010). Beliefs about the

empirical support of drug abuse treatment interventions: A survey of outpatient treatment

providers. Drug and Alcohol Dependence, 107(2-3), 202-208.

Bland, J. M., & Altman, D. G. (2000). The odds ratio. BMJ: British Medical Journal, 320(7247),

1468.

Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2013). Comprehensive Meta-Analysis

Version 3. Biostat: Englewood, NJ.

Borenstein, M., Hedges, L. V., Higgins, J., & Rothstein, H. R. (2009). Chapter 25: Multiple

comparisons within a study (pp. 239 – 242). John Wiley & Sons, Ltd.

Bradley, H. A., Rucklidge, J. J., & Mulder, R. T. (2017). A systematic review of trial registration

and selective outcome reporting in psychotherapy randomized controlled trials. Acta

CONTINGENCY MANAGEMENT META-ANALYSIS 24

Psychiatrica Scandinavica, 135(1), 65-77.

*Brooner, R. K., Kidorf, M. S., King, V. L., Stoller, K. B., Neufeld, K. J., & Kolodner, K.

(2007). Comparing adaptive stepped care and monetary-based voucher interventions for

opioid dependence. Drug and Alcohol Dependence, 88, S14-S23.

Burke, B. L., Arkowitz, H., & Menchola, M. (2003). The efficacy of motivational interviewing: a

meta-analysis of controlled clinical trials. Journal of Consulting and Clinical

Psychology, 71(5), 843.

Carroll, K. M., Kiluk, B. D., Nich, C., DeVito, E. E., Decker, S., LaPaglia, D., Duffey, D.,

Babuscio, T. A. & Ball, S. A. (2014). Toward empirical identification of a clinically

meaningful indicator of treatment outcome: features of candidate indicators and

evaluation of sensitivity to treatment effects and relationship to one year follow up

cocaine use outcomes. Drug and Alcohol Dependence, 137, 3-19.

Chambless, D. L., & Hollon, S. D. (1998). Defining empirically supported therapies. Journal of

Consulting and Clinical Psychology, 66(1), 7.

*Chudzynski, J., Roll, J. M., McPherson, S., Cameron, J. M., & Howell, D. N. (2015).

Reinforcement schedule effects on long-term behavior change. The Psychological

Record, 65(2), 347-353.

Davis, D. R., Kurti, A. N., Skelly, J. M., Redner, R., White, T. J., & Higgins, S. T. (2016). A

review of the literature on contingency management in the treatment of substance use

disorders, 2009–2014. Preventive Medicine, 92, 36-46.

Del Boca, F. K., & Noll, J. A. (2000). Truth or consequences: the validity of self‐ report data in

health services research on addictions. Addiction, 95(11s3), 347-360.

DePhilippis, D., Petry, N. M., Bonn-Miller, M. O., Rosenbach, S. B., & McKay, J. R. (2018).
CONTINGENCY MANAGEMENT META-ANALYSIS 25

The national implementation of Contingency Management (CM) in the Department of

Veterans Affairs: Attendance at CM sessions and substance use outcomes. Drug and

Alcohol Dependence, 185, 367-373.

Donovan, D. M., Bigelow, G. E., Brigham, G. S., Carroll, K. M., Cohen, A. J., Gardin, J. G.,

Hamilton, J. A., Huestis, M. A., Lindblad, R., & Marlatt, G. A. (2012). Primary outcome

indices in illicit drug dependence treatment research: Systematic approach to selection

and measurement of drug use end‐ points in clinical trials. Addiction, 107(4), 694-708.

Dutra, L., Stathopoulou, G., Basden, S. L., Leyro, T. M., Powers, M. B., & Otto, M. W. (2008).

A meta-analytic review of psychosocial interventions for substance use

disorders. American Journal of Psychiatry, 165(2), 179-187.

Egger, M., Davey Smith, G., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected

by a simple, graphical test. British Medical Journal, 315, 629–634.

*Epstein, D. H., Hawkins, W. E., Covi, L., Umbricht, A., & Preston, K. L. (2003). Cognitive-

behavioral therapy plus contingency management for cocaine use: findings during

treatment and across 12-month follow-up. Psychology of Addictive Behaviors, 17(1), 73.

Farronato, N. S., Dürsteler-MacFarland, K. M., Wiesbeck, G. A., & Petitjean, S. A. (2013). A

systematic review comparing cognitive-behavioral therapy and contingency management

for cocaine dependence. Journal of Addictive Diseases, 32(3), 274-287.

Griffith, J. D., Rowan-Szal, G. A., Roark, R. R., & Simpson, D. D. (2000). Contingency

management in outpatient methadone treatment: A meta-analysis. Drug and Alcohol

Dependence, 58(1-2), 55-66.

CONTINGENCY MANAGEMENT META-ANALYSIS 26

*Hagedorn, H. J., Noorbaloochi, S., Simon, A. B., Bangerter, A., Stitzer, M. L., Stetler, C. B., &

Kivlahan, D. (2013). Rewarding early abstinence in Veterans Health Administration

addiction clinics. Journal of Substance Abuse Treatment, 45(1), 109-117.

Herbeck, D. M., Hser, Y. I., & Teruya, C. (2008). Empirically supported substance abuse

treatment approaches: A survey of treatment providers' perspectives and

practices. Addictive Behaviors, 33(5), 699-712.

Higgins, J. P., & Green, S. (Eds.). (2011). Cochrane Handbook for Systematic Reviews of

Interventions (Vol. 4). John Wiley & Sons.

Higgins, S. T., Kurti, A. N., & Davis, D. R. (2019). Voucher-based contingency management is

efficacious but underutilized in treating addictions. Perspectives on Behavior Science,

42(3), 501-524.

Higgins, J. P., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency

in meta-analyses. BMJ: British Medical Journal, 327(7414), 557-560.

Hjorthøj, C. R., Hjorthøj, A. R., & Nordentoft, M. (2012). Validity of timeline follow-back for

self-reported use of cannabis and other illicit substances—systematic review and meta-

analysis. Addictive Behaviors, 37(3), 225-233.

*Iguchi, M. Y., Belding, M. A., Morral, A. R., Lamb, R. J., & Husband, S. D. (1997).

Reinforcing operants other than abstinence in drug abuse treatment: An effective

alternative for reducing drug use. Journal of Consulting and Clinical Psychology, 65(3),

421.

Irvin, J. E., Bowers, C. A., Dunn, M. E., & Wang, M. C. (1999). Efficacy of relapse prevention:

a meta-analytic review. Journal of Consulting and Clinical Psychology, 67(4), 563.

Jones, C. M., Campopiano, M., Baldwin, G., & McCance-Katz, E. (2015). National and state
CONTINGENCY MANAGEMENT META-ANALYSIS 27

treatment need and capacity for opioid agonist medication-assisted treatment. American

Journal of Public Health, 105(8), e55-e63.

*Jones, H. E., Wong, C. J., Tuten, M., & Stitzer, M. L. (2005). Reinforcement-based therapy: 12-

month evaluation of an outpatient drug-free treatment for heroin abusers. Drug and

Alcohol Dependence, 79(2), 119-128.

Kirby K. C., Benishek L. A., Dugosh K. L., & Kerwin. M. E. (2006). Substance abuse treatment

providers’ beliefs and objections regarding contingency management: Implications for

dissemination. Drug and Alcohol Dependence, 85, 19-27.

Korte, J. E., Magruder, K. M., Chiuzan, C. C., Logan, S. L., Killeen, T., Bandyopadhyay, D., &

Brady, K. T. (2011). Assessing drug use during follow-up: direct comparison of

candidate outcome definitions in pooled analyses of addiction treatment studies. The

American Journal of Drug and Alcohol Abuse, 37(5), 358-366.

Lee, E. B., An, W., Levin, M. E., & Twohig, M. P. (2015). An initial meta-analysis of

Acceptance and Commitment Therapy for treating substance use disorders. Drug and

Alcohol Dependence, 155, 1-7.

Lipsey, M. W., & Wilson, D. B. (2001). Practical Meta-analysis. SAGE publications, Inc.

Lundahl, B. W., Kunz, C., Brownell, C., Tollefson, D., & Burke, B. L. (2010). A meta-analysis

of motivational interviewing: Twenty-five years of empirical studies. Research on Social

Work Practice, 20(2), 137-160.

Lussier, J. P., Heil, S. H., Mongeon, J. A., Badger, G. J., & Higgins, S. T. (2006). A meta‐

analysis of voucher‐ based reinforcement therapy for substance use

disorders. Addiction, 101(2), 192-203.

Magill, M., & Ray, L. A. (2009). Cognitive-behavioral treatment with adult alcohol and illicit
CONTINGENCY MANAGEMENT META-ANALYSIS 28

drug users: a meta-analysis of randomized controlled trials. Journal of Studies on Alcohol

and Drugs, 70(4), 516-527.

Magill, M., Ray, L., Kiluk, B., Hoadley, A., Bernstein, M., Tonigan, J. S., & Carroll, K. (2019).

A meta-analysis of cognitive-behavioral therapy for alcohol or other drug use disorders:

Treatment efficacy by contrast condition. Journal of Consulting and Clinical

Psychology, 87(12), 1093.

Magura, S., & Kang, S. Y. (1996). Validity of self-reported drug use in high risk populations: a

meta-analytical review. Substance Use & Misuse, 31(9), 1131-1153.

*McDonell, M. G., Srebnik, D., Angelo, F., McPherson, S., Lowe, J. M., Sugar, A., Short, R. A.,

Roll, J. M., & Ries, R. K. (2013). Randomized controlled trial of contingency

management for stimulant use in community mental health patients with serious mental

illness. The American Journal of Psychiatry, 170(1), 94-101.

McGovern, M. P., Fox, T. S., Xie, H., & Drake, R. E. (2004). A survey of clinical practices and

readiness to adopt evidence-based practices: Dissemination research in an addiction

treatment system. Journal of Substance Abuse Treatment, 26(4), 305-312.

McLellan, A. T., Lewis, D. C., O'Brien, C. P., & Kleber, H. D. (2000). Drug dependence, a

chronic medical illness: implications for treatment, insurance, and outcomes

evaluation. JAMA, 284(13), 1689-1695.

Moher, D., Liberati, A., Tetzlaff, J., & Altman, D. G. (2009). Preferred reporting items for

systematic reviews and meta-analyses: the PRISMA statement. Annals of Internal

Medicine, 151(4), 264-269.

Neyeloff, J. L., Fuchs, S. C., & Moreira, L. B. (2012). Meta-analyses and Forest plots using a

microsoft excel spreadsheet: step-by-step guide focusing on descriptive data

CONTINGENCY MANAGEMENT META-ANALYSIS 29

analysis. BMC Research Notes, 5(1), 52.

Notley, C., Gentry, S., Livingstone‐ Banks, J., Bauld, L., Perera, R., & Hartmann‐ Boyce, J.

(2019). Incentives for smoking cessation. Cochrane Database of Systematic Reviews, (7).

Peace, M. R., Tarnai, L. D., & Poklis, A. (2000). Performance evaluation of four on-site drug-

testing devices for detection of drugs of abuse in urine. Journal of Analytical

Toxicology, 24(7), 589-594.

*Peirce, J. M., Petry, N. M., Stitzer, M. L., Blaine, J., Kellogg, S., Satterfield, F., Schwartz, M.,

Krasnansky, J., Pencer, E., Silva-Vazquez, L. & Kirby, K. C. (2006). Effects of lower-

cost incentives on stimulant abstinence in methadone maintenance treatment: A National

Drug Abuse Treatment Clinical Trials Network study. Archives of General

Psychiatry, 63(2), 201-208.

Petry, N. M. (2012). Contingency management for substance abuse treatment: A guide to

implementing this evidence-based practice. Routledge.

*Petry, N. M., Alessi, S. M., Barry, D., & Carroll, K. M. (2015). Standard magnitude prize

reinforcers can be as efficacious as larger magnitude reinforcers in cocaine-dependent

methadone patients. Journal of Consulting and Clinical Psychology, 83(3), 464.

*Petry, N. M., Alessi, S. M., Carroll, K. M., Hanson, T., MacKinnon, S., Rounsaville, B., &

Sierra, S. (2006). Contingency management treatments: Reinforcing abstinence versus

adherence with goal-related activities. Journal of Consulting and Clinical

Psychology, 74(3), 592-601.

*Petry, N. M., Alessi, S. M., & Ledgerwood, D. M. (2012b). A randomized trial of contingency

management delivered by community therapists. Journal of Consulting and Clinical

Psychology, 80(2), 286.

CONTINGENCY MANAGEMENT META-ANALYSIS 30

Petry, N. M., Alessi, S. M., Olmstead, T. A., Rash, C. J., & Zajac, K. (2017). Contingency

management treatment for substance use disorders: How far has it come, and where does

it need to go?. Psychology of Addictive Behaviors, 31(8), 897.

*Petry, N. M., Alessi, S. M., Marx, J., Austin, M., & Tardif, M. (2005). Vouchers versus prizes:

Contingency management treatment of substance abusers in community settings. Journal

of Consulting and Clinical Psychology, 73(6), 1005-1014.

*Petry, N. M., Barry, D., Alessi, S. M., Rounsaville, B. J., & Carroll, K. M. (2012a). A

randomized trial adapting contingency management targets based on initial abstinence

status of cocaine-dependent patients. Journal of Consulting and Clinical

Psychology, 80(2), 276-285.

Petry, N. M., DePhilippis, D., Rash, C. J., Drapkin, M., & McKay, J. R. (2014). Nationwide

dissemination of contingency management: The Veterans Administration initiative.

American Journal on Addictions, 23(3), 205-210.

Petry, N. M., Tedford, J., Austin, M., Nich, C., Carroll, K. M., & Rounsaville, B. J. (2004). Prize

reinforcement contingency management for treating cocaine users: How low can we go,

and with whom? Addiction, 99, 349–360.

*Petry, N. M., Weinstock, J., & Alessi, S. M. (2011). A randomized trial of contingency

management delivered in the context of group counseling. Journal of Consulting and

Clinical Psychology, 79(5), 686-696.

*Petry, N. M., Weinstock, J., Alessi, S. M., Lewis, M. W., & Dieckhaus, K. (2010). Group-based

randomized trial of contingencies for health and abstinence in HIV patients. Journal of

Consulting and Clinical Psychology, 78(1), 89-97. doi:10.1037/a0016778

Prendergast, M., Podus, D., Finney, J., Greenwell, L., & Roll, J. (2006). Contingency
CONTINGENCY MANAGEMENT META-ANALYSIS 31

management for treatment of substance use disorders: A meta‐

analysis. Addiction, 101(11), 1546-1560.

Preston, K. L., Silverman, K., Higgens, S. T., Brooner, R. K., Montoya, I., Schuster, C. R., &

Cone, E. J. (1998). Cocaine use early in treatment predicts outcome in a behavioral

treatment program. Journal of Consulting and Clinical Psychology, 66(4), 691.

*Preston, K. L., Umbricht, A., & Epstein, D. H. (2002). Abstinence reinforcement maintenance

contingency and one-year follow-up. Drug and alcohol dependence, 67(2), 125-137.

Rash, C. J., Alessi, S. M., & Zajac, K. (2020). Examining implementation of contingency

management in real-world settings. Psychology of Addictive Behaviors, 34(1), 89-98.

Rash, C. J., & DePhilippis, D. (2019). Considerations for implementing contingency

management in substance abuse treatment clinics: The Veterans Affairs initiative as a

model. Perspectives on Behavior Science, 42(3), 479-99.

Rash, C. J., DePhilippis, D., McKay, J. R., Drapkin, M., & Petry, N. M. (2013). Training

workshops positively impact beliefs about contingency management in a nationwide

dissemination effort. Journal of Substance Abuse Treatment, 45(3), 306-312.

Rash, C. J., Petry, N. M., Kirby, K. C., Martino, S., Roll, J., & Stitzer, M. L. (2012). Identifying

provider beliefs related to contingency management adoption using the Contingency

Management Beliefs Questionnaire. Drug & Alcohol Dependence, 121, 205-212.

Rash, C. J., Stitzer, M., & Weinstock, J. (2017). Contingency management: New directions and

remaining challenges for an evidence-based intervention. Journal of Substance Abuse

Treatment, 72, 10-18.

*Rawson, R. A., Huber, A., McCann, M., Shoptaw, S., Farabee, D., Reiber, C., & Ling, W.

(2002). A comparison of contingency management and cognitive-behavioral approaches

CONTINGENCY MANAGEMENT META-ANALYSIS 32

during methadone maintenance treatment for cocaine dependence. Archives of General

Psychiatry, 59(9), 817-824.

*Rawson, R. A., McCann, M. J., Flammino, F., Shoptaw, S., Miotto, K., Reiber, C., & Ling, W.

(2006). A comparison of contingency management and cognitive‐ behavioral approaches

for stimulant‐ dependent individuals. Addiction, 101(2), 267-274.

Ray, L. A., Meredith, L. R., Kiluk, B. D., Walthers, J., Carroll, K. M., & Magill, M. (2020).

Combined Pharmacotherapy and Cognitive Behavioral Therapy for Adults With Alcohol

or Substance Use Disorders: A Systematic Review and Meta-analysis. JAMA Network

Open, 3(6), e208279-e208279.

Roll, J. M., Higgins, S. T., & Badger, G. J. (1996). An experimental comparison of three

different schedules of reinforcement of drug abstinence using cigarette smoking as an

exemplar. Journal of Applied Behavior Analysis, 29(4), 495-505.

Roll, J. M., Madden, G. J., Rawson, R., & Petry, N. M. (2009). Facilitating the adoption of

contingency management for the treatment of substance use disorders. Behavior Analysis

in Practice, 2(1), 4-13.

*Roll, J. M., Petry, N. M., Stitzer, M. L., Brecht, M. L., Peirce, J. M., McCann, M. J., Blaine, J.,

MacDonald, M., DiMaria, J., Lucero, L., & Kellogg, S. (2006). Contingency

management for the treatment of methamphetamine use disorders. The American Journal

of Psychiatry, 163(11), 1993-1999.

Roozen, H. G., Boulogne, J. J., van Tulder, M. W., van den Brink, W., De Jong, C. A., &

Kerkhof, A. J. (2004). A systematic review of the effectiveness of the community

reinforcement approach in alcohol, cocaine and opioid addiction. Drug and Alcohol
CONTINGENCY MANAGEMENT META-ANALYSIS 33

Dependence, 74(1), 1-13.

Rosenthal, R. (1979). The ‘file drawer problem’ and tolerance for null results. Psychological

Bulletin, 86, 638–641.

Sayegh, C. S., Huey Jr, S. J., Zara, E. J., & Jhaveri, K. (2017). Follow-up treatment effects of

contingency management and motivational interviewing on substance use: A meta-

analysis. Psychology of Addictive Behaviors, 31(4), 403.

Schuler, M. S., Lechner, W. V., Carter, R. E., & Malcolm, R. (2009). Temporal and gender

trends in concordance of urine drug screens and self-reported use in cocaine treatment

studies. Journal of Addiction Medicine, 3(4), 211.

*Shoptaw, S., Reback, C. J., Peck, J. A., Yang, X., Rotheram-Fuller, E., Larkins, S., ... & Hucks-

Ortiz, C. (2005). Behavioral treatment approaches for methamphetamine dependence and

HIV-related sexual risk behaviors among urban gay and bisexual men. Drug and Alcohol

Dependence, 78(2), 125-134.

*Silverman, K., Robles, E., Mudric, T., Bigelow, G. E., & Stitzer, M. L. (2004). A randomized

trial of long-term reinforcement of cocaine abstinence in methadone-maintained patients

who inject drugs. Journal of Consulting and Clinical Psychology, 72(5), 839.

Stitzer, M. L., Peirce, J., Petry, N. M., Kirby, K., Roll, J., Krasnansky, J., Cohen, A., Blaine, J.,

Vandrey, R., Kolodner, K., & Li, R. (2009). Abstinence-based incentives in methadone

maintenance: Interaction with intake stimulant test results. Experimental and Clinical

Psychopharmacology, 15(4), 344.

Tiffany, S. T., Friedman, L., Greenfield, S. F., Hasin, D. S., & Jackson, R. (2012). Beyond drug

use: a systematic consideration of other outcomes in evaluations of treatments for

substance use disorders. Addiction, 107(4), 709-718.

CONTINGENCY MANAGEMENT META-ANALYSIS 34

Wampold, B. E. (2015). How important are the common factors in psychotherapy? An update.

World Psychiatry, 14(3), 270-277.

Willenbring, M. L., Kivlahan, D., Kenny, M., Grillo, M., Hagedorn, H., & Postier, A. (2004).

Beliefs about evidence-based practices in addiction treatment: A survey of Veterans

Administration program leaders. Journal of Substance Abuse Treatment, 26(2), 79 85

 CONTINGENCY MANAGEMENT META-ANALYSIS 35

Table 1.

Description of Studies and Study Conditions Included in the Meta-Analysis on Contingency Management
Escalation
Total Comparison condition CM Prize or with Immediate Reinforcer
Study N Treatment conditions (n) (n) Duration Voucher Reset Fading Frequencya Reinforcers Magnitudeb
Participants recruited from a setting other than a medication assisted treatment clinic

Alessi, 2007 103 CM (46) SC-IOP (57) 12 wks P Y Y 2 Y $248

CM (30) V Y N Y
CM (32) V Y N N
Chudzynski, 2015 119 CM (28) SC-Matrix (29) 16 wks V Y N 3 N $1155
Hagedorn, 2013 141 CM (71) SC-IOP (70) 8 wks V Y N 2 Y $157
Jones, 2005 130 CM living expenses (66) SC (64) 12 wks V N Y 7 N $2294
McDonell, 2013 176 CM (91) SC + NC prizes (85) 12 wks P Y N 3 Y $588
CM (53) V Y Y Y $456
Petry, 2005 142 CM (51) SC-IOP (38) 12 wks P Y Y 3 Y $543
Petry, 2006 84 CM (44) SC-IOP (40) 12 wks P Y Y 3 Y $470
Petry, 2010 170 CM (89) TSF + NC prizes (81) 24 wks P Y N 1 Y $400
Petry, 2011 239 CM (117) SC-IOP (122) 12 wks P Y N 2 Y $462
CM (35) P Y Y Y $250
Petry, 2012a 109 CM (40) SC-IOP (34) 12 wks P Y Y 3 Y $560
Petry, 2012a 2nd
sample 226 CM (118) SC-IOP (108) 12 wks P Y Y 3 Y $250
Rawson, 2006 118 CM (60) CBT (58) 16 wks V Y N 3 Y $989
Roll, 2006 113 CM (51) SC (62) 12 wks P Y N 2 Y $420
CM (29) 16 wks Y
CM 2 month (30) 8 wks Y
Roll, 2013 118 CM 1 month (30) SC-Matrix (29) 1 wk P Y N 3 Y $250
 CONTINGENCY MANAGEMENT META-ANALYSIS 36

Shoptaw, 2005 82 CM (42) CBT (40) 16 wks V Y N 3 Y $1278

Participants recruited from a medication assisted treatment clinic (all were methadone maintenance clinics)
Brooner, 2007 118 CM (59) SC (59) 24 wks V Y N 1 Y $3201
Epstein, 2003 96 CM (47) SC + NC vouchers (49) 12 wks V Y N 3 Y $1155
Iguchi, 1997 62 CM (27) SC (35) 12 wks V N N 3 N $180
Peirce, 2006 388 CM (190) SC (198) 12 wks P Y N 2 Y $400
Petry, 2012b 130 CM (71) SC (59) 12 wks P Y N 3 Y $381
CM (63) V Y N Y $900
CM (62) P Y N Y $900
Petry, 2015 240 CM (58) SC (57) 12 wks P Y N 3 Y $300
Preston, 2002 110 CM take home doses (55) NC take home doses (55) 12 wks V N N 3 N $360
Rawson, 2002 54 CM (27) SC (27) 16 wks V Y N 3 Y $1278
Silverman, 2004 52 CM take home doses (26) SC (26) 52 wks V Y N 3 Y $5800
Notes. aFrequency refers to the maximum number of times that reinforcement could be earned per week; bReinforcement magnitude
refers to the maximum monetary value that could be earned per participant during treatment; CBT = cognitive behavioral therapy; CM
= contingency management; IOP = intensive outpatient treatment; NC = noncontingent incentives; nr = not reported; SC = standard
care or treatment as usual; TSF = twelve-step facilitation;
CONTINGENCY MANAGEMENT META-ANALYSIS 37

Table 2.
Design Features of Studies Included in the Meta-Analysis and Assessment of their Study Quality
Long-Term Outcome Random Allocation Blinding of Overall
Comparison Type (weeks since Sequence Concealme Outcome Complete Study
Study Drug Condition CM discontinuation) Generation nt Assessors Data Quality

Participants recruited from a setting other than a medication assisted treatment clinic

Alessi, 2007 S CBCT PP (24) ? ? + + -

Chudzynski, 2015 S CBCT %neg (1-12) ? ? + - -

Hagedorn, 2013 S CBCT PP (46) + ? + + +

Jones, 2005 P NST PP (40) + ? + + +

McDonell, 2013 S NST PP (12) + ? + - -

Petry, 2005 S CBCT $PP (24) + ? + + +

Petry, 2006 P CBCT PP (24) + ? + + +

Petry, 2010 P PFST PP (28) + ? + + +

Petry, 2011 P CBCT PP (40) + ? + + +

Petry, 2012a S CBCT %neg (20-24) + ? + + +

Petry, 2012a 2nd
sample S CBCT %neg (20-24) + ? + + +

Rawson, 2006 S PFST PP (36) + ? + ? -

Roll, 2006 S NST %neg (12+24) ? ? + - -

Roll, 2013 S CBCT %neg (1-16) ? ? + - -

CONTINGENCY MANAGEMENT META-ANALYSIS 38

Shoptaw, 2005 S PFST PP (36) + ? + + +

Participants recruited from a medication assisted treatment clinic (all methadone)

Brooner, 2007 P NST %neg (1-12) ? ? + + -

Epstein, 2003 S NST %neg (1-26) ? ? + + -

Iguchi, 1997 P NST %neg (1-6) ? ? + ? -

Peirce, 2006 P NST PP (12) + + + - +

Petry, 2012b S NST PP (24) + ? + + +

Petry, 2015 S NST PP (40) + ? + - -

Preston, 2002 O NST PP (52) + ? + + +

Rawson, 2002 S NST %neg (1-26) ? ? + ? -

Silverman, 2004 S NST %neg (1-9) + ? + + +

Notes. + = low risk of bias/high study quality; - = high risk of bias/low study quality; ? = risk of bias is unclear; %neg = percentage of submitted
urine samples that were negative; CBCT = community based comprehensive therapy; NST = nonspecific therapy; O = opioids, P = poly
substance use; PP = point prevalence rates of abstinence; PFST = protocol focused specific therapy; S = stimulants
 CONTINGENCY MANAGEMENT META-ANALYSIS 39

Table 3
Results from Meta-Regressions of Several Possible Moderators with Contingency Management

Outcomes up to One Year Following Treatment

Point
Moderator (k) estimate 95% CI Z-value p-value
Participant age (24)
Slope 0.00 -0.03, 0.03 0.06 0.95
Intercept 0.16 -1.02, 1.34 0.27 0.78
Participant gender: Percentage female (24)
Slope 0.01 -0.00, 0.02 1.23 0.22
Intercept -0.14 -0.70, 0.42 -0.49 0.63
Participant race: Percentage White (21)
Slope -0.01 -0.02, 0.00 -1.11 0.26
Intercept 0.48 -0.07, 1.03 1.70 0.09
Publication year (24)
Slope -0.04 -0.07, -0.00 -2.03 0.04
Intercept 73.52 2.88, 144.17 2.04 0.04
Reinforcer frequency (24)
Slope -0.02 -0.19, 0.14 -0.29 0.77
Intercept 0.27 -0.20, 0.74 1.11 0.27
Reinforcer magnitude (24)
Slope 0.00 0.00, 0.00 1.98 0.05
Intercept 0.06 -0.19, 0.30 0.44 0.66
Time of follow up in weeks (24)
Slope -0.01 -0.02, 0.01 -1.22 0.22
Intercept 0.42 0.02, 0.82 2.08 0.04
Treatment duration (24)
Slope 0.03 0.00, 0.06 2.05 0.04
Intercept -0.25 -0.68, 0.18 -1.14 0.25
Note. Point estimates reflect the amount of increase or decrease in the log odds ratio effect sizes.
 CONTINGENCY MANAGEMENT META-ANALYSIS 40

Table 4
Results from Subgroup Analyses of Several Possible Moderating Variables with Contingency

Management Outcomes up to One Year Following Treatment

Moderator (k or number of studies) Odds Ratio 95% CI Q-value p-value

Control condition (24) 2.79 0.25
Nonspecific therapy (9) 1.15 0.90, 1.48
Community based comprehensive
therapy (12) 1.40 1.07, 1.83
Protocol focused specific therapy (3) 0.91 0.58, 1.43
Escalating reinforcers (24) 0.00 0.95
No (3) 1.21 0.74, 1.97
Yes (21) 1.23 1.02, 1.48
Fading reinforcers (24) 0.15 0.70
No (18) 1.24 1.03, 1.51
Yes (6) 1.14 0.78, 1.67
Immediate reinforcers (32) 1.03 0.31
No (5) 1.00 0.69, 1.46
Yes (27) 1.24 1.05, 1.48
Outcome type (24) 2.57 0.11
Percentage of negative samples (10) 1.49 1.08, 2.06
Point prevalence (14) 1.10 0.93, 1.31
Study recruitment at a Methadone
Treatment Clinic (24) 1.19 0.28
No (15) 1.16 0.97, 1.38
Yes (9) 1.46 1.00, 2.11
Study quality (24) 2.45 0.12
High (13) 1.08 0.88, 1.33
Low (11) 1.41 1.08, 1.84
Time of follow up in weeks (24) 0.40 0.52
< 3 months (6) 1.37 0.93, 2.00
3-12 months (18) 1.19 0.98, 1.44
Type of drug (24) 0.06 0.97
CONTINGENCY MANAGEMENT META-ANALYSIS 41

Opioids (1) 1.34 0.63, 2.84

Stimulants (15) 1.22 0.98, 1.51
Polysubstance use (8) 1.22 0.89, 1.67
Type of reinforcer (32) 0.40 0.53
Prizes (17) 1.16 0.98, 1.38
Voucher (15) 1.29 0.97, 1.73
CONTINGENCY MANAGEMENT META-ANALYSIS 42

Records identified through Additional records identified

database searching through other sources
(k = 5489) (k = 21)

Records after duplicates removed

(k = 4319)

Records excluded
Records screened
(k = 3331)
(k = 4319)

Abstract reviewed articles

excluded
Abstracts assessed for
eligibility (k = 526)

(k = 988)

Full-text articles excluded, with reasons (k = 439)

No urine toxicology outcome (k = 189)

Full-text articles Did not evaluate contingency management
assessed for eligibility targeting abstinence for illicit substance use
disorders (k = 73)
(k = 462) No long term follow up (k = 65)
Too few participants per condition (k = 48)
No random assignment (k = 41)
Did not focus on adults (k = 23)
Studies included in
qualitative synthesis
(k = 23 papers, 24
samples)

Figure 1.
Flowchart of records identified and reviewed
CONTINGENCY MANAGEMENT META-ANALYSIS 43

Study name Subgroup within study Statistics for each study Odds ratio and 95% CI
Odds Lower Upper
ratio limit limit Z-Value p-Value
Alessi 2007 CM 2.350 1.028 5.371 2.026 0.043
Brooner 2007 CM 1.780 0.841 3.766 1.508 0.132
Chudzynski 2015 Combined 0.912 0.675 1.230 -0.605 0.545
Epstein 2003 CM 2.270 0.941 5.476 1.825 0.068
Hagedorn 2013 CM 1.020 0.480 2.169 0.051 0.959
Iguchi 1997 CM 1.140 0.332 3.919 0.208 0.835
Jones 2005 CM living expenses 1.110 0.520 2.370 0.270 0.787
McDonell 2013 CM 1.570 0.858 2.873 1.463 0.143
Peirce 2006 CM 1.150 0.739 1.789 0.620 0.535
Petry 2005 Combined 1.633 0.893 2.985 1.592 0.111
Petry 2006 CM 0.470 0.159 1.390 -1.364 0.172
Petry 2010 CM 1.080 0.590 1.978 0.249 0.803
Petry 2011 CM 1.120 0.662 1.895 0.422 0.673
Petry 2012a Combined 1.099 0.555 2.176 0.271 0.786
Petry 2012a 2nd sample CM 0.820 0.484 1.390 -0.737 0.461
Petry 2012b CM 0.750 0.374 1.505 -0.810 0.418
Petry 2015 Combined 0.894 0.588 1.358 -0.527 0.598
Preston 2002 CM take home doses 1.340 0.632 2.840 0.764 0.445
Rawson 2002 CM 4.420 1.402 13.932 2.537 0.011
Rawson 2006 CM 0.860 0.360 2.052 -0.340 0.734
Roll 2006 CM 1.520 0.888 2.603 1.526 0.127
Roll 2013 Combined 1.712 1.108 2.644 2.424 0.015
Shoptaw 2005 CM 0.600 0.208 1.727 -0.947 0.344
Silverman 2004 CM take home doses 6.420 1.722 23.931 2.770 0.006
1.219 1.032 1.441 2.334 0.020
0.1 0.2 0.5 1 2 5 10

Favors Control Favors CM

Figure
Meta2. Analysis
Forest plot of effect sizes for contingency management treatment versus comparison conditions at long-term follow up
Notes. Values are truncated after the third decimal point. The last row in the forest plot represents the overall odds ratio effect size of
all studies. “Combined” indicates a study with multiple CM conditions that were summarized into one odds ratio effect size. CM =
contingency management