Short abstract
This systematic review synthesizes evidence on the effects of medication-assisted treatment for opioid use disorder on functional outcomes, including cognitive, physical, occupational, social/behavioral, and neurological function.
Keywords: Military Health and Health Care, Opioids, Pharmaceutical Drugs, Substance Abuse Treatment
Abstract
This systematic review addresses the question: What are the effects of medication-assisted treatment (MAT) that use buprenorphine, buprenorphine combined with naloxone, methadone, or naltrexone for opioid use disorder (OUD) on functional outcomes compared with wait-list, placebo, treatment without medication, any other comparator, or each other (e.g., buprenorphine versus naltrexone)?
Functional outcomes investigated included cognitive (e.g., memory), physical (e.g., fatigue), occupational (e.g., employment status), social/behavioral (e.g., criminal activity), and neurological (e.g., balance) function.
The authors searched five scientific research databases from inception to 2017 and reference mined existing reviews. Two independent literature reviewers screened 6,292 citations; 1,327 full-text publications were reviewed in detail and 37 studies met inclusion criteria. Critical appraisals assessed studies in detail, and quality of evidence was rated using established criteria. Results were synthesized in meta-analyses and presented in comprehensive evidence tables. Although MAT patients performed significantly better on some functional outcomes than persons with OUD who did not receive MAT, MAT patients performed worse on several cognitive measures than did matched “healthy” controls with no history of substance use disorder (SUD) or OUD. Because of the moderate-to-high risk of bias of most studies, quality of evidence is low or very low for all findings.
The small number of studies reporting on outcomes of interest and the weaknesses in the body of evidence prevent making strong conclusions about MAT effects on functional outcomes. The literature shows that more research is needed that targets functional outcomes specifically, and there is, in particular, a lack of research evaluating potential differences in functional effects among medication types, the route of administration, treatment modality, and length of treatment.
Introduction
In response to the growing epidemic of opioid misuse, federal agencies in the United States were directed to improve access to medication-assisted treatment (MAT). MAT is the use of approved medications combined with counseling, other behavioral therapies, and patient monitoring to treat opioid use disorder (OUD). Medications approved in the United States for MAT for OUD include methadone, buprenorphine, Suboxone (a combination of buprenorphine and naloxone), and naltrexone.
The Defense Centers of Excellence for Psychological Health and Traumatic Brain Injury commissioned a systematic review of the effects of MAT for OUD on functional outcomes. These outcomes include cognitive (e.g., memory), physical (e.g., fatigue), occupational (e.g., return to work), behavioral/social (e.g., family function), and neurological (e.g., balance) function. Such outcomes are important in determining whether active-duty service members can be deployed.
Key Questions
This review was guided by the following key question (KQ) and subquestions:
-
What are the effects of MAT (using buprenorphine, buprenorphine plus naloxone, methadone, or naltrexone) for OUD on functional outcomes compared with wait-list, placebo, treatment without medication, any other comparator, or each other (e.g., buprenorphine versus naltrexone)?
Do the effects vary by type of medication?
Do the effects vary by route of administration (e.g., oral versus injection versus implant)?
Do the effects vary by length of treatment, follow-up time, or later cessation of MAT?
Do the effects vary by treatment modality (e.g., methadone clinic versus prescription medication taken at home)?
Methods
To answer the KQs, we searched PubMed, PsycINFO, EMBASE, Cumulative Index to Nursing and Allied Health Literature, Cochrane CENTRAL, and Cochrane Database of Systematic Reviews from inception to January 2017, as well as bibliographies of existing systematic reviews to identify reports of English-language controlled trials, case-control studies, and cohort studies that compared two or more groups and reported baseline and follow-up data on functional outcomes. Cross-sectional studies were excluded, as were studies of MAT medications not approved for use for OUD in the United States. We included two types of studies: (1) those that compared MAT-treated patients with OUD to persons with OUD who were not treated with MAT (i.e., they received another treatment, placebo, treatment as usual, or no treatment) and (2) those that compared MAT-treated patients with OUD to matched controls with no history of substance use disorder (SUD).
Two independent reviewers screened 6,292 identified citations using predetermined eligibility criteria. Because functional outcomes are most often reported as secondary outcomes in studies of substance abuse treatment, we retrieved full-text copies of all studies that assessed the efficacy of MAT for OUD and therefore potentially could meet our inclusion criteria and combed the results sections for relevant outcomes. Reviewers abstracted prespecified study-level information and assessed each included study's risk of bias (ROB); all abstracted data were checked by the project lead for accuracy.
Meta-analyses were conducted using the Hartung-Knapp method for random-effects models when sufficient data were available. Continuous outcomes were expressed as standard mean differences (SMDs), and categorical outcomes were expressed as relative ratios (RRs) together with the 95-percent confidence intervals (CIs). The overall quality of evidence (QoE) was assessed using the Grades of Recommendation, Assessment, Development, and Evaluation approach, and we differentiated high, moderate, low, and very low confidence in summary results (conclusions).
Results
Despite a comprehensive literature search that identified 1,327 publications that were scrutinized as full text, only 37 studies (27 randomized controlled trials [RCTs] and ten observational studies reported in 41 articles) met inclusion criteria. No RCT was rated as having low ROB, primarily because of lack of participant blinding, high attrition rate, and lack of reporting of the method of randomization and allocation concealment. Several observational studies with low ROB were identified.
The studies reported on a large number of highly diverse functional outcome measures, including verbal memory, attention, insomnia, fatigue, and criminal activity. Functional measures were primary outcomes in only six of the RCTs; it is unclear if the other trials, which were statistically powered to detect differences in illicit use of opioids or treatment retention, had adequate power to detect differences in functional effects.
KQ 1: Effects of MAT on Functional Outcomes
We found that although MAT patients performed significantly better on some functional outcomes than persons with OUD who did not receive MAT, they performed worse on several cognitive measures than did matched “healthy” controls with no history of SUD or opioid use. Because of the limited number of studies identified and the moderate-to-high ROB of many of them, QoE is low or very low for all evidence statements.
Cognitive Function
Individuals with OUD using MAT versus healthy controls with no history of SUD.
A very large observational study conducted in France found that persons prescribed MAT had twice the risk of injurious traffic accidents than nonusers (low QoE). A cohort study measuring working memory found that OUD patients using either buprenorphine or methadone scored significantly worse than did matched controls with no history of SUD or opioid use (very low QoE). Another cohort study measured cognitive speed and found that OUD patients who used either buprenorphine or methadone scored significantly worse than did controls with no history of SUD or opioid use (very low QoE).
Two studies reported that MAT patients performed no worse than healthy controls in verbal memory tasks. One cohort study reported a difference in verbal memory favoring methadone patients, while another cohort study found no difference between methadone patients and healthy controls (very low QoE). Both studies found no difference between OUD patients taking buprenorphine and healthy controls (low QoE).
Two cohort studies showed no significant difference in attention between MAT-treated OUD patients and healthy controls with no history of SUD or opioid use (low QoE).
Individuals with OUD who were treated with MAT versus individuals with OUD not treated with MAT.
No studies that met our inclusion criteria compared cognitive outcomes between persons with OUD who were treated with MAT to persons with OUD who were not treated with MAT medications.
Occupational Function
Individuals with OUD who were treated with MAT versus individuals with OUD not treated with MAT.
Three RCTs and two observational studies that measured employment outcomes found no difference between MAT patients and persons with OUD treated for substance abuse without MAT (very low QoE).
Physical Function
Individuals with OUD who were treated with MAT versus controls with no history of SUD.
We identified no studies that reported physical function that met our inclusion criteria.
Individuals with OUD who were treated with MAT versus individuals with OUD not treated with MAT.
In one cohort study, a significantly lower percentage of patients with OUD who received buprenorphine reported fatigue than did untreated persons with OUD, while there was no difference in the rate of fatigue between persons with OUD who received methadone and these controls (very low QoE).
A meta-analysis of four RCTs found no difference in the percentage reporting insomnia between participants receiving MAT and those receiving a placebo (RR 1.02; CI 0.61, 1.71; moderate QoE). A methadone cohort study also reported no difference in insomnia (very low QoE).
One RCT comparing methadone versus a non-MAT intervention reported no difference in mean ASI medical scores. A cohort study comparing buprenorphine treatment with syringe exchange also reported no difference for this measure (very low QoE).
Behavioral/Social Function
Individuals with OUD who were treated with MAT versus controls with no history of SUD.
One small cohort study reported aggression outcomes; an RCT that randomized OUD patients to either buprenorphine or methadone found that individuals receiving either treatment scored significantly worse on aggressive responding than did controls with no history of SUD (very low QoE).
Individuals with OUD who were treated with MAT versus individuals with OUD not treated with MAT.
Studies reporting on crime (classified as a dysfunctional outcome) showed mixed results. Two RCTs reported that OUD patients on methadone spent fewer days engaged in criminal activity than those randomized to a placebo or wait-list (SMD −0.57; CI −1.00, −0.13; low QoE). However, a meta-analysis of RCTs that reported the percentage arrested or incarcerated found no statistically significant difference between patients randomized to MAT and those not randomized to MAT (RR 0.75; CI 0.46, 1.23; low QoE). An RCT and a cohort study that used a scale assessing illegal activities reported significantly better scores for OUD patients treated with MAT than for participants provided with psychosocially enhanced detox or syringe exchange. Another RCT and another cohort study reported no statistically significant difference for the mean number of charges or the mean number of arrests between MAT-treated patients and those not receiving treatment (all very low QoE).
One RCT reported no significant differences in mean family or psychiatric function scores between MAT and placebo groups; a cohort study also reported no statistically significant differences in a psychiatric function score (very low QoE).
Neurological Function
No studies that compared OUD patients who received MAT to those without MAT or that compared OUD patients who received MAT to controls with no history of SUD reported on neurological outcomes (e.g., hyporeflexia, balance, coordination).
KQ 1a: Effects by Type of Medication
Sixteen studies compared the effects of different MAT medications; of these, ten compared buprenorphine to methadone.
A meta-analysis of three RCTs that compared the effects of buprenorphine treatment on fatigue to that of methadone showed a significantly lower prevalence of fatigue in buprenorphine patients than in methadone patients (RR 0.62; CI 0.41, 0.95; moderate QoE). In absolute terms, 52 fewer buprenorphine patients per 1,000 reported fatigue compared with methadone patients.
Three RCTs that focused on cognitive function compared the effects of buprenorphine to methadone; no statistically significant differences in memory, cognitive speed and flexibility, attention, or vision were reported between treatments, with the exception of a small study where buprenorphine patients performed better than did methadone patients in vision tracking (low QoE for memory and attention because of high ROB and imprecision; very low QoE for cognitive speed, cognitive flexibility, and vision because of high ROB, imprecision, and lack of replication).
A meta-analysis of three RCTs reporting on insomnia found no statistical difference between buprenorphine and methadone groups (low QoE). Two RCTs reporting on pain perception found no significant difference between buprenorphine and methadone groups (low QoE). The only RCT of Suboxone versus methadone that reported functional outcomes found no significant difference in pain-rating scores at six months (very low QoE). One RCT of methadone versus naltrexone found no significant difference in the mean number of days patients engaged in illegal activity (very low QoE). A large nationally representative observational study found no differences in the increase in the proportion of participants who were employed and the number of arrests in the past 30 days among patients who received extended-release naltrexone administered by injection, oral naltrexone, Suboxone, or psychosocial treatment without medication (very low QoE).
KQ 1b: Effects by Route of Administration
We identified three RCTs that directly compared routes of administration of the same medication that reported on functional outcomes (physical and social function). None reported significant differences on these outcomes.
One RCT found no difference in risk of insomnia between oral Suboxone and Suboxone implant, and one RCT found no difference in effect on social function between methadone administered orally and by injection (very low QoE). Two RCTs found no difference in effects on mental health or physical health between methadone administered orally and that administered by injection (very low QoE). One large nationally representative observational study reported no differences in percentage employed or arrested during treatment among patients receiving Suboxone, oral naltrexone, injection naltrexone, or non-MAT treatment (very low QoE).
KQ 1c: Effects by Length of Treatment, Follow-Up, and Later Cessation
Three studies followed MAT patients longitudinally but none found an interaction effect of intervention by time on the reported results (ASI family component and psychiatric component) (very low QoE).
One RCT and one cohort study reported no statistically significant effect of length of treatment by treatment group on memory, attention, medical score, or legal score (very low QoE). A meta-regression across studies found no indication that RCTs with longer follow-up differed in effects on sleep (insomnia) or legal outcomes (arrests).
We identified only one study that addressed the question of how and when functional outcome effects change after cessation of MAT. The study compared MAT patients who remained in treatment to former patients no longer on MAT. Current MAT patients were significantly more likely to be not working in the past 30 days than former patients. The difference between current and former MAT treatment groups in terms of the percentage arrested in the past 30 days was not statistically significant (very low QoE).
KQ 1d: Effects by Treatment Modality
In the United States, methadone is traditionally dispensed under supervision at a methadone clinic, while buprenorphine is prescribed by a physician and can be distributed at a pharmacy. These two drugs are compared in KQ 1a. The results of studies that compared different modalities using the same medication are presented in the next section.
One cohort study with low ROB reported fewer nondrug-related crimes were committed by OUD patients prescribed methadone from a general practitioner's office than by those dispensed methadone at a traditional methadone clinic (very low QoE).
One RCT reported no differences in ASI psychiatric, legal, employment, or medical scores between a group take-home methadone program with twice a week distribution versus twice per month; however, the statistical power to detect differences in these functional outcomes is unknown (very low QoE).
Discussion
Some studies found significant effects in favor of MAT regarding the amount of criminal activity or legal status compared with persons with OUD not receiving MAT. In several studies, MAT patients performed significantly worse than matched controls with no history of SUD or opioid use on measures of aggression, working memory, and cognitive speed. However, it is unclear if the observed differences are because of MAT or because of long-term use of opioids in general. Although healthy controls are usually matched to patients on demographic and other characteristics, these individuals clearly differ in substance abuse history and may differ in unreported psychological, psychiatric, and family history characteristics that contribute to poor function. Quality of evidence for most outcomes was low or very low.
It is unclear in many instances if participants met standards required for military deployment. No studies were conducted on active-duty service members or reported performance on specific occupational tasks. No studies reported the current or former occupations of participants, and applicability of the outcome measures to successful military deployment was not discussed in any study.
Conclusions
Making clinical and policy recommendations is beyond the scope of the systematic review; the goal of this report was to summarize, synthesize, and assess the quality of the existing evidence. Despite an exhaustive and systematic search, the small number of studies that report on outcomes of interest and the weaknesses in the body of evidence prevent any strong conclusions about the effects of MAT on functional outcomes or differences in effects among medication types, route of administration, treatment modality, or length of treatment.