Predictors of Premature Discontinuation from Behavioral Health Services: A Mixed Methods Study Guided by the Andersen & Newman Model of Health Care Utilization
Doctor of Philosophy (Ph.D.)
Degree Granting Department
Bruce L. Levin, Dr.PH.
Russell Kirby, Ph.D.
Roger A. Boothroyd, Ph.D.
Dina Martinez-Tyson, Ph.D.
Oliver T. Massey, Ph.D.
mental health treatment, dropout, emerging adults, substance abuse treatment
Emerging adults, ages 18 to 25, demonstrate high prevalence of behavioral health illnesses, yet infrequently access treatment. For those who do access care, premature discontinuation rates from treatment are high, ranging from 30 to 50 percent. For the emerging adults who initiate contact with the systems of care by attending at least one session of therapy, there is an opportunity to engage them in treatment and prevent negative health outcomes in adulthood. This mixed methods study used the Andersen & Newman (1973) model of healthcare utilization as a guiding framework to explore and examine premature discontinuation from public sector behavioral health services among emerging adults in Florida.
To begin, a systematic literature review was conducted to guide measurement of the outcome variable for the quantitative study, which consisted of a secondary analysis of the Substance Abuse and Mental Health Information Systems (SAMHIS) dataset (N=107,565). The purpose of the quantitative study was to examine the elements of the A&N model that could be tested quantitatively in relationship to dropout. A qualitative study, which consisted of in-person semi-structured interviews (N=20), was conducted with emerging adults in the Tampa, Florida area who were attending the Healthy Transitions group program, administered by the Success 4 Kids and Families agency. The interviews were conducted to explore the conditions in the A&N model that could not be tested with SAMHIS data in relationship to dropout.
The findings of the literature review confirmed the measurement of premature discontinuation remains inconsistent. Across 28 studies, definitions of dropout consisted of duration-based measures, dosage thresholds, clinician determination of behavior change, substance abuse treatment program guidelines, and client self-report. The varying definitions, in combination with different terminology, data sources, determiners, modalities and settings, diverse samples, and a wide range of diagnoses, contributed to the varied measurement of this topic.
Given the inconsistency, for the quantitative analysis, dropout was defined as attending one session and missing a follow up session within 90 or 180 days of the initial encounter. These timeframes were selected in order to conduct a sensitivity analysis. The 90-day logistic regression produced a 17-variable model χ2 (df =16) = 4015.183, p<.001. (p=.001). Those in urban areas, with severe diagnoses, and more education, were factors most strongly associated with dropout. The 180-day model revealed trivial differences in significance and fit statistics. Dropout rates decreased from 37.5% at 90 days to 33.4% at 180-days. The variables remained constant in both models with the exception of residential stability which became nonsignificant at 180-days. Fit statistics at the 90-day model of a -2 Log Likelihood (138267.05) and a Nagelkerke R2 = .050 suggest the models explained a small proportion of the variance in the dataset.
The qualitative study findings suggest emerging adults are dropping out of care for reasons not accounted for in the A&N model. Specifically, emerging adults indicated logistical issues, such as timing conflicts, financial constraints, and transportation concerns, influenced their decision to leave care. Lack of a therapeutic relationship or bond, particularly during initial encounters was an emergent theme discussed in relationship to dropout. Emerging adults who experienced severe symptomology, such as suicidal tendencies, indicated they wanted to remain in care to prevent doing harm to themselves or others. However, when emerging adults interacted with a therapist and received a diagnosis, this was perceived as a stigmatizing and shameful label which caused them to want to leave care.
Perhaps more importantly than revealing reasons for dropout, emerging adults discussed why they were engaged in the Healthy Transitions program and how it differed in their perspective from traditional therapy. The Healthy Transitions group was free, transportation was provided through Uber, Lyft and ride sharing options, and the group sessions were held two evenings per week to reduce conflict with work hours. The setting for the Healthy Transitions group was a church which was perceived as less intimidating than a therapist office. The Healthy Transitions program was successful in terms of using peer educators for group leaders. Emerging adults in this sample did not feel judged or stigmatized by group leaders and felt their perspective and voices were incorporated into their care plan. Essentially, Healthy Transitions addressed nearly every reason for dropout, which is how they are successfully retaining emerging adults in treatment.
Dropout from behavioral health care continues to be a problem and rates have remained high for the past 50 years, particularly among emerging adults. The inconsistencies in measurement leads to confusion in terms of the scope of the problem, and in our understanding of the topic. This field of research could benefit from developing at least some standard indicators for measuring dropout.
The model fit in the quantitative study suggests factors aside from those in the Andersen and Newman model are accounting for the reasons emerging adults are leaving care. Future studies could shift the focus from measuring immutable sociodemographic variables in relationship to discontinuation, to identifying a predictive profile of those who are most likely to drop out. Measures to assess the relationship emerging adults have with their therapist as well as engagement level may be the variables most strongly associated with dropout.
The qualitative findings further suggest we shift the focus from dropout to researching the reasons emerging adult decide to remain in treatment. This shift in focus may present new ideas for engaging these individuals in their initial encounters, rather than focusing solely on why they decide to leave care. Most importantly, future studies conducted from the client perspective are needed to help us better understand this concept. This information could lead to the develop of interventions for emerging adults who attend at least one session of therapy.
Scholar Commons Citation
Green, Shawna M., "Predictors of Premature Discontinuation from Behavioral Health Services: A Mixed Methods Study Guided by the Andersen & Newman Model of Health Care Utilization" (2020). USF Tampa Graduate Theses and Dissertations.