Primary Outcomes for Adults Receiving the Unified Protocol after Hurricane Harvey in an Integrated Healthcare Setting

Abstract

The Unified Protocol for Transdiagnostic Treatment of Emotional Disorders (UP) has demonstrated efficacy for treating anxiety and depression. However, there are limited effectiveness data when conducted in real-world settings with diverse populations, including those with trauma. We evaluated treatment outcomes in a naturalistic, community setting among 279 adults who received UP following Hurricane Harvey. We examined change in overall clinical severity, depression and anxiety symptoms, functional impairment, and baseline outcome predictors (i.e., demographic characteristics, impact from Hurricane Harvey, co-occurrence of depression and anxiety symptoms). Global clinical severity, depression and anxiety symptoms, and functional impairment decreased by end-of-treatment. Participants experienced global symptom improvement to a lesser degree than demonstrated in efficacy trials. Participants who experienced greater storm impact reported larger reductions in anxiety symptoms than those less impacted by Harvey. Further studies evaluating the effectiveness of the UP post-disaster and with diverse samples are needed.

Introduction

Depressive and anxiety disorders are prevalent in adults and associated with significant distress and disability (Hanson & Young, 2017; Hendriks et al., 2014; Twenge & Joiner, 2020). Comorbidity among anxiety and depressive disorders is common and often associated with attenuated treatment outcomes and sustained illness chronicity (Almeida et al., 2012; Lamers et al., 2011).

First-line interventions for depression and anxiety include various cognitive-behavioral therapy (CBT) protocols and serotonin reuptake inhibitors (SRIs; Bandelow et al., 2017; Bauer et al., 2007; Watts et al., 2015). However, depression and anxiety comorbidity may complicate treatment delivery and outcomes (Almeida et al., 2012; Fava et al., 2008). CBT is typically delivered through single disorder protocols (SDPs), which can have disadvantages in the presence of substantial comorbidity. SDPs require clinicians to receive specialized training in multiple intervention protocols (e.g. Salomonsson et al., 2018), which may be time-consuming, costly, and difficult to obtain (Karlin & Cross, 2014; McHugh & Barlow, 2010), thereby limiting treatment access (Barlow et al., 2020). Using SDPs to address comorbid emotional disorders also requires clinical decision-making regarding selection and sequencing of multiple SDPs, and clear guidelines to inform these decisions are lacking. Thus, there is a need for more efficient treatment delivery strategies that both address underlying mechanisms associated with depressive and anxiety disorders (Sauer-Zavala et al., 2017) and alleviate burdens on mental health clinicians.

The Unified Protocol for Transdiagnostic Treatment of Emotional Disorders (UP; Barlow et al., 2017b) is a transdiagnostic, modular intervention that addressessome of the limitations of SDPs by addressing core psychopathological dysfunctions underlying emotional disorders (Ellard et al., 2010). Using terminology from the UP, an emotional disorder is characterized by the experience of intense negative emotions, negative evaluation of these emotions, and accompanying experiential avoidance. Depressive and anxiety disorders are the conditions most commonly designated as emotional disorders, though the term can also include diagnoses such as post-traumatic stress disorder (PTSD) and obsessive–compulsive disorder (OCD; Bullis et al., 2019). The UP primarily targets the temperamental vulnerability of neuroticism, which is defined by the presence of frequent and/or intense negative affect and an aversive response to the experience of heightened emotional states, often resulting in avoidant coping (Barlow et al., 2013). The UP may offer an efficient and effective way to treat multiple diagnoses and presenting problems and reduce the burden on clinicians who would otherwise have to attain competency in multiple SDPs.

In clinical trials, the UP has shown promise as an intervention for addressing both depressive and anxiety disorders (Barlow et al., 2017a; Farchione et al., 2012; Sakiris & Berle, 2019; Steele et al., 2018). A randomized controlled trial (RCT) found that individuals who received the UP demonstrated a similar treatment response at 6-month follow up compared to those treated with a SDP for anxiety (no longer met principal disorder criteria: 71% UP vs 63% anxiety SDP; no longer met criteria for any emotional disorders: 62% UP vs. 47% SDP). Individuals who received the UP were more likely to complete treatment (with 88% completing in the UP condition and 69% completing in the SDP condition; Barlow et al., 2017a). Using these same data, Steele and colleagues (2018) also examined the effectiveness of the UP in treating co-occurring disorders. The mean number of emotional disorder diagnoses decreased significantly from baseline to post-treatment and baseline to 12-month follow-up, though the UP and SDP conditions did not significantly differ in their effect on comorbid diagnoses (Steele et al., 2018).

Though a recent meta-analysis suggested that the UP was efficacious in a range of outpatient, hospital, and residential settings, effectiveness research with majority Hispanic samples is limited. Additional data are needed to understand how the UP fares in a variety of naturalistic clinical environments with diverse patient populations and presenting problems (Cassiello-Robbins et al., 2020). Community settings offer unique obstacles that may not arise in a controlled research context, and these challenges may influence treatment outcomes—making practice-based research critical for successful implementation (Westfall et al., 2007). One such obstacle is reduced session attendance in individuals seeking psychosocial care in community health clinics. For example, in a database of 6000 individuals, the average number of sessions attended was fewer than five (Hansen et al., 2002), and a study on CBT for anxiety disorders in a community health clinic also indicated that on average, patients attended fewer than five visits over a six-month period, with a median of three sessions (Wolitzky-Taylor et al., 2015). This is in contrast to controlled RCT efficacy studies, which typically involve 12 to 18 completed sessions over a specified period of time with ongoing quality assurance and fidelity checks (Cassiello-Robbins et al., 2020). There are also other practical obstacles that arise in real-world contexts, such as a lack of clinician expertise in evidence-based therapies, less structured and thorough assessments, and a higher patient volume in fast-paced clinical environments that may leave less time to ensure protocol adherence. Significant adaptions to the UP may be required to fit the needs of individuals treated in community settings in order to accommodate the lower number of treatment sessions that patients in these clinics are often able to attend, perhaps in the form of a manualized abbreviated protocol that is designed to be completed in 3–5 treatment sessions or additional materials for patient review outside of treatment sessions. Research into the UP’s effectiveness in demographically diverse populations is needed as well; thus far, the UP has primarily been examined in populations that are predominantly white and female (Cassiello-Robbins et al., 2020). The effectiveness of the UP in primarily Hispanic samples is unknown.

Research on the UP for the treatment of trauma-exposed populations is in the early stages. Some studies have shown limited effectiveness (e.g., Varkovitzky et al., 2018), though case studies have been promising and a small randomized controlled trial indicated that the UP may be efficacious for PTSD (Gallagher, 2017; O'Donnell et al., 2021). A recent study examining the UP in a community sample of 50 participants with trauma histories demonstrated a significant decrease in symptoms of anxiety and depression on average, and the UP was associated with similar rates of improvement in participants with past trauma exposure as in those without, even though trauma-exposed patients presented with higher baseline levels of anxiety and depression (Hood et al., 2021). The UP could be useful for trauma-exposed populations because these patients may have both higher levels of anxiety and depression and increased barriers to accessing care. To date, the UP has also not been investigated for symptoms that arise or exacerbate in the aftermath of large-scale crises such as natural disasters, where a flexible intervention could be particularly suitable.

In the present study, we examine treatment outcomes among a diverse sample of adults who received the UP in a naturalistic community health setting in the aftermath of Hurricane Harvey. The storm struck Houston in August 2017 and became the second most financially damaging hurricane in United States history, causing approximately $125 billion worth of damage (Blake & Zelinsky, 2018). In a December 2017 survey conducted by the Kaiser Family Foundation and the Episcopal Health Foundation, 18% of those affected by Hurricane Harvey reported that their mental health had declined since the storm, while 32% of the population affected by the storm felt they had suffered one of a range of negative mental health-related effects. Studies on psychological impacts in the aftermath of major flooding also indicate that mental health can be negatively affected by storms similar to Harvey (Orengo-Aguayo et al., 2019; Picou & Hudson, 2010; Stanke et al., 2012).

We had three primary goals for the present study: (1) to examine outcomes for adults receiving the UP in terms of change in overall clinical severity, change in anxiety and depressive symptoms, and improvement in idiographic “top problems” identified collaboratively by the client and therapist at the start of treatment; (2) to explore factors associated with improvement, such as demographic characteristics and Hurricane Harvey impact; and (3) to examine differential treatment response as a function of baseline symptoms and the presence of co-occurring elevated anxiety and depressive symptoms. Based on prior studies, we predicted that the UP would reduce clinical symptoms and severity and improve global functioning when provided in a naturalistic community health setting (Cassiello-Robbins et al., 2020) for a predominately Hispanic sample.

Methods

Participants

A total of 279 adults in the greater Houston area, who were emotionally impacted by Hurricane Harvey, received the UP. Table 1 presents demographics and baseline symptoms and includes means with standard deviations and frequencies to describe the sample. Participants were on average 46.2 years old (SD = 13.3) and the majority (75.3%) were female. The sample was racially and ethnically diverse: 59.6% were Hispanic/Latino, 25.3% white, 14.7% Black/African American, and 0.4% other. Participants’ primary language was either English or Spanish. 67% spoke English, 31% spoke Spanish, and 2% spoke another language (interpretation services were provided if needed). The majority of participants lived with other people and a majority were unemployed. The extent to which they were affected by Harvey ranged from experiencing the aftermath of the storm broadly to significant impacts from the hurricane (e.g., loss of housing, loss of a relative, trauma resulting from enduring the hurricane). In the present sample, 267 adults had complete data on Harvey impact (n = 11 had missing data, n = 1 did not endure the storm): 88.8% endured the storm (meaning that the participant was in the area and experienced the hurricane; n = 237), 6.7% were impacted (indicating that the participant was strongly or directly affected by the hurricane; n = 18), 1.1% were impacted with exacerbation of symptions (participant was impacted by the storm, with clear exacerbation of existing mental health symptoms; n = 3), 0.7% endured the aftermath of the storm only (meaning that the participant either left the Houston area to avoid the hurricane and returned in the aftermath, or moved to Houston immediately following the storm and experienced the aftermath; n = 2), and 2.6% provided a custom response (n = 7). Please see Table 2 for a full list of the specific storm impacts participants endorsed.

Table 1.

Demographics and baseline symptoms and clinical severity of treatment-seeking adults (n = 279)

	M ± SD or n(%)	N
Age	46.2 ± 13.3	277
Gender		279
Female	210 (75.3%)
Male	69 (24.7%)
Race/ethnicity		265
Hispanic	158 (59.6%)
White	67 (25.3%)
Black	39 (14.7%)
Other	1 (0.4%)
Employed, %	85 (33.5%)	254
Living Situation, % Living Alone	38 (14.2%)	268
Hurricane Harvey Impact Score	1.5 ± 1.1	208
Baseline PHQ-9	14.4 ± 6.0	278
Baseline GAD-7	8.0 ± 4.7	254
Baseline GAF	55.8 ± 3.0	269
Baseline Global Severity Scale	3.0 ± 0.3	268
Baseline Patient Top Problems Rating	8.2 ± 1.7	99
Baseline Clinician Top Problems Rating	8.4 ± 1.2	99

PHQ-9 Patient Health Questionnaire, GAD-7 Generalized Anxiety Disorder Screener, GAF Global Assessment of Functioning

Table 2.

Hurricane Harvey impacts

	n (%)	N
Did you get hurt during or right after Hurricane Harvey?	0 (0.0)	208
Were you scared someone in your family might get hurt or die due to Hurricane Harvey?	23 (11.1)	208
At any time during the hurricane, did you think you might die?	12 (5.8)	208
Did a family member or someone close to you get injured due to Hurricane Harvey?	1 (0.5)	208
Did a family member or someone close to you die due to Hurricane Harvey?	3 (1.4)	208
Did you know someone who was injured due to Hurricane Harvey?	0 (0.0)	208
Did you lose a job as a result of Hurricane Harvey?	5 (2.4)	208
Did you experience a loss of income as a result of Hurricane Harvey?	21 (10.1)	208
Have you had financial issues as a result of Hurricane Harvey?	20 (9.6)	208
Have you received mental health care as a result of Hurricane Harvey?	1 (0.5)	208
Have you had health issues as a result of Hurricane Harvey?	1 (0.5)	207
Did you experience a loss of social support as a result of Hurricane Harvey?	1 (0.5)	208
Was your home destroyed or damaged as a result of Hurricane Harvey?	33 (16.3)	203
Did you have to go through flood waters to escape?	7 (3.4)	207
Were you stuck in your home as a result of Hurricane Harvey?	176 (85.0)	207
Were you evacuated from your home as a result of Hurricane Harvey?	15 (7.2)	208

Procedure

Eligibility and Treatment

Hurricane Harvey struck Houston in August of 2017. Patients were eligible for Project Reach (a program that aimed to provide mental health treatment after Hurricane Harvey) if they were seeking mental health care at selected Harris Health Clinics between July 2018 and February 2020 and were emotionally affected by Hurricane Harvey. Whether a patient was emotionally impacted was determined through an unstructured conversation with the Project Reach intake coordinator; patients were invited to receive treatment through Project Reach if they endorsed emotional impact from the storm. Project Reach patients received free therapy and/or psychiatric care as needed. Coded data from Project Reach patients who received the UP was collected and analyzed for this study if the participant was between the ages of 6 and 88 years and had completed the Project Reach standard of care measures. Treatment included an initial intake assessment with a licensed master’s level clinician (first session) and subsequent sessions of the UP (session two onward). Treatment sessions occurred in person and were scheduled weekly or bi-weekly for a duration of 30 min. The UP principles were applied flexibly to address the clients’ needs. This flexibility included the shorter treatment sessions (30 min each); irregular spacing between sessions when required to fit the participant’s schedule; and highlighting the core components of the UP while focusing session content, in order to accommodate potentially fewer treatment sessions. The number of treatment sessions in this sample ranged from 2–21 (M = 5.22, SD = 4.76) and sessions occurred every 2–3 weeks, on average. As expected, given the community mental health clinic environment, the number of completed treatment sessions was lower than the 12–18 sessions that are typically provided in an efficacy study of the UP: 33.5% completed 2 sessions, 36.4% completed 3–5 sessions, 18.8% completed 6–10 sessions, and 11.3% completed greater than 10 sessions.

Intervention

Master’s level therapists (LCSW and LPC) were trained to implement the UP (Barlow et al., 2017b) via a 1.5-day training led by one of the developers (Jill Ehrenreich-May), which included reading materials and didactics. Therapists also received supervision and weekly case consultation from UP certified experts. Because it was anticipated that most participants would not be able to attend a full 12–18 sessions (Wolitzky-Taylor et al., 2015), therapy included UP modules, concepts, and skills flexibly applied to meet the client’s particular needs in the context of their availability. Treatment goals were established through the identification of top problems using a modified version of the Youth Top Problems scale, a collaborative effort between the therapist and the client that allows for the measurement of client progress throughout treatment on a 0–10 subjective rating scale (Weisz et al., 2011). Therapists assigned homework and reviewed homework completion at the following session. Harris Health System interpreters were used for patients who did not speak English, and all therapy materials were available in English and Spanish.¹

Measures

Therapists administered assessment measures to participants during each intake and treatment session. The Hurricane Harvey Impact questionnaire was delivered at intake, and the Global Assessment of Functioning, Patient Health Questionnaire (PHQ-9), General Anxiety Disorder Screener (GAD-7), Clinical Global Impressions Scales, and Top Problems Scale (all described below) were delivered at every session. Measures included the following:

Clinician Report Measures

Global Assessment of Functioning (GAF)

The GAF is an overall measure of psychological functioning that that is rated by a clinician on a scale of 0–100, where 100 indicates the highest level of functioning. It has demonstrated validity as a measure of overall mental health in adults with anxiety (Schwartz & Prete-Brown, 2003) and depressive disorders, among others (Pedersen & Karterud, 2012).

Global Severity Scale

A 0–7 rating scale adapted from the Yale-Brown Obsessive–Compulsive Scale (Y-BOCS) Clinical Global Impression—Obsessive Compulsive Scale (CGI-OCS), which is based on the Clinical Global Impression–Severity Scale (CGI-Severity; Goodman et al., 1989; Guy, 1976). The Global Severity Scale included the following anchors: 0 = no illness; 1 = illness slight, doubtful, transient; 2 = mild symptoms; 3 = moderate symptoms; 4 = moderate/severe symptoms; 5 = severe symptoms; 6 = extremely severe symptoms; and 7 = free of symptoms/in remission.

Clinical Global Impression Scale–Improvement

The Clinical Global Impression (CGI) Scales include a 1–7 rating scale of symptom improvement rated by clinician, where 1 = very much improved; 2 = much improved; 3 = minimally improved; 4 = no change; 5 = minimally worse; 6 = much worse; and 7 = very much worse (Guy, 1976). The CGI–Improvement has been validated as a measure of clinical outcomes in routine clinical use, including for depressive and anxiety disorders (Berk et al., 2008).

Patient Self-report measures

Hurricane Harvey Impact Questionnaire–Adult

The Hurricane Harvey Impact Questionnaire includes 18 questions developed by the research team. The first question determines the type of impact, with six options: “endured = was in the area for and experienced the hurricane”; “endured aftermath only = left the area for the hurricane and returned to aftermath, or moved to Houston immediately following hurricane and experienced aftermath”; “impacted = strongly or directly affected by the hurricane”; “impacted with exacerbation of symptoms = impacted by hurricane, with clear exacerbation of existing mental health symptoms”; “impacted with onset of new symptoms = impacted by hurricane, with new mental health symptoms”; “did not endure = was not present for hurricane or aftermath”; and “custom fill-in,” where individual circumstances could be noted. The following 17 yes/no questions elicited various types of hurricane impact, such as “Did a family member or someone close to you die due to Harvey?” and “Was your home destroyed or damaged as a result of Harvey?” A full list of these questions can be found in Table 2. The Hurricane Harvey Impact score was calculated by adding one point for each “yes” response on these items. Cronbach’s alpha was 0.43.

PHQ-9

The PHQ-9 is a nine-item measure that assesses the presence and severity of depressive symptoms over the past two weeks. Each item is rated on a 0–3 scale (where 3 indicates highest severity), for a total severity rating of 0–27. It has been shown to be a valid measure of depression severity (Kroenke et al., 2001). Cronbach’s alpha ranged from 0.84-0.90 across repeated administrations of the measure.

GAD-7

The GAD-7 is a seven-item questionnaire for assessing generalized anxiety disorder symptom presence and severity that has demonstrated validity (Spitzer et al., 2006). Each item is rated on a scale of 0–3 (where 3 indicates highest severity), with a total score of 0–21. Cronbach’s alpha ranged from 0.80–0.92 across administrations of the GAD-7.

Clinician and self-report measures

Top Problems Scale

The Top Problems Scale is a subjective 0–10 rating scale for the client’s top concerns, where 0 indicates no problem and 10 indicates maximum problem severity. Top Problems are identified through collaboration between therapist and client at intake and rated at every treatment session. In the case of disagreement between therapist and client’s selection of top problems, the client’s top problems were chosen. The top problems scale used in this study was based on the Youth Top Problems measure, a similar 0–10 rating scale (0 = “not at all” and 10 = “very, very much”; Weisz et al., 2011). A minimum of three top problems were identified for the current study. Each top problem was rated by the clinician and the client separately.

Data collection

Data were collected by clinical staff and entered into the patients’ electronic medical records. Staff reviewed these records and entered de-identified data into a secure database. Unique alpha-numeric identifiers were created for each participant to protect their identity. Entered data were reviewed for accuracy by a second staff member. The research team obtained approval to collect de-identified data from the Baylor College of Medicine and Harris Health System institutional review boards with a waiver of consent.

Statistical Analysis Plan

Descriptive statistics are reported as means (with standard deviations) and frequencies and used to describe our treatment-seeking sample. We first examined associations between the number of treatment sessions completed and demographic variables and baseline symptoms using t-tests or ANOVA for group comparisons (i.e., sex, race/ethnicity) and Pearson’s r for continuous variables (i.e., age, Harvey Impact Score, baseline PHQ-9, GAD-7, GAF, clinician/patient rated top problems). We then used latent growth curve models within a structural equation-modeling framework to examine repeated measures as observed indicators of growth over time. Latent variables reflect the baseline status (intercept). Rate of change (linear or quadratic) was used to describe trajectories and model inter-individual variations in the trajectory components. Two models were estimated. Model 1 was unconditional, contained no predictor variables, and examined the following outcome variables: PHQ9, GAD7, GAF, Global Severity Scale, CGI-Improvement, and clinician and patient-rated top problems. Model fit indices were used to select the best form of longitudinal change and examine change over time in clinical symptoms, severity, and improvement in specific areas of functioning. Of note, models incorporating a quadratic term did not converge and linear models incorporating both the intercept and slope growth factors are reported herein for the outcome variables.

Model 2, the conditional model, examined the impact of covariates on the intercept and growth factors. The covariates included age, sex, dummy-coded race/ethnicity (Hispanic as reference category), Hurricane Harvey Impact Score, therapist, employment status (employed or not), and living arrangement (lives with others or alone). Baseline anxiety and depression symptom level was included in all models, except for those examining PHQ-9 and GAD-7 as outcome variables. To this end, patients were categorized by whether they evidenced elevations or not on the PHQ-9 and GAD-7 (GAD-7 scores of 6 or above and PHQ-9 scores of 10 or higher were designated as elevated); these groups included: Both Elevated Anxiety and Elevated Depression (reference category), Elevated Anxiety Only, Elevated Depression Only, and No Elevations in Anxiety/Depression). Such models were used to examine differential improvement over the course of treatment. Models were tested in Mplus Version 8.1 (Muthén & Muthén, 1998-2017) using full information maximum likelihood estimation. Effect sizes were calculated and reported using Cohen’s d (Cohen, 1988).

Results

Adults with greater versus fewer treatment sessions did not differ on age, sex, race-ethnicity, Harvey Impact Score, or baseline PHQ-9, GAD-7, GAF, and clinician or patient-rated top problems, p’s > 0.05. There was a difference in session attendance for those with higher vs lower Global Severity Scale scores: participants with higher Global Severity Scale at baseline had more treatment sessions, r = 0.14, p < 0.05.

Following treatment, 38.4% (n = 93) of participants were rated as at least “minimally improved” by the end of therapy, with 11% (n = 27) “much improved” or “very much improved.” The majority of participants (59.1%; n = 143) were described as “no change” or that symptoms remained essentially unchanged from baseline. A small percentage (2.1%; n = 5) was rated “minimally worse” and one participant (0.4%) as “very much worse.”

Cut-off scores on the PHQ-9 and GAD-7 identified the following groups: 1) elevated anxiety but not elevated depression (n = 18, 7.1%), 2) elevated depression but not elevated anxiety (n = 87, 34.2%), 3) elevated anxiety and depression (n = 109, 42.9%), and 4) neither elevated anxiety nor elevated depression (n = 40, 15.8%). These groups were identified to assess the effectiveness of the treatment for individuals with comorbid depression and anxiety symptoms in comparison to individuals with depression or anxiety symptoms only, or non-elevated symptoms. Model fit statistics for the unconditional latent growth curve models are found in Table 3. Models incorporating a quadratic term did not converge; thus, the linear models provided the best fit for the seven outcome variables. Results for the conditional models are presented in Table 4.

Table 3.

Model fit indices for the unconditional models

	Model fit statistics
		AIC	BIC	SSA-BIC
PHQ9	Intercept	8355	8424	8364
	Slope	8176	8256	8186
GAD7	Intercept	6007	6075	6014
	Slope	5822	5900	5831
GAF	Intercept	5495	5564	5503
	Slope	5153	5232	5163
CGI-Improvement	Intercept	1958	2021	1964
	Slope	1539	1613	1546
Clinician-Top Problem	Intercept	4320	4385	4327
	Slope	3782	3857	3790
Patient-Top Problem	Intercept	4680	4745	4688
	Slope	4245	4320	4253

Global Severity Scale model did not converge

AIC Akaike Information Criteria, BIC Bayesian Information Criteria, SSA-BIC Sample-Size Adjusted Bayesian Information Criteria

Table 4.

Parameter estimates

	PHQ-9		GAD-7		GAF		CGI-Improvement		Clinician Top Problem Rating		Patient Top Problem Rating
	Effect Estimate (95% Confidence Interval)
	Intercept	Slope	Intercept	Slope	Intercept	Slope	Intercept	Slope	Intercept	Slope	Intercept	Slope
Age	0.01 (−0.03, 0.06)	0.00 (−0.00, 0.01)	0.02 (−0.01, 0.05)	0.00 (−0.00, 0.00)	−0.02 (−0.04, 0.00)	−0.00 (−0.00, 0.00)	0.00 (−0.00, 0.00)	0.00 (−0.00, 0.00)	0.00 (−0.01, 0.01)	0.00 (−0.00, 0.00)	0.01 (−0.01, 0.02)	0.00 (−0.00, 0.00)
Female	−0.85 (−2.17, 0.47)	−0.01 (−0.09, 0.07)	−0.39 (−1.18, 0.40)	0.02 (−0.03, 0.06)	0.13 (−0.51, 0.78)	−0.03 (−0.06, 0.01)	0.03 (−0.05, 0.11)	−0.01 (−0.02, 0.01)	0.18 (−0.11, 0.48)	−0.02 (−0.07, 0.03)	0.45 (0.04, 0.85)	−0.03 (−0.09, 0.03)
White	1.31 (−0.14, 2.76)	0.07 (−0.03, 0.18)	0.56 (−0.32, 1.43)	0.00 (−0.06, 0.06)	−0.23 (−0.95, 0.50)	0.01 (−0.03, 0.05)	0.05 (−0.04, 0.13)	0.00 (−0.02, 0.02)	−0.14 (−0.46, 0.18)	0.02 (−0.04, 0.07)	−0.09 (−0.54, 0.36)	0.00 (−0.07, 0.07)
Black	2.41 (0.70, 4.12) *	−0.01 (−0.12, 0.11)	0.57 (−0.46, 1.59)	0.03 (−0.03, 0.09)	0.48 (−0.37, 1.34)	0.01 (−0.04, 0.05)	−0.02 (−0.12, 0.08)	−0.02 (−0.04, 0.00)	0.10 (−0.28, 0.48)	0.01 (−0.06, 0.07)	0.31 (−0.22, 0.84)	0.00 (−0.08, 0.08)
Hispanic	Reference
Harvey Impact Score	0.95 (0.23, 1.66)*	−0.06 (−0.13, −0.00)	0.79 (0.34, 1.23)*	−0.06 (−0.09, −0.03)*	−0.07 (−0.45, 0.31)	0.03 (0.00, 0.05)	0.01 (−0.04, 0.05)	0.01 (−0.01, 0.02)	0.13 (−0.05, 0.31)	−0.01 (−0.04, 0.02)	0.23 (−0.02, 0.48)	−0.02 (−0.06, 0.012)
Employed	−0.80 (−2.28, 0.69)	−0.02 (−0.12, 0.09)	−0.40 (−1.30, 0.50)	−0.04 (−0.11, 0.03)	0.14 (−0.68, 0.96)	0.02 (−0.03, 0.07)	0.07 (−0.01, 0.15)	0.00 (−0.02, 0.02)	−0.29 (−0.61, 0.03)	0.05 (−0.01, 0.11)	−0.07 (−0.55, 0.41)	0.01 (−0.04, 0.10)
Living Alone	−0.00 (−1.54, 1.53)	−0.01 (−0.11, 0.09)	0.96 (0.01, 1.90)	−0.08 (−0.14, −0.02)*	0.22 (−0.55, 1.00)	0.03 (−0.02, 0.07)	−0.06 (−0.16, 0.03)	0.02 (−0.00, 0.03)*	0.83 (0.47, 1.19)*	−0.11 (−0.17, −0.05)*	0.71 (0.20, 1.22)*	−0.10 (−0.18, −0.03)*
Therapist	0.06 (−0.74, 0.86)	0.04 (−0.02, 0.10)	2.52 (2.03, 3.00)*	0.01 (−0.03, 0.04)	0.34 (−0.08, 0.76)	0.02 (−0.01, 0.04)	0.00 (−0.05, 0.05)	0.02 (0.01, 0.03)	−1.13 (−1.34, −0.93)*	−0.02 (−0.06, 0.01)	−1.59 (−1.86, −1.31)*	−0.00 (−0.04, 0.04)
Elevated Anxiety	–	–	–	–	−0.37 (−1.62, 0.88)	0.07 (0.01, 0.14)	0.01 (−0.14, 0.14)	0.01 (−0.02, 0.03)	−0.05 (−0.64, 0.54)	0.01 (−0.08, 0.10)	−0.44 (−1.20, 0.32)	0.01 (−0.09, 0.11)
Elevated Depression	–	–	–	–	−0.10 (−0.88, 0.69)	0.05 (0.00, 0.09)	−0.06 (−0.16, 0.03)	0.01 (−0.01, 0.03)	0.29 (−0.09, 0.67)	−0.01 (−0.07, 0.05)	−0.08 (−0.60, 0.44)	0.01 (−0.06, 0.08)
No Elevation	–	–	–	–	1.45 (0.52, 2.38)*	0.05 (−0.00, 0.10)	−0.07 (−0.19, 0.04)	0.03 (0.01, 0.06)*	0.17 (−0.28, 0.62)	−0.10 (−0.17, −0.02)*	−0.25 (−0.86, 0.37)	−0.10 (−0.18, −0.01)
Elevated Anxiety & Depression	–	–	–	–	Reference

PHQ-9 Patient Health Questionnaire, GAD-7 General Anxiety Disorder Screener, GAF Global Assessment of Functioning

^*p < .05

Latent Growth Curve Model for the PHQ-9

Participant scores on the PHQ-9, on average, began in the moderately depressed range (mean intercept = 13.3, p < 0.001) and evidenced considerable variability in starting level (mean intercept variance = 26.8, p < 0.001). Baseline values were followed by decreases in PHQ-9 scores at an average rate of −0.17 units per treatment session (mean slope = −0.17, p < 0.001); there was significant variability in the rate of change in PHQ-9 scores over the course of treatment (mean slope variance = 0.03, p < 0.01). The intercept and slope factors were not correlated (r = −0.05, p = 0.65). The effect was medium (d = 0.56).

With the intercept and slope factors regressed on to the predictors and covariates, Black participants evidenced higher PHQ-9 scores at baseline (B = 2.41, p < 0.05) relative to Hispanic participants. In addition, higher scores on the Harvey impact scale were associated with higher PHQ-9 at baseline (B = 0.95, p < 0.05). No other predictors of intercept or slope were found.

Latent growth curve model for the GAD-7

Similar to the PHQ-9, scores on the GAD-7 on average began in the mild-to-moderate anxiety range (mean intercept = 7.69, p < 0.001) with significant variability in starting level (mean intercept variance = 16.95, p < 0.001) and reduced by an average of −0.06 units per session (mean slope = −0.06, p < 0.001). There was significant variability in rate of change in GAD-7 scores over time (mean slope variance = 0.01, p < 0.001). The intercept and slope factors were not correlated. The effect was small in magnitude (d = 0.31). As for predictors of the intercept and slope, higher scores on the Harvey impact scale were correlated with higher GAD-7 at baseline (B = 0.79, p < 0.01) as well as greater reductions in GAD-7 scores relative to patients with lower Harvey impact scores (B = −0.06, p < 0.01). Therapist was a predictor of baseline GAD-7 scores (B = 2.52, p < 0.001) such that one therapist saw patients with significantly higher anxiety. Furthermore, living with others, relative to living alone, was associated with greater reductions in GAD-7 (B = −0.08, p < 0.05) scores over the course of treatment.

Latent growth curve model for the GAF

The mean intercept (B = 55.64, p < 0.001) and mean slope (B = 0.02, p < 0.05) were significant on the GAF, suggesting moderate difficulties in functioning at baseline, and that functioning gradually improved over time by approximately 0.02 units per treatment session. Mean intercept and the slope growth factors correlated to suggest that participants with higher GAF scores at baseline evidenced a slower rate of change in GAF scores throughout the course of treatment (r = −0.08, p < 0.01). The effect was very small (d = 0.01). Furthermore, intercept (B = 7.32, p < 0.001) and slope variance (B = 0.01, p < 0.001) on the GAF were both significant. When the predictors and covariates were entered into the model, participants with No Elevations in Anxiety/Depression had higher GAF scores at baseline relative to patients with Both Elevated Anxiety/Depression (B = 1.45, p < 0.01).

Latent growth curve model for Global Severity Scale and CGI–Improvement

For the improvement domain of the CGI, patients unsurprisingly had no change in improvement by the second intervention session (mean intercept = 3.02, p < 0.001). Gains in improvement were observed over the course of treatment at a rate of 0.04 units per treatment session (mean slope = 0.04, p < 0.001). Both the intercept (B = 0.03, p < 0.05) and slope variance (B = 0.002, p < 0.001) were significant on the CGI-Improvement. The intercept and slope growth factors were non-significantly correlated (r = −0.002, p = 0.30). As for predictors of initial improvement and rate of change in improvement over time only symptom group (No Elevations vs Anxiety/Depression Elevations) was a predictor of change in improvement over time (B = 0.03, p < 0.05). Escalations in both PHQ-9 and GAD-7 symptoms at baseline were associated with greater increases in CGI-Improvement over the course of treatment relative to patients with No Anxiety/Depression elevation. Therapist was also a predictor of change over time in CGI–Improvement (B = 0.02, p < 0.01).

Global Severity Scale analyses were not conducted because the model would not converge to establish the best form of change.

Latent Growth Curve Model for the Clinician and Patient-Rated Top Problems Scale

For the clinician-rated top problems, the mean intercept (B = 9.23, p < 0.001) and mean slope (B = −0.12, p < 0.001) were significant, indicating that top problems were rated high and reduced by an average of 0.12 units per treatment session. Intercept (B = 1.92, p < 0.001) and slope (B = 0.02, p < 0.001) variance on the clinician-rated top problems scale were significant. The intercept and slope growth factor means were not significantly correlated (r = −0.02, p = 0.29). Cohen’s d was small (d = 0.19).

Living with others (B = 0.83, p < 0.001) was associated with higher clinician-rated top problems at baseline and greater reductions in clinician-rated top problem scores (B = −0.11, p < 0.01) compared to patients who lived alone. Therapist was a significant predictor of clinician-rated top problems (B = −1.13, p < 0.001). Furthermore, patients with Both Elevated Anxiety/Depression had greater change over time in clinician-rated top problems compared to adults with no elevations in these symptoms (B = −0.10, p < 0.05).

The mean intercept (B = 9.13, p < 0.001) and slope (B = −0.12, p < 0.001) were significant for client-rated top problems and followed a similar trajectory to that of the clinician-rated top problems. The intercept and slope factors were not significantly correlated (r = −0.02, p = 0.44). Intercept (B = 2.86, p < 0.001) and slope variance (B = 0.02, p < 0.01) were both significant. The effect size was small (d = 0.14).

Similar to the clinician-rated top problems, living with others was associated with higher patient-reported scores at baseline (B = 0.71, p < 0.05) and greater reductions in patient-rated (B = −0.10, p < 0.01) top problems scores over time. Therapist was a predictor of patient-reported top problem scores (B = −1.59, p < 0.001) at baseline.

Discussion

The present study investigated treatment outcomes among 279 adults who received the UP, beginning approximately one year after Hurricane Harvey. Our findings offer insight into the effectiveness of the UP in a naturalistic, community health setting among a majority Hispanic population. This demographic focus addresses a gap in the UP literature, which has disproportionately represented white participants (Cassiello-Robbins et al., 2020). The results of this study also highlight the potential of the UP in a trauma-exposed population, an area where data are currently limited—though case studies and a small randomized controlled trial have been promising (Gallagher, 2017; Hood et al., 2021; O'Donnell et al., 2021).

Overall, treatment was associated with significant decreases in depressive symptoms and more modest reductions in anxiety symptoms. These outcomes coincide with a meta-analysis of UP for adults in a variety of settings (Cassiello-Robbins et al., 2020). Functional impairment and the severity of the top problems that participants identified at baseline (the concerns they most wanted to address) also decreased across treatment. Participants demonstrated global clinical symptom improvement throughout treatment as measured by CGI–Improvement scores; 38% of participants were rated as at least “minimally improved” by the end of therapy, with 11% “much improved” or “very much improved” by the end of therapy. The level of improvement demonstrated in this sample is lower than in efficacy trials and RCTS for the UP (e.g., Barlow et al., 2017a). These findings may be explained by the context of a naturalistic community sample, where participants received fewer treatment sessions relative to participants in clinical efficacy settings. Participants in this sample were also exposed to particularly high stressors, both related to Harvey and in general, which may have contributed to symptom persistence. On a positive note, the beneficial effects for depressive symptoms within relatively few sessions is promising. More modest results for anxiety symptoms may reflect limited use of core treatment components, namely exposure therapy.

Participants with higher Harvey impact scores evidenced greater anxiety and depression symptoms at baseline. Furthermore, those with higher Harvey impact scores reported greater reductions in anxiety symptoms over the course of treatment relative to those less impacted by Harvey. These results indicate that the UP may be an effective intervention for anxiety in the aftermath of a natural disaster and under-score the pervasive impact such an event is likely to have on overall functioning. However, it is important to note that those who experienced higher impact from the hurricane may also have had greater room for improvement in anxiety symptoms. The negative mental health effects of disasters like Harvey have been well documented (e.g. Orengo-Aguayo et al., 2019; Picou & Hudson, 2010; Stanke et al., 2012).

Several other factors were also associated with attenuated treatment response and rate of symptom improvement. Living with others (as opposed to living alone) was linked to greater improvement in anxiety symptoms and top problem scores, and participants with baseline elevations in both anxiety and depression symptoms experienced greater improvements in top problem ratings than those who had neither elevated anxiety nor depression at baseline. On the other hand, participants who initially presented with less functional impairment experienced slower rates of improvement in functional impairment. This may be because there was less room for impairment to improve, especially since the intervention was brief. Further research exploring the use of UP in the aftermath of a disaster could investigate effectiveness for acute cases versus those who experience subtler impacts or exacerbation of existing symptoms. In addition, different outcome variables may be relevant in patients with more acute verses more mild symptoms; for example, skill development, self-efficacy, or general stress levels may be more helpful measures for participants with low baseline impairment.

Participants who presented with co-occurring elevated anxiety and depression symptoms at baseline did not experience differential rates of improvement across outcomes compared with those who had only elevated anxiety symptoms or elevated depression symptoms only, though participants without elevated depression or and anxiety symptoms experienced slower rates of improvement. This suggests that the UP may be able to address symptoms that cut across multiple problem areas relating to emotional disorders. The effectiveness of the UP for participants with co-occurring anxiety and depression symptoms could indicate that patients presenting with both anxiety and depression had underlying dysfunctions that were effectively targeted by the UP, as hypothesized by the transdiagnostic approach to emotional disorders that underlies the UP (Barlow et al., 2013).

Baseline clinical presentation was associated with several participant characteristics. As expected, the presence of co-occurring anxiety and depression symptoms at baseline were associated with increased baseline functional impairment when compared to participants who did not have anxiety and depression symptoms. Race-ethnicity also emerged as a factor associated with baseline depression severity: participants who identified as Black or African American reported higher PHQ-9 scores at baseline relative to Hispanic adults. This finding is consistent with previous research demonstrating that African American patients report more severe depression symptoms than white patients (Williams et al., 2007). Rates of depression also appear to be higher among African Americans and Hispanics than among whites—though when controlling for factors such as disparities in health, healthcare, and economic resources, white and Hispanic participants report roughly equal rates of depression and African American patients report lower rates than whites or Hispanics (Dunlop et al., 2003). Female gender, living alone, and being employed were associated with more severe clinician-rated top problems at baseline.

There are several limitations to the current study. Our sample was majority female. The study did not include a comparison group or blinded interview-rated assessments. Because the study began approximately one year after Hurricane Harvey struck Houston, the results may not extend to populations receiving treatment directly after a disaster. There was also a high drop-out rate among the adults followed, with one-third of participants discontinuing treatment after the second session and 19% dropping out after three sessions. This issue may have been due to treatment barriers (e.g., transportation, cost of time away from work, childcare, etc.), as indicated by descriptive reports from participants. Challenges with patient retention in community health settings have been documented in previous research as well (e.g., Wolitzky-Taylor et al., 2015), including studies that have used the UP (Cassiello-Robbins et al., 2020). This emphasizes the need for flexible interventions that can be utilized in settings associated with low patient retention, such as community health. Future studies evaluating the use of the UP in community health settings may wish to identify barriers to completing treatment. Given the high drop-out rate observed together with the relatively impressive effects (particularly for depression), future research might also explore a flexible, brief version of the UP that could be readily tailored to address client presenting problems and needs in a data-driven manner. It is also important to consider how best to adapt the UP for community health settings. Further studies examining UP effectiveness in the immediate aftermath of a natural disaster are needed as well; in the present study, treatment did not begin until at least one year after Hurricane Harvey.

The present study examined outcomes among a racially diverse sample of 279 adults who received UP in a naturalistic, community health setting. These outcomes occurred in the aftermath of Hurricane Harvey, with overall improvements observed for those with and without substantial impact from Harvey. Our first aim was to evaluate change in overall clinical severity, change in anxiety and depression symptoms, and any improvements in top problems. On the whole, the UP was associated with meaningful improvement in depression symptoms, more modest effects on anxiety, and enhanced functioning. Clinical severity and top problem severity also decreased over the course of treatment. Regarding the second study aim (factors associated with improvement), results indicated that those with greater hurricane impact experienced increased improvement in anxiety symptoms, that participants with both elevated anxiety and depression symptoms at baseline experienced greater improvements than those without either, and that participants who lived with others experienced increased improvements in anxiety symptoms and top problems. Our third aim was to investigate any differential treatment response as a function of baseline symptoms and the presence of co-occurring elevated depression and anxiety symptoms; findings indicated that those with both depression and anxiety symptoms did not demonstrate any difference in treatment response than those with depression symptoms only or anxiety symptoms only. Though high drop-out rates were a challenge in this study, overall, clinical benefit was observed. Future research should explore the use of the UP in demographically diverse populations and as a flexible, brief intervention in healthcare settings with characteristically low patient retention.