Termination From Substance Use Disorder Treatment in the United States: Residential and Outpatient Settings

Abstract

Objective:

As rates of noncompletion in substance use treatment remain high and consequences of noncompletion can be severe, research regarding individual and environmental factors associated with specific types of discharge is crucial. The current study used data from the Treatment Episodes Dataset–Discharge (TEDS-D) 2015–2017 in the United States to investigate the impact of social determinants of health on discharge from treatment due to termination by the treatment facility in both outpatient/intensive outpatient and residential treatment settings.

Method:

A multinomial logistic regression was conducted examining the likelihood of discharge due to termination compared with discharge because of dropout or incarceration.

Results:

Results revealed differences in termination based on treatment setting, race, income, criminal justice referral, and mental health diagnoses, among others. Broadly across settings, people of color were significantly more likely to be terminated from treatment than to drop out compared with their White counterparts. Further, with little exception, individuals with less financial security (i.e., those who were unemployed, those with low/no income, those with no insurance) were less likely to drop out and more likely to be discharged due to termination across treatment settings.

Conclusions:

The results of the current study further solidify the need for nuanced examination of the reason individuals do not complete substance use treatment and extend the impact of social determinants of health to involuntary termination from substance use treatment.

Rates of noncompletion in substance use treatment in the United States can range anywhere from 17% to 67% and vary across treatment settings (Brorson et al., 2013). The consequences of noncompletion and/or ongoing substance use can include both individual- (e.g., poorer health, increased risk of relapse) and community-level (e.g., increase in crime, spread of HIV) effects (Brorson et al., 2013; Simpson, 1979, 1981; Stark, 1992; United Nations Office on Drugs and Crime, 2009). Ultimately, because substance use treatment noncompletion rates reach higher than 50% in many programs, many clients are also not receiving adequate treatment to recover (Brorson, 2013; Stark, 1992).

Factors associated with retention and attrition in substance use treatment have been widely investigated (for a review, see Brorson et al., 2013). Brorson and colleagues (2013) found a variety of individual and environmental factors associated with treatment dropout, including cognitive deficits, the presence of a personality disorder, low treatment alliance, and younger age. Recent research has also revealed the role of sex, race, and criminal justice referral on substance use treatment completion (Gallagher et al., 2015; Mennis & Stahler, 2016; Petry & Bickel, 2000). Specifically, a study by Mennis and Stahler (2016) found that African American and Hispanic individuals were less likely to complete their treatment episode compared with their White counterparts, and that the level of this disparity varied based on one's substance of choice. Overall, retention and attrition research has largely focused on noncompletion attributable to dropout or the examination of noncompletion as a whole (i.e., combining all reasons for noncompletion in analyses compared with treatment completion), excluding specific reasons for discharge such as termination by the treatment facility. Further, much of the research has focused on a single treatment setting (either outpatient/intensive outpatient [IOP] or residential). Because there are many circumstances that may lead to treatment noncompletion, there remains a significant gap in understanding the specific reasons for treatment noncompletion beyond dropout (Marotta et al., 2020; Nunes et al., 2010; Wormith & Olver, 2002).

Given the real-life distinctions between reasons for discharge that are attributable to the client's own decision (e.g., dropout) and those that are decided by an external entity (e.g., termination by facility, incarceration), research examining factors associated with specific types of discharge from substance use treatment is necessary. However, prior research overwhelmingly merges reasons for noncompletion when examining predictors or exclusively focuses on noncompletion because of dropout (Brorson et al., 2013; Mennis & Stahler, 2016; Petry & Bickel, 2000; Syan et al., 2020), excluding specific reasons for discharge such as client termination by the treatment facility, and incarceration. Examination of noncompletion has also largely focused on a single treatment setting with fewer studies examining predictors across different levels of care (Brorson et al., 2013).

No known research has examined the factors associated with termination from substance use treatment by the treatment facility. To fill this gap, the current study uses a social determinants of health (SDOHs) framework to understand their influence on the likelihood of discharge from substance use treatment as a result of termination by the treatment facility. SDOHs are defined as “conditions in the environments where people are born, live, learn, work, play, worship, and age that affect a wide range of health, functioning, and quality-of-life outcomes and risks” (Healthy People 2030, 2023). As outlined by the U.S. Department of Health and Human Services, SDOH can include a wide range of individual, community, and societal conditions, including racism, discrimination, income, education, and safe neighborhoods. Recently, scholars have expanded this framework to begin to understand how SDOH interact with marginalization and contribute to the known health inequity present in health and treatment services (Baah et al., 2019).

The current study uses this framework to examine how SDOH may be associated with the decision by the treatment facility to discharge someone from substance use treatment. Specifically, we use sociodemographic, criminal justice, and mental health predictors to compare the likelihood of termination by facility to (a) dropping out of treatment and (b) discharge due to incarceration in outpatient/IOP and residential treatment settings.

Method

Data

The current study included a secondary analysis from the Treatment Episodes Dataset–Discharge (TEDS-D) 2015–2017. The TEDS-D includes data collected by the Substance Use and Mental Health Services Administration (SAMHSA) from licensed substance use treatment facilities across the United States. TEDS-D data are collected and recorded as part of standard service delivery at every substance use treatment facility and include information regarding individuals discharged from substance use treatment during each calendar year regardless of the year they were admitted to treatment.

Measures

Dependent variable—reason for discharge. The reason-for-discharge variable was used as the dependent variable in the current study to identify the reason given for each individual's discharge from treatment. Reason-for-discharge options included successful treatment completion, transferred to another treatment facility, dropped out/withdrew from treatment, terminated by facility, incarcerated, death, or other. The TEDS-D codebook defines terminated by facility as “treatment terminated by action of facility, generally because of client non-compliance or violation of rules, laws, or procedures” (SAMHSA, 2019). Discharge due to dropped out/withdrew (i.e., dropout) and incarceration were used as comparative groups. Individuals with other discharge options were excluded from the current analyses (e.g., death, transferred to another treatment facility, successful treatment completion).

Sociodemographic factors. Sociodemographic factors include information regarding individuals' education, age at admission, sex, race, and ethnicity. Also included in the analyses are the number of days an individual waited to enter treatment, whether they reported being homeless at admission, employment status, primary source of income, type of health insurance, previous episodes of substance use treatment, and frequency of attendance at self-help groups in the month before discharge. Race was categorized as Black, American Indian/Alaska Native (AI/AN), Asian Pacific Islander (Asian/PI), multiracial, other single race, or White. Ethnicity included options for Hispanic or non-Hispanic. Employment status was categorized as full time (35 or more hours per week), part time (fewer than 35 hours per week), unemployed (looking for work in the past 30 days), and not in the labor force (e.g., not looking for work in the past 30 days, student, homemaker, etc.). Primary source of income options included wages/salary, public assistance, retirement, disability or pension, and none. Insurance type was categorized as either private, Medicare, Medicaid/other, and none. Frequency of attendance at self-help groups was categorized as either none, low (1–7 days in the past month), or high (8–30 days in the past month).

Criminal justice referral. The current analysis dichotomized treatment referral source to examine criminal justice referrals compared with all other types of referrals. Criminal justice referrals include individuals referred to substance use treatment by the court system, probation/parole, prison, a formal adjudication process, other recognized legal entity, diversionary programs, or as a result of a driving-while-intoxicated charge. Non–criminal justice referrals include those made by the individual (i.e., self-referral), substance use provider, school, employer, community agency, or other health care provider.

Mental health and primary substance use. Information regarding whether an individual entered treatment with any present mental health diagnosis was included as a predictor in the current analyses. The TEDS-D primary substance of use at admission variable was used to create dichotomized variables indicating whether the primary substance of use at intake was alcohol, cocaine/crack, marijuana, heroin, or methamphetamine. Finally, whether an individual entered treatment with more than one substance of use was dichotomized to indicate the presence of multiple substances of use at admission (polysubstance use).

Participants

The current study used cases from the TEDS-D who were identified as discharging from either outpatient/IOP (i.e., non-intensive outpatient/intensive outpatient) or residential (i.e., rehab/residential short term, rehab/residential long term) treatment. In addition, cases were excluded if they were less than 18 years of age or if they didn't indicate any substance of use at admission to services. In total, data from 232,416 individual cases from both residential and outpatient/IOP settings were included in the current analyses.

Statistical analyses

To examine factors associated with termination from treatment by facility compared with other discharge types, the current exploratory analysis included two multinomial logistic regressions. The first used individuals from outpatient/IOP treatment services, and the second used individuals from residential treatment services. When running the analyses, termination by facility was used as the reference category for all comparisons. Next, to capture which factors were associated with increased likelihood of termination compared with other types of discharge, all odds ratios and their confidence intervals were inversed using the 1/OR (odds ratio) formula to inverse the reference category in logistic regression analyses (OR[inv]; Bland and Altman, 2000). Last, given the large sample size, a Bonferroni p-value correction was calculated, and significance of the results was determined at the p = .0001 level or lower (Feise, 2002; Jafari & AnsariPour, 2019).

Results

Descriptive

Information regarding 232,416 individuals was included in the current analyses (171,617 individuals in outpatient or IOP treatment; 60,799 individuals in residential substance use treatment services). Full results regarding outpatient/IOP and residential treatment settings and demographics are outlined in Tables 1 and 2, respectively.

Table 1.

Multinomial logistic regression predicting reason for discharge among outpatient substance use treatment (N = 171,617)

Variable	n	%	Termination vs. incarcerated			Termination vs. dropout
			OR (inv)	95% CI lower	95% CI upper	OR (inv)	95% CI lower	95% CI upper
Education			1.02	1.05	1.00	0.99	1.00	0.97
8 years or less	8,690	5.1
9-15 years/GED	153,837	89.6
16 or more	9,090	5.3
Age at admission			1.01	1.02	1.00	0.99	1.00	0.99
18-20 years	10,404	6.1
21-29 years	56,382	32.9
30-39 years	53,069	30.9
40-49 years	29,457	17.1
50-64 years	21,156	12.3
≥65 years	1,149	0.7
Sex
Female	62,348	37.6	0.65**	0.68	0.62	0.99	1.02	0.97
Male	103,538	62.4	ref.			ref.
Race
Black	3,155	1.9	1.13	1.29	0.98	1.30**	1.41	1.19
AI/AN	4,714	2.8	1.52**	1.74	1.33	1.27**	1.38	1.18
Asian/PI	960	0.6	2.00**	2.61	1.54	1.63**	1.88	1.41
Multiracial	5,404	3.3	1.25**	1.38	1.13	1.58**	1.68	1.48
Other (single race)	31,356	18.9	1.25**	1.32	1.18	1.13**	1.17	1.09
White	120,297	72.5
Ethnicity
Hispanic	16,885	10.2	0.94	1.01	0.88	0.99	1.03	0.94
Non–Hispanic	149,001	89.8	ref.			ref.
Days waiting			1.00	1.02	0.98	0.90**	0.92	0.89
0	99,050	57.7
1-30	66,676	38.9
31 or more	5,891	3.4
Employment status
Part time	18,152	10.9	0.96	1.04	0.88	0.90*	0.95	0.85
Not in labor force	62,825	37.9	1.07	1.18	0.98	0.97	1.02	0.91
Unemployed	45,769	27.6	0.95	1.01	0.90	0.95	0.98	0.92
Full time	39,140	23.6	ref.			ref.
Homeless
Yes	16,972	10.2	0.82**	0.88	0.77	0.78**	0.82	0.75
No	148,914	89.8	ref.			ref.
Income source
Public assistance	20,453	12.3	1.05	1.13	0.97	1.01	1.06	0.97
Retirement, disability, or pension	16,094	9.7	1.07	1.17	0.98	1.12*	1.18	1.06
None	14,704	8.9	1.42**	1.59	1.28	1.07	1.14	1.00
Other	59,028	35.6	1.28**	1.42	1.15	1.18**	1.25	1.11
Wages/salary	55,607	33.5	ref.			ref.
Insurance type
Medicare	69,539	41.9	1.14	1.23	1.05	0.80**	0.84	0.76
Medicaid, other	12,899	7.8	1.13**	1.19	1.08	0.90**	0.93	0.87
Uninsured	67,278	40.6	1.04	1.12	0.97	1.14**	1.19	1.09
Private	16,170	9.7	ref.			ref.
Self–help attendance frequency
None	131,229	79.1	1.59**	1.71	1.47	1.06	1.12	1.01
Low (1-7 days)	22,260	13.4	1.18*	1.29	1.09	1.10	1.16	1.03
High (8-30 days)	12,397	7.5	ref.			ref.
Previous SUD treatment
Yes	112,323	67.7	1.30**	1.37	1.25	0.99	1.01	0.96
No	53,563	32.3	ref.			ref.
Referral source
Criminal justice	65,613	39.6	0.32**	0.34	0.31	1.59**	1.63	1.55
Other referral source	100,273	60.4	ref.			ref.
Mental health diagnosis
Present	84,715	51.1	1.32**	1.38	1.27	1.13**	1.16	1.10
Absent	81,171	48.9	ref.			ref.
Primary substance
Alcohol	48,171	29.0	1.37**	1.48	1.27	0.87**	0.91	0.83
All other substances	117,715	71.0	ref.			ref.
Cocaine/crack	7,116	4.3	0.94	1.05	0.84	0.96	1.03	0.89
All other substances	158,770	95.7	ref.			ref.
Marijuana	29,761	17.9	1.30**	1.41	1.20	0.89**	0.93	0.85
All other substances	136,125	82.1	ref.			ref.
Heroin	39,878	24.0	0.85**	0.91	0.78	0.95	1.00	0.91
All other substances	126,008	76.0	ref.			ref.
Methamphetamine	24,137	14.6	0.79**	0.86	0.73	0.92	0.97	0.88
All other substs	141,749	85.4	ref.			ref.
Polysubstance use
Yes	99,904	60.2	0.97	1.01	0.93	1.07**	1.10	1.04
No	65,982	39.8	ref.			ref.

Notes: OR (inv) = odds ratio inversed; CI = confidence interval; GED = General Educational Development credential; ref. = reference; AI/AN = American Indian/Alaska Native; PI = Pacific Islander; days waiting = number of days waiting to enter treatment. Termination by Facility is the reference category in all presented analyses; SUD = substance use disorder.

^*p = .0001;

^**p < .0001.

Table 2.

TABLE 2. Multinomial logistic regression predicting reason for discharge among residential substance use treatment (N = 60,799)

Variable	n	%	Termination vs. incarcerated			Termination vs. dropout
			OR (inv)	95% CI lower	95% CI upper	OR (inv)	95% CI lower	95% CI upper
Education			0.99	1.05	0.93	1.06**	1.09	1.04
8 years or less	5,300	8.7
9-15 years/GED	52,958	87.1
16 or more	2,541	4.2
Age at admission			0.93**	0.96	0.90	0.96**	0.97	0.95
18-29 years	32,737	40.7
30-49 years	29,821	49.1
50-64 years	6,059	10
≥65 years	182	0.3
Sex
Female	18,743	39.35	0.90	1.02	0.79	1.10*	1.15	1.05
Male	28,884	60.65	ref.			ref.
Race
Black	1,045	2.19	0.70	0.97	0.51	1.19	1.38	1.03
AI/AN	1,413	2.97	1.00	1.43	0.70	1.22	1.40	1.06
Asian/PI	301	0.63	1.32	2.76	0.63	1.12	1.46	0.86
Multiracial	1,521	3.19	1.09	1.44	0.83	1.39**	1.57	1.23
Other (single race)	6,627	13.91	1.27	1.53	1.06	1.26**	1.34	1.18
White	36,720	77.10	ref.			ref.
Ethnicity
Hispanic	4,517	9.48	1.04	1.28	0.84	0.93	1.01	0.86
Non–Hispanic	43,110	90.52
Days waiting			1.05	1.10	1.00	1.00	1.02	0.99
0	30,383	50.0
1-30	26,799	44.0
31 or more	3,617	5.9
Employment status
Part time	1,045	2.19	0.96	1.30	0.71	1.06	1.21	0.94
Not in labor force	18,473	38.79	0.79	1.13	0.55	1.15	1.35	0.98
Unemployed	26,037	54.67	1.21	1.37	1.07	1.32**	1.38	1.26
Full time	2,072	4.35	ref.			ref.
Homeless
Yes	10,988	23.07	1.20	1.40	1.03	1.16**	1.22	1.10
No	36,639	76.93	ref.			ref.
Income source
Public assistance	4,033	8.47	1.08	1.32	0.88	0.86	0.93	0.79
Retirement, disability, or pension	3,768	7.91	0.98	1.29	0.75	0.93	1.03	0.84
None	3,196	6.71	1.64	2.26	1.19	1.26*	1.40	1.12
Other	5,591	11.74	1.08	1.46	0.79	1.01	1.13	0.90
Wages/salary	31,039	65.17	ref.			ref.
Insurance type
Medicare	24,231	50.88	1.12	1.40	0.89	0.92	1.00	0.84
Medicaid, other	4,554	9.56	1.29*	1.48	1.13	0.98	1.03	0.93
Uninsured	15,429	32.40	1.03	1.28	0.83	0.73**	0.79	0.67
Private			ref.			ref.
Self–help attendance frequency
None	16,834	35.35	0.78	0.90	0.68	0.69**	0.72	0.65
Low (1-7 days)	13,720	28.81	0.92	1.07	0.79	0.69**	0.73	0.65
High (8-30 days)	17,073	35.85	ref.			ref.
Previous SUD treatment
Yes	37,528	78.80	1.13	1.31	0.97	0.89**	0.94	0.84
No	10,099	21.20	ref.			ref.
Referral source
Criminal justice	12,009	25.21	0.24**	0.27	0.21	1.78**	1.87	1.69
Other referral source	35,618	74.79	ref.			ref.
Mental health diagnosis
Present	19,811	41.60	1.08	1.22	0.96	0.97	1.01	0.92
Absent	27,816	58.40	ref.			ref.
Primary substance
Alcohol	10,560	22.17	0.98	1.23	0.79	0.88	0.96	0.81
All other substances	37,067	77.83	ref.			ref.
Cocaine/crack	3,383	7.10	1.39	1.90	1.02	0.98	1.09	0.88
All other substances	44,244	92.90	ref.			ref.
Marijuana	2,750	5.77	0.85	1.10	0.65	1.13	1.25	1.01
All other substances	44,877	94.23	ref.			ref.
Heroin	17,766	37.30	1.10	1.35	0.90	1.03	1.11	0.96
All other substances	29,861	62.70	ref.			ref.
Methamphetamine	8,086	16.98	0.80	1.00	0.64	0.74**	0.81	0.68
All other substances	39,541	83.02	ref.			ref.
Polysubstance use
Yes	33,129	69.56	1.11	1.26	0.98	1.12**	1.17	1.06
No	14,498	30.44	ref.			ref.

Notes: OR (inv) = odds ratio inversed; CI = confidence interval; GED = General Educational Development credential; ref. = reference; AIAN = American Indian/Alaska Native; PI = Pacific Islander; days waiting = number of days waiting to enter treatment. Termination by Facility is the reference category in all presented analyses; SUD = substance use disorder.

^*p = .0001;

^**p < .0001.

When looking at the differences between outpatient/IOP and residential samples, both shared relatively similar demographic make-up, including similar distribution on the identification of sex, race, and ethnicity. The majority of both samples identified as male, White, and non-Hispanic. Larger differences were found when examining employment status, frequency of attendance at self-help groups, and a present mental health diagnosis. In particular, more individuals in outpatient/IOP settings were reported to have full-time employment, attend fewer self-help groups, and have a present mental health diagnosis.

Sociodemographic factors

Several patterns arose among sociodemographic factors when examining the likelihood of discharge from treatment because of termination by the treatment facility. In outpatient/IOP, almost all people of color (Black, AI/AN, Asian/PI, multiracial, other single race) were more likely to be terminated from treatment than drop out or be incarcerated when compared with their White counterparts. There was one exception to this finding in that there was no significant difference in the likelihood of Black and White individuals to be terminated from treatment compared with being incarcerated. These results were reflected to a lesser degree in residential treatment, where individuals identifying as multiracial or other single race were more likely to be terminated than drop out compared with White individuals. No other significant racial differences were found among those in residential treatment. Of note, there were no significant differences based on ethnicity (Hispanic, non-Hispanic) on any analyses in either treatment setting.

Regarding sex, results were mixed. In outpatient/IOP treatment, women were significantly less likely to be terminated compared with incarcerated men, and there were no sex differences in the likelihood of termination or dropout. In residential treatment, the likelihood of termination or incarceration was not significantly different based on sex. Also in residential treatment, it was found that women were more likely to be terminated than drop out when compared with men.

Homelessness was found to have a mixed effect on the likelihood of termination based on treatment setting. In outpatient/IOP treatment, individuals who were identified as homeless were less likely to be terminated from treatment and more likely to be either incarcerated or drop out of treatment compared with their housed counterparts. In residential treatment, individuals identified as homeless were more likely to be terminated from treatment than to drop out. There was no significant difference in the likelihood of being incarcerated or terminated based on homelessness in residential treatment.

In many instances, individuals with no income and/or no insurance were more likely to be terminated from treatment than to drop out or be incarcerated at discharge compared to those with wages/salary and/or private insurance. In outpatient/IOP treatment, individuals reported as having no income were more likely to be terminated than incarcerated and individuals reported as having an “other” source of income were more likely to be terminated than incarcerated or drop out compared with individuals noted as having income from wages/salary. This pattern was similar with a smaller effect size in residential treatment, where individuals with no income were more likely to be terminated than to drop out compared with those noted as having wages/salary. No other comparisons based on income were significant.

Individuals who were reported as uninsured had different likelihoods of different types of discharges based on treatment setting. Specifically, in outpatient/IOP treatment, individuals noted as uninsured were more likely to be terminated than to drop out compared to those with private insurance. This effect was flipped for those in residential treatment with no reported insurance, who were less likely to be terminated and more likely to drop out compared to their counterparts with private insurance.

Finally, self-help attendance was found to have the most significant effect on the likelihood of dropping out or being terminated in residential treatment. People who reported having none, or low levels of self-help group attendance in residential treatment were more likely to drop out than be terminated compared to those with high levels of self-help attendance.

Criminal justice referral

Across treatment settings, individuals referred to treatment by the criminal justice system were significantly more likely to be incarcerated than terminated from treatment and significantly more likely to be terminated from treatment than to drop out when compared with individuals referred to treatment from other sources.

Mental health and substance use

In outpatient/IOP treatment, individuals reported as having a present mental health diagnosis were more likely to be terminated than incarcerated or drop out of treatment. This comparison was nonsignificant among individuals in residential treatment.

Dropout from residential treatment was more likely than termination if an individual reported previous treatment attempts. However, incarceration was more likely than termination from outpatient/IOP if previous treatment attempts were reported.

Regarding primary substance of use, in outpatient/IOP treatment, individuals with primary alcohol or marijuana use were more likely to be terminated than incarcerated and more likely to drop out than be terminated compared with those with any other primary substance. Individuals with heroin or methamphetamine as their primary substance of use were more likely to be incarcerated than terminated from treatment compared to those with any other primary substance. Last, individuals in both residential and outpatient/IOP treatment who reported more than one substance of use at admission to treatment were more likely to be terminated than to drop out of treatment compared to those with only one reported substance of use.

Discussion

The current study aimed to explore factors associated with termination from substance use treatment in outpatient/IOP and residential settings using an SDOH lens. Results revealed unique sociodemographic, criminal justice, and mental health predictors in each setting associated with the likelihood of an individual to be terminated by their treatment facility. Taken together, the current results further solidify the influence of SDOH on both the decision to drop out of treatment and the decision to terminate an individual from treatment.

Sociodemographic factors

Among those in outpatient/IOP treatment settings, with few exceptions, people of color were more likely to be terminated by their treatment facility compared with their White counterparts, even when controlling for other demographic and environmental factors. Of note, all people of color in outpatient treatment were more likely to be terminated than to drop out. Although much of the current literature focuses on the need to intervene in the high dropout rate of people of color (Brorson et al., 2013; Mennis & Stahler, 2016), the current finding highlights the need for clinical and scholarly attention to racial differences in termination as determined by the treatment facilities. Although the cause of this disparity is beyond the scope of the current article, this result is consistent with previous literature revealing racial differences in punishment in educational and legal settings (Forsyth et al., 2015; Owens & McLanahan, 2019; Pettit & Gutierrez, 2018). In particular, Black individuals in various settings are more likely to receive harsher punishment and punishment using the criminal justice system compared with White individuals (Pettit & Gutierrez, 2018).

An additional consideration is the known impact of treatment alliance on the completion of substance use treatment (Brorson et al., 2013). Despite efforts to increase cultural competency among substance use professionals, racial disparities in treatment alliance and/or relationships in treatment remain (Maharaj et al., 2021). The continued acknowledgment of racial discrimination within substance use treatment (Earnshaw, 2020), coupled with the racial disparities in the current study, point to an ongoing need for efforts to ensure equitable treatment in substance use treatment programs.

With little exception, individuals with less financial security (i.e., being unemployed, having low/no income, having no insurance) were less likely to drop out and more likely to be discharged due to termination across treatment settings. In residential treatment settings, individuals who were unemployed and/or had no income at admission were largely more likely to be discharged due to termination compared with those who were employed full time and/or had wages or a salary. The importance of meaningful employment and access to consistent income in completing substance use treatment is well known (Melvin et al., 2012; Room, 1998), and the current results further amplify the influence of employment not only in successful treatment completion but also in preventing termination by the treatment facility.

Insurance type was also found to be a significant predictor of discharge due to termination, although this effect varied greatly based on one's treatment setting. Among those in outpatient/IOP treatment, individuals with no insurance coverage were more likely to be terminated from treatment than to drop out compared to those with private insurance. In residential settings, this effect was nonsignificant in many comparisons or reversed. Specifically, individuals with no insurance were less likely to be terminated and more likely to drop out of residential treatment compared to those with private insurance.

This pattern of financial insecurity increasing one's likelihood of termination was not found when looking at homelessness. Among those in outpatient/IOP treatment settings, individuals who were homeless when they entered treatment were less likely to be terminated and more likely to drop out or be incarcerated compared with their housed counterparts. In residential settings, homelessness did not have an impact on one's likelihood of termination compared with incarceration, but homeless individuals were more likely to be terminated from treatment than drop out compared with their housed counterparts.

Criminal justice referral

Referral to treatment from the criminal justice system had a similar impact on discharge type across treatment settings. In both residential and outpatient/IOP settings, individuals were more likely to be discharged because of incarceration when referred to treatment by the criminal justice system. Individuals in both samples were also more likely to be terminated from treatment by their treatment facility than they were to be discharged due to dropping out of treatment. This finding is consistent with previous literature noting the low rate of dropouts among individuals referred by the criminal justice system, in part because of the consequences of noncompletion (DeFulio et al., 2013; Longinaker & Terplan, 2014; Velazquez-Millings, 2017). However, further research is still necessary to best understand the possible racial disparities within this conclusion (Sahker et al., 2015) as well as the intersection of criminal justice referral and termination.

Mental health and substance use

Overall, the impact of a mental health diagnosis on discharge type depended heavily on treatment setting. In residential treatment, having a mental health diagnosis was not significantly related to discharge type. Conversely, in outpatient/IOP treatment, individuals noted as having a mental health diagnosis were more likely to be discharged because of termination compared with dropping out or incarceration. Of note, individuals in outpatient/IOP treatment more frequently reported having a mental health diagnosis (51.1%) compared with those in residential treatment (41.6%). Given the noted difference between settings, this finding furthers the need for future research examining the impact of facility-level policies and procedures in decisions of termination.

Previous literature regarding the influence of primary substance on treatment discharge is mixed (Brorson et al., 2013; Mennis & Stahler, 2016). The current findings found significant differences in the likelihood of termination primarily among individuals in outpatient/IOP treatment. Individuals reporting to treatment with more than one substance of use (polysubstance use) were more likely to be terminated from both treatment settings than they were to discharge from treatment because of dropout. Because the majority of people in both settings reported more than one substance of use, adaptive and effective treatments addressing more than one substance are crucial to meeting the needs of those in treatment for substance use.

Recovery capital

Taken together, the results of the current study may also be understood through the lens of “recovery capital.” Recovery capital encompasses the intrapersonal, interpersonal, and environmental resources available to any one person that help to sustain their recovery from substances (Hennessey, 2017). The current findings make it clear that access to these resources—including but not limited to education, income, and social capital—also serves as a protective factor against the likelihood of being involuntarily terminated from substance use treatment by the facility. Taken together, these findings further solidify the need for examination of disparities in the distribution of recovery capital, as well as facility-level cultures and policies that impact termination decisions in substance use treatment.

Limitations and future directions

This study has several limitations. First, limited information was available on the differences in rates of termination based on each treatment facility. Research is encouraged to examine programmatic factors influencing the decision to discharge an individual from treatment through termination (e.g., policies, rules, geographic setting, staff makeup, etc.). Second, and relatedly, data regarding specifics of why an individual was terminated were not available in the current data. Future research regarding differences in reasons for termination would expand on the current results and aid in the potential identification of specific program rules and structures that may result in disproportionate termination among some groups. Third, because the TEDS-D includes de-identified data from discharged cases, it is possible that an individual could have multiple treatment episodes included in the current data. Finally, because the TEDS-D only allowed service providers to identify individuals as either male or female, the current results were unable to examine the influence of nonbinary gender identities. Future research is necessary to examine the influence of nonbinary and transgender identities on termination from substance use treatment.

Conclusion

Because the noncompletion rate among individuals in substance use treatment remains high, there is a need to examine the factors associated with reasons for discharge, including termination by the treatment facility. The current study aimed to investigate both individual and environmental factors associated with termination from substance use treatment in both outpatient/IOP and residential treatment settings. Results revealed several patterns, including the influence of race, employment, and criminal justice referral, among others. Findings highlight the need for further investigation into decisions regarding termination of individuals from substance use treatment and identification of interventions that may increase treatment success.