Abstract
Background and aims
Screening, brief intervention and ‘referral to treatment’ programs have been widely promoted as US federal policy. Little is known about the efficacy of the RT component (referral to treatment) of brief intervention for motivating patients with unhealthy drug use identified by screening to use addiction treatment. This study aimed to compare receipt of addiction treatment following two types of brief intervention for drug use vs a no-intervention control group among primary care patients screening positive for drug use.
Design
Secondary analyses from a single-site randomized controlled trial.
Setting
Massachusetts, USA
Participants
528 adults with Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) drug specific scores ≥4.
Interventions
Random assignment to: (1) a 10-15 minute brief negotiated interview (BNI) conducted by health educators (n=174), (2) a 30-45 minute adaptation of motivational interviewing by Masters-level counselors (MOTIV) (n=177), or 3) no BI (n=177). All received a list of treatment and mutual help resources; both intervention protocols included dedicated staff for treatment referrals.
Measurements
Receipt of any addiction treatment within 6 months after study entry, assessed in a statewide database and hospital electronic medical record linked to trial data
Findings
Among 528 participants, the main drugs used were marijuana (63%), cocaine (19%), and opioids (17%); 46% met past-year drug dependence criteria (short form Composite International Diagnostic Interview); and 10% of MOTIV, 18% of BNI, and 17% of control participants had any addiction treatment receipt within 6 months after study entry. There was no significant difference in addiction treatment receipt for BNI vs control (AOR 1.11; 95%CI 0.57, 2.15, Hochberg adjusted p=0.76). The MOTIV group had lower odds of linking to treatment (AOR 0.36, 95%CI: 0.17, 0.78, Hochberg adjusted p=0.02) compared with the no BI group.
Conclusion
Brief intervention delivered in primary care for screen-identified drug use did not significantly increase addiction treatment receipt; a motivational interviewing approach appeared to be counterproductive.
Keywords: health services research, drug use, primary care
Introduction
Unhealthy alcohol and other drug use, ranging from risk of harmful consequences to diagnosis of substance use disorder (SUD), is not adequately addressed in health care settings. Brief interventions for drugs in primary care settings have no or at best modest effect1,2,3 and for those who have a disorder, approaches rely on referral to treatment.4,5 Few people with SUDs (11.6%) receive treatment.6 Furthermore, referral from medical care settings is not a common source for such treatment.6
Although brief intervention (BI) may be inadequate as a stand-alone intervention, it is possible that BI is effective for facilitating linkage to addiction treatment for those with more severe drug use. To date, there has been relatively little in the way of randomized controlled trials (RCT) of BI for drug use on addiction treatment entry in primary care. This despite the widespread implementation of Screening, Brief Intervention, and Referral to Treatment (SBIRT programs) to screen and conduct BI for patients with unhealthy drug use and refer those with SUDs to treatment.5 The few studies investigating treatment receipt after BI for drug use have had mixed results ranging from greater addiction treatment access in an observational study7 to no effect on treatment entry in 2 randomized trials, one of which was in urgent care.2,8 Glass et al. found in a systematic review that unhealthy alcohol use had no effect on receipt of treatment, although BIs in almost half of the RCTs did not include any referral specific processes.9 Another limitation is the exclusion of patients with more severe drug use who might benefit from specialty treatment.1 The absence of objectively assessed treatment entry as an outcome and insufficient data on primary care patients are other limitations. Along with studies of the effectiveness of BI with referral-specific efforts for drug use on treatment entry,10 a better understanding is needed of who with screen-identified substance use in primary care is likely to receive addiction treatment to support coordination of primary care with substance use and mental health care.11,12,13
The primary objective of the current study was 1) to compare addiction treatment receipt following two different types of motivational BIs for drug use versus a no-intervention control group, among primary care patients with drug use identified by screening. Both interventions included referral to treatment (RT) when appropriate. Given the gaps in what is known about treatment utilization among primary care patients with screen-identified drug use, an exploratory aim was 2) to identify patient-level factors associated with addiction treatment receipt.
Methods
Study design
We present a pre-specified secondary analysis of treatment receipt from the Assessing Screening Plus Brief Intervention's Resulting Efficacy to stop drug use (ASPIRE) study, a randomized controlled trial (RCT) of two types of BI in primary care for screen-identified unhealthy drug use (clinicaltrials.gov Identifier: NCT00876941). Previous analyses showed no reductions by intervention group in drug use or drug-associated consequences.3 The primary objective of the current study was to test whether addiction treatment receipt differed across study arms. We also conducted exploratory analyses examining potential associations between patient-level factors and treatment receipt. The study was approved by the Institutional Review Boards of the Boston University Medical Campus and the Massachusetts Department of Public Health.
Participants
Recruitment occurred in the waiting room of an urban, hospital-based primary care practice. ASPIRE study participants, described in greater detail in Saitz et al.,3 were adults presenting for an appointment with a primary care clinician (PCC) with recent (i.e. past- 3 month) unhealthy drug use. Screening for unhealthy drug use was conducted by a health educator employed by the hospital (Boston Medical Center) or a research assistant trained in the same role. Eligibility criteria included a drug-specific Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) score of 4 or greater, indicating weekly or more drug use (including prescription misuse) or less frequent use with a consequence.14 Patients were excluded if they did not speak English or Spanish, planned to leave the Boston area that would prevent research participation, could not provide contact information for tracking purposes, or were pregnant.
Randomization and Intervention
After study enrollment, ASPIRE participants were randomly assigned to receive: 1) a single 15-minute Brief Negotiated Interview (BNI) by a health educator;8 2) 30-45 minute in-person motivational interviewing (MOTIV) with an offered booster session (in-person or by telephone) conducted by Masters-degree level counselors; or 3) no BI. Participants were randomly assigned to treatment arms using a stratified permuted block randomization procedure to balance random assignment by past-year drug dependence (short form Composite International Diagnostic Interview15 [CIDI-SF]) and main drug (self-identified drug of most concern at enrollment). All received printed information about treatment resources including phone numbers/web addresses for mutual help groups, the hospital behavioral health clinic and emergency department (ED), city SUD triage and a state treatment locator line. Brief Treatment (BT) -- 12 sessions of motivational interviewing/cognitive behavioral therapy -- by a primary care-based social worker was available. RT for BNI participants was made by health educators and for MOTIV participants, the BT specialist.
Referrals were made based on an assessment of severity (e.g. ASSIST score, past-year drug dependence [CIDI-SF], inability to abstain or cut down) and motivation. Participant interest in a referral and plans were communicated to the PCC in the electronic medical record (EMR). Participants could be referred to BT, an Office-based Opioid Treatment (OBOT) Program (also located in primary care), or outside treatment programs (via Boston Medical Center [BMC] treatment referral specialists available by phone or in-person in the ED). Participants were accompanied by counselors/educators to BT or OBOT for a “warm handoff.” All participants had access to the same treatment resources.
Both types of interventions were audio-recorded and coded using the Motivational Interviewing Treatment Integrity (MITI) scale and a study-developed instrument measuring study fidelity.3 Both interventions had scores consistent with proficiency.
Measures
Outcome: receipt of addiction treatment
We linked two data sources to the standardized in-person research interviews: 1) BMC EMR and 2) Massachusetts Bureau of Substance Abuse Services (BSAS) treatment data, which includes discharge data from all specialty programs that contract with the state for public funding. This dataset includes utilization in programs receiving Medicaid-reimbursed care or block grant funds for treatment resources not reimbursed by Medicaid. ASPIRE enrollees were matched to BSAS admission and discharge records on birthdate, gender, first and last name initials, and social security number.
The outcome, addiction treatment receipt within a 6-month period after study enrollment was identified in the BSAS and BMC EMR data. Modalities captured in the BSAS data included detoxification, outpatient treatment, residential treatment, medication-assisted treatment, transitional services, and a range of support services.16 Treatment services provided at BMC included addiction pharmacotherapy, ascertained by computerized search of participants' medication lists in the EMR.
Main independent variables and covariates
For the primary analysis, randomization group was the main independent variable. The covariates were past treatment utilization (6-months prior to study entry) and randomization stratification factors were (past-year drug dependence (CIDI-SF) and main drug).
To select main independent variables for the exploratory analysis of predictors of addiction treatment receipt, we used the Behavioral Model for Vulnerable Populations,17 which characterizes healthcare utilization as a function of (1) need for care; (2) predisposing sociodemographic vulnerabilities that influence access to care; and (3) enabling factors that facilitate or impede receipt of care.
We examined the following need factors: main drug type, drug-related consequences (Short Inventory of Problems-Drug),18 alcohol-specific ASSIST score ≥27 (alcohol use risk score consistent with dependence), drug-specific ASSIST score ≥27 (drug use risk score consistent with drug dependence), and severity of total substance involvement (Global ASSIST for drugs).14
Predisposing factors included: age, sex, race/ethnicity, living with children, employment, homelessness (any night in shelter/street in the past 3 months), criminal justice involvement, depressive symptoms (Patient Health Questionnaire (PHQ-9),19 overall health status (EuroQol),20 and severe pain (average level of ≥7 on 1 to 10 scale over the past 3 months).
Enabling factors included: primary language, health insurance, substance use within peer-group, past addiction treatment receipt, and ED utilization (past 3 months).
Statistical analysis
To describe the cohort, we calculated proportions, means, standard deviations, medians, and interquartile ranges as appropriate for all variables.
For the RCT analysis, the two main comparisons of interest were between each BI vs. control. Analyses were conducted using the intention-to-treat principle, analyzing participants according to randomized group. There were no missing values for intervention status, covariates, or outcomes.
This primary analysis of treatment receipt used multivariable logistic regression models, adjusted for the randomization stratification factors stated above. To assess the degree of confounding by addiction treatment receipt before study entry, we report “partially” adjusted models in which past addiction treatment receipt was not included in regression models. As in the main trial results, we corrected for multiple comparisons within outcome using the Hochberg procedure.21
To explore potential effect modification, we tested a series of interactions between randomization group and four separate factors: main drug, past-year drug dependence (CIDI-SF), past addiction treatment, and main drug-specific ASSIST score of ≥27. Our original data analysis plan was to conduct subsequent stratified analyses for interactions with p<0.15 in order to describe how intervention effects appeared to differ across groups. As a result of peer review, we conducted post-hoc subgroup analyses among those with: 1) ASSIST score ≥27 and 2) ASSIST score ≥27 whose main drug was not marijuana.
For the exploratory analysis identifying predictors of addiction treatment receipt, we used separate logistic regression models for each main independent variable of interest with the following covariates in models where the variable was not the main predictor: age, sex, race/ethnicity, primary language, depressive symptoms, main drug, and past treatment utilization. No pair of covariates and main independent variables had correlation greater than 0.40.
Additional confirmatory analyses adjusted for randomization group and health insurance in models where the main independent variable was significant at p<0.05. Finally, a single logistic regression model was fit with all significantly associated patient-level factors and core covariates. The full model did not include Global ASSIST score due to correlation with main drug (rho=0.51). We did not adjust for multiple comparisons due to the hypothesis-generating nature of the analysis. All analyses were completed using SAS/STAT software, Version 9.3 (SAS Institute).
Results
Study participants
Table 1 shows the characteristics of the study sample overall and by randomization group. The most common main drug was marijuana (63%) followed by cocaine (19%), and opioids (17%). Almost half had past-year drug dependence (46%) based on the CIDI-SF, but only 18% had a current ASSIST drug risk score consistent with dependence. Randomization arms were balanced for all variables including past addiction treatment receipt. Addiction treatment receipt for the cohort was 14% (73/528) within a 6-month period before study entry and 15% (78/528) within 6 months after study entry.
Table 1. Characteristics of primary care patients with unhealthy drug use identified by screening enrolled in a randomized trial of 2 types of brief interventionsa.
| Characteristic, n (%) | Overall n=528 | BNI n=174 | MOTIV n=177 | No BI n=177 |
|---|---|---|---|---|
| Age, mean years (STD) | 41 (12) | 40 (12) | 43 (12) | 41 (13) |
| Male sex | 369 (70%) | 124 (71%) | 126 (71%) | 119 (67%) |
| Race/ethnicity, non-white | 423 (80%) | 142 (82%) | 140 (79%) | 141 (80%) |
| Currently employed | 145 (27%) | 51 (29%) | 41 (23%) | 53 (30%) |
| Homeless b | 88 (17%) | 24 (14%) | 32 (18%) | 32 (18%) |
| Criminal justice involvement | ||||
| Incarceration | 23 (4%) | 6 (3%) | 5 (3%) | 12 (7%) |
| Probation/parole | 36 (7%) | 12 (7%) | 9 (5%) | 15 (8%) |
| None | 469 (89%) | 156 (90%) | 163 (92%) | 150 (85%) |
| Depressive symptoms (PHQ-9 ≥ 10) c | 189 (36%) | 63 (36%) | 66 (37%) | 60 (34%) |
| Health insurance | ||||
| Medicaid/Medicare | 429 (81%) | 138 (79%) | 153 (86%) | 138 (78%) |
| Private/Commercial | 69 (13%) | 24 (14%) | 18 (10%) | 27 (15%) |
| None | 30 (6%) | 12 (7%) | 6 (3%) | 12 (7%) |
| Primary Language | ||||
| Not English | 40 (8%) | 11 (6%) | 9 (5%) | 20 (11%) |
| English | 488 (92%) | 163 (94%) | 168 (95%) | 157 (89%) |
| Emergency Department visit past 3 months | 189 (36%) | 65 (37%) | 59 (33%) | 65 (37%) |
| Main drug d | ||||
| Cocaine | 98 (19%) | 32 (18%) | 33 (19%) | 33 (19%) |
| Opioids | 90 (17%) | 31 (18%) | 28 (16%) | 31 (18%) |
| Marijuana | 331 (63%) | 109 (63%) | 111 (63%) | 111 (63%) |
| Other (sedative, amphetamine) | 9 (2%) | 2 (1%) | 5 (3%) | 2 (1%) |
| Past year drug dependence (CIDI-SF) e | 245 (46%) | 80 (46%) | 83 (47%) | 82 (46%) |
| Drug-specific ASSIST score ≥ 27 f | 97 (18%) | 29 (17%) | 28 (16%) | 40 (23%) |
| Alcohol ASSIST score ≥ 27 f | 55 (10%) | 14 (8%) | 20 (11%) | 21 (12%) |
| Past addiction treatment g | 73 (14%) | 22 (13%) | 26 (15%) | 25 (14%) |
| Addiction treatment receipt within 6 months after study entry | 78 (15%) | 31 (18%) | 17 (10%) | 30 (17%) |
Two different brief interventions: MOTIV, a 30-45 minute adaptation of motivational interviewing conducted by Masters-degree level counselors and a 15 minute Brief Negotiated Interview (BNI) by a health educator. All characteristics assessed at baseline except where indicated.
Any night in the past 3 months in a shelter or on the street
Patient Health Questionnaire; higher scores indicate worse depressive symptoms (over past 2 weeks)
Main drug was drug of most concern in past month, as determined by participant
Composite International Diagnostic Interview Short Form
Alcohol, Smoking, and Substance Involvement Screening Test score for the main drug (or alcohol): Substance-specific scores range from 0-39; 27+ indicates likely severe use disorder
Past addiction treatment defined as any addiction treatment in the 6 months before study entry.
Intervention effects on addiction treatment receipt
There were no significant differences in addiction treatment receipt between the BNI and control groups in the partially or fully adjusted models: BNI vs control (AOR 1.11, 95% CI: 0.57, 2.15, Hochberg adjusted p-value 0.76). Participants in the MOTIV group had lower odds of receiving treatment in the partially and fully adjusted models compared to those in the control group (AOR 0.36, 95% CI: 0.17, 0.78, Hochberg adjusted p-value 0.02) (Table 2).
Table 2. Multivariable logistic regression analyses evaluating the effect of 2 types of brief interventions for drug use on addiction treatment receipt.
| MOTIV vs no BI AOR (95%CI) | p-valuel | BNI vs no BI AOR (95%CI) | p-valuej | |
|---|---|---|---|---|
| Overall analysis | ||||
| Unadjusted a | 0.52 (0.28, 0.98) | 0.09 | 1.06 (0.61, 1.85) | 0.83 |
| Partially adjusted b | 0.46 (0.23, 0.91) | 0.05 | 1.08 (0.58, 2.00) | 0.82 |
| Fully adjusted c | 0.36 (0.17, 0.78) | 0.02 | 1.11 (0.57, 2.15) | 0.76 |
|
| ||||
| Stratified by drug dependence d | ||||
| Yes e | 0.21 (0.08, 0.54) | 0.003 | 0.97 (0.43, 2.18) | 0.94 |
| No f | 1.41 (0.35, 5.76) | 0.63 | 1.64 (0.43, 6.30) | 0.63 |
| Stratified by past addiction treatment g | ||||
| Yes h | 0.15 (0.03, 0.64) | 0.01 | 0.93 (0.19, 4.40) | 1.00 |
| No i | 0.67 (0.26, 1.70) | 0.69 | 1.18 (0.52, 2.70) | 0.69 |
| Subgroup analyses | ||||
| Main drug with ASSIST 27+ j | ||||
| Unadjusted | 0.57 (0.22, 1.66) | 0.55 | 1.35 (0.51, 3.56) | 0.55 |
| Adjusted for past addiction treatment | 0.25 (0.07, 0.95) | 0.08 | 0.85 (0.26, 2.76) | 0.79 |
| Main drug other than marijuana with ASSIST 27+ k | ||||
| Unadjusted | 0.45 (0.14, 1.45) | 0.36 | 1.38 (0.44, 4.37) | 0.58 |
| Adjusted for past addiction treatment | 0.28 (0.07, 1.05) | 0.12 | 0.93 (0.26, 3.36) | 0.92 |
Logistic regression models examining the effect of enhanced brief intervention (MOTIV) vs no brief intervention (BI) and a structured brief negotiated interview (BNI) vs no BI on addiction treatment receipt.
Partially adjusted model includes main drug type and past year drug dependence (Composite International Diagnostic Interview short form)
Fully adjusted model includes main drug type, past year drug dependence and any past addiction treatment in the 6 months before study entry
Stratified by past-year drug dependence, adjusted for main drug type and past addiction treatment
N=245. Significant covariates: past addiction treatment AOR (95%CI) 8.88 (4.01, 19.67)
N=283 Significant covariates: past addiction treatment AOR (95%CI) 9.05 (1.71, 48.02)
Past addiction treatment defined as any addiction treatment in the 6 months before study entry. Models adjusted for main drug type and past year drug dependence
N=73. No significant covariates
N=455, Significant covariates: main drug type, marijuana vs benzo/amphet AOR (95%CI) 0.14 (0.02, 0.77)
Analytic sample restricted to participants with a main drug ASSIST score of 27 or greater (n=97)
Analytic sample restricted to participants with a main drug that was not marijuana with an ASSIST score of 27 or greater (n=70)
Adjusted for multiple comparisons using the Hochberg procedure
We identified potential interactions between intervention group and (CIDI-SF-assessed) drug dependence and prior addiction treatment. Stratified analyses suggested that among those with past-year drug dependence, the MOTIV intervention results in significantly lower odds of receiving treatment (AOR 0.21, 95% CI: 0.08, 0.54). In contrast, for those without past-year drug dependence, the MOTIV intervention appeared to increase the odds of receiving treatment (AOR 1.41, 95%CI: 0.35, 5.76), although the results were not statistically significant.
For those with prior treatment experience, the MOTIV intervention resulted in significantly lower odds of receiving addiction treatment (AOR 0.15, 95%CI: 0.03, 0.64), whereas among those without prior treatment experience, the magnitude of effect appeared smaller and not statistically significant (AOR 0.67, 95%CI: 0.26, 1.70).
Subgroup analyses of participants with ASSIST scores of ≥27 (n=97), and those with ASSIST scores of ≥27 for drugs other than marijuana (n=70), did not support the hypothesis that BI was associated with increased treatment utilization.
Patient-level factors associated with receipt of addiction treatment
In adjusted analyses of need characteristics (Table 3), the odds of addiction treatment receipt were lower for those with marijuana as the main drug and higher for those with an alcohol-specific ASSIST score ≥27. Greater severity of total substance involvement (Global ASSIST score), but not the drug-specific ASSIST score for the main drug was significantly associated with higher odds of treatment receipt although the effect size was small (AOR 1.14 95%CI 1.06, 1.22 per 5-unit increase in Global ASSIST score).
Table 3. Need factors associated with addiction treatment a.
| Received addiction treatment | ||||
|---|---|---|---|---|
| Predictor | Yes (n=78) | No (n=450) | AOR (95% CI) | p-value |
| Main drugb | ||||
| Cocaine | 21 (27%) | 77 (17%) | 3.90 (1.69, 8.99) | 0.001 |
| Opioids | 37 (47%) | 53 (12%) | 6.84 (3.13, 14.95) | <.001 |
| Other (sedative, amphetamine) c | 3 (4%) | 6 (1%) | 11.73 (2.14, 64.24) | 0.005 |
| Marijuana | 17 (22%) | 314 (70%) | 1 | -- |
| SIP-Drug (tertiles)d | ||||
| Highest | 48 (62%) | 128 (29%) | 1.20 (0.43, 3.32) | 0.73 |
| Middle | 21 (27%) | 164 (37%) | 1.77 (0.71, 4.38) | 0.22 |
| Lowest | 8 (10%) | 156 (35%) | 1 | -- |
| Alcohol ASSIST score | ||||
| Greater than or equal to 27 e | 19 (24%) | 36 (8%) | 2.71 (1.21, 6.05) | 0.02 |
| Less than 27 | 59 (76%) | 414 (92%) | 1 | -- |
| Drug specific ASSIST score f | ||||
| Greater than or equal to 27 | 35 (45%) | 62 (14%) | 1.33 (0.66, 2.67) | 0.42 |
| Less than 27 | 43 (55%) | 388 (86%) | 1 | -- |
| Global ASSIST for Drugs, median (IQR) g | 33 (19, 52) | 15 (10, 24) | 1.14 (1.06, 1.22) | <0.001 |
Addiction treatment receipt between baseline and 6 months. Separate logistic regression models for each independent variable of interest The following variables were included as covariates in models where the variable was not the main predictor: age, sex, race/ethnicity, primary language, depressive symptoms, baseline past addiction treatment receipt, and main drug type.
Main drug used in the past month
Results for cocaine, opioid, and marijuana were similar in models that did not include the small number of participants with sedative/ampheamine as the main drug
Drug use consequences
Alcohol, Smoking, and Substance Involvement Screening Test Substance-specific scores range from 0-39; 27+ indicates likely substance use disorder
Measure of severity of total drug use involvement. Global scores range from 0-273. Higher scores indicate riskier use/greater severity.
Severity of total drug involvement
Older age was significantly associated with lower odds of addiction treatment (AOR 0.83 per 5-year increase, 95%CI: 0.72, 0.96) (Table 4). We did not detect differences for other predisposing characteristics including sex, race/ethnicity, employment, housing status, criminal justice involvement, depression, overall health status, or chronic pain.
Table 4. Predisposing factors associated with receipt of addiction treatmenta.
| Received addiction treatment | ||||
|---|---|---|---|---|
| Predictor | Yes (n=78) | No (n=450) | AOR 95%CI | p-value |
| Age (mean, STD) b | 38 (11) | 42 (13) | 0.83 (0.72, 0.96) | 0.01 |
| Sex | ||||
| Female | 23 (29%) | 136 (30%) | 0.77 (0.40, 1.50) | 0.44 |
| Male | 55 (71%) | 314 (70%) | 1 | |
| Race/ethnicity | ||||
| Non-white | 46 (59%) | 377 (84%) | 0.80 (0.40, 1.61) | 0.54 |
| White | 32 (41%) | 73 (16%) | 1 | |
| Living with children | ||||
| Yes | 6 (8%) | 109 (24%) | 0.41 (0.15, 1.13) | 0.09 |
| No | 72 (92%) | 341 (76%) | 1 | -- |
| Currently employed | 13 (17%) | 132 (29%) | 0.60 (0.28, 1.27) | 0.18 |
| Homeless, past 3 months | 27 (35%) | 61 (14%) | 1.87 (0.91, 3.80) | 0.09 |
| Criminal justice involvement: | ||||
| Incarceration | 8 (10%) | 15 (3%) | 2.16 (0.72, 6.52) | 0.17 |
| Probation/Parole | 12 (15%) | 24 (5%) | 1.50 (0.56, 4.05) | 0.42 |
| None | 58 (75%) | 411 (91%) | 1 | |
| Depressive symptomsc | ||||
| PHQ9 score > =10 | 40 (51%) | 149 (33%) | 1.35 (0.74, 2.46) | 0.33 |
| PHQ9 score < 10 | 38 (49%) | 301 (67%) | 1 | -- |
| Overall health status (EUROQOL) | ||||
| Lowest tertile | 27 (35%) | 146 (32%) | 0.59 (0.26, 1.35) | 0.21 |
| Middle tertile | 31 (40%) | 171 (38%) | 0.90 (0.42, 1.91) | 0.78 |
| Highest tertile | 20 (26%) | 133 (30%) | 1 | |
| Severe pain d | ||||
| Yes | 38 (49%) | 220 (49%)
|
0.99 (0.54, 1.82) | 0.97 |
| No | 40 (51%) | 230 (51%) | 1 | -- |
Addiction treatment receipt between baseline and 6 months. Separate logistic regression models for each independent variable of interest The following variables were included as covariates in models where the variable was not the main predictor: age, sex, race/ethnicity, primary language, depressive symptoms, baseline past addiction treatment receipt, and main drug type.
Per 5-year increase
Patient Health Questionnaire, questions reference past 2 weeks
Pain (average in past 3 months, 7 -10 vs less than 7)
Past addiction treatment was the only enabling characteristic significantly associated with treatment receipt after baseline (AOR 5.77, 95%CI: 2.94, 11.34) (Table 5). Models that included health insurance status and intervention arm as well as a full regression model with statistically significant factors were not materially different (in magnitude, direction, or statistical significance).
Table 5. Enabling factors associated with receipt of addiction treatmenta.
| Predictor | Received addiction treatment N=78 | No addiction treatment receipt N=450 | AOR 95%CI | p-value |
|---|---|---|---|---|
| Primary language | ||||
| Other Language | 3 (4%) | 37 (8%) | 0.33 (0.08, 1.30) | 0.11 |
| English | 75 (96%) | 413 (92%) | 1 | -- |
| Health insurance | ||||
| Medicaid/Medicare | 69 (89%) | 360 (80%) | 0.96 (0.37, 2.51) | 0.93 |
| None vs Private | 2 (3%) | 28 (6%) | 0.33 (0.05, 2.15) | 0.25 |
| Private/Commercial | 7 (9%) | 62 (14%) | 1 | |
| Number of peers problem substance users | ||||
| Half or more | 41 (53%) | 144 (32%) | 1.31 (0.71, 2.42) | 0.39 |
| None to a few | 37 (47%) | 306 (68%) | 1 | -- |
| Past addiction treatment | ||||
| Yes | 41 (53%) | 32 (7%) | 5.77 (2.94, 11.34) | <.001 |
| No | 37 (47%) | 418 (93%) | 1 | -- |
| ED visit, past 3 months | ||||
| Yes | 33 (42%) | 156 (35%) | 1.05 (0.57, 1.92) | 0.88 |
| No | 45 (58%) | 294 (65%) | 1 | -- |
Addiction treatment receipt between baseline and 6 months. Separate logistic regression models for each independent variable of interest The following variables were included as covariates in models where the variable was not the main predictor: age, sex, race/ethnicity, primary language, depressive symptoms, baseline past addiction treatment receipt, and main drug type.
Discussion
In this cohort of primary care patients with screen-identified drug use, neither BI method resulted in greater addiction treatment receipt compared to no intervention. MOTIV, the enhanced BI, resulted in lower odds of addiction treatment receipt. Stratified analyses suggested that the negative effect of MOTIV was among those with past-year drug dependence and those with previous addiction treatment utilization.
This study is one of the very few randomized trials of BI for screen-identified drug use in primary care on treatment receipt. Both BI protocols had fairly robust referral to treatment services that included active efforts by interventionists to link participants to treatment and dedicated staff for treatment referrals, unlike other studies of BI on treatment receipt.9 We examined referral to treatment in an environment in which health insurance was largely not a factor in obtaining some exposure to specialty treatment because treatment entry is funded, when necessary, through contracts with the Massachusetts Bureau of Substance Abuse Services. The lack of a positive effect of BI found in this study is aligned with results of two RCTs of drug use screening and BI that found no effect on specialty treatment utilization.2,8 An observational study in primary care found that SBIRT was associated with more referrals to addiction treatment, but not actual treatment receipt.22 These studies call into question reliance on BI as the sole method of ensuring addiction treatment receipt by patients in primary care settings.
Our study's results may reflect the fact that accessing treatment is a challenging process in primary care.12 Typically the patient is presenting for another reason, has drug use identified by screening (i.e., is not raising drug use as a concern), may not be motivated to change, and has no expectation that the clinician will address drug use. Referrals are time-consuming (e.g., identifying treatment resources, calling treatment programs, obtaining insurance approval, coordinating concrete needs such as transportation) and treatment capacity is often inadequate.23, 24 It is possible that BI helped patients to realize why they might seek treatment, but not how to access treatment or help them feel empowered to seek treatment.
The study's null findings are not likely to have been affected by provision of information about treatment services to control participants, given that the simple provision of phone numbers is not thought to have large effects on treatment receipt and overall rates were very low in all groups. Lack of effects may be due to the study design, in that there were no exclusions for heavy drinking, multiple drug use, or other comorbidities. However, these high-risk characteristics are the norm in other primary care samples, and thus would be the proper group to study rather than exclude. Although we cannot rule out the potential effect of assessment reactivity on control participants, these “high-risk” characteristics make it unlikely that assessment effects would have had a strong influence on treatment receipt for the control group. Furthermore, since treatment receipt did not increase in any group compared to the period before study entry, it is difficult to postulate any significant assessment reactivity.
Although the differential effect of BI on those with and without drug dependence is intriguing, and the decreased likelihood of treatment receipt for the MOTIV group unexpected, the finding should be interpreted with caution. It is possible that treatment receipt was related to the semi-structured format of the MOTIV intervention. As motivational interviewing is a collaborative approach to change, the intervention structure was guided by patient responses and may not have emphasized specialty treatment in each BI. Another possible factor is that in the BNI arm, the health educator, located in primary care, conducted screening, BNI, made referrals to treatment and was available to see participants on subsequent primary care visits. The MOTIV interventionist was not part of the primary care practice and referred participants interested in addiction treatment to the BT counselor. Having one person perform these tasks may be beneficial for treatment access.7,8
An exploratory objective beyond the RCT was to identify patient-level factors associated with addiction treatment receipt for primary care patients with drug use identified by screening. Older age was a negative factor, as in other studies.25 The odds of treatment receipt were higher for those with likely alcohol dependence, and lower for those with marijuana as the main drug. Patients and clinicians may perceive marijuana use disorder to be less severe than other SUDs, even when objective signs of severity (e.g., disorder criteria, ASSIST scores) are similar.
The study results should be interpreted with the following limitations. Although referral was part of the BI, it was not mandated for all, and although a motivational intervention was used for referral, there was not a singular focus on an intensive referral. Addiction treatment utilization was not available from Veterans Affairs institutions, private facilities that never receive state funds, or programs out of state, although these were unlikely sources of care for our population. Additionally, we did not have data from psychiatric settings that do not contract with the state of Massachusetts as addiction treatment facilities or data from 12-step mutual help groups; our aim was addiction treatment receipt.
We did not evaluate treatment adequacy or clinical outcomes of treatment, though only a minority received any. Moreover, this study's findings may not extend to study samples with a higher prevalence of severe SUDs or more use of drugs such as cocaine or opioids (although these were the main drugs for almost 40% of the study sample). It is notable that more people received treatment in the U.S. for marijuana than for any other substance.6
Drug type was defined as self-reported drug of most concern at enrollment. Patients may not have been concerned about a drug with clinically more serious implications that would enhance likelihood of a treatment admission. However, 93% of participants reported their drug of main concern to be the one for which they scored highest on the ASSIST. Moreover, main drug type was entered as a covariate, not as an independent predictor.
Finally, generalizability of results from an urban academic center with a diverse but largely low-income population may be limited. On the other hand, results are likely applicable to similar centers, of which there are many in the US. We did not find, however, that homelessness, overall health status or depression symptoms, factors that could vary by setting, were associated with treatment receipt. Findings could also differ based on access and quality of treatments that were available locally.
In addition to being one of the few RCT's of BI for drug use in primary care on addiction treatment entry, there are several other study strengths. Two types of BI were examined: one offered in routine clinical care and an “enhanced” BI intended to address efficacy considerations. Patients with more severe drug use who might benefit from specialty treatment were not excluded and random assignment was balanced by drug dependence and main drug. Treatment receipt was assessed objectively.
This study's findings add to the growing number of studies that suggest that the “RT” component of SBIRT does not increase receipt of addiction treatment. The RCT results presented here apply to both the most commonly implemented modes of BI in US federally-supported SBIRT programs, as well as to more intensive and more difficult to implement motivational interviewing based BI. The cohort analysis presented here suggest that particular attention may be needed for older patients, those with marijuana use as their main drug, and those without concomitant alcohol use disorders. The findings suggest that one should not rely on RT in the context of SBIRT as a sole method of addressing SUDs and that research is needed on better ways to increase treatment receipt with referral and to increase treatment receipt regardless of referral (e.g. in medical settings). Both of these will likely require greater effort and new methods that are not part of the current SBIRT dissemination.
Acknowledgments
Sources of Funding: The ASPIRE study was funded by an award from the National Institute on Drug Abuse (R01 DA025068), a portion of which was funded by the Center for Substance Abuse Treatment, Substance Abuse and Mental Health Services Administration (SAMHSA). BNI counselors, training, and supervision were supported by a contract from the SAMHSA (#TI018311) to the Massachusetts Department of Public Health (MADPH) Bureau of Substance Abuse Services which contracted with Boston Medical Center to deliver BNI services (contract #INTF2330M04W76112076). The study was also supported by the National Center for Research Resources, award number UL1RR025771. The funders had no role in study design, in collection or analysis of data, writing the report or in the decision to to submit the article.
Footnotes
Trial Registration: http://clinicaltrials.gov, Registration number NCT00876941, URL http://clinicaltrials.gov
Conflict of interest: Dr. Saitz is a paid author and editor for Springer, UpToDate, and the BMJ. Wolters Kluwer and BMJ have supported conference travel to an editors' meeting and Springer (BioMed Central) and UpToDate supported attendance at a local editorial meeting. He has also been paid to serve as an expert witness in malpractice cases related to the management of alcohol and other drug disorders. Dr. Cheng serves on data monitoring committees for Janssen Research & Development.
Refer to web version in PubMed Central for supplemental material
References
- 1.Gelberg L, Andersen RM, Afifi AA, Leake BD, Arangua L, Vahidi M, et al. Baumeister SE. Project QUIT (Quit Using Drugs Intervention Trial): A randomized controlled trial of primary care-based multi-component brief intervention to reduce risky drug use. Addiction. 2015;110:1777–90. doi: 10.1111/add.12993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Roy-Byrne P, Bumgardner K, Krupski A, Dunn C, Ries R, Donovan D, et al. Zarkin GA. Brief interventions for problem drug use in safety-net primary care settings. JAMA. 2014;312:492–501. doi: 10.1001/jama.2014.7860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Saitz R, Palfai TPA, Cheng DM, Alford DP, Bernstein JA, Lloyd-Travaglini C, et al. Samet JH. Screening and brief intervention for drug use in primary care: The ASPIRE randomized clinical trial. JAMA. 2014;312:502–513. doi: 10.1001/jama.2014.7862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Babor TF, McRee BG, Kassebaum PA, Grimaldi PL, Ahmed K, Bray J. Screening, brief intervention, and referral to treatment (SBIRT): toward a public health approach to the management of substance abuse. Subst Abus. 2007;28:7–30. doi: 10.1300/J465v28n03_03. [DOI] [PubMed] [Google Scholar]
- 5.Department of Health and Human Services. Centers of Medicare and Medicaid Services, Medicare Learning Network Series: Screening and Brief Intervention and Referral to Treatment. [Accessed October 14, 2016]; Available at https://www.cms.gov/Outreach-and-Education/Medicare-Learning-Network-MLN/MLNProducts/downloads/SBIRT_Factsheet_ICN904084.pdf.
- 6.Substance Abuse and Mental Health Services Administration. Receipt of services for behavioral health problems: results from the 2014 National Survey on Drug Use and Health: NSDUH Data Review. [Accessed September 7, 2016];2015 Available at http://www.samhsa.gov/data/ Archived at http://www.webcitation.org/6kL1yC61a.
- 7.Krupski A, Sears JM, Joesch JM, Estee S, He L, Dunn C, et al. Ries R. Impact of brief interventions and brief treatment on admissions to chemical dependency treatment. Drug and Alcohol Depend. 2010;110:126–136. doi: 10.1016/j.drugalcdep.2010.02.018. [DOI] [PubMed] [Google Scholar]
- 8.Bernstein J, Bernstein E, Tassiopoulos K, Heeren T, Levenson S, Hingson R. Brief motivational intervention at a clinic visit reduces cocaine and heroin use. Drug Alcohol Depend. 2005;77:49–59. doi: 10.1016/j.drugalcdep.2004.07.006. [DOI] [PubMed] [Google Scholar]
- 9.Glass JE, Powell BJ, Perron BE, Brown RT, Ilgen MA. Specialty substance use disorder services following brief alcohol intervention: A meta-analysis of randomized controlled trials. Addiction. 2015;110:1404–15. doi: 10.1111/add.12950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hingson R, Compton WM. Screening and brief intervention and referral to treatment for drug use in primary care: back to the drawing board. JAMA. 2014;312:488–9. doi: 10.1001/jama.2014.7863. [DOI] [PubMed] [Google Scholar]
- 11.Institute of Medicine. Improving the quality of health care for mental and substance-use conditions: quality chasm series. Washington DC: The National Academies Press; 2006. [PubMed] [Google Scholar]
- 12.Gurewich D, Prottas J, Sirkin JT. Managing care for patients with substance abuse disorders at community health centers. J Subst Abuse Treat. 2014;46:227–31. doi: 10.1016/j.jsat.2013.06.013. [DOI] [PubMed] [Google Scholar]
- 13.Substance Abuse and Mental Health Services Administration. SAMHSA-HRSA Center for Integrated Health Solutions. Integrating Behavioral Health into Primary Care. [Accessed September 7, 2016]; Available at http://www.integration.samhsa.gov/integrated-care-models/behavioral-health-in-primary-care. archived URL http://www.webcitation.org/6kL03MRvg.
- 14.Humeniuk R, Ali R, Babor TF, Farrell M, Formigoni ML, Jittiwutikarn J, et al. Simon S. Validation of the Alcohol, Smoking Substance Involvement Screening Test (ASSIST) Addiction. 2008;103:1039–47. doi: 10.1111/j.1360-0443.2007.02114.x. [DOI] [PubMed] [Google Scholar]
- 15.Kessler RC, Andrews G, Mroczek D, Ustun TB, Wittchen HU. The World Health Organization Composite International Diagnostic Interview Short Form (CIDI-SF) Int J Methods Psychiatr Res. 1998;7:171–185. [Google Scholar]
- 16.Bernstein J, Derrington TM, Belanoff C, Cabral HJ, Babakhanlou-Chase H, Diop H, et al. Kotelchuck M. Treatment outcomes for substance use disorder among women of reproductive age in Massachusetts: a population-based approach. Drug Alcohol Depend. 2015;147:151–9. doi: 10.1016/j.drugalcdep.2014.11.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gelberg L, Andersen RM, Leake BD. The Behavioral Model for Vulnerable populations: application to medical care use and outcomes for homeless people. Health Serv Res. 2000;34:1273–1302. [PMC free article] [PubMed] [Google Scholar]
- 18.Alterman AI, Cacciola JS, Ivey MA, Habing B, Lynch KG. Reliability and validity of the alcohol short index of problems and a newly constructed drug short index of problems. J Stud Alcohol Drugs. 2009;70:304–307. doi: 10.15288/jsad.2009.70.304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kroenke K, Spitzer RL, Williams JBW. The PHQ-9 validity of a brief depression severity measure. J Gen Intern Med. 2001;16:606–13. doi: 10.1046/j.1525-1497.2001.016009606.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.EuroQol Group. EurQol: a new facility for the measurement of health-related quality of life. Health Policy. 1990;16:199–208. doi: 10.1016/0168-8510(90)90421-9. [DOI] [PubMed] [Google Scholar]
- 21.Hochberg Y. A sharper Bonferroni procedure for multiple tests of significance. Biometrika. 1988;75:800–802. [Google Scholar]
- 22.Dwinnells R. SBIRT as a vital sign for behavioral health identification, diagnosis, and referral in community health care. Ann of Fam Med. 2015;13:261–63. doi: 10.1370/afm.1776. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Quanbeck A, Wheelock A, Ford JH, Pulvermacher A, Cappoccia V, Gustafson D. Examining access to addiction treatment: scheduling processes and barriers. J Subst Abuse Treat. 2013;44:343–348. doi: 10.1016/j.jsat.2012.08.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Urada D, Teruya C, Gelberg L, Rawson R. Integration of substance use disorder services with primary care: health center surveys and qualitative interviews. Subst Abuse Treat, Prev, and Policy. 2014;9:15. doi: 10.1186/1747-597x-9-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gurewich D, Sirkin JT, Shepard DS. On-site provision of substance abuse treatment services at community health centers. J Subst Abuse Treat. 2012;42:339–345. doi: 10.1016/j.jsat.2011.09.012. [DOI] [PubMed] [Google Scholar]