When is Psychotherapy Brief? Considering Sociodemographic Factors, Problem Complexity, and Problem Type in U.S. Adolescents

Abstract

Objective:

When psychotherapy is brief (1–2 sessions), “early dropout”—defined as premature treatment discontinuation due to financial or structural barriers—is a commonly-assumed cause. However, there are several possible reasons why treatment may be brief, including youth-level factors such as psychopathology complexity or problem type. Better characterizing whether factors beyond financial and structural barriers predict adolescents’ receipt of briefer (versus longer-term) treatment may guide efforts to retain specific youth in longer-term services—and disseminate intentionally-brief interventions to youth potentially positioned to benefit.

Method:

Using data from the 2017 SAMHSA National Survey on Drug Use and Health, we examined whether sociodemographic disadvantage (minority race, low-income, government assistance), perceived problem type, and psychopathology complexity (1 versus multiple problem types) related to psychotherapy length (1–2 versus 3–24+ sessions) among adolescents receiving outpatient psychotherapy (N=1,601; ages 12–17; 60.59% white; 64.50% female).

Results:

Among adolescents beginning outpatient psychotherapy, 23.36% ended treatment after 1–2 sessions. Psychopathology complexity predicted greater likelihood of receiving >2 sessions, after adjusting for specific problem type (χ² = 75.14, p<.001, OR=1.80). Further, although certain problem types (e.g., depression, anxiety, and anger control) were associated with increased likelihood of greater treatment length, these findings did not hold after accounting for psychopathology complexity. No sociodemographic factors significantly predicted treatment length.

Conclusions:

Structural and financial barriers alone may not explain when and why youth psychotherapy is brief. Additional factors, such as psychopathology complexity, may be important and potentially primary contributors to treatment duration among youth who access outpatient services. Future research may examine whether youth with less comorbidity differentially benefit from intentionally-brief interventions, along with strategies for retaining youth who might benefit from longer-term care—such as those with multiple co-occurring problems—in treatment.

Many youth who experience mental health problems do not receive services (Kataoka, Zhang, & Wells, 2002; Konrad, Ellis, Thomas, Holzer, & Morrissey, 2009) with some estimates suggesting that up to 80% of youth with mental health needs go untreated each year (Costello, He, Sampson, Kessler, & Merikangas, 2014). When youth do receive mental health treatment, the mean length of treatment tends to be brief, and many terminate after just a handful of sessions (Harpaz-Rotem, Leslie, & Rosenheck, 2004). Indeed, the modal number of outpatient psychotherapy sessions attended by youth is one (Hoyt, Bobele, Slive, Young, & Talmon, 2018), with an average of approximately 3.9, per national insurance reimbursement data (Harpaz-Rotem et al., 2004). There are several possible reasons why treatment may be brief, including monetary costs to families; the time-intensive nature of most evidence-based treatments; myriad access barriers (e.g., transportation challenges); and, for some, the possibility that brief treatment was sufficient to effect clinical change (see Schleider & Weisz, 2017, for a meta-analysis suggesting that even single-session interventions may reduce youth psychopathology). However, the relative roles of youth- and sociodemographic-level characteristics of adolescents most likely to receive “small doses” of psychotherapy remains largely unexamined. Knowledge of these characteristics is needed to better serve youth most likely to receive just a few sessions of therapy. Such work may inform targeted efforts to disseminate youth-directed supports, including psychotherapy retention efforts for youth at greatest risk of premature termination and intentionally-brief programs that may benefit certain youth seeking care. In the current study, we examine number of outpatient psychotherapy sessions attended among a nationally representative sample of adolescents. Number of sessions attended may reflect early “drop-out” or actual treatment dose; A smaller number of sessions attended is henceforth referred to as brief psychotherapy but is meant to encompass all of these possibilities.

Impacts of Sociodemographic Factors on Treatment Length

Factors found to predict briefer treatment-lengths have included lower socioeconomic status and cumulative environmental adversity (Larson et al., 2013; Rowan, McAlpine, & Blewett, 2013). Youth race and ethnicity may also play a role; African-American and Hispanic adolescents have attended fewer psychotherapy sessions compared to Caucasian youth across a variety of mental health service settings (Elster, Jarosik, VanGeet, & Fleming 2003; Kataoka et al., 2002)—although these patterns may reflect racial and ethnic differences in access to mental health services, not just decisions about continuing care once it is initiated. Nonetheless, ethnic disparities in mental health utilization occur even among those youth who have initiated treatment (e.g., Alegria et al., 2007; Alegria et al., 2008). Hispanic and non-Hispanic black adolescents, relative to Caucasian adolescents, are also significantly less likely to receive frequent, and consistent, therapy services (Merikangas et al., 2011).

Although numerous sociodemographic factors may influence a youth’s duration of psychotherapy, multiple questions remain. First, the studies noted above generally tested the effects of sociodemographic factors (race/ethnicity; socioeconomic status) on treatment length individually rather than simultaneously, rendering each factor’s relative impact on service duration unclear. Second, several studies have conflated treatment access (whether or not a youth begins care) with treatment length (number of sessions attended upon initiating care), obscuring sociodemographic factors’ effects on treatment length specifically among youth who do initially access services. Third, many studies of sociodemographic influences on treatment length have not accounted for potentially relevant youth-level factors—that is, features of a youth’s clinical presentation—on the duration of care (with some exceptions; e.g., Merikangas et al., 2011). Understanding the impacts of such features may inform targeted dissemination of tailored supports—such as intentionally brief interventions or dropout prevention efforts—to youth most likely to discontinue therapy quickly. Two candidate youth-level factors, problem type and psychopathology complexity, are discussed below.

Impacts of Problem Type and Complexity on Treatment Length

There is reason to expect that the presence of co-morbidity (i.e., greater psychopathology complexity), and potentially type of mental health problem, may relate to psychotherapy duration among treatment-seeking youth. With respect to psychopathology complexity, youth presenting with multiple concurrent problems may remain in treatment for longer periods of time, as they may have more needs to be addressed. Indeed, in one nationally representative sample of 6,483 youth ages 13 to 18 years, endorsing more than one problem type predicted greater number of sessions attended, even after accounting for sociodemographic influences on treatment length (Merikangas et al., 2011). In contrast to psychopathology complexity, the role of youth problem type on psychotherapy duration is less clear. To our knowledge, no study has systematically examined whether certain youth problem types receive or require longer treatment, on average, compared to others. In fact, manualized evidence-based protocols for the treatment of youth anxiety, depression, and behavioral difficulties tend to contain a similar number of sessions (typically between 12 and 16; Weisz et al., 2017).

To our knowledge, the impacts of youth problem type and psychopathology complexity on treatment length have not been thoroughly explored among youth who have initiated outpatient psychotherapy. Thus, understanding problem type and complexity status among youth who receive shorter versus longer treatment is an essential next step. It may be that youth with more complex needs tend to stay in treatment longer. Alternatively, youth with more complex psychopathology may be more likely to receive small therapy dosages, either due to doubts that psychotherapy will be effective, dropout due to limited progress early in treatment, or co-occurring financial or access barriers that families with more mental health needs might face. As such, exploring the relative impact of sociodemographic, financial, and problem type factors on treatment length is critical in order to better characterize which youth are more likely to receive brief treatments. This information can then be used to better target intentionally brief intervention (or, if warranted, retention supports) to address client needs within the anticipated treatment length timeframe.

Current Study

Many youth who need mental health treatment do not receive services. Even when they do receive services, the treatment tends to be brief. Shorter-than-expected treatment length has been attributed to a variety of financial, sociodemographic, and access factors, but the relative influences of sociodemographic, financial, and youth-level variables on treatment length in U.S. adolescents remains unknown. Understanding these factors’ respective links to treatment length is critical: If clinicians are guided towards the factors most likely to predict brief treatment duration for new youth clients, they may be better positioned to structure and target the earliest stages of treatment for those clients to fit their immediate needs. For instance, if therapists can reliably identify new clients at high-risk for receiving just a few sessions of care, they may choose to (1) immediately emphasize retention efforts for youth likely to require long-term services, or (2) deliver of intentionally brief interventions for youth positioned to benefit from short-term services (Schleider & Weisz, 2017; Schleider, Dobias, Sung, & Mullarkey, 2019).

As a first-step toward this goal, we examined the association between sociodemographic disadvantage, reasons for receiving services (youth problem type), and psychopathology complexity (number of co-occurring youth problems) on length of outpatient psychotherapy using a nationally-representative sample of treatment-seeking adolescents from the 2017 SAMHSA National Survey on Drug Use and Health (NSDUH; Center for Behavioral Health Statistics and Quality, 2018). All adolescents in this sample had accessed some form of outpatient mental health service within the previous 12 months. We tested whether youth-level factors related to clinical presentation, in addition to sociodemographic factors, might help account for when and for whom youth psychotherapy is brief, defined here as < 3 sessions of outpatient treatment (based on the mean number of sessions U.S. youth attend prior to terminating care—between 3 and 4; Harpaz-Rotem et al., 2004). We hypothesized that both sociodemographic and youth-level factors would predict treatment length, such that greater sociodemographic disadvantage and less psychopathology complexity (i.e., fewer co-morbid problems) would predict receipt of brief treatment for youth. We also predicted that greater psychopathology complexity would be incrementally related to treatment length, such that adolescents reporting greater complexity, or more co-occurring problems, would stay in treatment for longer periods of time. Finally, we explored whether specific problem types (anxiety; anger; depression) or different combinations of problem types (internalizing only; externalizing only; co-occurring internalizing and externalizing) related to briefer versus longer-term treatment. Because evidence-based treatments for various mental health problems are often designed to include similar numbers of sessions (12–16, on average; Weisz et al., 2017), we had no specific predictions as to whether certain presenting problems (depression, anxiety, anger, etc.) or problem type combinations would predict psychotherapy length.

Methods

The National Survey on Drug Use and Health (NSDUH) is an annual survey that assesses mental health, substance use, and treatment-seeking behaviors among individuals in the United States (Center for Behavioral Health Statistics and Quality, 2018). The 2017 dataset examined 68,032 individuals ages 12 and older. Because the NSDUH assesses sensitive information, all identifying factors were removed following data collection, including personal and geographic factors. Additionally, researchers used subsampling to remove a portion of records randomly selected from the data file. Both substitution and subsampling were used to reduce the probability of identifying a person from the data file. For sensitive questions (such as those related to mental health treatment), participants used audio computer-assisted self-interviewing, which gave pre-recorded prompts in which participants listened to questions and types their answers directly onto a computer. All other questions were collected using computer-assisted personal interviewing (CAI) by a field interviewer. In the 2017 survey, there was a weighted screening response rate of 75.08% and a weighted interview response rate for the CAI of 67.12%. Further information and details about the NSDUH can be found in the NSDUH administration manual (Substance Abuse and Mental Health Services Administration, 2018).

Participants

This study included youth ages 12–17 years old from the 2017 public use file (https://www.datafiles.samhsa.gov/study-dataset/national-survey-drug-use-and-health-2017-nsduh-2017-ds0001-nid17939; n = 13,722). We included only those adolescents who reported having accessed outpatient treatment in the prior 12 months, including individual psychotherapy at outpatient community mental health clinics and outpatient psychotherapy. Our final sample included 1,601 youth ages 12–17 years old who received at least some outpatient psychotherapy in the past year. Among this sample, 60.59% were Caucasian, 9.18% were Black, 1.56% Native American, 0.37% were Hawaiian/Pacific Islander, 2.69% were Asian, 6.68% were more than one race, and 18.93% were Hispanic. Thirty-five percent were male. Median total family income was $40,000–49,000 and 23.36% of families participated in government-assisted programming.

Measures

Sociodemographic disadvantage.

Measures of race, income, number of children in the home, insurance gap in the past 12 months, and government assistance were included in this study, consistent with prior work exploring socioeconomic barriers to treatment length (e.g., Larson et al., 2013; Rowan et al., 2013). Note that family income was reported in $10,000 increments (1 = < $10,000, 2 = $10,000-$19,999, 3 = $20,000-$29,999, 4 = $30,000-$39,999, 5 = $40,000-$49,999, 6 = $50,000-$74,999, 7 = ≥ 75,000). Insurance gap was coded dichotomously (1=gap, 0=no gap) and government assistance was coded as participated in one or more government assisted program (1=yes, 0=no). The NSDUH codebook lists other variables included in the NSDUH dataset, but not utilized here (Center for Behavioral Health Statistics and Quality, 2018).

Perceived problem type.

Youth completed a survey regarding mental health service utilization to address behavior and emotional problems not attributed to substance use. Youth reported whether they received outpatient psychotherapy to address any (or multiple) of the following problems in the previous year: depression, anxiety, suicidality, eating problems, anger control, rule-breaking behaviors, and engaging in physical fights. Specifically, youth were asked to respond to separate questions inquiring if they had received outpatient psychotherapy services in the past year because they had thought about or tried to kill themselves, were depressed, were afraid/tense, broke rules or “acted out”, had eating problems, had problems controlling anger, or were in physical fights. Each ‘perceived problem type’ variable was rated on a yes/no (yes = 1, no = 0) scale, with “yes” indicating that the youth did perceive that problem as a reason they received outpatient psychotherapy, and “no,” that they did not perceive that problem as a reason they received outpatient psychotherapy. For secondary analyses, we created a three category variable to explore specific combinations of psychopathology with 0 = least one externalizing problem type (anger control, rule-breaking behaviors, physical fights), and no internalizing problems; 1 = at least one internalizing problem type (depression, suicidality, anxiety, and eating problems), and no externalizing problems; and 3 = at least one externalizing problem and at least one internalizing problem.

Psychopathology complexity.

Complexity of psychopathology (i.e., perceived comorbidity) was assessed using information regarding youth’s perceived problem type(s). We created a binary “complexity” variable based on whether youth selected just one perceived problem type or reason for receiving treatment (reflecting little to no co-morbidity; coded ‘0’) versus more than one perceived problem types or reasons for receiving treatment (reflecting co-morbidity; coded ‘1’).

Treatment length.

Youth self-reported length of outpatient psychotherapy, based on number of outpatient psychotherapy sessions received in the prior year. Youth were asked to indicate how many times an outpatient mental health service was visited in the past year. The total number of visits was calculated by summing number of visits and the original ‘treatment length’ variable was coded based on whether a youth reported receiving 1, 2, 3–6, 7–24, or ≥25 sessions. Here, our primary interest was in characteristics of youth who ended services immediately or shortly after accessing care. Thus, we recoded this variable as 1–2 sessions (0, reflecting ‘brief psychotherapy’) versus 3–24+ sessions (1, reflecting ‘longer psychotherapy’). Because nearly one quarter (22.66%) of the sample received only 1–2 sessions of outpatient services, and given that the mean number of outpatient psychotherapy sessions U.S. youth attend is approximately 3 (Harpaz-Rotem et al., 2004), this division was optimally informative to the study’s objectives. For secondary analyses examining the association between psychopathology complexity and treatment length, we used the original treatment length variable described above (1, 2, 3–6, 7–24, or ≥25 sessions).

Data Analytic Plan

We first ran descriptive analyses to characterize variable distributions and percentages. Table 1 presents percent of endorsement for all socioeconomic and psychotherapy predictors. We then ran two logistic regressions to test the association between the categorical outcome of psychotherapy length (1–2 sessions versus 3–24+ sessions) and (1) socioeconomic disadvantage variables, (2) problem type and psychopathology complexity. In the first model, psychotherapy length was regressed on predictors of income bracket (in intervals of $20,000), number of children in household, government assistance, and ethnicity in order to determine unique effects of the predictors. The second model was conducted as a hierarchical logistic regression in two steps. In Step 1, psychotherapy length was regressed on identified problem types of suicidal behaviors, depression, anxiety, rule-breaking behaviors, eating disorder behaviors, anger control problems, and physical fighting. In Step 2, psychopathology complexity was added to determine (a) whether psychopathology complexity added additional unique variance to the model and (b) whether any specific problem types were associated with treatment length after accounting for the variance added by psychopathology complexity. All analyses were conducted in SPSS 26. To adjust for multiple tests with 16 planned comparisons, Bonferroni adjustments were applied (alpha = .0031, or .05/16).

Table 1.

Response Rates for Mental Health Treatment and Socioeconomic Variables

	Response	Percent Endorsed N(%)1
Depression	Yes	945 (61.40%)
	No	594 (38.60%)
Suicidality	Yes	455 (29.64%)
	No	1080 (70.36%)
Anxiety	Yes	485 (31.60%)
	No	1050 (68.40%)
Anger Control	Yes	170 (11.10%)
	No	1361 (88.90%)
Rule-Breaking Behaviors	Yes	273 (17.80%)
	No	1261 (82.20%)
Eating Problems	Yes	170 (11.10%)
	No	1361 (88.90%)
Physical Fights	Yes	38 (2.49%)
	No	1490 (97.51%)
Psychotherapy Complexity	1 problem	678 (43.83%)
	≥2 problems	869 (56.17%)
Treatment Length	1–2 sessions	374 (23.36%)
	3–24+ sessions	1227 (76.64%)
Received Treatment at a Community Mental Health Center	Yes	497 (31.06%)
	No	1103 (68.94%)
Received Outpatient Therapy from a Therapist	Yes	1472 (92.00%)
	No	128 (8.00)
Received Outpatient Therapy from a Social Worker	Yes	549 (34.29%)
	No	1004 (62.71%)
Participated in government assisted program	Yes	372 (23.24%)
	No	1229 (76.76%)
Number of children <18 years in household	1	558 (34.85%)
	2	578 (36.10%)
	≥3	465 (29.04%)
Insurance Gap (past 12 months)	Yes	58 (4.39%)
	No	1480 (95.61%)
	Caucasian	970 (60.59%)
Race	Black/African American	147 (9.18%)
	Native American/Alaska Native	25 (1.56%)
	Pacific Islander	6 (0.37%)
	Asian	43 (2.69%)
	More than one race	107 (6.68%)
	Hispanic	303 (18.93%)
Total Family Income	Less than $10,000	65 (4.06%)
	$10,000-$19,999	163 (10.18%)
	$20,000-$29,999	155 (9.68%)
	$30,000-$39,999	129 (8.06%)
	$40,000-$49,999	131 (8.18%)
	$50,000-$74,999	222 (13.87%)
	$75,000 or more	736 (45.97%)

¹Percents calculated out of total percent valid

Results

Sample Characteristics

Approximately half of youth in this sample self-reported receiving treatment for 1 problem (43.83%) versus multiple problems (56.17%). Reasons for treatment were depression (61.40%), anxiety (31.60%), suicidality (29.64%), rule breaking behaviors (17.80%), anger control (11.10%), eating problems (11.10%), and physical fighting (2.49%). Full demographic information and rates of service use are reported in Table 1.

Correlates of Brief Versus Longer-term Treatment

Sociodemographic disadvantage.

We first tested whether sociodemographic disadvantage predicted treatment length (1–2 sessions versus 3 or more sessions; Table 2, Model 1). No significant (p < .0031) effects emerged, suggesting that no sociodemographic factor was significantly associated with treatment length in this sample.

Table 2.

Results of Logistic Regressions Testing Psychopathology Complexity, Problem Types, and Sociodemographic Variables as Predictors of Treatment Length

	B	SE	Wald Statistic	Odds Ratio	χ2
Model 1					24.93#
Participated in government assisted program	−.06	.18	.12	.94
No. of children at home	−.09	.08	1.28	.92
Insurance gap (past 12 months)	.09	.29	.11	1.10
Caucasian	1.01	.39	6.76	2.76
Black	.84	.42	4.03	2.32
Multiracial	.93	.45	4.39	2.55
Hispanic	1.03	.40	6.60	2.81
Asian	.30	.50	.37	1.36
Family income	.08	.04	4.00	1.08
Model 2					75.14**
Step 1
Depression Treatment	.47	.14	12.07	1.60*
Suicidality Treatment	.48	.17	8.45	1.62
Anxiety Treatment	.49	.15	10.72	1.64*
Anger Control Treatment	.95	.28	11.37	2.58*
Rule-Breaking Behaviors Treatment	.04	.17	.06	1.04
Eating Problems Treatment	.35	.25	1.90	1.42
Physical Fights Treatment	−.54	.49	1.21	.58
Step 2
Depression Treatment	.27	.15	3.32	1.32
Suicidality Treatment	.28	.18	3.36	1.32
Anxiety Treatment	.30	.16	3.39	1.35
Anger Control Treatment	.72	.29	6.27	2.07
Rule-Breaking Behaviors Treatment	−.06	.18	.13	.94
Eating Problems Treatment	.24	.26	.90	1.28
Physical Fights Treatment	−.62	.49	1.60	.54
Psychopathology Complexity	.59	.18	10.53	1.80*

NB: with Bonferroni corrections, significance level set to α = .003

^#p<.005,

^*p<.003,

^**p<.001

Psychopathology complexity and problem type.

We then tested whether psychopathology complexity and problem type predicted treatment length (1–2 versus ≥3 sessions; Table 2, Model 2). In Step 1, receiving treatment for depression (B = .47, p = .001, OR=1.60), anxiety (B = .49, p = .001, OR = 1.64), and anger control (B = .95, p = .001, OR = 2.58) predicted significantly increased likelihood of staying in treatment for ≥3 sessions. Specifically, 75% of youth seeking treatment for depression, 75% of youth seeking treatment for anxiety, and 83% of youth seeking treatment for anger control issues received ≥3 sessions of outpatient services. In Step 2, psychopathology complexity (number of co-occurring problems) was added to the model. Greater psychopathology complexity predicted significantly increased likelihood of receiving ≥3 sessions of treatment, as opposed to 1–2 sessions (χ² = 75.14, p < .001, OR = 1.80). Compared to youth who endorsed only 1 problem, youth who endorsed multiple problems were 1.80 times more likely to report longer treatment durations. Specifically, 76% of youth with multiple problems, versus 64% of youth with 1 problem, stayed in treatment for 3 or more sessions. Receiving treatment for depression (B = .27, p = .07, OR=1.32), anxiety (B = .30, p = .12, OR = 1.32), and anger control (B = .72, p = .01, OR = 2.07) were not significantly related to treatment length after accounting for psychopathology complexity and applying statistical corrections.

Secondary sensitivity analyses: Relation between psychopathology complexity and treatment length.

To better parse the link between psychopathology complexity and treatment length, and to assess for the potential presence of a ‘dose-response’ effect (whereby greater complexity is related to longer treatment length in a step-wise fashion), we conducted a univariate analysis of variance to examine how number of mental health problems related to total number of psychotherapy sessions. We specified the outcome as total number of problem types endorsed (continuous); the factor was the four-category treatment length variable (1–2 sessions, 3–6 sessions, 7–24 sessions, and ≥25 sessions; this was the finest-grained division of treatment length available in the dataset). The variance in total number of sessions explained by number of mental health problems was incrementally stronger across the categories of treatment length (F = 47.76, p <.001), suggesting a potential ‘dose-response’ effect. Bonferroni-corrected pairwise comparisons indicated the youth who received 7–24 sessions reported more problem types than youth who received 3–6 sessions (group mean difference = .40, p <.001) and fewer problem types than youth who received ≥25 sessions (mean difference = −.67, p <.001). Youth who received 3–6 sessions reported more problem types those who received 1–2 sessions (mean difference = .35), although this difference fell short of statistical significance (p = .02).

Secondary exploratory analysis: Relation of comorbidity patterns to treatment length.

To further examine how psychopathology complexity might relate to treatment length, we conducted a secondary, exploratory univariate analysis of variance to examine whether and how specific problem comorbidities (internalizing-only, externalizing-only, or simultaneous internalizing and externalizing) might relate differently to treatment length. As stated previously, we had no specific hypothesis as to patterns of results. The outcome variable was treatment length (1–2 sessions verses ≥ 3 sessions); the factor was the three category problem combination (externalizing only, internalizing only, both externalizing or internalizing). Youth who endorsed at least one externalizing problem were more likely to end treatment after 1–2 sessions compared to youth who endorsed both externalizing and internalizing (p <.001).

Discussion

Psychopathology complexity, but neither problem type nor sociodemographic disadvantage, related to duration of outpatient mental health treatment in a nationally-representative sample of adolescents receiving outpatient services. These results are consistent with past research suggesting that psychiatric comorbidity (an index of psychopathology complexity) is associated with greater service utilization and number of psychotherapy sessions attended (e.g., Merikangas et al., 2011). However, results contrast with prior work suggesting that sociodemographic factors such as family income, youth race/ethnicity, and participation in government-assisted programming (indicating low-income status) were unrelated to treatment length among adolescents who accessed care. Given the large size and diversity of the present sample, along with our consideration of financial, sociodemographic, and clinical contributors to service duration, our results provide strong evidence that adolescents’ early treatment termination may not be attributable to logistical and financial barriers alone. Rather, clinical presentation may play an important role: Here, youth with a larger number of total co-occurring problems, as well as youth with both internalizing and externalizing problems versus externalizing problems alone (but not internalizing problems alone), tended to receive more sessions of therapy. In contrast, youth with fewer presenting problems, and those presenting with externalizing problems only, were more likely to end treatment after 1 to 2 sessions. Secondary sensitivity analyses suggested that number of youth problems showed an approximate “dose response” association with treatment length, adding nuance to the result that youth with greater psychopathology complexity may stay in treatment for longer periods of time.

Present results appear to contrast findings from previous work that have shown that sociodemographic factors suggesting that financial and structural barriers may impede participation in treatment (e.g., Kazdin, Holland, & Crowley., 1997). One possible explanation is that youth in this sample had access to treatment and were already actively engaged in psychotherapy. Logistical and financial barriers may be more salient barriers in initially accessing care—and less to persisting with services once they are initiated. Sociodemographic correlates of low service utilization include availability of health insurance, income, and family structure (Cunningham & Freiman, 1996; Larson et al., 2013; Rowan et al., 2013). However, our results suggest that, among those who do successfully initiate outpatient treatment, those same logistical and financial barriers may not impact treatment length.

It is notable that specific adolescent problem types did not predict treatment length above and beyond psychotherapy complexity. Certainly, this pattern may not apply to all individual youth, as some problem presentations may call for lengthier treatment than others: As one example, some research suggests that aggressive, impulsive, and disruptive behavioral problems in youth largely drive recurrent, long-term mental health care utilization (Merikangas et al., 2011). In this sample, however, youth endorsing both internalizing and externalizing problems stayed in treatment for longer compared to those with externalizing problems alone. One possible explanation is that intentionally-brief interventions have shown larger mean effect sizes for reducing externalizing and conduct problems (meta-analytic g = .52 for single-session interventions) than for depression symptoms (meta-analytic g = .21 for single-session interventions; Schleider & Weisz, 2017). Further, treatments for youth externalizing problems often involve caregivers, sometimes as primary participants in therapy. However, caregivers are not always available or open to participating in behavioral parent training interventions, which may lead to premature treatment dropout when externalizing problems are the main or only target of therapy (e.g., Chacko et al., 2016). Taken together, it appears that the complexity of the problem may drive treatment length, rather specific type of mental health concern.

Additionally, we were unable in this study to account for the severity of presenting problems; it remains possible that youth with just one problem type of considerable intensity or severity might need equally lengthy treatment compared to peers with multiple presenting problems. Nonetheless, present results do suggest—although certain problem types may be associated with treatment length under some circumstances, which we could not test in this study—youth presenting with a greater number of clinical problem types are likely to engage in longer-term treatment, perhaps due a greater number of sessions needed to address all challenges and needs.

Of note, 22% of adolescents who reported more than one presenting problem did end treatment after 1–2 sessions, regardless of sociodemographic disadvantage. Due to limitations inherent in the present dataset, we were unable to assess why individual youth ended treatment prematurely; although some may have terminated treatment before symptom reductions were achieved, it is possible that some subset of these youth actually benefited from briefer durations of treatment (Schleider & Weisz, 2017; Schleider et al., 2019). Emerging research suggests that, for some, very brief and even single-session interventions might help reduce youth psychological distress of multiple types (e.g., Schleider & Weisz, 2018; Schleider & Weisz, 2017; Schleider et al., 2019). As such, it remains possible, and worthy of formal investigation, that some youth with co-occurring mental health needs may still benefit from brief forms of treatment—and, if so, whether it might be possible to predict which youth are best-positioned to benefit from these intentionally brief interventions, and administer these interventions accordingly. Another possibility, also worthy of future exploration, is that youth who report less psychopathology complexity (i.e., a smaller number of total presenting problems) may be especially well-positioned to benefit from intentionally brief interventions; present findings suggest that these youth were especially likely to receive just 1–2 sessions of therapy. That is, perhaps brief treatments carry greater potential to confer noticeable benefits when presenting problems are fewer and more specific. Although we cannot test these possibilities directly within the current dataset (which does not include data on youth’s responses to outpatient therapy), our results suggest the direct clinical utility of these and related next-steps.

The clinical reality that adolescent psychotherapy is often briefer-than-planned is often assumed to stem from logistical or financial barriers to accessing care, which certainly contribute in some cases (Larson et al., 2013; Owens et al., 2002). However, our results suggest additional possibilities: For some, duration of treatment may be associated with the complexity of presenting problems, with youth endorsing a larger number of problem types being more likely to stay in treatment for more sessions, even after accounting for specific problem types and structural and financial barriers. Of note, there are several other reasons why youth may end treatment after only 1 or 2 sessions, beyond psychopathology complexity. These include a mismatch between therapist and patient; patient dissatisfaction with services; the possibility that brief services might be sufficient, in some cases, for clinical benefit; or referrals to alternative or more intensive services. For example, youths’ expectations about treatment—including length (i.e., how long treatment should be) or modality (i.e., how treatment should be delivered)— may have influenced treatment duration and engagement. In fact, mismatch between treatment expectations and treatment actually received is associated with premature termination (Nock & Kazdin, 2001). Our study cannot conclude reasons for early termination; nonetheless, findings do suggest that financial and sociodemographic factors are neither solely nor primarily responsible for when and why psychotherapy is brief for U.S. adolescents. Given the prevalence of unplanned, “low-dose” psychotherapy in treatment settings across the country, future work unpacking its causes and implications may have direct clinical utility.

Limitations and Future Directions

Our study has certain methodological limitations that warrant attention and may suggest directions for future research. First, the cross-sectional nature of the study design precludes drawing causal implications for the relation between personal and structural barriers and treatment length. It is possible that youth may over- or under-estimate the number of psychotherapy sessions they attended over the past year. For example, youth who had positive sessions and were engaged in therapy may be more likely to accurately remember their experiences. In fact, children often have more coherent and personally significant memories of emotionally-salient (both positive and negative) events (e.g., Greenhoot, Sun, Bunnell, & Lindboe, 2013). Nonetheless, in the current study, we were not able to objectively assess number of sessions attended (e.g., via health records). Second, all data is a retrospective self-report, introducing possible informant bias, potentially inexact estimates of treatment length, and questionable validity of the treatment utilization markers. Third, all youth in this sample were over the age of 12 years old; patterns of treatment utilization and prevalence of mental health problems may differ for younger children. Further, there are many other factors that may influence treatment length, including beliefs (from both parent and child) about psychotherapy. Parent beliefs about psychotherapy can directly shape children’s access to and retention in mental health treatment, as parents are often responsible for initiating services for their children. However, the NSDUH database does not include questions on caregivers’ beliefs about psychotherapy. Separately, we had no index of adolescents’ perceptions of whether (and to what degree) outpatient treatment benefited them; such information will be important to collect in future studies and may provide more nuanced evidence regarding the utility of briefer interventions for youth with diverse clinical problems. It is also important to note that type of therapy received may influence likelihood of ending treatment after only 1 or 2 sessions. However, we did not have data on type of therapy received (e.g., CBT, behavioral therapy) and thus we were not able to explore this possibility. Finally, we were unable to distinguish between the number of times youth may have started or “restarted” therapy. It is possible that youth receiving multiple treatment sessions did so across multiple episodes of care and possibly addressing different factors with different treatment plans. Despite these limitations, our study has considerable strengths. The large, nationally-representative adolescent sample, stringent statistical corrections, and consistency of the overall pattern of results lend credence to the observed effects. Similar work may help to guide future dissemination efforts to youth most likely to benefit from services.

Conclusion

Overall, many youth who begin outpatient therapy do not complete what is commonly considered to be a “full course” of treatment. Thus, brief psychotherapy is exceptionally common in real-world clinical settings. The cause of brief treatment is often attributed to logistical, structural, or financial barriers that impede continuation of care. However, this study provides preliminary support that certain youth-level factors—namely, psychopathology complexity—may be important and even primary contributors to treatment duration among adolescents who access services. Given that psychopathology complexity appears to impact on adolescent treatment length, identifying whether intentionally-brief, targeted interventions can spur initial symptom reductions in multi-problem youth—thereby alleviating immediate distress and potentially improving motivation to continue treatment—may hold great practical value.