If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Advocate Aurora Behavioral Health Services, Advocate Aurora Health, 1220 Dewey Ave, Wauwatosa, WI 53213, USAAdvocate Aurora Research Institute, Advocate Aurora Health, 960 N 12th St, Milwaukee, WI 53233, USA
Advocate Aurora Research Institute, Advocate Aurora Health, 960 N 12th St, Milwaukee, WI 53233, USAChicago Medical School, Rosalind Franklin University, 3333 Green Bay Rd, North Chicago, IL 60064, United States of America
Advocate Aurora Research Institute, Advocate Aurora Health, 960 N 12th St, Milwaukee, WI 53233, USAAAH Ed Howe Center for Health Care Transformation, Advocate Aurora Health, 960 N 12th St, Milwaukee, WI 53233, USA
SUD treatment programs had high levels of attrition and few care step-downs.
•
Patients were less likely to relapse if they stepped down in care within 14 days.
•
Patient-level variables predicted odds of stepping down in care after discharge.
Abstract
Introduction
Substance use disorder is often a chronic condition, and its treatment requires patient access to a continuum of care, including inpatient, residential, partial hospitalization, intensive outpatient, and outpatient programs. Ideally, patients complete treatment at the most suitable level for their immediate individual needs, then transition to the next appropriate level. In practice, however, attrition rates are high, as many patients discharge before successfully completing a treatment program or struggle to transition to follow-up care after program discharge. Previous studies analyzed up to two programs at a time in single-center datasets, meaning no studies have assessed patient attrition and follow-up behavior across all five levels of substance use treatment programs in parallel.
Methods
To address this major gap, this retrospective study collected patient demographics, enrollment, discharge, and outcomes data across five substance use treatment levels at a large Midwestern psychiatric hospital from 2017 to 2019. Data analyses used descriptive statistics and regression analyses.
Results
Analyses found several differences in treatment engagement based on patient-level variables. Inpatients were more likely to identify as Black or female compared to lower-acuity programs. Patients were less likely to step down in care if they were younger, Black, had Medicare coverage were discharging from inpatient treatment, or had specific behavioral health diagnoses. Patients were more likely to relapse if they were male or did not engage in follow-up SUD treatment.
Conclusions
Future studies should assess mechanisms by which these variables influence treatment access, develop programmatic interventions that encourage appropriate transitions between programs, and determine best practices for increasing access to treatment.
The prevalence of substance use disorders (SUD) among Americans is high. According to the 2020 Substance Abuse and Mental Health Services Administration (SAMHSA) National Survey on Drug Use and Health (NSDUH), 40.3 million individuals age 12 and older within the United States (14.5 % of the population) had an SUD within the past year (
Who seeks care where? Utilization of mental health and substance use disorder treatment in two national samples of individuals with alcohol use disorders.
Prevalence and correlates of past-year recovery from DSM-5 alcohol use disorder: Results from national epidemiologic survey on alcohol and related conditions-III.
SUD is a chronic, often relapsing disease with estimates of 30 % of individuals relapsing within the first three months of recovery and 50 %–94 % of individuals relapsing within the first year (
). Evidence in populations receiving formal treatment indicates that treatment completion can be critical to sustaining recovery, whereas treatment program attrition is linked to relapse (
). Nevertheless, many people with SUD who do not enter formal treatment still achieve recovery, although this type of recovery is more common in those with lower severity SUDs (
Prevalence and correlates of past-year recovery from DSM-5 alcohol use disorder: Results from national epidemiologic survey on alcohol and related conditions-III.
From a clinical perspective, relapse is often defined as a return to a previous substance use behavior following an episode of abstinence, despite the intention of an individual to remain abstinent (
). Increasing utilization and completion of SUD treatment is a significant strategy to attempt to decrease relapses, morbidity, and mortality. However, the progression or lack of progression across the SUD continuum of care by patients has received limited attention in the literature. Understanding the predictors of individuals' progression through treatment services may aid in developing strategies to reduce attrition rates across that continuum.
As a result of disease chronicity and temporal changes in acuity, various levels of SUD treatment are available, ranging from the more intensive inpatient admission to the lowest intensity of outpatient visits. Between those, levels include residential programs, partial hospitalization programs (PHP), and intensive outpatient programs (IOP). Each treatment type is meant to address different treatment and skill acquisition needs of patients and should, ideally, be matched to the patient's current severity of substance use, number of substance use diagnoses, co-occurring mental health disorders, and treatment history (
in: Mee-Lee D. Shulman G.D. Fishman M.J. Gastfriend D. Miller M.M. Provence S.M. The ASAM criteria: Treatment criteria for addictive, substance-related, and co-occurring conditions. 3rd ed. The Change Companies,
2013: 105-126
This continuum of care allows for a flexible approach to treatment, with patients able to ramp up or step down intensity as appropriate, thereby improving outcomes (
). However, patients discharging from a treatment program commonly do not engage in stepping down to the next recommended level of treatment (see review by
The objectives of this study were to: 1) Describe the progression, transition, and outcomes (attrition, length of stay, relapse rates) of patients admitted for SUD treatment across the continuum of care; and 2) Explain the contribution of patient-level demographics (e.g., age, race, SUD diagnoses types, insurance type,) and treatment-level variables (e.g., treatment program type) to attrition and appropriate step downs within 14 days of treatment program discharge.
2. Materials and methods
2.1 Design and sample
This study used a retrospective descriptive correlational study design. We collected data on eligible subjects from the electronic health records (EHR). Eligible patients were at least 18 years old and treated for any DSM-5 SUD diagnosis at a large, Midwestern behavioral health center using inpatient, residential (American Society of Addiction Medicine [ASAM] Level 3.1), PHP, IOP, or outpatient treatment programs from January 1, 2017, through December 31, 2019. For SUD treatment, the campus included >60 inpatient beds across three mental health/SUD inpatient units, 28 residential beds in one residential unit, 30 PHP slots across three groups, 40 IOP slots across four groups, and an outpatient treatment building for both adult SUD and mental health. Additional mental health–focused treatment programs were available on the same campus, including a building dedicated to child/adolescent behavioral health treatment, a private, alternative school for grades 8–12, 70 PHP slots for mental health, and 40 IOP slots for mental health.
Residential, SUD PHP, and SUD IOP programs were strictly SUD treatment programs; therefore, analyses included all patients from those programs. In contrast, the center's inpatient and outpatient programs admit patients for any behavioral health need (e.g., schizophrenia, suicidal ideation, SUD); therefore, the study limited inpatient and outpatient data to those treated for any SUD or those treated for co-occurring mental health and SUDs. The Advocate Aurora Health Institutional Review Board provided institutional review board approval, #19.130ET.
2.2 Measurement and procedures
Retrospectively pulled EHR data were the source of all data for the study, and the study used code targeting discrete EHR fields to extract data for all variables except reason for discharge. Data for attrition required manual extraction due to formatting as free text within patient records. Based on EHR notes, the study coded attrition as either completed program or attrition. Attrition was scored if the discharge record included the phrases “Discharged against medical advice,” or “DAMA,” or any notation that the discharge was earlier than clinically recommended or occurred due to attendance issues, insurance issues, transfer to jail, or patient election in the absence of any statement about clinician agreement or medical stability. We coded patients as having completed the treatment program if EHR notes indicated they completed treatment, they reached maximum benefit from that level of care, they felt ready to step down to a lower level of care and were stable, they had transferred to a more clinically appropriate level of care, or they were considered medically stable. For the few subjects who had no clear indicator for the reason for discharge, we left this field blank.
After manual coding for attrition, an independent coder blind to the initial coding results randomly selected 200 encounters for re-extraction. The study calculated interrater reliability between the initial and independent discharge datasets using Cohen's kappa (K). Results indicated high interrater reliability, with 96.5 % raw agreement and K of 0.913.
Study staff extracted data into two groups: patient-level variables and encounter-level variables. We pulled patient variables for each patient at the time of their index SUD treatment encounter during 2017–2019. Variables included age, sex, race/ethnicity, comorbid disorders, insurance type, SUD diagnoses, and mental health disorder diagnoses. During the period being assessed (2017–2019), the EHR offered sex, race, and ethnicity as single-choice data elements, meaning multiple choices were available but only a single selection was allowed. Selections were made at the time of first encounter within the EHR. At the time of data extraction, sex was a binary variable representing biological sex and required a selection between male and female. Race was categorical and included options for American Indian or Alaska Native, Asian, Black/African American, Native Hawaiian, Other Pacific Islander, Two or More Races, or White. Ethnicity was binary and include options for Hispanic/Latino Origin or Not of Hispanic/Latino Origin.
Given the chronicity of SUD, mental health disorders, and comorbid disorders, the study determined diagnoses by the presence of mental health, SUD, or comorbid ICD-10 codes input during psychiatric program encounters at any time during a 5-year lookback in hospital and professional billing records. Encounter-level variables included treatment program type, date of program admission and discharge, reason for program discharge, insurance type, and distance between the treatment program and patient's home address.
The study scored two variables based on combinations of other variables. A step down in care was scored if a patient had at least one completed visit in a lower-level program within 14 days of discharge from the previous program (e.g., one IOP visit within 14 days of discharge from PHP). We scored it as relapse if a patient returned to a higher level of treatment after discharge from the previous program within 6 months or was admitted twice consecutively within 6 months to inpatient, residential, or PHP. For example, we labeled admission to PHP within 6 months after discharging from IOP or PHP as a relapse, whereas readmission from IOP to IOP was not.
Coders extracted patient addresses at the time of each encounter. Each encounter address was processed through an in-house machine learning algorithm that corrected malformed address attributes, while an in-house geocoding application determined the longitude and latitude for each encounter address. Study staff imported each encoded longitude and latitude value into ArcGIS (version 10.3.1.4959) to conduct Euclidean distance calculations to the behavioral health center.
2.3 Analyses
Statistical software, including SAS 9.4 software (SAS Institute Inc.) and GraphPad Prism version 8.0.0 for Windows (GraphPad Software), performed statistical analyses. We summarized descriptive characteristics using means and standard deviations for continuous variables and frequency and percentages for categorical variables. Pearson's chi-square test or Fisher's exact test compared categorical variables, as appropriate. A parametric analysis of variance (ANOVA) compared continuous variables. Multiple logistic regression analyses analyzed variables that were significant on univariate analyses or designated as potentially significant based on expert opinion. For the regression analyses, the study counted only the patient's first encounter for each treatment type, and we excluded any patients missing regression-required data elements. All tests were two-tailed, and we considered a p < .05 statistically significant.
3. Results
3.1 Program demographics
Data analytics pulled records for 6633 unique adult patients (Table 1). At the time of first visit into any program, mean patient age was 39.0 years (SD = ±13.1; Supplementary Table 2). A majority of patients identified as male and White, and the most common insurance categories were commercial and Medicaid. Demographic distributions varied significantly across programs. Inpatient and outpatient programs had higher frequencies of women, with significantly different distributions of patients by sex, χ2 (4, N = 10,120) = 59.2, p < .001. A chi-square test of independence indicated a large association between race/ethnicity and treatment program, wherein inpatient had lower frequencies of White patients and higher frequencies of Black patients than all other programs, χ2 (12, N = 10,120) = 568.2, p < .001. Insurance type was also significantly associated with program type, χ2 (12, N = 10,120) = 959.4, p < .001. Commercial insurance accounted for the majority of payors in residential, PHP, IOP, and outpatient admissions, whereas commercial and Medicaid each made up around 40 % of payers for inpatient. Most patients with Medicare were < 65 years old.
Table 1Patient demographics and outcomes across programs.
Note. Patient demographic variables were counted only once per program at the time of first program encounter, whereas treatment outcomes were counted for each program admission. Patients with no data available for a particular variable were excluded from that variable category. AI/AN = American Indian or Alaskan Native; AAPI = Asian American Pacific Islander; IOP = intensive outpatient program; LOS = length of stay; PHP = partial hospitalization program; SD = standard deviation; SHA = sedative, hypnotic, or anxiolytic; UD = use disorder; UNK = unknown. Program attrition is the number of patients who discharged from the index program before completing program treatment. Step down is the number of patients who stepped down to a lower level of care within 14 days. Relapse is the number of patients who returned to inpatient, residential, or PHP treatment within 6 months outside of a step down. P values for categorical variables were calculated using chi-square goodness of fit tests, and age was compared using a parametric analysis of variance (ANOVA).
Further, there were significant group distribution differences for substance use diagnoses, χ2 (20, N = 24,429) = 634.5, p < .001, and mental health diagnoses, χ2 (16, N = 18,844) = 576.8, p < .001. Alcohol use disorder was the most common SUD across programs except for outpatient, where opioid use disorder was the most common SUD. Compared to the other four programs, inpatient had lower rates of opioid use disorder and higher rates of cannabis use disorder. All programs had high rates of depressive and anxiety disorders, whereas psychotic disorders were considerably higher in inpatient than all other programs.
3.2 Program completion and follow-up care
The percentage of patients completing any given treatment program were highest in the residential program (78.0 %) and lowest in the PHP program (60.4 %; Table 1), and percentages across programs were significantly different. Fig. 1 shows data reflecting postdischarge transitions among treatment programs within 60 days. Transitions included stepping up to a higher level of care, readmitting to the same level, or stepping down to a lower level of care. Patients discharging from inpatient and IOP had low admission rates to any other treatment program within 60 days (Fig. 1, Panels A and D), whereas residential and PHP had higher rates of program transitions (Fig. 1, Panels B and C). Inpatients were most frequently readmitted as inpatients, although the study found an increased frequency of transition to PHP if the transition occurred within 14 days. Residential-discharged patients frequently stepped down to PHP or IOP programs, and PHP-discharged patients frequently stepped down to IOP. The timing of transitions also differed among programs. Across all programs, patients were most likely to step down in care if their transition occurred within 14 days of discharge.
Fig. 1Transitions after SUD Program Discharge: Location of and Latency from Index Discharge to Follow-Up
Note. Bars display the percentage of patients admitted and/or had at least one completed visit to another program within 60 days of current program discharge. Data include latency to next program. Striped bar indicates the percentage of patients who did not transition to any other program within 60 days.
When specifically measuring step-down transitions, defined as when a patient stepped down to a lower level of follow-up care within 14 days of discharge from their index program, inpatient and IOP had low transition frequencies, 17.1 % and 10.7 %, respectively (Table 1). In contrast, almost half of residential patients stepped down in care after discharge.
Multiple logistic regression analyses identified several variables predictive of a patient stepping down to a lower level of care across programs (Table 2). Younger patients (ages 18–27) were less likely to step down than their older counterparts, males had 1.2 odds of stepping down compared to females, and patients of all race/ethnicity categories were significantly more likely to step down than Black patients. Medicare patients were significantly less likely to step down in care compared to Exchange, commercial, or Medicaid-insured patients. Importantly, patients who did not complete their index treatment program were less likely to step down in care following their early discharge. Only two behavioral health diagnoses predicted lack of step down within 14 days: bipolar and psychotic disorders. In contrast, patients with SUD diagnoses, except cocaine or cannabis use disorders, were more likely to step down.
Table 2Multiple logistic regression analyses predicting odds of step down within 14 days.
Data included the index admission for each unique patient during the study timeline. Variables included in the model were: age group, sex, race/ethnicity, program type, insurance type, history of substance use diagnoses, history of mental health diagnoses, program treatment completion, and distance. Diagnoses included in the model are not mutually exclusive. AAPI, Asian American or Pacific Islander; AI/AN = American Indian or Alaska Native; IOP = intensive outpatient program; PHP = partial hospitalization program; ref. = reference; SHA = sedative, hypnotic, or anxiolytic; UD = use disorder. N = 5594 unique patients.
The study found program-specific differences in predictors in inpatient and PHP-specific regressions (Supplementary Table 3). For example, most variable categories included significant predictors of step downs in inpatients, with the insurance category producing the highest odds ratios and patients with Medicare having the lowest odds of stepping down. In contrast, PHP patients only had a few significant predictors, including sex, program completion, and distance from home. Patients who completed PHP programming had an odds ratio of 15.1 for stepping down in care versus those who did not complete, whereas inpatients who completed treatment had an odds ratio of 1.8 for stepping down in care.
3.3 Program relapses
Relapses within 6 months after discharge from index treatment program ranged from 21.9 % after IOP to 34.7 % after inpatient (Table 1). Several variables were predictive of relapse within 6 months (Table 3). In the multiple logistic regression model, sex was the only patient demographic predictor of relapse, with males having slightly higher relapse rates than females. Patients with any mental health or substance use diagnoses had higher rates of relapse compared to the absence of those diagnoses. Patients were less likely to relapse if they were in IOP treatment (compared to inpatient) or stepped down to the next level of treatment within 14 days after discharge. Inpatient and PHP-specific regressions produced similar results (Supplementary Table 4), with some differences in predictors between programs. Failure to step down in care predicted relapse across both populations.
Table 3Multiple logistic regression analyses predicting odds of relapse within 6 months.
Data included the index admission for each unique patient during the study timeline. Variables included in the model were: age group, sex, race/ethnicity, program type, insurance type, history of substance use diagnoses, history of mental health diagnoses, program treatment completion, and step down transition from program. Diagnoses included in the model are not mutually exclusive. AAPI = Asian American or Pacific Islander; AI/AN = American Indian or Alaska Native; IOP = intensive outpatient program; PHP = partial hospitalization program; ref. = reference; SHA = sedative, hypnotic, or anxiolytic; UD = use disorder. N = 5761 unique patients.
Research examining SUD treatment completion has primarily focused on discharge from a single level of treatment type (e.g., inpatient, residential, PHP). This study is unique in that it provides new information on transitions between SUD treatment programs and respective patient outcomes across the continuum of care by comparing different SUD treatment programs with varying acuity levels available within a single behavioral health system and on a shared physical campus. This multiprogram comparison offers insights into patient progression and needs in the context of a continuum of care. Patient characteristics and outcomes differed across programs, suggesting that unique patient populations are being referred to or accessing specific levels of treatment.
4.1 Length of stay
Prior reports found that inpatient SUD treatment had an average length of stay between 2.8 and 5.1 days (
Expanding access to addiction treatment services through Section 1115 waivers for substance use disorders: Experiences from Virginia and Maryland. AcademyHealth.
). Both range averages are consistent with the lengths of stay found here. PHP and IOP average lengths of stay in the literature were shorter, with PHP averaging 14.3 days (
). Differences in lengths of stay may be due to either the use of different treatment modalities or preprogrammed intensity differences aligned with ASAM Criteria.
4.2 Care transitions
Although patients in this study had concurrent access to several different acuity levels of SUD follow-up treatment, only 10.7 %–49.6 % of patients stepped down to a lower level of care within 14 days. This result is consistent with extant literature in which follow-up rates range from 16 % to 79 % within 7–30 days of inpatient discharge (see review by
). Of patients who stepped down in care within at least 60 days, the majority did so within 14 days, suggesting that there is a window in which step-down transitions are most likely to occur. If a patient returned to care after 14 days, they typically were admitted to inpatient treatment, indicating that treatment access beyond the 14-day window may represent relapse as opposed to continuity of care.
found that 50 % of patients who had engaged in follow-up treatment within 14 days of inpatient discharge did so within 48 h, suggesting the window for targeting follow-up transitions may be even more transient than indicated by our study findings.
Programs also differed in terms of where patients went after discharge. Although this result is relatively consistent with other literature, previous studies looking at transitions were limited both by the number of discharge programs being assessed as well as the number of follow-up programs available, thus limiting their applicability. For example, one study found that patients discharging from inpatient treatment were unlikely to engage in any follow-up care; however, if they did follow up, they were more likely to do so at residential or recovery housing as opposed to outpatient treatment (
found that patients discharging from detoxification were more likely to follow up at outpatient or IOP as opposed to residential treatment. Again, many patients either did not follow up at all or re-engaged in inpatient care (
). However, the proximity and availability of residential treatment options in both studies is questionable, whereas all treatment programs assessed in the current study were available simultaneously and on the same campus. Therefore, the data presented herein provide unique insights into the transition types and rates when patients have concurrent and consistent access to a full continuum of care.
Other studies looking at discharge from a single treatment level found somewhat similar results on predictors of step-down care engagement. Patients discharged from inpatient opioid overdose treatment were more likely to attend follow-up care if they were male, White, and were previously receiving treatment for anxiety, with Black patients half as likely to follow up as White patients (
). Ilgen et al. (2008) found that veteran inpatients with schizophrenia were less likely to step down in SUD care, whereas those with depression or anxiety were more likely to do so. In contrast, regression analyses presented here did not identify anxiety diagnoses as a predictor of follow-up, and sex was a predictor in PHP but not inpatient. Further, inpatients with depression were less likely to follow up, which contrasts with Ilgen et al. (2008). These differences are potentially a function of studying different SUD populations, as this study included all patients with SUD and was not limited by veteran status or SUD type. Nevertheless, differences in significant predictors between programs further highlight that different acuity programs, even on the same campus, treat unique patient populations and suggests the need for program-specific interventions to keep patients engaged in care.
Although this study did not qualitatively assess programmatic differences which may have contributed to patient likelihood of stepping down in care or relapsing, clinic variables are known to correlate with or predict patient outcomes. For example, patients are more likely to engage in follow-up care in clinics that are larger and offer single-provider treatment and involve family/peers, inpatient psychotherapy/education, warm hand-offs between levels of care, and small financial incentives for attendance (
). General patient population makeup may also impact individuals' desire to stay in treatment. For example, many psychiatric hospitals co-house patients with SUD and patients with mental health disorders on the same inpatient unit; however, patients receiving SUD treatment found the behavior of mental health patients disruptive enough to initiate early discharge (
Additional barriers to accessing a continuum of care include insurance gaps and/or incarceration. This study was not able to identify the number of patients who stopped treatment due to incarceration or insurance versus nonadherence or renewed substance use. However, this population is known to face relatively high justice-involvement rates (
Release from incarceration, relapse to opioid use and the potential for buprenorphine maintenance treatment: A qualitative study of the perceptions of former inmates with opioid use disorder.
Mat Working Group Of CJ-DATS Medication-assisted treatment in criminal justice agencies affiliated with the Criminal Justice-Drug Abuse Treatment Studies (CJ-DATS): Availability, barriers, and intentions.
). Even outside of incarceration, many people with a perceived need for SUD and mental health treatment do not access treatment due to a lack of health care coverage (
Comparing barriers to mental health treatment and substance use disorder treatment among individuals with comorbid major depression and substance use disorders.
Medicaid and CHIP Payment and Access Commission Access to substance use disorder treatment in medicaid (June 2018 report to congress on medicaid and CHIP).
). Given these potentially major barriers interrupting access to a continuum of SUD care, future studies should assess how many patients do not transition to the next level of care due to these barriers versus treatment nonadherence or substance use renewal.
The insurance issue is especially critical, given that only 12 states have a Medicaid plan covering the full continuum of SUD treatment, often leaving gaps in coverage for residential and PHP (
Medicaid and CHIP Payment and Access Commission Access to substance use disorder treatment in medicaid (June 2018 report to congress on medicaid and CHIP).
), which would restrict some patients' ability to step down in care to programs potentially matched to their acuity needs. The deficit in step downs by inpatients with Medicare is consistent with other studies where Medicare recipients were less likely to access substance use treatment than other insurance groups (
). In this study, inpatient and outpatient programs did not have major insurance restrictions, whereas residential, PHP, and IOP had some insurance limitations. Specifically, residential did not accept Medicaid or Medicare coverage, whereas PHP and IOP accepted all insurance except Medicare and Medicaid preferred provider organizations (PPOs). In other words, PHP and IOP accepted Medicare/Medicaid, but only their Health Maintenance Organization (HMO) plans. Therefore, it is possible that insurance had a direct impact on the rate of step downs; however, insurance was not a predictor of step downs for PHP patients, suggesting the influence of other, population-specific barriers at the inpatient level. For example, the demographics data show that most patients in the dataset with Medicare were under age 65, suggesting this population largely comprised patients with disabilities. This population may have unique social and physiological barriers to treatment such as medical disabilities, finances, social stigma concerns, and lack of transportation (
). The Medicaid population is known to have higher rates of acute readmissions and lower engagement in preventative care as a result of barriers at three levels: provider, system, and patient (
Kaiser Family Foundation Medicaid enrollees are sicker and more disabled than the privately-insured. KCMU Analysis of MEPS 3-Year Pooled Data, 2004-2006.
). Patient-level barriers include generally poorer health or higher rates of disabilities, financial stressors, medication adherence gaps, access issues, and housing instability.
The positive correlation between completing index treatment program and successfully stepping down in care is especially intriguing, given the large population of patients who do not complete treatment (22.0 %–39.6 % across programs). The demonstrated correlation suggests a potential target for intervention, although we don't know whether the relationship is functional versus merely correlational. For example, patients who drop out of index treatment early may be less likely to engage in follow-up care because they have personal barriers (e.g., employment or family obligations, financial concerns), perceive stigma, dislike clinical settings, or are bored (
). However, the failure to step down may be a direct result of failure to complete care, if patients who drop out early fail to receive discharge case management, miss opportunities to become familiarized with next-step treatment programs, or have less opportunity for relationship building with clinical teams. Many of these barriers are potentially surmountable, and future research should identify which variables can be targeted by the treatment provider to increase completion rates and engagement in follow-up care. Several interventions have been tested to increase follow-up engagement and showed promising results. These include Assertive Continuing Care (
Results showed that the inpatient level of treatment had a much higher proportion of Black patients than any other racial/ethnic category, yet Black patients were also less likely to step down than any other group. In the literature, Black patients are up to 8.2 times more likely to be admitted as inpatients for behavioral health treatments as their White counterparts (
Substance use disorders and comorbid Axis I and II psychiatric disorders among young psychiatric patients: Findings from a large electronic health records database.
). These racial/ethnic differences in inpatient treatment likely result from several factors related to both engagement and access, such as socioeconomic status, clinical biases or cultural competency gaps, distrust toward healthcare, cultural beliefs and stigma, perceived racism, and/or greater co-occurring severe mental health diagnoses (
A systematic review of inequalities in the mental health experiences of Black African, Black Caribbean and Black-mixed UK populations: Implications for action.
in: Medlock M.M. Shtasel D. Trinh N.-H.T. Williams D.R. Racism and psychiatry: Contemporary issues and interventions. Springer International Publishing,
2019: 105-137https://doi.org/10.1007/978-3-319-90197-8_7
Substance use disorders and comorbid Axis I and II psychiatric disorders among young psychiatric patients: Findings from a large electronic health records database.
). However, the higher rate of psychotic diagnoses in Black patients is hypothesized to be a function of differences in trauma, symptom reporting, and clinical interpretations during diagnosis (for reviews see
). Overall, more research needs to determine why Black patients are so overrepresented in this context, whether the group step-down rates are reasonable, and/or whether interventions need to be performed at the patient and/or system level to address these relatively low engagement rates.
Although transitioning down to a lower and standardized level of care is often considered to be ideal, many patients do not engage in the recommended follow-up care. As a result, additional support services can be made available to patients facing barriers to standard treatment. For example, clinics may utilize case management or harm reduction strategies, which focus on avoiding/reducing the negative effects of drug use such as death, virus transmission (Hepatitis C and HIV), and legal issues (
Relapse prevention is a critical issue in SUD treatment. The percentage of patients in this study who relapsed early post-treatment is consistent with other reports showing that 37 %–86 % of individuals relapse within 12 months, although relapse rates vary by relapse definition, drug type, time of follow-up, and treatment type (
). The pattern of relapse that this study found is consistent with the same pattern (when comparing alcohol, heroin, and tobacco) as demonstrated in 1971, with many patients returning to substance use within 3 months of completing treatment and <30 % remaining in recovery at one year (
). These data showing high relapse rates are consistent with long-term data on time to remission, wherein mean time to remission is 4–22 years and remission is preceded by multiple relapses (
). Individuals who transitioned down to the next level of care within 14 days and/or completed their index treatment program had reduced relapse risk, similar to other studies (
Nevertheless, relapse is a construct with widely varying operational definitions across studies and clinics, and with definitions varying across measures, drug type, time, frequency, and/or intensity (
). Given that chronic diseases are defined by the very fact that they recur and require maintenance treatment and also that many people recover from alcohol use disorders while continuing to engage in low-risk drinking (
), this study did not measure relapse based on drug use/abstinence. Instead, this study defined relapse as any step up to a higher level of care or readmission to an acute care program (inpatient, residential, or PHP) within 6 months. However, patients may re-enter treatment for reasons beyond recurrence of drug use or may step down in care even through continued or reoccurring drug use.
Given the high rates of relapse, we have a great need to provide public health education regarding treatment options for SUD and co-occurring disorders so that individuals understand their early treatment options to avoid the need for more intensive treatments, such as inpatient or residential care. Future research should address key demographics (e.g., race and insurance coverage) and clinical characteristics (e.g., co-occurring disorders and subtype of substance use disorder) in the development of interventions to address retaining patients in the treatment continuum for at least 90 days. Addressing equity in health insurance coverage for individuals who do not have commercial insurance seems critical to decreasing barriers for those who seek treatment, as this funding issue is identified as a major barrier to care (
Medicaid and CHIP Payment and Access Commission Access to substance use disorder treatment in medicaid (June 2018 report to congress on medicaid and CHIP).
) and is associated with lower rates of treatment step downs. Nevertheless, it bears noting that some patients respond without treatment and yet not every patient responds to every type of treatment. Many patients will relapse even as their substance use improves over time (
). Therefore, access to and engagement in appropriate, formal, holistic substance use treatment can be an important step for many people but does not currently guarantee full and permanent recovery for all.
4.4 Limitations
The study was focused on SUD treatment within a single health system, which may limit the generalizability of results and also limit the data available for analyses. For example, although patients discharging from treatment had several acuities of follow-up programs available within the same health system and all on the same campus, some patients likely chose to receive treatment outside the system. Therefore, the number of patients receiving follow-up care is likely at least somewhat higher than shown in Fig. 1. Similarly, limiting the study to SUD programs further limited the follow-up data. Specifically, patients discharged from inpatient treatment for co-occurring SUD and mental health diagnoses would have had the option to follow up with any of the following: SUD residential, mental health PHP, SUD PHP, mental health IOP, SUD IOP, or outpatient treatment. Mental health program data were not included in the study dataset so, again, it is likely that the number of inpatients receiving follow-up care is higher than we show in Fig. 1. Additionally, some programs may have had waitlists in place, especially at the outpatient level, and patients may have attempted to engage in follow-up care but were delayed beyond 14 days due to being waitlisted, which would result in some patients being scored as not stepping down despite having attempted to do so. Nevertheless, given how few patients engaged in follow-up care within 60 days, this possibility likely represents relatively few patients.
This study utilized a large dataset, which by its nature has its own limitations. Specifically, large datasets often produce outcomes that are statistically significant but may not be clinically significant or may introduce other types of errors (
). For example, several odds ratios were statistically significant yet relatively small, thus limiting their clinical significance. A further limitation is that the data for the study were from the EHR and thus subject to the accuracy and completeness of data input by clinical staff. Data manually pulled from the EHR on completion of treatment were coded as binary variables (completed versus did not complete), and therefore analyses were not able to further clarify the varying reasons for program noncompletion. This inability to determine exactly why patients did not continue to access the continuum of care (e.g., financial barriers, incarceration, substance use renewal, or other reasons) limits the ability to identify specific barriers to inform future interventional studies.
5. Conclusion
The results of this study, which focused on SUD patient treatment access and relapse across five treatment program levels within a continuum of care, demonstrated that many patients do not engage in SUD follow-up care after program discharge, even when the follow-up program was located on the same campus as the index program. Predictors of failure to step down included Black race, Medicare insurance, program noncompletion, and a variety of comorbid behavioral health diagnoses. These data can serve to identify potential targets or subpopulations in need of interventions aimed at increasing transitions to follow-up care. However, to develop targeted interventions, future studies should determine why these variables are predictive in this population; for example, if follow-up studies find that race/ethnicity are predictive of patient engagement in SUD care because of a lack of culturally competent care, then the intervention should target increasing cultural competency followed by measurement of outcomes. Although programs should increase step-down care transitions in general, inpatients have the greatest need for step-down intervention to avoid a cycle of relapses and readmissions. Thus, developing interventions for inpatients is a critical target for future studies.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
We thank those who manually pulled a substantial amount of discharge-related data from the electronic health records, including Jon Phillips, Catherine Warner, Aspen Duffin, Elizabeth Wanninger, and Logan Friedrich. We also thank Joe Grundle for his feedback on the manuscript and Dawn Reese for her insights into the clinical programs.
Expanding access to addiction treatment services through Section 1115 waivers for substance use disorders: Experiences from Virginia and Maryland. AcademyHealth.
A systematic review of inequalities in the mental health experiences of Black African, Black Caribbean and Black-mixed UK populations: Implications for action.
Who seeks care where? Utilization of mental health and substance use disorder treatment in two national samples of individuals with alcohol use disorders.
Prevalence and correlates of past-year recovery from DSM-5 alcohol use disorder: Results from national epidemiologic survey on alcohol and related conditions-III.
Release from incarceration, relapse to opioid use and the potential for buprenorphine maintenance treatment: A qualitative study of the perceptions of former inmates with opioid use disorder.
Medication-assisted treatment in criminal justice agencies affiliated with the Criminal Justice-Drug Abuse Treatment Studies (CJ-DATS): Availability, barriers, and intentions.
in: Medlock M.M. Shtasel D. Trinh N.-H.T. Williams D.R. Racism and psychiatry: Contemporary issues and interventions. Springer International Publishing,
2019: 105-137https://doi.org/10.1007/978-3-319-90197-8_7
in: Mee-Lee D. Shulman G.D. Fishman M.J. Gastfriend D. Miller M.M. Provence S.M. The ASAM criteria: Treatment criteria for addictive, substance-related, and co-occurring conditions. 3rd ed. The Change Companies,
2013: 105-126
Comparing barriers to mental health treatment and substance use disorder treatment among individuals with comorbid major depression and substance use disorders.
Substance use disorders and comorbid Axis I and II psychiatric disorders among young psychiatric patients: Findings from a large electronic health records database.
All categories had p < .001.
00004-2&id=gr1.jpg){kind=link}