An estimated 13% of U.S. children have a mental disorder (
1); however, only half of the youths who may be in need of services receive them and, of these, many do not stay in treatment (
1,
2). According to a meta-analysis, the mean treatment dropout rate in child samples is 47% (
3), and this rate may be even higher among children treated in inner-city clinics (
4). Children’s utilization of services is of high importance for public health because obtaining adequate treatment in childhood can reduce not only current impairment but also the likelihood of having psychiatric problems later in life (
5,
6).
Several child and parent factors have been identified as predictors of retention in mental health services. Studies have consistently shown that higher socioeconomic status (
7–
9), younger child age (
10), higher parent education level (
11,
12), and living in a two-parent household (
9,
13,
14) are associated with retention. Parents’ recognition of a psychiatric concern and parent burden resulting from child psychiatric problems are also associated with increased service utilization (
15). Several studies have found racial or ethnic minority status to be associated with lower retention rates (
9,
10,
16,
17); a few studies, however, have found no effect of race or ethnicity on retention (
8,
13,
18). This inconsistency may be due to variability in the samples’ demographic and clinical characteristics, methodology, and service use definition. Studies examining treatment retention across multiple treatment sites are rare (
13,
18). Improved understanding of predictors of treatment retention should ultimately help to improve access to and engagement in mental health services.
Because parents usually make the decision to access services for children, parents’ perceptions of mental health services are important determinants of whether a child receives and stays in treatment (
9,
19,
20). Early models of health behavior, such as Donabedian’s (
21) structure-process-outcomes model, suggest that both individual patient characteristics and patient-practitioner interpersonal processes influence quality of care. Building on such models, Kazdin and colleagues’ (
9) barriers-to-treatment model indicates that, in addition to sociodemographic factors, families’ perceptions of the treatment or of the therapeutic relationship further affect the likelihood of retention. Parents’ perceptions of relevance, or match, of a treatment to their children’s needs have received little attention in the service-retention literature, although the scant extant research suggests that children are less likely to continue treatment if it does not match their parents’ expectations (
9,
22). Among a sample of parents with low incomes from racial or ethnic minority groups (typically factors indicative of high risk for dropout), those whose motivations for initiating treatment matched the goals of treatment were more likely to continue treatment compared with parents whose motivations did not match treatment goals (
23). Better understanding of these perceptions, and the interplay of these perceptions and sociodemographic variables, may point to factors that can be targeted to improve engagement in mental health services and increase service retention.
Because of its large sample size and clinical and demographic heterogeneity, the Longitudinal Assessment of Manic Symptoms (LAMS) provides a rich longitudinal data set with which to examine the effects of sociodemographic characteristics, clinical factors, and parents’ perceptions of treatment on treatment retention. Previous analyses of treatment retention in the LAMS sample found that 435 of 573 (76%) participants continued to receive services at six-month follow-up (
13). These earlier analyses indicated that living with both biological parents and perceived benefit of treatment by parents were associated with retention at six months.
This study explored possible racial-ethnic disparities in retention as well as associations between treatment retention and sociodemographic variables, child clinical characteristics, and perceived treatment match across an extended period of 12 months. We sought to determine whether perceived treatment match contributed to the prediction of treatment retention above and beyond child and family sociodemographic and clinical characteristics.
Methods
Participants
The LAMS, a prospective study of youths with elevated symptoms of mania (ESM), recruited participants ages six through 12 years during their first visit to one of nine outpatient clinics associated with one of four LAMS academic medical centers; recruitment took place between November 2005 and November 2008. Of 1,111 youths who were screened and considered eligible, 621 agreed to participate in the longitudinal portion of LAMS; an additional 86 children without ESM matched for sex, age, race, ethnicity, and insurance status also agreed to participate, for a total of 707 youths who participated in the baseline assessment. A total of 675 (95%) youths remained eligible after 12 months, and of these, 416 (62%) had received outpatient medication or psychotherapy services at baseline and had 12-month treatment data available and were included in these analyses. LAMS study design, participant selection and eligibility criteria, and demographic characteristics of the sample have been reported previously (
24,
25). Children were reassessed by trained research assistants every six months by phone or in person (depending on family preference) with the state-of-the-art measures described below.
All study procedures were approved by the institutional review board at each study site. Caregivers and children provided informed consent and assent before completing any study measures.
Measures
Demographic characteristics.
Child sex, age, race, ethnicity, insurance status, family structure (whether the child was living with both biological parents), and caregiver education were recorded at baseline. To maintain consistency with previous analyses of service use in the LAMS sample (
26), baseline age was categorized as younger (six or seven years old) or older (eight to 12 years old) in these analyses.
Parenting stress.
The Parent Stress Survey (PSS) (
27), a 25-item parent-report questionnaire, assessed parents’ experiences with several stressful events related to parenting, familial matters, and children’s mental health needs. Each item is rated as 0, no, or 1, yes, and scores are a sum of all items. Possible scores range from 0 to 25, with higher scores indicating greater parenting stress.
Mental health services use.
Caregivers completed the parent version of the Services Assessment for Children and Adolescents (SACA), a valid and reliable instrument, at each interview (
28). The SACA surveys inpatient, outpatient, and school service use. Information on the type of treatment used (therapy, medication, both, other, and unknown) and parents’ perception of how well the child’s treatment matched his or her needs (not well, somewhat well, or very well) was also collected (
28). Only data on outpatient services use were included in these analyses. Children were identified as having been retained in treatment at the 12-month assessment if they continued to receive the same outpatient services—for example, medication and psychotherapy—that they had used at baseline.
Children’s clinical characteristics.
Children and caregivers were interviewed at baseline by using the Schedule for Affective Disorders and Schizophrenia for School-Aged Children–Present and Lifetime (K-SADS-PL) (
29) with additional questions about mood symptoms from the Washington University in St. Louis K-SADS (
30). The Children’s Global Assessment Scale (CGAS) was used to assess children’s overall functioning at home, at school, and with peers at baseline (
31). Scores range from 1 to 100, with lower scores indicating more significant impairment.
Analyses
We conducted bivariate comparisons at 12-month follow-up of children who continued to receive the outpatient services they had received at baseline versus those who had discontinued a baseline service; chi square tests were used for categorical variables and t tests were used for continuous variables. Hierarchical logistic regression examined the relationship between participant characteristics and treatment retention. Baseline sociodemographic and clinical variables entered in the first block included child sex, ethnicity, race, age (six or seven versus eight to 12), and insurance status (Medicaid versus self-pay or private insurance), primary caregiver education, family structure, CGAS score, number of diagnoses, presence of a depressive or bipolar disorder, most recent type of outpatient treatment (medication or psychotherapy only versus a combination of medication and psychotherapy), and PSS score. Parents’ perceived treatment match at baseline, according to ratings on the SACA (not well or somewhat versus very well), was entered in the second block to determine if this factor contributed to the prediction of treatment retention above and beyond effects of baseline sociodemographic and clinical variables (indicated by a significant change from block 1, according to chi square results). To assess whether treatment match moderated the effect of race, the treatment match × race interaction term was added to the model in a third block.
[A table showing correlations among all independent variables as well as bivariate relationships between the independent variables and treatment retention is available as an online supplement to this article.]
All analyses used SPSS 22; p<.05 was the cutoff for statistical significance.
Results
Two-thirds (N=289, 69%) of the 416 participants who were using outpatient services at baseline and had complete service use data at the 12-month follow-up interview continued to use the services they had used at baseline. A majority of participants were male (N=284, 68%) and white (N=269, 65%), with a mean±SD age at baseline of 9.4±1.9. Children who continued to use their baseline outpatient services were more likely than children who discontinued a baseline service to be white (p<.001) and to have greater functional impairment (p=.023), more diagnoses (p=.003), a bipolar disorder diagnosis (p=.038), and a caregiver who reported more parenting stress (p=.003) at baseline [see online supplement].
The results of logistic regression analyses examining the relationship between participant characteristics at baseline and treatment retention are presented in
Table 1. In both the first and second models, white race (compared with racial minority groups) was associated with greater odds of continued use of baseline outpatient services at 12 months (p<.001). Greater functional impairment was also associated with greater odds of treatment retention, although the odds ratio (OR) was close to 1.0 in both model 1 (p=.040) and model 2 (p=.024). The addition of perceived treatment match in the second model significantly contributed to the prediction of outpatient services retention; thus after controlling for sociodemographic and clinical factors, the analyses indicated that perceived treatment match was significantly related to retention (p=.001). The treatment match × race interaction did not significantly contribute to the model, nor was it significantly associated with retention.
Discussion
Outpatient treatment retention in the LAMS sample was fairly stable between the six- and 12-month follow-up interviews (76% and 70%, respectively) (
13). Preliminary analyses indicated that several clinical factors, including more impaired global functioning, number of diagnoses, bipolar disorder diagnosis, and more parenting stress at baseline, were associated with 12-month treatment retention. Global functioning was associated with the odds of retention (OR=.97) after the analyses controlled for other demographic and clinical variables. The effects of the number of diagnoses, bipolar disorder diagnosis, and parenting stress were nonsignificant after the analyses controlled for other variables, likely because of shared variance among them [see
online supplement] .
Primary analyses suggested that white race and perceived treatment match were independent predictors of 12-month treatment retention. Of the sociodemographic variables, white race alone was significantly associated with greater odds of 12-month treatment retention, even after the analyses controlled for the effects of other sociodemographic characteristics, clinical severity, and perceived treatment match. As such, the effect of race on 12-month treatment retention in this sample cannot be explained by differences in socioeconomic status.
Parents’ perceived treatment match—assessed early in the child’s treatment—also predicted outpatient service retention over 12 months above and beyond several sociodemographic and clinical factors. Results are similar to a finding by Kazdin and colleagues (
11) in reference to a sample of youths being treated for oppositional and aggressive behavior. They found that at the end of treatment, perceived irrelevance of treatment significantly contributed to treatment dropout. The current study adds to this literature by elucidating the relationship between parent perceptions and treatment retention in a more diagnostically heterogeneous group. The effect of perceived treatment match was not moderated by race, indicating that low perceived match affected retention similarly across racial groups. Together with earlier findings from the LAMS sample indicating that living in a two-parent household and parents’ perceived benefit of treatment at baseline predicted retention at six months (
13), the current findings support parent perceptions of treatment as robust, stable predictors of retention. LAMS service retention findings are also consistent with theoretical models that identify individual characteristics and treatment perceptions as predictors of service use and quality (
9,
21).
As noted above, race remained a significant factor in treatment retention after the analyses controlled for treatment match, suggesting that treatment match did not completely explain or moderate racial group differences in retention, although it appeared to be an important influence on retention. Prior work has found differences among racial-ethnic groups in attitudes toward pharmacotherapy and psychotherapy and about whether particular emotions and behaviors among children require professional intervention (
20,
32,
33). In addition, previous research has indicated that parents from racial and ethnic minority groups may be more likely to feel stigmatized by using mental health services (
34) and that youths from minority groups are less likely to receive adequate mental health care (
7), which could increase the likelihood of dropping out of treatment. These factors, which are related to service use but were not included in these analyses, may explain racial-ethnic disparities in retention better than perceived treatment match. Current study results suggest that an improved match between parents’ treatment expectations and the execution of treatment could improve retention across racial groups. Future research should investigate other potential targets for intervention to improve engagement in outpatient services among families from racial minority groups.
These results should be interpreted in light of some limitations. Data regarding children’s service use included parent report and, therefore, were susceptible to potential recall errors. In addition, the study design did not allow examination of other aspects of treatment that might serve as predictors of retention—for example, the treatment modality used by a therapist, the appropriateness of the treatment modality chosen, and the provider’s credentials. It is important to note that, for some youths, discontinuation of an outpatient treatment may have been clinically indicated; as noted above, better global functioning was associated with dropping out of treatment. However, the main analyses controlled for baseline global functioning.
In addition, LAMS primarily enrolled children with ESM and, therefore, may not be representative of all children initiating outpatient services; this sample may have had more severe symptomatology, which could have made them more likely to remain engaged in treatment. Last, although LAMS is a multisite study in which children were recruited from a total of nine outpatient clinics, all sites are located in the Midwest; therefore, results may not be generalizable to children across the United States and elsewhere.
Conclusions
Parents’ perceptions early in treatment and family sociodemographic characteristics were important, robust indicators of whether their children continued to receive mental health services. Previous research has often focused on demographic factors and barriers associated with initiating treatment; current results demonstrate that families who were able to overcome potential barriers to begin treatment may encounter further difficulties that prevent them from fully engaging in treatment. Quality improvement strategies should be developed and implemented to promote treatment match and family-provider communication, including early discussion of treatment plans with parents and goals for incorporating parents into the treatment plan, explanation of the rationale for selected treatment modalities, a choice of treatments when feasible, and mechanisms for assessing and addressing concerns that may prevent families from engaging in treatment.