This study aims to analyze the reach and efficacy of the Mental Health Collaborative (MHC) at Casa de Salud quantitatively, using demographic data and self-reported patient scores on various mental health screenings, and qualitatively, using patient feedback.

Part 1. Mental Health Outcomes

Data Description

When a patient is referred to the MHC either internally or from one of our partners, they are called to complete an intake, which includes a health history questionnaire and baseline screenings for depression (PHQ-9), anxiety (GAD-7), and trauma (PCL5-DSMIV or PCL5-DSM5 (if after 2021)). After their first therapy appointment, patients are called every three months to complete follow-up screenings and provide feedback. This occurs for every course of treatment a patient receives.

For the first two years of the program (February 2018 - September 2021), patient data was stored in Microsoft Access in three files.

  1. Access Full Demographics - patient demographics and health history questionnaires (n=1033)
  2. Access Baseline - baseline mental health screening scores completed at intake (n=1037)
  3. Access Scores - follow up mental health screening scores (n=801)

Since September 20, 2021, patient data has been stored in Ninja/Collaborate, Casa’s EMR/case management system. Casa volunteers are slowly inputting historical records from Access into Ninja in reverse chronological order, but they have only reached January 2021. To avoid duplicate records across the two systems in this analysis, patients who had their intake records on Access had all of their follow-up scores consolidated on Access, even if the follow ups occurred after the clinic transitioned to Ninja.

  1. Ninja Intakes - intakes completed between February 20, 2020 and July 20, 2022 (334 observations of 24 variables).

    • Most of these observations are duplicates of Access intake records. Only new patients who had their intakes completed in Ninja and who had follow-up scores were maintained (n=38).

  2. Ninja Scores - follow-up screenings completed between September 28, 2021 and July 21, 2022 (38 observations of 26 variables).

Both Ninja files were combined to make a Master Ninja dataset (n=38), and all Access files were combined to make a Master Access dataset (n=791). Both master datasets only include patient intakes that received follow-up. In other words, only patients who had subsequent mental health screening scores were included.

Finally, both master files were merged to create a single dataset called Master MHC (n= 829) with which to answer our three main research questions:

  1. How do mental health outcomes (change in depression, anxiety, and trauma scores) vary by type of therapist?
  2. How do mental health outcomes vary by therapy delivery mode?
  3. How do mental health outcomes vary by interpreter use?

Outcome variables:

Variable Description Type Values
Delta.PHQ9 The difference in depression score from baseline continuous -20 to 15
PHQchange Interpreting the difference in depression score from baseline categorical

1. Improved (change <0)

2. No change (change =0)

3. Worsened (change >0)
Delta.GAD7 The difference in anxiety score from baseline continuous -23 to 15
GADchange Interpreting the difference in anxiety score from baseline categorical

1. Improved (change <0)

2. No change (change =0)

3. Worsened (change >0)
Delta.PCL4

The change in trauma (PCL5-DSMIV)** score from intake

Examined in subset of Master Access file (n=436; all patients without PCL5-DSMIV scores were removed)

 continuous -21 to 16
PCL4change

Interpreting the change in PCL5-DSMIV score from baseline

Examined in subset of Master Access file (n=436; all patients without PCL5-DSMIV scores were removed)

 categorical

1. Improved (change <0)

2. No change (change =0)

3. Worsened (change >0)
ClientSatisfaction Rating from 1-10 (Clients after 2021 use 5-point Likert scale; these were converted to numerical values and transformed to match Access scale) discrete 1-10

**The MHC switched from PCL5-DSMIV to the newest PCL5 screening tool (PCL5-DSM5) in 2018. Given that these screening tools employ different scales, they cannot be reconciled and must be analyzed separately. In this data, the sample of patients who have PCL5-DSM5 scores is insufficient for statistical analysis. Thus, only patients from 2018-2021 who had PCL5-DMSIV scores will be analyzed using a subset of the Master Access file.

Key predictor variables of interest:

Variable Description Type Values
Interpret Is an interpreter used during therapy sessions? categorical

1. Yes

2. No
Theratype Type of therapist categorical

1. Student

2. Licensed
Theramode Therapy delivery mode categorical

1. Telehealth

2. In-person

3. Hybrid
Secondary predictors of interest:
Variable Description Type Values
Age Client age at intake continuous 0-99
Gender Client gender categorical

1. Male

2. Female
Race/ethnicity Client race/ethnicity categorical

1. Latino

2. White (Not Hispanic)

3. Black (Not Hispanic)

4. Asian or Pacific Islander

5. Other (including Middle Eastern, Native American, and Mixed Race)

6. Prefer not to respond
Immigrant Whether or not client was born out of the country categorical

1. Yes

2. No
Refugee/Asylum Whether client self-identifies as a refugee or asylum seeker categorical

1. Yes

2. No
Language Client’s preferred language categorical

1. Spanish

2. English

3. Other (includes Arabic, Swahili, French & Vietnamese)

4. Bilingual (English/Spanish, English/Swahili, English/Arabic)
Suicide.Thoughts Client has experienced or is currently experiencing suicidal thoughts categorical

1. Yes

2. No
Suicide.Attempts Client has a history of suicide attempts categorical

1. Yes

2. No
Suicide.Hotline.Referral Client was given the Behavioral Health Response crisis line number categorical

1. Yes

2. No
MHCMilestone Follow up month categorical

1. 3

2. 6

3. 9

4. 12

5. 15+ (15 to 51 month f/u)

Data Cleaning

For more meaningful analysis, discharges completed within 1 month or less of starting therapy were removed (n=8), and minors under age 16 (n=9) were also removed from the sample, as the depression and anxiety screening tools used are not validated for this age group. The sole respondent who identified their gender as “Other” was also removed, resulting in a final sample size of 811. Data classified as ‘Prefer not to respond’ in demographic variables (such as nation of origin) was kept, as without it the sample size would have been further reduced. Categories within the variables Race/Ethnicity, County, and MHC Follow up Milestone that had few respondents were pooled. For Race/Ethnicity, the “Other” category consists of Middle Eastern (n=26), Native American (n=4), Prefer not to respond (n=5), and multiracial/ethnic (n=7) folks. For the variable MHC Follow up Milestone, all follow-ups after 12 months were pooled. Finally, Theratype was recoded as a binary variable, licensed vs. student.

Master MHC Results

Who are MHC patients?

Table 1: Casa de Salud MHC Screenings - Demographics, 2018-2022 (n=811)
Overall
(N=811)
Age in years (Median, IQR) 37.0, 18.0
Race/Ethnicity
Asian or Pacific Islander 14 (1.7%)
Black (Not Hispanic) 77 (9.5%)
White (Not Hispanic) 90 (11.1%)
Latino 588 (72.5%)
Other 42 (5.2%)
Gender
Female 599 (73.9%)
Male 212 (26.1%)
County
Madison 31 (3.8%)
St. Clair 17 (2.1%)
St. Charles 68 (8.4%)
St. Louis County 375 (46.2%)
St. Louis City 299 (36.9%)
Other in IL 5 (0.6%)
Other in MO 16 (2.0%)
Language Spoken
Spanish 473 (58.3%)
English 249 (30.7%)
Bilingual 67 (8.3%)
Other 22 (2.7%)
Born Outside of the Country?
No 200 (24.7%)
Yes 611 (75.3%)
Refugee or Asylum Seeker
No 732 (90.3%)
Yes 79 (9.7%)

Summary: The mental health screenings conducted from 2018 to July 2022 (n=811) belonged to 414 unique patients with a median age of 37 years (IQR=18). The majority of scores belonged to female (74%), Latino (72.5%) patients living in St. Louis County (46%), followed by St. Louis City (37%). Most scores (58%) correspond to Spanish-speaking patients. Aside from English (31%) and English-Spanish bilingualism (n=64), the most common language spoken was Arabic (n=16, 2%). 75% of scores belonged to patients who were born out of the country, but only 9.7% to refugees.

Table 2: Casa de Salud MHC Screenings - Extended Health History, 2018-2022 (n=811)
Overall
(N=811)
Previously Received Mental Health Treatment
No 374 (46.1%)
Yes 437 (53.9%)
Previous or Existing Mental Health Diagnosis
No 554 (68.3%)
Yes 257 (31.7%)
Family History of Mental Health Issues
No 571 (70.4%)
Yes 240 (29.6%)
Taking Medication for Mental Health
No 602 (74.2%)
Yes 209 (25.8%)
History of Eating Disorder
No 771 (95.1%)
Yes 40 (4.9%)
Experienced or Experiencing Suicidal Thoughts
No 485 (59.8%)
Yes 326 (40.2%)
Attempted Suicide
No 720 (88.8%)
Yes 91 (11.2%)
Referred to Behavioral Health Response Crisis Line (BHR)
No 660 (81.4%)
Yes 151 (18.6%)
Drug or Alcohol Use
NA 4 (0.5%)
No 443 (54.6%)
Yes 364 (44.9%)
History of Substance Abuse
NA 4 (0.5%)
No 715 (88.2%)
Yes 92 (11.3%)
Family History of Substance Abuse
NA 4 (0.5%)
No 547 (67.4%)
Yes 260 (32.1%)

Summary: Over half of scores (54%) corresponded to patients who had received mental health treatment previously, but the majority did not have a mental health diagnosis (68%), a family history of mental health issues (70.4%), nor a prescription for their mental health issue (74.2%). Only ~5% of scores were from patients who had a history of eating disorders. 40% of scores belonged to patients who reported having had or currently having suicidal thoughts, but only 11.2% had previously attempted suicide. In most instances (81.4%), patients were not referred to the BHR crisis hotline. Most patients reported that they did not use drugs or alcohol, have a history of substance abuse, or a family history of substance abuse (~55%, 88%, 67%, respectively). 0.5% of scores corresponded to patients under 18, who were not asked the substance abuse questions.

Where are MHC patients from?

Table 3: MHC Clients Place of Origin, Casa de Salud MHC Screenings 2018-2022 (n=772)
Overall
(N=772)
Place of Origin
Europe 3 (0.4%)
Asia 9 (1.2%)
Africa 15 (1.9%)
Caribbean 16 (2.1%)
Middle East 23 (3.0%)
South America 53 (6.9%)
Central America 75 (9.7%)
USA 198 (25.6%)
Mexico 380 (49.2%)

Summary: Excluding the scores from patients who did not provide their nation of origin (n=39), we see that MHC patients represent 34 distinct countries. Nearly 50% of scores correspond to patients from Mexico, and just over a quarter from the United States. Roughly 10% came from Central America, notably Honduras (5%), Guatemala (1.6%), Nicaragua (1.4%), and El Salvador (1.2%). 7% came from South America, particularly Colombia (3%), Venezuela (2%), and Peru (1.3%). 3% came from the Middle East, 2% from the Caribbean, 2% from African countries, and 1.2% from Asian countries including China, Vietnam, and India.

Summary: MHC clients come from 13 different counties and 84 zip codes across MO and IL. Nearly half of the total sample (n=811) live in St. Louis County, and 37% live in St. Louis City. Of the top 10 zip codes (see figure below), the most common zip code was 63116, which corresponds to the neighborhoods south of Tower Grove Park, including Tower Grove South, Bevo, and parts of Dutchtown. 9.5% of clients live in 63118, which includes Benton Park neighborhood and is just south of Casa de Salud. 8.5% of patients live in 63114, which includes Overland and is just south of St. Louis Lambert Airport.

MHC Operations and Outcomes

Table 4: Casa de Salud MHC Screening Scores - Operations-Related Questions, 2018-2022 (n=811)
Overall
(N=811)
MHC Follow Up Milestone
3 295 (36.4%)
6 189 (23.3%)
9 106 (13.1%)
12 67 (8.3%)
15+ 154 (19.0%)
Discharge Screening?
No 611 (75.3%)
Yes 200 (24.7%)
Therapist Organization
Come As You Are Counseling 10 (1.2%)
St. Francis Community Services (SFCS) 31 (3.8%)
Behavioral Health Response (BHR) 10 (1.2%)
Bilingual International Assistant Services (BIAS) 155 (19.1%)
Casa de Salud 77 (9.5%)
Feel Good Counseling 32 (3.9%)
Great Circle 7 (0.9%)
Lutheran Family & Children's Services of Missouri 31 (3.8%)
Mercy Professional Services 214 (26.4%)
Mindful Coping Solutions 2 (0.2%)
Multicultural Counseling and Research Center (MCRC) 44 (5.4%)
Private Practice 65 (8.0%)
South City Counseling 23 (2.8%)
Spring to Life Counseling, LLC 9 (1.1%)
St. Louis Psychoanalytic Institute 81 (10.0%)
St. Louis University Medical Family Therapy Program (SLUMFTP) 20 (2.5%)
Therapy Delivery Mode
Hybrid 36 (4.4%)
In-person 354 (43.6%)
Telehealth 421 (51.9%)
Type of Therapist
Licensed 446 (55.0%)
Student 365 (45.0%)
Interpreted Therapy
No 630 (77.7%)
Yes 181 (22.3%)

Summary: Of all screening scores (n=811), 36% were 3 month follow-ups, 23% were 6 month assessments, 13% were 9 month follow-ups, 8% were 12 months, and ~19% were follow-ups from 15 to 51 months. Nearly 25% of scores were discharges. Scores corresponded to patients receiving treatment from 70 different therapists, most commonly Rosario Bobadilla (~16%), Margaret Jackson (8%), Bob Fyfe (3.9%), Katherine Lozano (3.8%), and Sandra Schulte (3.7%). The organizations most represented were Mercy Professional Services (26%), Bilingual International Assistant Services (19%), St. Louis Psychoanalytic Institute (10%), and Casa de Salud (9.5%). 8% of scores were from patients receiving therapy from a private practice provider. Telehealth was the most common mode of therapy delivery (~52%), followed by in-person (~44%). Most follow up assessments were from patients receiving therapy from a licensed provider (55%) and ~78% denied using an interpreter during therapy.

Change in Mental Health Scores by Follow-Up Month
Table 5: Change in PHQ-9 & GAD-7 Scores by Follow Up Month, Casa de Salud MHC Screenings 2018-2022 (n=811)
3
(N=295)		 6
(N=189)		 9
(N=106)		 12
(N=67)		 15+
(N=154)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake (Median, IQR) -4.00, 9.00 -3.00, 9.00 -3.50, 11.0 -4.00, 8.00 -3.00, 9.75 -4.00, 9.00
Change in GAD-7 Score Since Intake (Median, IQR) -4.00, 8.00 -3.00, 7.00 -3.00, 9.00 -5.00, 9.00 -3.00, 9.00 -3.00, 8.00

Interpretation: The median change in depression score at 3 months (n=295) and 12 months (n=67) was an improvement of 4 points, with a 1-1.5 point dip in median progress during the 6 month (n=189) and 9 month (n=106) marks. The highest improvement in anxiety scores was seen at 12 months (improvement of 5 points (IQR=9), followed by 3 months (median 4 point improvement (IQR=8)). All other months saw a median improvement of 3 points in anxiety scores.

Visualizing the Relationship between Demographic Predictors and Mental Health Scores

Race/Ethnicity
Change in Depression (PHQ-9) Scores by Race/Ethnicity, Casa MHC Screenings 2018-2022 (n=811)
Asian or Pacific Islander
(N=14)		 Black (Not Hispanic)
(N=77)		 White (Not Hispanic)
(N=90)		 Latino
(N=588)		 Other
(N=42)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 9 (64.3%) 40 (51.9%) 57 (63.3%) 424 (72.1%) 25 (59.5%) 555 (68.4%)
No change in Depression 2 (14.3%) 5 (6.5%) 4 (4.4%) 32 (5.4%) 1 (2.4%) 44 (5.4%)
Worsened Depression 3 (21.4%) 32 (41.6%) 29 (32.2%) 132 (22.4%) 16 (38.1%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -3.50, 9.50 -1.00, 9.00 -2.00, 7.00 -4.00, 9.00 -1.50, 7.75 -4.00, 9.00

Interpretation: Across all racial groups, over 51% of depression scores (n=811) improved. Latino (n=588) was the racial/ethnic group that had the highest percent of improved scores (72%), followed by Asian or Pacific Islander (64% improved, n=14), White (Not Hispanic) (n=90; 63% improved), and Other (n=47, 60% improved). The racial group that experienced the least improvement in depression scores was Black (Not Hispanic) (n=77). Only 52% of Black (Not Hispanic) patients’ scores improved, and 42% of scores worsened.

Change in Anxiety (GAD-7) Scores by Race/Ethnicity, Casa MHC Screenings 2018-2022 (n=811)
Asian or Pacific Islander
(N=14)		 Black (Not Hispanic)
(N=77)		 White (Not Hispanic)
(N=90)		 Latino
(N=588)		 Other
(N=42)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 10 (71.4%) 48 (62.3%) 65 (72.2%) 411 (69.9%) 30 (71.4%) 564 (69.5%)
No change in Anxiety 1 (7.1%) 8 (10.4%) 4 (4.4%) 41 (7.0%) 4 (9.5%) 58 (7.2%)
Worsened Anxiety 3 (21.4%) 21 (27.3%) 21 (23.3%) 136 (23.1%) 8 (19.0%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -4.50, 8.50 -3.00, 9.00 -3.00, 6.00 -4.00, 8.00 -3.00, 7.00 -3.00, 8.00

Interpretation: Roughly 70% of anxiety scores improved across all racial groups except Black (Not Hispanic). Only 62% of scores within for Black (not Hispanic) patients improved and almost 30% of scores worsened. In all other groups, the percentage of scores that worsened was less than 23%.

Gender
Change in Depression Scores by Gender, Casa MHC Screenings 2018-2022 (n=811)
Female
(N=599)		 Male
(N=212)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 417 (69.6%) 138 (65.1%) 555 (68.4%)
No change in Depression 27 (4.5%) 17 (8.0%) 44 (5.4%)
Worsened Depression 155 (25.9%) 57 (26.9%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -4.00, 9.00 -3.00, 9.00 -4.00, 9.00

Interpretation: Of the depression scores corresponding to female patients (n=599), 70% improved, 26% worsened, and 4.5% remained at baseline. Scores belonging to males (n=212) were slightly worse; 65% improved, 27% worsened, and 8% remained at baseline.

Change in Anxiety Scores by Gender, Casa MHC Screenings 2018-2022 (n=811)
Female
(N=599)		 Male
(N=212)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 413 (68.9%) 151 (71.2%) 564 (69.5%)
No change in Anxiety 47 (7.8%) 11 (5.2%) 58 (7.2%)
Worsened Anxiety 139 (23.2%) 50 (23.6%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -3.00, 8.00 -3.50, 9.00 -3.00, 8.00

Interpretation: Of the anxiety scores corresponding to female patients (n=599), 69% improved, 23% worsened, and 7.8% remained at baseline. 71% of men’s anxiety scores (n=212) improved, 24% worsened, and 5% remained unchanged.

Language
Change in Depression Scores by Language, Casa MHC Screenings 2018-2022 (n=811)
Spanish
(N=473)		 English
(N=249)		 Bilingual
(N=67)		 Other
(N=22)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 340 (71.9%) 161 (64.7%) 45 (67.2%) 9 (40.9%) 555 (68.4%)
No change in Depression 26 (5.5%) 17 (6.8%) 1 (1.5%) 0 (0%) 44 (5.4%)
Worsened Depression 107 (22.6%) 71 (28.5%) 21 (31.3%) 13 (59.1%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -4.00, 9.00 -3.00, 8.00 -2.00, 8.00 1.00, 5.00 -4.00, 9.00

Interpretation: Among Spanish-speakers (n=473), 72% of depression scores improved, 23% worsened, and 5.5% did not change from baseline. Bilingual patients (n=67) had the second highest percent of improved scores (67%). English-speakers (n=249) had slightly worse outcomes; only 65% of patient depression scores improved, 28.5% worsened, and ~7% remained at baseline. Patients who spoke other languages (n=22) were the only group to report worse depression symptoms during/after treatment than at baseline. Only 41% of scores from these patients improved, while 59% worsened. None remained the same as intake.

Change in Anxiety Scores by Language, Casa MHC Screenings 2018-2022 (n=811)
Spanish
(N=473)		 English
(N=249)		 Bilingual
(N=67)		 Other
(N=22)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 332 (70.2%) 177 (71.1%) 41 (61.2%) 14 (63.6%) 564 (69.5%)
No change in Anxiety 36 (7.6%) 15 (6.0%) 4 (6.0%) 3 (13.6%) 58 (7.2%)
Worsened Anxiety 105 (22.2%) 57 (22.9%) 22 (32.8%) 5 (22.7%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -4.00, 9.00 -3.00, 7.00 -1.00, 10.0 -1.00, 4.75 -3.00, 8.00

Interpretation: The English-speakers’ anxiety scores (n=249) improved the most of all language groups. 71% of their anxiety scores showed improvement, followed by Spanish-speakers (70%, n=473), ‘Other’ language speakers (64%, n=22), and bilingual patients (61%, n=67). 33% of bilingual speakers’ anxiety scores worsened, the highest percentage across all language groups. The highest percentage of unchanged scores was seen in the “other” languages group (14%).

Immigrant Status
Change in Depression Scores by Immigrant Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=200)		 Yes
(N=611)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 126 (63.0%) 429 (70.2%) 555 (68.4%)
No change in Depression 10 (5.0%) 34 (5.6%) 44 (5.4%)
Worsened Depression 64 (32.0%) 148 (24.2%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -2.00, 9.00 -4.00, 9.00 -4.00, 9.00

Interpretation: Foreign-born patients (n=611) had better depression outcomes than native patients (n=200). 70% of scores improved for immigrant patients, 24% worsened, and 5.6% remained unchanged from baseline. Of the scores of native patients, only 63% improved, 32% worsened, and 5% experienced no change from baseline.

Change in Anxiety Scores by Immigrant Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=200)		 Yes
(N=611)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 142 (71.0%) 422 (69.1%) 564 (69.5%)
No change in Anxiety 13 (6.5%) 45 (7.4%) 58 (7.2%)
Worsened Anxiety 45 (22.5%) 144 (23.6%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -3.00, 8.00 -3.00, 8.00 -3.00, 8.00

Interpretation: Native patients (n=200) had slightly better anxiety outcomes than foreign-born patients. 71% of scores improved for native patients, 23% worsened, and 6.5% experienced no change. Of the scores of foreign-born patients (n=611), 69% improved, 24% worsened, and 7% experienced no change from baseline.

Refugee Status
Change in Depression Scores by Refugee Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=732)		 Yes
(N=79)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 511 (69.8%) 44 (55.7%) 555 (68.4%)
No change in Depression 41 (5.6%) 3 (3.8%) 44 (5.4%)
Worsened Depression 180 (24.6%) 32 (40.5%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -4.00, 8.00 -1.00, 10.0 -4.00, 9.00

Interpretation: The score outcomes of self-reported refugee/asylum seekers (n=79) were worse than those of non-refugee/asylum seekers (n=732), with an median improvement of only 1 point (IQR=10). Only 56% of refugees’ depression scores improved, 41% worsened, and 4% remained unchanged. Meanwhile, 70% of scores from patients who were not refugee/asylum seekers improved, 25% worsened, 5.6% did not change from baseline, and the median improvement was of 4 points (IQR=8).

Change in Anxiety Scores by Refugee Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=732)		 Yes
(N=79)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 515 (70.4%) 49 (62.0%) 564 (69.5%)
No change in Anxiety 49 (6.7%) 9 (11.4%) 58 (7.2%)
Worsened Anxiety 168 (23.0%) 21 (26.6%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -3.00, 8.00 -2.00, 8.50 -3.00, 8.00

Interpretation: Like their depression scores, the anxiety score outcomes for refugee/asylum seekers (n=79) were worse than those of patients who were not refugee/asylum seekers (n=732). 62% of their anxiety scores improved by a median 2 points, 27% worsened, and 11% did not change from baseline, while 70% of scores for patients who did not identify as refugee/asylum seekers improved by a median of 3 points, 23% worsened, and 7% remained at baseline.

Suicidal Ideation
Change in Depression Scores by Suicidal Ideation, Casa MHC Screenings 2018-2022 (n=811)
No
(N=485)		 Yes
(N=326)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 328 (67.6%) 227 (69.6%) 555 (68.4%)
No change in Depression 31 (6.4%) 13 (4.0%) 44 (5.4%)
Worsened Depression 126 (26.0%) 86 (26.4%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -3.00, 9.00 -4.00, 10.0 -4.00, 9.00

Interpretation: Depression outcomes were very slightly worse for patients who denied having or currently experiencing suicidal thoughts (n=485). 68% of scores in this group improved by a median of 4 points, which is 2% less than the percentage of scores that improved for patients who were historically or currently suicidal (n=326). For both groups, ~26% of depression scores worsened. 4% of scores remained the same for suicidal patients, compared to 6% for non-suicidal patients.

Change in Anxiety Scores by Suicidal Ideation, Casa MHC Screenings 2018-2022 (n=811)
No
(N=485)		 Yes
(N=326)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 344 (70.9%) 220 (67.5%) 564 (69.5%)
No change in Anxiety 31 (6.4%) 27 (8.3%) 58 (7.2%)
Worsened Anxiety 110 (22.7%) 79 (24.2%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -4.00, 8.00 -3.00, 8.00 -3.00, 8.00

Interpretation: 71% of anxiety scores for patients who denied having suicidal thoughts (n=485) improved by a median of 4 points (IQR=8). 23% of scores for this group worsened, and 6% remained the same. This is slightly better than the outcomes seen for patients who admitted having had or currently having suicidal thoughts (n=326). 67% of scores for this group improved by a median of 3 points, 24% worsened, and 8% remained the same as baseline.

History of Attempted Suicide
Change in Depression Scores by History of Attempted Suicide, Casa MHC Screenings 2018-2022 (n=811)
No
(N=720)		 Yes
(N=91)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 490 (68.1%) 65 (71.4%) 555 (68.4%)
No change in Depression 40 (5.6%) 4 (4.4%) 44 (5.4%)
Worsened Depression 190 (26.4%) 22 (24.2%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -3.00, 9.00 -4.00, 9.50 -4.00, 9.00

Interpretation: Depression outcomes did not vary greatly between patients with a history of attempted suicide (n=91, ~11% of sample) and those without (n=720). For both groups, roughly 70% of depression scores improved. 24% of scores from those who had a history of attempted suicide worsened, and \4% remained the same as baseline, compared to 26% and 6%, respectively, for patients without a history of attempted suicide.

Change in Anxiety Scores by History of Attempted Suicide, Casa MHC Screenings 2018-2022 (n=811)
No
(N=720)		 Yes
(N=91)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 507 (70.4%) 57 (62.6%) 564 (69.5%)
No change in Anxiety 50 (6.9%) 8 (8.8%) 58 (7.2%)
Worsened Anxiety 163 (22.6%) 26 (28.6%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -3.00, 8.00 -3.00, 9.00 -3.00, 8.00

Interpretation: Anxiety outcomes for patients with a history of attempted suicide (n=91, ~11% of sample) were worse than that of those without a history of attempted suicide (n=720). Only 63% of their anxiety scores improved, ~29% of scores worsened, and ~9% remained the same as baseline. For patients who have not attempted suicide, 70% of scores improved, 23% worsened, and 7% did not change from baseline. The median change in score for both groups was -3 points.

Behavioral Health Crisis Line Referral
Change in Depression Scores by BHR Referral Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=660)		 Yes
(N=151)		 Overall
(N=811)
Change in PHQ-9 Score Since Intake
Improved Depression 437 (66.2%) 118 (78.1%) 555 (68.4%)
No change in Depression 41 (6.2%) 3 (2.0%) 44 (5.4%)
Worsened Depression 182 (27.6%) 30 (19.9%) 212 (26.1%)
Change in PHQ-9 Score Since Intake (Median, IQR) -3.00, 8.00 -6.00, 10.0 -4.00, 9.00

Interpretation: The depression scores of patients who were referred to BHR (n=151, ~19% of sample) were much better than those of patients who did not receive the BHR number (n=660). The median change in score for the latter group was an improvement of 6 points (IQR=10), double that of the latter group. 78% of depression scores for patients who were referred to BHR improved from baseline, 20% worsened, and 2% remained the same. For patients who were not referred, only 66% of scores improved, 28% worsened, and 6% did not change from intake.

Change in Anxiety Scores by BHR Referral Status, Casa MHC Screenings 2018-2022 (n=811)
No
(N=660)		 Yes
(N=151)		 Overall
(N=811)
Change in GAD-7 Score Since Intake
Improved Anxiety 447 (67.7%) 117 (77.5%) 564 (69.5%)
No change in Anxiety 47 (7.1%) 11 (7.3%) 58 (7.2%)
Worsened Anxiety 166 (25.2%) 23 (15.2%) 189 (23.3%)
Change in GAD-7 Score Since Intake (Median, IQR) -3.00, 9.00 -5.00, 9.00 -3.00, 8.00

Interpretation: Following the trend seen in the depression assessments, the anxiety score outcomes for patients who were referred to BHR (n=151, ~19% of sample) were better than those of patients who did not receive the BHR number (n=660). 78% of anxiety scores for patients who were referred to BHR improved by a median of 5 points (IQR=9), 15% worsened, and 7% remained the same. For patients who were not referred, ~68% of scores improved by a median of 3 points (IQR=9), 25% worsened, and 7% did not change from intake.

Trauma (PCL5-DSMIV) Outcomes (Access Data, 2018-2021)

The subset of Master Access data that was used to evaluate trauma score outcomes had a final sample size of 432. Due to their small sample sizes, the “Asian or Pacific Islander” (n=6) and “Prefer not to respond” racial categories were collapsed into “Other”. Therapy delivery mode was analyzed as a binary variable (telehealth vs. in-person) after the hybrid category (n=4) was removed.

Table 1: Casa de Salud MHC Trauma Scores - Demographics, 2018-2021 (n=432)
Overall
(N=432)
Age (years) 38.3, 11.5
Race/Ethnicity
Black (Not Hispanic) 34 (7.9%)
White (Not Hispanic) 56 (13.0%)
Latino 320 (74.1%)
Other 22 (5.1%)
Gender
Female 306 (70.8%)
Male 126 (29.2%)
Preferred Language
Spanish 262 (60.6%)
English 129 (29.9%)
Bilingual 36 (8.3%)
Other 5 (1.2%)
Born Outside of the Country?
No 109 (25.2%)
Yes 323 (74.8%)
Refugee or Asylum Seeker
No 390 (90.3%)
Yes 42 (9.7%)
Table 2: Casa de Salud MHC Trauma Scores - Extended Health History, 2018-2021 (n=432)
Overall
(N=432)
Previously Received Mental Health Treatment
No 181 (41.9%)
Yes 251 (58.1%)
Previous or Existing Mental Health Diagnosis
No 297 (68.8%)
Yes 135 (31.3%)
Taking Medication for Mental Health
No 331 (76.6%)
Yes 101 (23.4%)
Family History of Mental Health Issues
No 310 (71.8%)
Yes 122 (28.2%)
History of Eating Disorder
No 409 (94.7%)
Yes 23 (5.3%)
Experienced or Experiencing Suicidal Thoughts
No 244 (56.5%)
Yes 188 (43.5%)
Attempted Suicide
No 368 (85.2%)
Yes 64 (14.8%)
Referred to Behavioral Health Response Crisis Line (BHR)
No 344 (79.6%)
Yes 88 (20.4%)
Drug or Alcohol Use
No 228 (52.8%)
Yes 204 (47.2%)
History of Substance Abuse
No 397 (91.9%)
Yes 35 (8.1%)
Family History of Substance Abuse
No 289 (66.9%)
Yes 143 (33.1%)
Table 3: Casa de Salud MHC Trauma Scores - Operations-Related Questions, 2018-2021 (n=432)
Overall
(N=432)
Follow Up Month Milestone
3 168 (38.9%)
6 111 (25.7%)
9 62 (14.4%)
12 38 (8.8%)
15+ 53 (12.3%)
Discharge
No 339 (78.5%)
Yes 93 (21.5%)
Mode of Therapy Delivery
In-person 167 (38.7%)
Telehealth 265 (61.3%)
Type of Therapist
Licensed 239 (55.3%)
Student 193 (44.7%)
Interpreter Use
No 332 (76.9%)
Yes 100 (23.1%)
Table 4: Change in Trauma Scores by Month, Casa de Salud MHC 2018-2021 (n=432)
3
(N=168)		 6
(N=111)		 9
(N=62)		 12
(N=38)		 15+
(N=53)		 Overall
(N=432)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -3.04, 5.39 -2.84, 6.02 -2.68, 5.61 -2.66, 6.72 -2.94, 7.77 -2.89, 6.01
Trauma Outcomes
Improved Trauma 114 (67.9%) 72 (64.9%) 39 (62.9%) 23 (60.5%) 35 (66.0%) 283 (65.5%)
No change in Trauma 17 (10.1%) 9 (8.1%) 9 (14.5%) 5 (13.2%) 1 (1.9%) 41 (9.5%)
Worsened Trauma 37 (22.0%) 30 (27.0%) 14 (22.6%) 10 (26.3%) 17 (32.1%) 108 (25.0%)

Summary: The same demographic patterns seen in the MasterMHC dataset were seen in this subset of Access data. Scores belonged to patients who were predominantly Latino (74%), female (71%), Spanish-speaking (61%), immigrant (75%), and non-refugees (90%). The average age was 38 years old (sd=11.5). The majority of patients had received mental health treatment previously, but no diagnosis or medication. The majority did not have a family history of mental health issues or substance abuse. The majority of these patients also did not have history of eating disorders, substance use, substance abuse, suicidal ideation, or suicidal attempts. Almost 40% of scores are from 3 months, and 78.5% were not discharges. Most patients were receiving therapy via telehealth, working with licensed therapists, and not using interpreters.

Analysis of Mental Health Outcomes at 3 Months

To measure outcomes after 3 months of services, the data was subset by follow up month in order to know both the exact conditions (i.e. what kind of therapy mode, therapist type, and interpreter status) the patient had and their mental health scores at that particular follow-up time. Subsetting by month also ensured that patients are not represented multiple times (within a single course of treatment), as they should only complete one follow up assessment every 3 months, thus meeting the independent observations criteria required for most statistical analyses. Racial categories were recoded in this subset so that “Prefer not to respond,” which only had 2 respondents, and “Asian or Pacific Islander” (n=6) became part of “Other.” Patients with duplicate screenings at 3 months (n=3) were also removed. In the Access data subset used to analyze trauma scores, the bilingual and “other” categories were recoded to “Bilingual or Other”.

3 Month Follow Up Data Summary
Table 1: Casa de Salud MHC Screening Scores at 3 Months Follow-Up, 2018-2022 (n=293)
Overall
(N=293)
Discharge
No 206 (70.3%)
Yes 87 (29.7%)
Gender
Female 218 (74.4%)
Male 75 (25.6%)
Race/Ethnicity
Other 23 (7.8%)
Black (Not Hispanic) 28 (9.6%)
White (Not Hispanic) 28 (9.6%)
Latino 214 (73.0%)
Language Spoken
Spanish 165 (56.3%)
English 87 (29.7%)
Bilingual 29 (9.9%)
Other 12 (4.1%)
Born Outside of the Country?
No 65 (22.2%)
Yes 228 (77.8%)
Refugee or Asylum Seeker
No 258 (88.1%)
Yes 35 (11.9%)
Experienced or Experiencing Suicidal Thoughts
No 180 (61.4%)
Yes 113 (38.6%)
Attempted Suicide
No 256 (87.4%)
Yes 37 (12.6%)
Referred to Behavioral Health Response Crisis Line (BHR)
No 240 (81.9%)
Yes 53 (18.1%)
Therapy Delivery Mode
Hybrid 14 (4.8%)
In-person 177 (60.4%)
Telehealth 102 (34.8%)
Type of Therapist
Licensed 143 (48.8%)
Student 150 (51.2%)
Interpreter Use
No 209 (71.3%)
Yes 84 (28.7%)
Change in GAD-7 Scores Since Intake (Mean, SD) -4.12, 6.03
Change in PHQ-9 Scores Since Intake (Mean, SD) -4.27, 6.80

Summary: 293 3-month follow up screenings for patients over 16 years old were completed from 2018-2022. Of these, 30% were discharges, and 206 continued with treatment. The demographic and health history patterns observed in the overall MHC data hold for this 3 month subset (i.e., patients were predominantly Latino (73%), female (74%), Spanish-speaking (56.3%), immigrants (78%), non-refugee/asylum seekers (88%); 61% denied suicidal ideation, 87% denied attempting suicide, and 82% did not receive a BHR referral (Table 1)). Of the 293 scores, the majority (60%) correspond to patients who were receiving in-person therapy, followed by telehealth (35%). More patients were working with students (51%) than with licensed therapists, and 71% of sessions were not interpreted. The average change in PHQ-9 score after 3 months of therapy was an improvement of 4.3 points (sd= 7). The average change in GAD-7 score was an improvement of 4.1 points (sd=6).

The same demographic patterns were also seen in the Master Access file in which trauma scores were analyzed (Table 1.2 below). Of the 167 trauma scores, the majority (54%) were from patients working with a student therapist, not using an interpreter (72%), and working with their therapist in-person (56%).

Table 1.2: Casa de Salud MHC Trauma Scores at 3 Months Follow-Up, 2018-2021 (n=167)
Overall
(N=167)
Gender
Female 121 (72.5%)
Male 46 (27.5%)
Race/Ethnicity
Other 10 (6.0%)
Black (Not Hispanic) 14 (8.4%)
White (Not Hispanic) 21 (12.6%)
Latino 122 (73.1%)
Language Spoken
Bilingual or Other 20 (12.0%)
Spanish 91 (54.5%)
English 56 (33.5%)
Born Outside of the Country?
No 43 (25.7%)
Yes 124 (74.3%)
Refugee or Asylum Seeker
No 148 (88.6%)
Yes 19 (11.4%)
Experienced or Experiencing Suicidal Thoughts
No 98 (58.7%)
Yes 69 (41.3%)
Attempted Suicide
No 139 (83.2%)
Yes 28 (16.8%)
Referred to Behavioral Health Response Crisis Line (BHR)
No 133 (79.6%)
Yes 34 (20.4%)
Discharge
No 123 (73.7%)
Yes 44 (26.3%)
Therapy Delivery Mode
In-person 93 (55.7%)
Telehealth 74 (44.3%)
Type of Therapist
Licensed 77 (46.1%)
Student 90 (53.9%)
Interpreter Use
No 120 (71.9%)
Yes 47 (28.1%)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -3.05, 5.41

Understanding the Relationship between Demographic Predictors and Mental Health Scores at 3 Month Follow-Up

Race/Ethnicity Numeric Outcomes
Table 5: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Race/Ethnicity at 3 Months Follow Up, Casa de Salud MHC 2018-2022 (n=293)
Other
(N=23)		 Black (Not Hispanic)
(N=28)		 White (Not Hispanic)
(N=28)		 Latino
(N=214)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -1.91, 6.04 -5.68, 7.08 -2.71, 5.02 -4.54, 6.99 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -1.78, 6.11 -5.36, 7.06 -2.96, 5.34 -4.36, 5.92 -4.12, 6.03

Interpretation: At three months, the depression scores of Black (Not Hispanic) patients (n=28) improved by 5.7 points (sd=7) on average, the highest mean reduction in scores from intake. Latino patients (n=214) followed, with an average 4.5 point reduction in depression score from intake. Latino patients’ scores were the most variable, ranging from a 23 point reduction in PHQ-9 to a 13 point increase. Other (n=23) and White (Not Hispanic, n=28) had the least improvement in depression scores at 3 months. The same pattern was observed for anxiety scores, with Black patients improving by an average of 5.4 points, and Latinos by 4.4 points. Patients who identified as Other or White (Not Hispanic) improved the least in anxiety scores.

Table 4.3: Change in Trauma (PCL5-DSMIV) Scores by Race/Ethnicity, Casa de Salud MHC 2018-2021 (n=167)
Other
(N=10)		 Black (Not Hispanic)
(N=14)		 White (Not Hispanic)
(N=21)		 Latino
(N=122)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -4.80, 8.20 -3.36, 4.68 -4.10, 4.97 -2.69, 5.30 -3.05, 5.41

Interpretation: The scores of patients who identified as “Other”, including Asian/Pacific Islander, Middle Eastern, and biracial people, changed the most from intake. The average change was an improvement of 4.8 points (sd=8). White (Not Hispanic) patients’ scores (n=21) followed, with a mean improvement of 4.1 points (sd=5). The average change in trauma score seen in Black (Not Hispanic) patients’ assessments (n=14) was an improvement of 3.4 points (sd=4.7). Latino patients had the lowest average change in score, improving only 2.6 points (sd=5.3).

Analysis of Variance: Race/Ethnicity

ANOVA tests will be used to examine the relationship between the change in scores (continuous variables, Delta.PHQ9, Delta.GAD7, Delta.PCL4) and race/ethnicity.

  • Independent observations: yes, only one observation per patient
  • Normality within groups: yes for anxiety; no for depression and trauma. they failed normality assumption
  • Homogeneity of variances: Yes for anxiety and trauma, Levene’s test had p value > 0.01, which is not enough to reject the null hypothesis that variances are equal across groups. depression failed homogeneity
  • Continuous variable and 3+ independent groups: yes, Delta.PHQ9 is continuous and race/ethnicity has 3 independent groups

Anxiety scores met the assumptions for ANOVA. The null hypothesis that the mean change in anxiety scores is the same across racial groups was tested. Alternative hypothesis is that the mean change in score is not the same across racial groups.

Result: There was no significant difference in the mean change in anxiety score across racial/ethnic groups [(F(3, 289) = 2.03; p=0.1)].

The relationship between depression/trauma scores and therapy delivery mode failed the normality assumption for the ANOVA statistical test. The Kruskal-Wallis test was conducted as a result with the null and alternative hypotheses:

H0: The mean rank of the change in depression and trauma scores is the same across racial groups.

HA: The mean rank of the change in depression and trauma scores is not the same across racial groups.

Results: We fail to reject the null hypothesis. There are no statistically significant differences between the mean ranks of the change in depression scores across racial groups [H(3) = 6.0; p = .11], nor between the mean ranks of the change in trauma score across race/ethnicity groups [H(3) = 2.9; p = .41].

Race/Ethnicity Categorical Outcomes
Table 5.2: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Race/Ethnicity at 3 Month Follow Up, Casa de Salud MHC 2018-2022 (n=293)
Other
(N=23)		 Black (Not Hispanic)
(N=28)		 White (Not Hispanic)
(N=28)		 Latino
(N=214)		 Overall
(N=293)
Depression Outcomes
Improved Depression 14 (60.9%) 19 (67.9%) 20 (71.4%) 160 (74.8%) 213 (72.7%)
No change in Depression 1 (4.3%) 3 (10.7%) 0 (0%) 6 (2.8%) 10 (3.4%)
Worsened Depression 8 (34.8%) 6 (21.4%) 8 (28.6%) 48 (22.4%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 14 (60.9%) 22 (78.6%) 21 (75.0%) 157 (73.4%) 214 (73.0%)
No change in Anxiety 3 (13.0%) 2 (7.1%) 0 (0%) 17 (7.9%) 22 (7.5%)
Worsened Anxiety 6 (26.1%) 4 (14.3%) 7 (25.0%) 40 (18.7%) 57 (19.5%)
Table 5.3: Change in Trauma (PCL5-DSMIV) Scores by Race/Ethnicity, Casa de Salud MHC 2018-2021 (n=167)
Other
(N=10)		 Black (Not Hispanic)
(N=14)		 White (Not Hispanic)
(N=21)		 Latino
(N=122)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 6 (60.0%) 10 (71.4%) 18 (85.7%) 79 (64.8%) 113 (67.7%)
No change in Trauma 1 (10.0%) 2 (14.3%) 0 (0%) 14 (11.5%) 17 (10.2%)
Worsened Trauma 3 (30.0%) 2 (14.3%) 3 (14.3%) 29 (23.8%) 37 (22.2%)

Interpretation: The anxiety scores of 58% of patients across all racial groups improved by 3 months. Latinos (n=214) improved the most in depression scores (75%), followed by White (Not Hispanic) patients (n=28), Black (Not Hispanic) patients (n=28), Asian/Pacific Islanders (n=6), and Other (n=17). More patients who identified as “Other” (which includes Middle Eastern, mixed race, and prefer not to respond) worsened than in any other racial group. As for anxiety screenings, Black (Not Hispanic) patients, followed by White (Not Hispanic) patients (75%) and Latino patients (73%), had the best anxiety outcomes. 79% of Black patients improved and only 14% worsened. Only 58% of patients who identified as “Other” improved, while 29% worsened, and 12% remained at baseline.

Chi-squared Analysis: Race/Ethnicity

H0: There is no association between race/ethnicity and mental health outcomes (change in PHQ-9, GAD-7 & PCL5-DSMIV scores).

HA: There is an association between race/ethnicity and mental health outcomes (change in PHQ-9, GAD-7 & PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: No, for all the variables, 6 out of 15 groups do not have at least 5 observations.

Since are the last assumption is not met for all three variables, Fisher’s exact test will be used instead.

Fisher’s exact test assumptions:

  • 2 categorical variables: yes

  • Levels of variables are mutually exclusive: yes

  • Independent observations: yes

  • Data are counts, not percentages or transformed values: yes

H0: There is no association between race/ethnicity and mental health outcomes (change in PHQ-9, GAD-7 & PCL5-DSMIV scores).

HA: There is an association between race/ethnicity and mental health outcomes (change in PHQ-9, GAD-7 & PCL5-DSMIV scores).

Results: The null hypothesis is retained; there is no statistically significant relationship between race/ethnicity and depression outcomes [Fisher’s exact, p = 0.16]. There is also no statistically significant relationship between race/ethnicity and anxiety outcomes (Fisher’s exact, p = 0.51) or trauma scores (Fisher’s exact, p=0.47).

Gender – Numeric Outcomes
Table 6: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Gender at 3 Month Follow Up, Casa de Salud MHC 2018-2022 (n=293)
Female
(N=218)		 Male
(N=75)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -3.93, 6.83 -5.25, 6.65 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -3.84, 6.07 -4.93, 5.89 -4.12, 6.03

Interpretation: The average change in depression scores from baseline to 3 months for females (n=218) was an improvement of 4 (sd=7), while men (n=75) improved 5.3 points (sd=7) on average. Men also saw a mean improvement of 5 points (sd=6) compared to 3.8 points (sd=6) for females.

Table 6.2: Change in Trauma (PCL5-DSMIV) Scores by Gender, Casa de Salud MHC 2018-2021 (n=167)
Female
(N=121)		 Male
(N=46)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -3.28, 5.12 -2.43, 6.13 -3.05, 5.41

Interpretation: The average change in trauma score was an improvement of 3.3 points (sd=5) for female patients (n=121) and 2.4 points (sd=6) for males (n=46).

T-test: Change in Scores by Gender at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety scores between genders at 3 months.
  • HA: There is a difference in the mean change in depression/anxiety scores between genders at 3 months.

Result: There is no significant difference in the mean change in depression score between male and female patients [Welch’s t-test, t(132)= 1.47, p=0.14] or the mean change in anxiety score between male and female patients [Welch’s t-test, t(132)= 1.37, p=0.17]. There is also no significant difference between the change in trauma scores between males and females [Mann-Whitney U t-test, p=0.41].

Gender – Categorical Outcomes
Table 6.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Gender at 3 Month Follow Up, Casa de Salud MHC 2018-2022 (n=293)
Female
(N=218)		 Male
(N=75)		 Overall
(N=293)
Depression Outcomes
Improved Depression 156 (71.6%) 57 (76.0%) 213 (72.7%)
No change in Depression 7 (3.2%) 3 (4.0%) 10 (3.4%)
Worsened Depression 55 (25.2%) 15 (20.0%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 157 (72.0%) 57 (76.0%) 214 (73.0%)
No change in Anxiety 17 (7.8%) 5 (6.7%) 22 (7.5%)
Worsened Anxiety 44 (20.2%) 13 (17.3%) 57 (19.5%)
Table 6.4: Change in Trauma (PCL5-DSMIV) Scores by Gender, Casa de Salud MHC 2018-2021 (n=167)
Female
(N=121)		 Male
(N=46)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 85 (70.2%) 28 (60.9%) 113 (67.7%)
No change in Trauma 13 (10.7%) 4 (8.7%) 17 (10.2%)
Worsened Trauma 23 (19.0%) 14 (30.4%) 37 (22.2%)

Interpretation: Of the 3 month depression scores corresponding to female patients (n=218), 72% improved, ~25% worsened, and 3% remained at baseline. Scores belonging to males (n=75) were slightly better; 76% improved, 20% worsened, and 4% did not change from baseline. The same pattern was observed for anxiety scores: a greater percentage of male patients improved (76%) and 17% worsened, compared to 72% of females (72%) improving and 20% worsening. As for trauma scores, 70% of scores from females (n=121) improved compared to 61% of scores from males (n=46). 305% of scores from males worsened compared to 19% of female trauma scores.

Chi-squared Analysis: Gender

H0: There is no association between gender and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between gender and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 out of 6 (83%) of groups have at least 5 observations.

Results: We fail to reject the null hypothesis; there is no statistically significant relationship between gender and depression outcomes (\(\chi^2\)(2) = 0.89; p = 0.64) or gender and anxiety outcomes \(\chi^2\)(2) = 0.45; p = 0.79). The relationship between gender and trauma outcomes is also not significant (\(\chi^2\)(2) = 2.5; p = 0.28).

Language – Numeric Outcomes
Table 7: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Preferred Language at 3 Month Follow Up, Casa de Salud MHC 2018-2022 (n=293)
Spanish
(N=165)		 English
(N=87)		 Bilingual
(N=29)		 Other
(N=12)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -4.59, 7.30 -4.48, 5.75 -3.21, 6.50 -0.917, 7.06 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -4.56, 6.07 -4.37, 5.93 -2.93, 5.51 0.833, 5.44 -4.12, 6.03

Interpretation: Spanish-speaking patients (n=165) had the highest mean improvement in anxiety and depression scores (~4.6 points), followed by English-speakers (n=87) and Bilingual patients (n=29). Patients who spoke other languages (n=12) were the only group who experienced an average increase (worsening) of anxiety scores after 3 months of treatment (~1 point more than baseline, sd= 5).

Table 7.2: Change in Trauma (PCL5-DSMIV) Scores by Language, Casa de Salud MHC 2018-2021 (n=167)
Bilingual or Other
(N=20)		 Spanish
(N=91)		 English
(N=56)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -0.950, 6.08 -3.01, 5.37 -3.86, 5.10 -3.05, 5.41

Interpretation: The average change in trauma score was best for English-speaking patients’ score (an improvement of 3.9 points (sd=5)), followed by Spanish-speakers (n=91) (improvement of 3 points (sd=5.4). Bilingual patients (n=18) improved the least (<2 points). Patients who spoke other languages (n=2) worsened by 1.5 points on average (sd=9).

Analysis of Variance: Language

ANOVA tests will be used to examine the relationship between the change in scores (continuous variables, Delta.PHQ9, Delta.GAD7, Delta.PCL4) and language.

  • Independent observations: yes, only one observation per patient
  • Normality within groups: yes for anxiety; no for depression and trauma. they failed normality assumption
  • Homogeneity of variances: Yes for anxiety and trauma, Levene’s test had p value > 0.01, which is not enough to reject the null hypothesis that variances are equal across groups. depression failed homogeneity
  • Continuous variable and 3+ independent groups: yes, Delta.PHQ9 is continuous and race/ethnicity has 3 independent groups

The relationship between depression/anxiety/trauma scores and language failed the normality assumption for the ANOVA statistical test. The Kruskal-Wallis test was conducted as a result with the null and alternative hypotheses:

H0: The mean rank of the change in depression and trauma scores is the same across language groups.

HA: The mean rank of the change in depression and trauma scores is not the same across language groups.

Result: We fail to reject the null hypothesis. There are no statistically significant differences between the mean ranks of the change in depression scores across racial groups [H(3) = 4.5; p = .21], between the mean ranks of the change in anxiety score across language groups [H(2) = 4.6; p = .09], nor between the mean ranks of the change in trauma score across language groups [H(2) = 3.1; p = .2].

Language – Categorical Outcomes
Table 7.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Preferred Language at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
Spanish
(N=165)		 English
(N=87)		 Bilingual
(N=29)		 Other
(N=12)		 Overall
(N=293)
Anxiety Outcomes
Improved Anxiety 120 (72.7%) 69 (79.3%) 19 (65.5%) 6 (50.0%) 214 (73.0%)
No change in Anxiety 14 (8.5%) 4 (4.6%) 2 (6.9%) 2 (16.7%) 22 (7.5%)
Worsened Anxiety 31 (18.8%) 14 (16.1%) 8 (27.6%) 4 (33.3%) 57 (19.5%)
Depression Outcomes
Improved Depression 123 (74.5%) 65 (74.7%) 19 (65.5%) 6 (50.0%) 213 (72.7%)
No change in Depression 5 (3.0%) 5 (5.7%) 0 (0%) 0 (0%) 10 (3.4%)
Worsened Depression 37 (22.4%) 17 (19.5%) 10 (34.5%) 6 (50.0%) 70 (23.9%)
Table 7.4: Change in Trauma (PCL5-DSMIV) Scores by Language, Casa de Salud MHC 2018-2021 (n=167)
Bilingual or Other
(N=20)		 Spanish
(N=91)		 English
(N=56)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 12 (60.0%) 59 (64.8%) 42 (75.0%) 113 (67.7%)
No change in Trauma 2 (10.0%) 11 (12.1%) 4 (7.1%) 17 (10.2%)
Worsened Trauma 6 (30.0%) 21 (23.1%) 10 (17.9%) 37 (22.2%)

Interpretation: Roughly ~75% of depression scores for both Spanish and English-speaking patients improved at 3 months, followed by bilingual patients (n=29) and patients who speak other languages (n=12). English-speaking patients’ anxiety scores also responded very well to treatment. At 3 months, 79% of English-speaking patients had improved, 73% of Spanish-speakers, 66% of bilingual patients, and 50% of other language speakers.

Chi-squared Analysis: Language

H0: There is no association between preferred language and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between preferred language and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 10 out of 12 groups have at least 5 observations.

Result: The null hypothesis is retained (\(\chi^2\)(6) = 9.66; p = 0.14); there is no statistically significant relationship between preferred language and depression outcomes. There is also no statistically significant relationship between preferred language and anxiety outcomes (\(\chi^2\)(6) = 6.8, p = 0.34) or trauma outcomes (\(\chi^2\)(4) = 2.6, p = 0.62).

Immigrant – Numeric Outcomes
Table 8: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Immigrant Status at 3 Month Follow Up, Casa de Salud MHC 2018-2022 (n=293)
No
(N=65)		 Yes
(N=228)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -3.55, 6.46 -4.47, 6.89 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -4.25, 6.23 -4.09, 5.99 -4.12, 6.03

Interpretation: At 3 months follow-up, the depression scores of foreign-born patients (n=228) improved on average 4.5 points (sd=7), ~9 points more than the average improvement seen for US-born patients (n=65). The anxiety scores of US-born patients, however, improved by a mean 4.25 points (sd=6), which is slightly more than the average 4 point improvement in anxiety scores seen for foreign-born patients.

Table 8.2: Change in Trauma (PCL5-DSMIV) Scores by Immigrant Status, Casa de Salud MHC 2018-2021 (n=167)
No
(N=43)		 Yes
(N=124)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -4.19, 5.04 -2.65, 5.50 -3.05, 5.41

Interpretation: The average change in trauma score was an improvement of 4 points (sd=5) for non-immigrants (n=43) and 2.7 points (sd=6) for immigrants (n=124).

T-test: Change in Scores by Immigrant Status at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety scores between immigrant status groups at 3 months.
  • HA: There is a difference in the mean change in depression/anxiety scores between immigrant status groups at 3 months.

Result: There is no significant difference in the mean change in depression score between immigrant and non-immigrant patients [Welch’s t-test, t(109)= 0.998, p=0.32] nor the mean change in anxiety score between immigrant and non-immigrant patients [Welch’s t-test, t(100)= -0.18, p=0.86]. There is also no significant difference between the change in trauma scores between males and females [Mann-Whitney U t-test, p=0.15].

Immigrant – Categorical Outcomes
Table 8.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Immigrant Status at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=65)		 Yes
(N=228)		 Overall
(N=293)
Depression Outcomes
Improved Depression 44 (67.7%) 169 (74.1%) 213 (72.7%)
No change in Depression 3 (4.6%) 7 (3.1%) 10 (3.4%)
Worsened Depression 18 (27.7%) 52 (22.8%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 50 (76.9%) 164 (71.9%) 214 (73.0%)
No change in Anxiety 3 (4.6%) 19 (8.3%) 22 (7.5%)
Worsened Anxiety 12 (18.5%) 45 (19.7%) 57 (19.5%)
Table 8.4: Change in Trauma (PCL5-DSMIV) Scores by Immigrant, Casa de Salud MHC 2018-2021 (n=167)
No
(N=43)		 Yes
(N=124)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 33 (76.7%) 80 (64.5%) 113 (67.7%)
No change in Trauma 4 (9.3%) 13 (10.5%) 17 (10.2%)
Worsened Trauma 6 (14.0%) 31 (25.0%) 37 (22.2%)

Interpretation: At three months, 74% of depression scores for patients who were born outside of the country (n=228) improved compared to 67% of patients born in the U.S. (n=65). 23% of foreign-born patients worsened, compared to 28% of U.S.-born patients. The reverse was seen for anxiety scores; 77% of scores from US-born patients improved compared to 72% of scores from foreign-born patients. Roughly 20% worsened in both groups. As for trauma scores, 77% of the 124 scores belonging to non-immigrant patients improved compared to 65% of immigrant patients’ scores. 25% of immigrant patients’ scores worsened vs. 14% of native-born patients’ scores.

Chi-squared Analysis: Immigrant

H0: There is no association between immigrant status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between immigrant status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 out of 6 groups have at least 5 observations.

Result: We fail to reject the null hypothesis [\(\chi^2\)(2) = 1.15; p = 0.56]; there is no statistically significant relationship between immigrant status and depression outcomes. There is also no statistically significant relationship between immigrant status and anxiety outcomes (\(\chi^2\)(2) = 1.15, p = 0.56) or trauma scores (\(\chi^2\)(2) = 2.51, p = 0.26).

Refugee Status – Numeric Outcomes
Table 9: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Refugee Status at 3 Months Follow Up, Casa de Salud MHC 2018-2022 (n=293)
No
(N=258)		 Yes
(N=35)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -4.41, 6.91 -3.26, 5.90 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -4.17, 6.07 -3.77, 5.79 -4.12, 6.03

Interpretation: The mean change in depression scores for non-refugees (n=258) was a 4.4 point reduction in depression severity (sd=7). The depression scores of refugees (n=35) only improved 3.3 points from intake on average. At 3 months, the improvement in anxiety scores for non-refugees was 0.4 points higher than that of refugees (4.2 vs 3.8, respectively).

Table 8.2: Change in Trauma (PCL5-DSMIV) Scores by Refugee/Asylum Seeker Status, Casa de Salud MHC 2018-2021 (n=167)
No
(N=148)		 Yes
(N=19)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -2.87, 5.21 -4.42, 6.81 -3.05, 5.41

Interpretation: From 2018-2021, the average change in trauma score was an improvement of 4.4 points (sd=7) for refugees (n=19) and 2.9 points (sd=5) for non-refugees (n=148).

T-test: Change in Scores by Refugee Status at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety/trauma scores between refugee status groups at 3 months.
  • HA: There is a difference in the mean change in depression/anxiety/trauma between refugee status groups at 3 months.

Result: There is no significant difference in the mean change in depression score between refugee and non-refugee patients [Welch’s t-test, t(47)= -1.06, p=0.295] nor the mean change in anxiety score between these groups [Mann-Whitney U t-test,p=0.58]. There is also no significant difference between the change in trauma scores between refugees and non-refugees [Mann-Whitney U t-test, p=0.19].

Refugee Status – Categorical Outcomes
Table 9.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Refugee Status at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=258)		 Yes
(N=35)		 Overall
(N=293)
Depression Outcomes
Improved Depression 190 (73.6%) 23 (65.7%) 213 (72.7%)
No change in Depression 7 (2.7%) 3 (8.6%) 10 (3.4%)
Worsened Depression 61 (23.6%) 9 (25.7%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 191 (74.0%) 23 (65.7%) 214 (73.0%)
No change in Anxiety 15 (5.8%) 7 (20.0%) 22 (7.5%)
Worsened Anxiety 52 (20.2%) 5 (14.3%) 57 (19.5%)
Table 9.4: Change in Trauma (PCL5-DSMIV) Scores by Refugee Status, Casa de Salud MHC 2018-2021 (n=167)
No
(N=148)		 Yes
(N=19)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 100 (67.6%) 13 (68.4%) 113 (67.7%)
No change in Trauma 16 (10.8%) 1 (5.3%) 17 (10.2%)
Worsened Trauma 32 (21.6%) 5 (26.3%) 37 (22.2%)

Interpretation: From intake to three months, 74% of patients who are not refugees (n=258) improved in depression scores, 3% remained at baseline, and 24% worsened. Refugee patients had worse depression outcomes at 3 months. 66% of refugee patients improved, 26% worsened, and 8.6% did not change from intake. A similar pattern held for anxiety scores: 66% of refugee scores improved and 20% remained at baseline vs. 74% of non-refugee scores improved and 6% did not change. The only difference is that only 14% of refugee patients’ anxiety scores worsened compared to 20% of non-refugees. Roughly 68% of trauma scores improved regardless of refugee/asylum seeker status. 26% of scores worsened for refugees(n=19) compared to 22% for non-refugees (n=148).

Chi-squared Analysis: Refugee/Asylum

H0: There is no association between refugee status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between refugee status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 of 6 groups have at least 5 observations.

Results: We fail to reject the first null hypothesis (\(\chi^2\)(2) = 3.42; p = 0.2); there is no statistically significant relationship between refugee status and depression outcomes or trauma outcomes (\(\chi^2\)(2) = 0.68; p = 0.7). However, there is a statistically significant relationship between refugee status and anxiety outcomes (\(\chi^2\)(2) = 9.1; p = 0.01). This result is driven by the fact that there were more non-refugees than expected in the improved anxiety category (n = 191; std res = -0.19) and fewer refugees than expected in the improved anxiety category (n=23, std res= -0.51). Also, more refugees than expected experienced no change (n=7, std res= 2.7). Although statistically significant, the relationship between refugee status and anxiety outcomes is weak (V = .11).

Suicidal Ideation – Numeric Outcomes
Table 10: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Suicidal Ideation at 3 Months Follow Up, Casa de Salud MHC 2018-2022 (n=293)
No
(N=180)		 Yes
(N=113)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -3.77, 6.36 -5.07, 7.41 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -3.95, 6.14 -4.40, 5.86 -4.12, 6.03

Interpretation: Almost 40% of the 3 month follow up sample affirmed having had or currently experiencing suicidal ideation. These patients (n=113) had slightly larger mean improvements in both depression and anxiety scores than patients who did denied suicidal ideation (n=180). Their mean change in scores at three months was 5 points (sd=7) for depression and 4.4 points for anxiety (sd=5.9), which is 1.3 and 0.5 points larger than the mean change of depression and anxiety scores, respectively, for patients who denied suicidal thoughts.

Table 10.2: Change in Trauma (PCL5-DSMIV) Scores by Suicidal Ideation, Casa de Salud MHC 2018-2021 (n=167)
No
(N=98)		 Yes
(N=69)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -2.08, 4.61 -4.42, 6.15 -3.05, 5.41

Interpretation: From 2018-2021, the average change in trauma score was an improvement of 4.4 points (sd=6) for patients who admitted suicidal ideation (n=69) and 2.1 points (sd=5) for patients who denied suicidal thoughts (n=98).

T-test: Change in Scores by Suicidal Ideation at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety/trauma scores between patients who admit suicidal ideation and those who do not at 3 months.
  • HA: There is a difference in the mean change in depression/anxiety/trauma between patients who admit suicidal ideation and those who do not at 3 months.

Result: There is no significant difference in the mean change in depression score between suicidal and non-suicidal patients [Welch’s t-test, t(44)= -0.38, p=0.71] nor the mean change in anxiety score between these groups [Welch’s t-test, t(246)= 0.63, p=0.53]. However, there is a significant difference between the change in trauma scores between patients who admit suicidal ideation and those who do not [Mann-Whitney U t-test, p=0.01]. Though significant, the effect size is small, r=0.2, indicating a weak relationship between history of attempted suicide and change in trauma score.

Suicidal Ideation – Categorical Outcomes
Table 10.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Suicidal Ideation at 3 Months Follow Up, Casa de Salud MHC 2018-2022 (n=293)
No
(N=180)		 Yes
(N=113)		 Overall
(N=293)
Depression Outcomes
Improved Depression 130 (72.2%) 83 (73.5%) 213 (72.7%)
No change in Depression 7 (3.9%) 3 (2.7%) 10 (3.4%)
Worsened Depression 43 (23.9%) 27 (23.9%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 131 (72.8%) 83 (73.5%) 214 (73.0%)
No change in Anxiety 11 (6.1%) 11 (9.7%) 22 (7.5%)
Worsened Anxiety 38 (21.1%) 19 (16.8%) 57 (19.5%)
Table 10.4: Change in Trauma (PCL5-DSMIV) Scores by Suicidal Ideation, Casa de Salud MHC 2018-2021 (n=167)
No
(N=98)		 Yes
(N=69)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 61 (62.2%) 52 (75.4%) 113 (67.7%)
No change in Trauma 15 (15.3%) 2 (2.9%) 17 (10.2%)
Worsened Trauma 22 (22.4%) 15 (21.7%) 37 (22.2%)

Interpretation: Depression outcomes at 3 months were similar for both patients who denied (n=180) and admitted (n=113) suicidal ideation. 74% of suicidal patients improved, 24% worsened, and 3% remained at baseline. Of the patients who denied suicidal thoughts, 73% improved, 24% worsened, and 4% remained at baseline. Anxiety scores were similar: 74% of patients who affirmed suicidal ideation improved, 10% did not change, and 17% worsened. Of the patients who denied suicidal thoughts, 73% improved, 6% did not change, and 21% worsened. As for trauma scores, 75% of the 69 scores from patients who admitted suicidal ideation improved, which is 13 percentage points higher than the percent of improved scores from patients who denied suicidal ideation.

Chi-squared Analysis: Suicidal Ideation

H0: There is no association between suicidal ideation and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between suicidal ideation and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 out of 6 groups do have at least 5 observations.

Results: The null hypothesis is partially rejected; there is a statisically significant relationship between suicidal ideation and trauma score outcomes (\(\chi^2\)(2) = 0.32; p = 0.9). Based on the standardized residuals, this result is driven by the fact that there were more improved scores and fewer unchanged scores for patients who admitted suicidal ideation than expected. There were also fewer improved scores and more unchanged scores than expected for patients who denied suicidal ideation. Though this relationship is significant, the effect size is weak to moderate (V = .20). There is no statistically significant relationship between suicidal ideation and depression outcomes (\(\chi^2\)(2) = 0.32; p = 0.9) or anxiety outcomes (\(\chi^2\)(2) = 1.88; p = 0.4).

History of Attempted Suicide – Numeric Outcomes
Table 11: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by History of Attempted Suicide at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=256)		 Yes
(N=37)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -4.11, 6.74 -5.35, 7.22 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -4.11, 6.09 -4.24, 5.65 -4.12, 6.03

Interpretation: From intake to 3 months, the mean change in depression score for patients who had a history of attempted suicide (n=37) was a reduction of 5.4 points (sd=7.2), which is slightly more than the improvement patients who denied a history of attempted suicide (n=256) experienced on average (4 points (sd=6.7). The mean change in anxiety score at 3 months was similar between the two groups; patients who had not attempted suicide improved 4.1 points (sd=6) and patients who had improved 4.2 (sd=6) points on average.

Table 11.2: Change in Trauma (PCL5-DSMIV) Scores by History of Attempted Suicide, Casa de Salud MHC 2018-2021 (n=167)
No
(N=139)		 Yes
(N=28)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -2.63, 5.10 -5.11, 6.47 -3.05, 5.41

Interpretation: From 2018-2021, the average change in trauma score was an improvement of 5.1 points (sd=6.5) for patients who had attempted suicide (n=28) and 2.6 points (sd=5) for patients who denied attempted suicide (n=139).

T-test: Change in Scores by History of Attempted Suicide at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety/trauma scores at three months between patients who have attempted suicide and those who have not.
  • HA: There is a difference in the mean change in depression/anxiety/trauma scores at three months between patients who admit suicidal ideation and those who have not.

Result: There is no significant difference in the mean change in depression score between suicidal and non-suicidal patients [Mann-Whitney U t-test, p=0.34] nor the mean change in anxiety score between these groups [Mann-Whitney U t-test, p=0.75]. There is also no significant difference between the change in trauma scores for patients who have attempted suicide and those who have not [Mann-Whitney U t-test, p=0.07].

History of Attempted Suicide – Categorical Outcomes
Table 11.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by History of Attempted Suicide at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=256)		 Yes
(N=37)		 Overall
(N=293)
Depression Outcomes
Improved Depression 184 (71.9%) 29 (78.4%) 213 (72.7%)
No change in Depression 10 (3.9%) 0 (0%) 10 (3.4%)
Worsened Depression 62 (24.2%) 8 (21.6%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 187 (73.0%) 27 (73.0%) 214 (73.0%)
No change in Anxiety 18 (7.0%) 4 (10.8%) 22 (7.5%)
Worsened Anxiety 51 (19.9%) 6 (16.2%) 57 (19.5%)
Table 11.4: Change in Trauma (PCL5-DSMIV) Scores by History of Attempted Suicide at 3 Months, Casa de Salud MHC 2018-2021 (n=167)
No
(N=139)		 Yes
(N=28)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 94 (67.6%) 19 (67.9%) 113 (67.7%)
No change in Trauma 14 (10.1%) 3 (10.7%) 17 (10.2%)
Worsened Trauma 31 (22.3%) 6 (21.4%) 37 (22.2%)

Interpretation: At three months follow-up, 72% of the 256 depression scores of patients who denied attempting suicide improved, 24% worsened, and 4% remained at baseline. Depression outcomes for patients who admitted a history of attempted suicide (n=37) were slightly better; 78% improved, 22% worsened, and none remained at baseline. There were no large differences in the percent of improved anxiety or trauma scores between the groups.

Chi-squared Analysis: Suicide Attempts

H0: There is no association between having a history of attempted suicide and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between having a history of attempted suicide and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 out of 6 groups have 5 or more observations.

Results: The null hypothesis cannot be rejected; there is no statistically significant relationship between history of attempted suicide and depression outcomes (\(\chi^2\)(2) = 1.72; p = 0.42), anxiety outcomes (\(\chi^2\)(2) = 0.84; p = 0.66), or trauma outcomes at 3 months (\(\chi^2\)(2) = 0.01; p = 0.99).

Behavioral Health Crisis Line Referral – Numeric Outcomes
Table 12: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by BHR Referral Status at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=240)		 Yes
(N=53)		 Overall
(N=293)
Change in PHQ-9 Scores Since Intake (Mean, SD) -3.79, 6.41 -6.43, 8.06 -4.27, 6.80
Change in GAD-7 Scores Since Intake (Mean, SD) -3.84, 6.14 -5.42, 5.39 -4.12, 6.03

Interpretation: At 3 months, patients who received the BHR number (n=53) had larger average improvements in both depression and anxiety scores than the patients who were not referred to BHR (n=240). At 3 months, the mean change in depression scores for patients who received the BHR crisis number was -6.4 (sd=8), which is 2.6 points more improvement than the mean change patients who did not receive the BHR number experienced. The mean change in anxiety score for patients who received the BHR crisis number was -5.4 (sd=5), which is nearly two points more than the mean reduction in anxiety severity of patients who were not referred to BHR.

Table 12.2: Change in Trauma (PCL5-DSMIV) Scores by BHR Referral Status, Casa de Salud MHC 2018-2021 (n=167)
No
(N=133)		 Yes
(N=34)		 Overall
(N=167)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -2.69, 5.31 -4.44, 5.66 -3.05, 5.41

Interpretation: From 2018-2021, the average change in trauma score was an improvement of 4.4 points (sd=6) for patients who were referred to BHR (n=34) and 2.7 points (sd=5) for patients who were not (n=133).

T-test: Change in Scores by BHR Referral at 3 Months
  • H0: There is no difference in the mean change in depression/anxiety/trauma scores between patients who were referred to BHR and those who were not.
  • HA: There is a difference in the mean change in depression/anxiety/trauma scores between patients who were referred to BHR and those who were not.

Results: The null hypothesis is partially rejected. There is a significant difference between the change in depression scores between patients who are referred to BHR and those who are not [Mann-Whitney U t-test, p=0.01]. Though significant, the effect size is small, r=0.1, indicating a weak relationship between BHR referral and change in depression score. Though very close, there were no significant differences in the mean change in anxiety score [Mann-Whitney U t-test, p=0.06] or trauma score [Mann-Whitney U t-test, p=0.05] between patients who were and were not referred to BHR.

Behavioral Health Crisis Line Referral – Categorical Outcomes
Table 12.3: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by BHR Referral Status at 3 Months, Casa de Salud MHC 2018-2022 (n=293)
No
(N=240)		 Yes
(N=53)		 Overall
(N=293)
Depression Outcomes
Improved Depression 170 (70.8%) 43 (81.1%) 213 (72.7%)
No change in Depression 9 (3.8%) 1 (1.9%) 10 (3.4%)
Worsened Depression 61 (25.4%) 9 (17.0%) 70 (23.9%)
Anxiety Outcomes
Improved Anxiety 173 (72.1%) 41 (77.4%) 214 (73.0%)
No change in Anxiety 18 (7.5%) 4 (7.5%) 22 (7.5%)
Worsened Anxiety 49 (20.4%) 8 (15.1%) 57 (19.5%)
Table 12.4: Change in Trauma (PCL5-DSMIV) Scores by BHR Referral Status at 3 Months, Casa de Salud MHC 2018-2021 (n=167)
No
(N=133)		 Yes
(N=34)		 Overall
(N=167)
Trauma Outcomes
Improved Trauma 87 (65.4%) 26 (76.5%) 113 (67.7%)
No change in Trauma 16 (12.0%) 1 (2.9%) 17 (10.2%)
Worsened Trauma 30 (22.6%) 7 (20.6%) 37 (22.2%)

Interpretation: Patients who were referred to BHR (n=53) had better depression and anxiety outcomes at three months than those who were not referred to BHR (n=240). For patients who were referred, 81% of depression scores and 77% of anxiety scores improved compared to 70% of depression scores and 72% of anxiety score for patients who were not referred to BHR. A smaller percentage of both depression and anxiety scores worsened for patients who were referred than those who were not. The same pattern was observed for trauma scores; 76% of scores improved for patients referred to BHR compared to only 65% of scores for those who were not referred.

Chi-squared Analysis: BHR Referral

H0: There is no association between BHR referral status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

HA: There is an association between BHR referral status and mental health outcomes (change in PHQ-9, GAD-7, PCL5-DSMIV scores).

Assumption checking:

  • The variables are nominal: yes

  • Independence of observations: Yes, within the three month subset, there are no duplicate patient screenings. Only one observation per patient.

  • At least 80% of the groups have 5 or more observations: Yes, 5 of 6 groups have at least 5 observations.

Results: The null hypothesis cannot be rejected; there is no statistically significant relationship between BHR referral status and depression outcomes (\(\chi^2\)(2) = 2.37; p = 0.3), anxiety outcomes (\(\chi^2\)(2) = 0.80; p = 0.67), or trauma outcomes (\(\chi^2\)(2) = 2.7; p = 0.26).

Analysis of Mental Health Outcomes at Discharge

The discharge subset (n=196) includes all scores labelled as discharges from any follow-up month. Therapy mode was analyzed as a binary variable in this subset, as the hybrid responses had to be removed due to low sample size. In addition to understanding the relationship between the therapy-related factors aforementioned and mental health outcomes at the end of treatment, this analysis aims is to answer two key questions:

  • At what month do most patients discontinue therapy?
  • What is the average change in anxiety, depression, and trauma scores at time of discharge?
Table 1: Casa de Salud MHC Discharge Assessments - Demographics, 2018-2022 (n=196)
Overall
(N=196)
Race/Ethnicity
Other 17 (8.7%)
Black (Not Hispanic) 15 (7.7%)
White (Not Hispanic) 15 (7.7%)
Latino 149 (76.0%)
Gender
Female 137 (69.9%)
Male 59 (30.1%)
Preferred Language
Spanish 120 (61.2%)
English 53 (27.0%)
Bilingual 18 (9.2%)
Other 5 (2.6%)
Born Outside of the Country?
No 35 (17.9%)
Yes 161 (82.1%)
Refugee or Asylum Seeker
No 172 (87.8%)
Yes 24 (12.2%)
MHC Follow Up Month Milestone
3 88 (44.9%)
6 52 (26.5%)
9 20 (10.2%)
12 13 (6.6%)
15+ 23 (11.7%)
Table 1.2: Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores at Discharge, Casa de Salud MHC 2018-2022 (n=196)
3
(N=88)		 6
(N=52)		 9
(N=20)		 12
(N=13)		 15+
(N=23)		 Overall
(N=196)
Change in PHQ-9 Score Since Intake
Mean (SD) -6.07 (6.91) -4.88 (6.97) -6.25 (6.91) -2.23 (5.26) -3.83 (9.17) -5.26 (7.15)
Change in GAD-7 Score Since Intake
Mean (SD) -5.66 (6.20) -3.77 (5.67) -5.35 (5.33) -2.08 (7.48) -3.83 (7.60) -4.67 (6.29)
Depression Outcomes
Improved Depression 71 (80.7%) 36 (69.2%) 15 (75.0%) 6 (46.2%) 13 (56.5%) 141 (71.9%)
No change in Depression 1 (1.1%) 4 (7.7%) 3 (15.0%) 3 (23.1%) 3 (13.0%) 14 (7.1%)
Worsened Depression 16 (18.2%) 12 (23.1%) 2 (10.0%) 4 (30.8%) 7 (30.4%) 41 (20.9%)
Anxiety Outcomes
Improved Anxiety 70 (79.5%) 39 (75.0%) 17 (85.0%) 9 (69.2%) 13 (56.5%) 148 (75.5%)
No change in Anxiety 5 (5.7%) 3 (5.8%) 1 (5.0%) 0 (0%) 2 (8.7%) 11 (5.6%)
Worsened Anxiety 13 (14.8%) 10 (19.2%) 2 (10.0%) 4 (30.8%) 8 (34.8%) 37 (18.9%)

Summary: Of the 196 discharge assessments completed for patients over age 16 from 2018-2022, 76% are Latino, 8% are Black (Not Hispanics) or White (Not Hispanics), and 9% are “Other”. 88 patients (45% of all discharges) were discharged at 3 months, followed by 6 months (n=52), 15+ months (n=23), 9 months (n=20), and 12 months (n=13). All patients improved in both anxiety and depression by time of discharge. The highest average improvement in depression scores is of 6.25 points (sd=7) at 9 months. The second-highest improvement (6 points (sd= 7)) was seen at 3 months, followed by 6 months (4.9 point improvement (sd=7). After 12 months, the average improvement in depression score was small (<3.9 points) and more variable, as is the percentage of people achieving improved scores. A higher percentage of patients discharged within the first 9 months of therapy had improved anxiety and depression outcomes than the percentage of patients who improved at or after 12 months.

Table 1.3: Casa de Salud MHC Discharge Assessments - Trauma Demographics, 2018-2021 (n=93)
Overall
(N=93)
Race/Ethnicity
Black (Not Hispanic) 9 (9.7%)
White (Not Hispanic) 9 (9.7%)
Latino 68 (73.1%)
Other 7 (7.5%)
Gender
Female 64 (68.8%)
Male 29 (31.2%)
Preferred Language
Spanish 52 (55.9%)
English 31 (33.3%)
Bilingual 10 (10.8%)
Born Outside of the Country?
No 20 (21.5%)
Yes 73 (78.5%)
Refugee or Asylum Seeker
No 82 (88.2%)
Yes 11 (11.8%)
MHC Follow Up Month Milestone
3 44 (47.3%)
6 24 (25.8%)
9 13 (14.0%)
12 7 (7.5%)
15+ 5 (5.4%)
Table 1.4: Change in Trauma (PCL5-DSMIV) Scores at Discharge, Casa de Salud MHC 2018-2021 (n=93)
3
(N=44)		 6
(N=24)		 9
(N=13)		 12
(N=7)		 15+
(N=5)		 Overall
(N=93)
Change in PCL5-DSMIV Score Since Intake (Mean, SD) -4.00, 5.90 -3.13, 5.14 -6.08, 4.94 -2.00, 3.16 2.40, 9.07 -3.57, 5.78
Trauma Outcomes
Improved Trauma 34 (77.3%) 18 (75.0%) 11 (84.6%) 5 (71.4%) 3 (60.0%) 71 (76.3%)
No change in Trauma 4 (9.1%) 1 (4.2%) 2 (15.4%) 1 (14.3%) 0 (0%) 8 (8.6%)
Worsened Trauma 6 (13.6%) 5 (20.8%) 0 (0%) 1 (14.3%) 2 (40.0%) 14 (15.1%)

Summary: The same demographic patterns seen in the overall discharge population hold for the discharges with trauma scores from 2018-2021 (n=93); most patients are Latino, female, Spanish-speaking, foreign-born, non-refugees. Most patients in this subset also discharged at 3 months, followed by 6 and 9 months. There are less than 13 respondents at the 12 and 15+ marks combined. Across all time points, at the end of treatment, patients’ trauma scores had improved 5.3 points from intake (sd=7). The greatest change in trauma score was seen at 9 months, followed closely by 3 months. After 9 months, the average improvement was less than 4 points, and the percentage of scores improving was also smaller, while the percentage of people worsening was higher than that seen in the earlier follow up months.

Time of Discharge

Interpretation: Most discharges from 2018 to 2022 occurred at 3 months (n=90), followed by 6 months (n=52), 9 months (n=20), 15+ months (n=15) and 12 months (n=13). Patients improved the most at 3 months for both depression scores (median reduction in PHQ-9 severity = -6) and anxiety scores (median 5 point improvement). Of the 93 discharges from 2018-2021 who had trauma scores, most were discharged at 3 months, followed by 6 and 9 months. The largest change in trauma score was seen at nine months, with a 6 point reduction in severity (sd=5). After 12 months, the average improvement for patients was less than 3 points, but the sample sizes for these later follow up months are quite small (n=<10).

Analysis of Variance: Time of Discharge

ANOVA tests will be used to examine the relationship between the change in depression, anxiety, and trauma scores (continuous variables, Delta.PHQ9, Delta.GAD7, Delta.PCL4) and the time of discharge (MHC Follow Up Month).

  • Independent observations: yes, only one observation per patient per course of treatment, but also no, because at least 5 patients have multiple courses of treatment. Report with caution.
  • Normality within groups: no
  • Homogeneity of variances: Yes, Levene’s test had p value > 0.01, which is not enough to reject the null hypothesis that variances are equal across groups
  • Continuous variable and 3+ independent groups: yes, continuous and MHC follow up month has 3+ independent groups

Anxiety scores met the assumptions for ANOVA. The null hypothesis that the mean change in anxiety scores is the same across follow up months was tested. Alternative hypothesis is that the mean change in score is not the same across follow up months.

Result: There was no significant difference in the mean change in anxiety score across follow up month [(F(4, 191) = 1.54; p=0.19)].

The relationship between depression/trauma scores and MHC follow up months failed the normality assumption for the ANOVA statistical test. The Kruskal-Wallis test was conducted as a result with the null and alternative hypotheses:

H0: The mean rank of the change in depression and trauma scores is the same across MHC follow up months.

HA: The mean rank of the change in depression and trauma scores is not the same across MHC follow up months.

Result: We fail to reject the null hypothesis. There are no statistically significant differences between the mean ranks of the change in depression scores across MHC follow up months [H(4) = 6.0; p =0.2], nor between the mean ranks of the change in trauma score across MHC follow up months [H(4) = 5.8; p = .22].

MHC Patient Satisfaction Overall (2018-2022)

During every follow-up screening, patients are asked to rate their satisfaction with counseling on a 10-point scale (prior to 2021) and 5-point Likert scale (after 2021). Values after 2021 were transformed to match the Access rating scale, for a final sample of 777 satisfaction ratings.

Satisfaction by Demographic Variables

Satisfaction by Age Group:
Under 20
(N=34)		 20-29
(N=148)		 30-39
(N=253)		 40-49
(N=182)		 50-59
(N=113)		 60 and over
(N=47)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 10.0, 1.00 9.00, 2.00 10.0, 1.00 10.0, 1.00 10.0, 1.00 10.0, 1.00 10.0, 1.00

Satisfaction by Race/Ethnicity:
Asian or Pacific Islander
(N=13)		 Black (Not Hispanic)
(N=76)		 White (Not Hispanic)
(N=88)		 Latino
(N=559)		 Other
(N=41)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 9.00, 3.00 10.0, 2.00 10.0, 2.00 10.0, 1.00 10.0, 1.00 10.0, 1.00

Satisfaction by Gender:
Female
(N=574)		 Male
(N=203)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 10.0, 1.00 10.0, 2.00 10.0, 1.00

Satisfaction by Language:
Spanish
(N=451)		 English
(N=240)		 Bilingual
(N=66)		 Other
(N=20)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 10.0, 0 10.0, 2.00 9.50, 2.00 10.0, 1.25 10.0, 1.00

Satisfaction by Immigrant Status:
No
(N=194)		 Yes
(N=583)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 10.0, 2.00 10.0, 1.00 10.0, 1.00

Satisfaction by Refugee Status:
No
(N=702)		 Yes
(N=75)		 Overall
(N=777)
Client Satisfaction (Median, IQR) 10.0, 1.00 10.0, 2.00 10.0, 1.00

Satisfaction by Place of Origin:
Mexico
(N=367)		 United States
(N=194)		 Other Latin America
(N=134)		 Middle East
(N=23)		 Africa
(N=14)		 Asia
(N=8)		 Europe
(N=3)		 Overall
(N=743)
Client Satisfaction (Median, IQR) 10.0, 1.00 10.0, 2.00 10.0, 0 10.0, 1.00 9.50, 3.00 10.0, 1.50 9.00, 0.500 10.0, 1.00

Part 2: Qualitative Analysis

Patient Feedback

MHC Feedback (2018-2022)

Of the 414 MHC patients surveyed from 2018-2022, only 160 people provided additional comments to the question “If you could change anything about your treatment, what would it be?” and “Do you have any other feedback for us?”. Their comments were parsed using conventional content analysis, and 256 observations of 28 codes, or themes, emerged from the text. These fell into three broad categories (Figure 1) and were further collapsed into 16 codes (Table 1).

Table 1. Detailed Themes in Patient Feedback, MHC Screenings 2018-2022
Overall
(N=257)
Themes
Pleased with services 95 (37.0%)
Improved or benefited from therapy 33 (12.8%)
Staff/therapist is kind and attentive 7 (2.7%)
Therapist is skillful and competent 6 (2.3%)
Feel comfortable, safe, or relaxed in therapy 5 (1.9%)
Would prefer in-person therapy 22 (8.6%)
Would prefer a counselor who speaks the same language 11 (4.3%)
Would like longer or more frequent sessions 8 (3.1%)
Would like medication 6 (2.3%)
Barriers to treatment 4 (1.6%)
Little to no progress made on presenting issues 16 (6.2%)
Issue with therapist's methods 14 (5.4%)
Matching issue with therapist 11 (4.3%)
Dislike switching counselors frequently 8 (3.1%)
Displeased by Casa's communication or policies 8 (3.1%)
Unhelpful, not for me 3 (1.2%)

Of the 257 feedback codes, the most common was that patients were satisfied with services and desired no changes in their therapy (37%). 13% of comment codes signaled that patients had improved or benefited from therapy, either by increasing their self-awareness or gaining new social-emotional tools. Other positive comments included feeling comfortable and relaxed in therapy (2%) and feeling that their therapist was very competent (2.3%) and kind (3%).

The most frequent negative comment code was from patients who felt they had made little to no progress on their presenting issue(s), but only 3 of these patients were disinterested in continuing treatment. The second-most frequent code was displeasure with their therapist’s methods (24% of negative comments (n=60), but only 5.4% of entire sample). This was driven by patients who felt their therapist did not provide enough specific, actionable advice or tools for them to try outside of counseling. 4% felt they were not well matched with their therapist due to gender, race, age/lived experience, religion, and/or personality. Other negative comment codes were about disliking having to switch counselors frequently (3%) and displeasure with the communication from Casa’s receptionists and/or MHC policies (e.g. the no-show policy).

The remaining codes fell into a neutral category of desired changes or unmet needs. The #1 change patients wanted was to have in-person (9%) therapy. Many patients stated that they felt unable to talk freely about their problems over the phone or through Zoom because they were either not in a safe space or were around the people who were driving their problems. Other patients preferred in-person therapy because they were uncomfortable with technology and/or because they found it difficult to be vulnerable and connect with their therapist remotely. Only 2 patients expressly stated that they enjoyed telehealth services. In addition to changing the mode of therapy, patients asked for longer or more frequent sessions with their therapist, as well as to move away from using interpreters, with one couple noting that working with one was “very uncomfortable” and another patient questioning the fidelity of interpretation. Six patients’ comments concerned their mental health medications, with most needing refills or adjustment by a psychiatrist. Finally, 2% of comment codes described difficulties people face in attending appointments, such as unreliable transportation, childcare, cost, or scheduling conflicts.

Patient Satisfaction Survey for Casa Overall

Casa Clinic and GUIA patients are invited to complete an online patient satisfaction survey after every appointment. In the MHC, the survey is distributed via text every 6 months to minimize duplicates given the recurrent nature of therapy. This dataset contains all responses collected from July 20, 2021 to July 25th, 2022.

This dataset has 255 observations across 9 variables. Qualitative responses were analyzed using conventional content analysis.2

Variable Description Type Values
Timestamp

Time survey was submitted

(serves as Case ID / row count)

 categorical date/time
PreferredLanguage Clients’ Language categorical

  • English

  • Spanish/Portuguese
ServicesUsed Casa service client receives categorical

  • Mental Health Only

  • Clinic Only

  • GUIAs Only

  • All of them

  • Clinic & GUIAs

  • Clinic & Mental Health

  • Mental Health & GUIAs
WouldRecommend Likelihood patient would recommend Casa services categorical

  • Strongly Disagree

  • Disagree

  • Neutral

  • Agree

  • Strongly Agree
WouldStillChoose Likelihood patient would still choose Casa services if they had other options available categorical

  • Strongly Disagree

  • Disagree

  • Neutral

  • Agree

  • Strongly Agree
Cultura How sensitive Casa has been to patients’ cultural and linguistic needs categorical
  • Strongly Disagree
  • Disagree
  • Neutral
  • Agree
  • Strongly Agree
LikertComments Clients’ free-text comments on Casa overall categorical, qualitative

Individual comments were organized into 2 themes and sub-themes:

  • Positive

    • Kind and helpful staff

    • Quality of services: Good care and attention

    • Enjoy receiving services; beneficial; thankful

  • Negative

    • Difficult to schedule and receive services

    • Negative experience with provider or staff
OtherServices Clients’ free-text suggestions on services they would like to receive at Casa categorical, qualitative Individual comments were organized into themes (described below)
Table 1: Casa de Salud Patient Satisfaction Survey, 2021-2022 (n=255)
Spanish/Portuguese
(N=219)		 English
(N=36)		 Overall
(N=255)
Services Used
Mental Health only 30 (13.7%) 26 (72.2%) 56 (22.0%)
Clinic only 135 (61.6%) 6 (16.7%) 141 (55.3%)
GUIAs 6 (2.7%) 0 (0%) 6 (2.4%)
Clinic & GUIAs 17 (7.8%) 1 (2.8%) 18 (7.1%)
All of them 12 (5.5%) 0 (0%) 12 (4.7%)
Clinic & Mental Health 18 (8.2%) 3 (8.3%) 21 (8.2%)
Mental Health & GUIAs 1 (0.5%) 0 (0%) 1 (0.4%)
Would recommend Casa services
Strongly Disagree 3 (1.4%) 0 (0%) 3 (1.2%)
Disagree 1 (0.5%) 0 (0%) 1 (0.4%)
Neutral 12 (5.5%) 4 (11.1%) 16 (6.3%)
Agree 10 (4.6%) 7 (19.4%) 17 (6.7%)
Strongly Agree 193 (88.1%) 25 (69.4%) 218 (85.5%)
Would still choose Casa services even among other options
Strongly Disagree 2 (0.9%) 0 (0%) 2 (0.8%)
Disagree 1 (0.5%) 0 (0%) 1 (0.4%)
Neutral 19 (8.7%) 5 (13.9%) 24 (9.4%)
Agree 9 (4.1%) 6 (16.7%) 15 (5.9%)
Strongly Agree 188 (85.8%) 25 (69.4%) 213 (83.5%)
Casa is culturally sensitive
Strongly Disagree 1 (0.5%) 0 (0%) 1 (0.4%)
Disagree 2 (0.9%) 1 (2.8%) 3 (1.2%)
Neutral 12 (5.5%) 3 (8.3%) 15 (5.9%)
Agree 11 (5.0%) 3 (8.3%) 14 (5.5%)
Strongly Agree 193 (88.1%) 29 (80.6%) 222 (87.1%)

Summary: Of the patients surveyed (n=255), most used clinic services (n=192), followed by mental health services at the MHC (n=90). The majority of patients (n=218) spoke Spanish (only 1 Portuguese respondent). Of the English-speaking patients surveyed (n=36), 72.2% were MHC patients. All GUIA patients surveyed (n=37), save for one person, were Spanish-speakers. Nearly 86% of patients would recommend Casa services to others. 90% of Spanish/Portuguese-speakers and 86% English-speaking patients would continue to choose Casa services even if other options were available to them (“Agree” and “Strongly Agree”). 93% of patients believe Casa is sensitive of their cultural and linguistic needs (“Agree” and “Strongly Agree”).

Table 2: Casa de Salud Patient Satisfaction Survey -- Additional Comments, 2021-2022 (n=51)
Overall
(N=51)
Comment Themes
Kind and helpful staff 10 (19.6%)
Quality of service: Good care and attention 25 (49.0%)
Enjoy receiving services, benefit from them, thankful 9 (17.6%)
Difficult to schedule and receive services 5 (9.8%)
Negative experience with provider or staff 2 (3.9%)

Summary: Among the 51 additional comments patients provided, two main themes emerged: positive experiences with Casa and negative ones, with three and two sub-themes, respectively. 86% of comments were positive. The most common positive feedback was regarding the quality of service, rating it as good and attentive care. 20% of respondents noted how kind, warm, and helpful the staff is, with one patient saying that “todos desde la señora que hace la limpieza allí son muy cariñosas.” ~17.6% of patients said they enjoyed receiving services, had benefited from them, and/or were grateful for Casa services. The most compelling comment of gratitude was “My time at Casa has been invaluable and life-changing. I could not afford it otherwise. Thank you!!” Most negative comments (10%) were complaints about how difficult it is to reach the front desk via phone to schedule an appointment and the long-wait time for appointments. The remaining 4% (n=2) were about negative experiences patients had had with providers or staff who were described as “not very personable” or “desagradable.”

Figure 1. The most common words used in patient feedback (yellow= most frequent, red= frequent, blue= at least once). Amables was the word most used by patients in their comments about Casa.

Figure 2. Other services patients would like to see at Casa. Of the 43 suggestions provided, dentistry was the most common request (n=29), followed by vision (n=4), dermatology (n=2), and services for kids (n=2). The most unique requests were for yoga, tax-prep help, and applying for disability.

Analysis of Presenting Issues

During an intake screening, patients are asked what issues brought them to counseling / what they would they hope to achieve in counseling. These responses are recorded by volunteers and used to triage and match patients to the most appropriate counselor. Sometimes the volunteer enters a direct quote from a patient; other times, a summary of their issue is recorded. Below are the most 75 most common words used to describe patient presenting issues:

The presenting issues for the 414 unique patients in the MasterMHC dataset were coded using conventional content analysis. 28 themes, or codes, and 48 subcodes emerged from the text. Each patient had 1-5 codes/subcodes assigned to them, depending on the richness of their comment, resulting in 891 observations of 82 different combinations of presenting issue codes in total (Table 1). In R, codes were deaggregated per patient, and subcodes were collapsed into 29 broader categories (Table 2).

Table 1. Presenting Issues of MHC Clients, 2018-2022
Overall
(N=891)
All codes and subcodes
Anger issues -- unspecified 20 (2.2%)
Anger issues -- management 11 (1.2%)
Anxiety -- unspecified 103 (11.6%)
Anxiety -- COVID 1 (0.1%)
Anxiety -- panic attacks 8 (0.9%)
Anxiety -- for family in home country 3 (0.3%)
Depression -- unspecified 122 (13.7%)
Depression -- COVID 1 (0.1%)
Depression -- post-partum 5 (0.6%)
Domestic abuse 21 (2.4%)
Eating disorder 5 (0.6%)
Family problems -- unspecified 37 (4.2%)
Family problems -- estrangement 3 (0.3%)
Family problems -- separation, particularly from child 11 (1.2%)
Grief 32 (3.6%)
Guilt 6 (0.7%)
Insomnia 14 (1.6%)
Loneliness 9 (1.0%)
Toll of managing physical illness 16 (1.8%)
Marital problems -- unspecified 45 (5.1%)
Marital problems -- infidelity 10 (1.1%)
Marital problems -- separation/divorce 14 (1.6%)
Marital problems -- intimacy or reproductive issue 3 (0.3%)
Marital problems -- substance abuse 2 (0.2%)
Medication -- desired 4 (0.4%)
Medication -- other 5 (0.6%)
Memory problems 6 (0.7%)
Mood disorder -- unspecified 3 (0.3%)
Mood disorder -- mood swings 5 (0.6%)
Parenting issue 21 (2.4%)
Poor concentration 6 (0.7%)
Poor concentration -- ADHD 2 (0.2%)
Psychiatric evaluation -- disorder suspected 2 (0.2%)
Psychosis -- hallucinations 3 (0.3%)
Psychosis -- paranoia 5 (0.6%)
Psychosis -- schizophrenia 2 (0.2%)
Psychosomatic symptoms 11 (1.2%)
Quarantine 2 (0.2%)
Relationship problems -- romantic, friendships & other unspecified (excludes marriage & family) 25 (2.8%)
Relationship problems -- difficulty connecting with others 9 (1.0%)
Relationship problems -- romantic break-up 6 (0.7%)
Relationship problems -- sexual issue 2 (0.2%)
Issues relating to oneself -- other 10 (1.1%)
Issues relating to oneself -- body image issues 2 (0.2%)
Issues relating to oneself -- regulating emotions 7 (0.8%)
Issues relating to oneself -- identity issue, including sexual orientation 2 (0.2%)
Issues relating to oneself -- seeking advice on self-actualization (goals, etc) 4 (0.4%)
Issues relating to oneself -- improving self-esteem 20 (2.2%)
Stress -- unspecified 29 (3.3%)
Stress -- COVID 2 (0.2%)
Stress -- due to care-taking or loved one's affliction 11 (1.2%)
Stress -- immigration 16 (1.8%)
Stress -- finances 11 (1.2%)
Stress -- parenting 3 (0.3%)
Stress -- work/school 6 (0.7%)
Suicidal ideation or behavior 20 (2.2%)
Substance abuse -- unspecified 9 (1.0%)
Substance abuse -- alcohol 5 (0.6%)
Substance abuse -- drugs 3 (0.3%)
Trauma -- unspecifed 25 (2.8%)
Trauma -- abuse 11 (1.2%)
Trauma -- assault 9 (1.0%)
Trauma -- childhood 11 (1.2%)
Trauma -- incarceration 2 (0.2%)
Trauma -- past relationship 2 (0.2%)
Trauma -- PTSD 3 (0.3%)
Trauma -- reproductive 2 (0.2%)
Trauma -- war 2 (0.2%)
Undisclosed 2 (0.2%)
Court-mandated counseling 6 (0.7%)
Other external reason for counseling 4 (0.4%)
Referred by Casa clinic 21 (2.4%)
Referred by partner organization 4 (0.4%)
Referred by other 6 (0.7%)
Table 2. Broader Themes in Patient Feedback, MHC Screenings 2018-2022
Overall
(N=891)
Themes
Anger issues 31 (3.5%)
Anxiety 115 (12.9%)
Depression 128 (14.4%)
Domestic abuse 21 (2.4%)
Eating disorder 5 (0.6%)
Family problems 51 (5.7%)
Grief 32 (3.6%)
Insomnia 14 (1.6%)
Loneliness 9 (1.0%)
Toll of managing physical illness 16 (1.8%)
Marital problems 74 (8.3%)
Medication 9 (1.0%)
Memory problem 6 (0.7%)
Mood disorder 8 (0.9%)
Parenting issue 21 (2.4%)
Poor concentration 8 (0.9%)
Psych evaluation 2 (0.2%)
Psychosis 10 (1.1%)
Psychosomatic symptoms 11 (1.2%)
Quarantine 2 (0.2%)
Relationship problems 42 (4.7%)
Issues relating to oneself 51 (5.7%)
Stress 78 (8.8%)
Suicidal ideation or behavior 20 (2.2%)
Substance abuse 17 (1.9%)
Trauma 67 (7.5%)
Undisclosed 2 (0.2%)
Mandated counseling for legal reason 10 (1.1%)
Referral 31 (3.5%)

Summary: The most common presenting issues were depression and anxiety, followed by stress, trauma, and family problems. A patients’ comments were coded as “depression” if they explicitly used that word or if they described its symptoms as their chief complaint. For example, patients who described frequent crying, low motivation, and “feeling fragile” were grouped into this theme. The main sub-codes of depression were post-partum and COVID-19-related. Anxiety included patients who used any derivative of the word anxious, as well as those who described feeling fearful, nervous/worried, and panicked.

Stress had numerous sub-categories, the most common being stress secondary to immigration. This broad subcategory captured all people who were having difficulty adjusting or assimilating to their new environment, having trouble with the language barrier, stressing about family dispersal and safety (including the threat of deportation), and those who were stressed about how their children would grow up in an environment entirely different from what they knew. This subcategory demonstrates how stress can go hand-in-hand with anxiety, as most of these concerns could have been listed under either category. The desperation felt by patients was best captured by one patients’ comment: “Is it worth it being in the U.S.?”

The fourth-most common presenting issue was marital problems, most commonly due to separation/divorce and infidelity. Other marital problems were due to financial concerns, differing parenting styles, substance abuse of one partner, and reproductive/intimacy issues, including impotence, sexual frustration, and disagreement on having children.

While many patients did not describe their trauma in great detail during an intake, those that did revealed that MHC patients have been exposed to a high level of inter-personal violence. Of the 67 trauma-related presenting issues, abuse (including sexual, physical, verbal, and emotional abuse), assault (sexual and physical assault) perpetuated by anyone other than an intimate partner (abuse/assault perpetuated by an intimate partner was captured under the separate domestic abuse category), and childhood trauma (unspecified) were the most common trauma-related comment codes. Less common traumas included incarceration, abortions, miscarriages, and war.

Family separation was the most common sub-theme within family problems, and this included any undesired separation from their child (e.g., people whose children were taken by DHS, by violence (e.g., kidnapping, n=1), etc), and families unable to immigrate together to the U.S. at the same time. The seventh-most common presenting issue was “Issues relating to oneself,” which captures all comments that expressed difficulty with self-perception, self-regulation, self-actualization, and identity.

Some patients came to the MHC because their physical health problems were causing mental health issues, including stress and depression (“toll of managing physical illness”). Others came because they believed their mental illness was manifesting physically as pain, shaking, and vision problems (“psychosomatic symptoms”). Many of these patients were referred from Casa (n=21) while 10 others were referred from other local organizations.

Many patients had had previous mental health treatment in their country of origin and had standing or suspected diagnoses, but until finding Casa, had been unable to pursue treatment due to financial barriers and more pressing stressors. Patients expressed the desire to retake their treatment, in hopes that counseling would provide relief and direction, as well as potentially lead to physical outputs such as a formal psychiatric evaluation and/or mental health prescription. Interestingly, very few people (n < 6) brought concerns about work or school. The majority of presenting issues were more serious, traumatic, family/relationships-related issues.

Figure 2: The most common themes in patients’ presenting issues.

Discussion & Recommendations

Overall, MHC outcomes are very positive. There was no instance where a larger percentage of scores worsened than improved, across any factor. Overall, patients improved 4 points in depression score, 3 points in anxiety, and 2.9 points in trauma score on average, across all follow-up time points. Improvement was strong at 3 months for all three mental health screenings, but no other consistent pattern applicable to all three screenings across all the follow-up milestones was observed.

We found that mental health outcomes do not vary by therapist type. The changes in depression, anxiety, and trauma scores of patients working with students and patients working with therapists were very similar between both groups overall, and they were not significantly different at three months or at discharge. This result matches those seen by Ost et al.,5 who found that students practicing cognitive behavioral therapy (one of the many modalities practiced in the MHC) under close supervision of experienced therapists (as student interns are at the MHC) can have treatment effects that are comparable to those achieved by unsupervised experienced therapists. It is worth noting that this result may also be a reflection of the fact that licensed therapists receive more severe/complex cases than students do and thus their patients’ scores may change less than patients with less severe conditions.

We also found that depression, anxiety, and trauma outcomes did not vary by interpreter use. The changes in scores were similar between both groups overall, and they were not significantly different at three months or at discharge. While few studies have explored the effect of interpreter use in psychotherapy, studies have shown that professional interpreters in outpatient clinical settings have been shown to elevate the clinical care received by patients with limited English proficiency to “approach or equal” that of patients with language-concordant providers.3 This result may speak to the high quality of interpreter services employed at Casa, as well as to the success of matching the majority of patients to the bilingual/language-concordant providers available at Casa.

Therapy delivery mode was the only factor that significantly influenced mental health outcomes. While the numeric change in depression, anxiety, and trauma score did not differ significantly between therapy modes, the percent of depression and trauma scores that improved did. There were significantly more improved depression and trauma scores for in-person patients than telehealth patients than expected. The greater improvement in depression scores for in-person patients than telehealth may be explained by the fact that the act of physically going to therapy and receiving positive reinforcement during sessions is a form of behavioral activation, which has been found to be an effective treatment for depression.1 As for trauma scores, best practices for trauma-informed therapy, in-patient or outpatient, state that clients need to feel connected, valued, informed, and hopeful of recovery,4 which may be easier to achieve when in direct contact with a therapist rather than through a screen or over the phone.

Limitations

This retrospective analysis is for internal use only and should not be used to inform clinical decisions. It should also be used with caution when informing programmatic decisions due to threats to internal validity due to bias and measurement error in data collection. The largest threat to validity is that patients receive different doses of therapy (i.e., differing number of appointments with their counselor) between follow-up calls, and many also switch between student and licensed therapists, as well as between therapy delivery modes between follow-up assessments. To minimize the effect of the latter, a higher level of granularity was used when classifying patients as telehealth, hybrid, or in-person, taking care to match the format of therapy to that most used around the time of the follow-up call using the MHC calendars and Remote Appointments spreadsheet as cross-references. Further concerns to internal validity include self-response bias and social-desirability bias, as well as the risk of patients not understanding the questions asked due to linguistic/cultural barriers (there were several comments noting that interpreters mentioned to the volunteer that the patient may have misunderstood the format of the screening questions). Finally, given the disruptions of COVID-19 to both therapy delivery and volunteer recruitment/retention, follow-up calls were not completed for many patients in 2020 and 2021. Thus, there is an inconsistent number of calls per follow-up time point.

As for the qualitative analysis, it is worth noting that presenting issues are interpreted, summarized, and transcribed by volunteers, so there is a risk that issues may be misrepresented. Given data collection procedures, inter-reliability checks are not possible between volunteers to know if volunteers are interpreting issues the same way. For example, one volunteer might hear that a patient is “tired, low energy” and enter “depression” into their chart without additional details; others may type the patient’s comment verbatim; and still a minority might wager a guess at a physical problem and enter “poor sleep.” Generally, volunteers write more details than less, which is helpful in evaluating the issue. However, given the richness of the comments, saturation was not reached, as new subcodes continued to emerge even in the third round of coding.

Recommendations

Appendix

A repository of detailed tables and alternate versions of graphs, as requested by LGK.

1. Change in Depression (PHQ-9) and Anxiety (GAD-7) Scores by Type of Therapist by Discharge Month, Casa de Salud MHC 2018-2022 (n=196)
Licensed
Student
3
(N=48)		 6
(N=20)		 9
(N=9)		 12
(N=10)		 15+
(N=18)		 3
(N=40)		 6
(N=32)		 9
(N=11)		 12
(N=3)		 15+
(N=5)
Depression Outcomes
Improved Depression 39 (81.3%) 13 (65.0%) 7 (77.8%) 5 (50.0%) 9 (50.0%) 32 (80.0%) 23 (71.9%) 8 (72.7%) 1 (33.3%) 4 (80.0%)
No change in Depression 1 (2.1%) 2 (10.0%) 2 (22.2%) 2 (20.0%) 3 (16.7%) 0 (0%) 2 (6.3%) 1 (9.1%) 1 (33.3%) 0 (0%)
Worsened Depression 8 (16.7%) 5 (25.0%) 0 (0%) 3 (30.0%) 6 (33.3%) 8 (20.0%) 7 (21.9%) 2 (18.2%) 1 (33.3%) 1 (20.0%)
Anxiety Outcomes
Improved Anxiety 39 (81.3%) 14 (70.0%) 8 (88.9%) 7 (70.0%) 9 (50.0%) 31 (77.5%) 25 (78.1%) 9 (81.8%) 2 (66.7%) 4 (80.0%)
No change in Anxiety 4 (8.3%) 2 (10.0%) 0 (0%) 0 (0%) 1 (5.6%) 1 (2.5%) 1 (3.1%) 1 (9.1%) 0 (0%) 1 (20.0%)
Worsened Anxiety 5 (10.4%) 4 (20.0%) 1 (11.1%) 3 (30.0%) 8 (44.4%) 8 (20.0%) 6 (18.8%) 1 (9.1%) 1 (33.3%) 0 (0%)
2. Change in Trauma (PCL5-DSMIV) Scores by Therapist Type by Discharge Month, Casa de Salud MHC 2018-2021 (n=93)
Licensed
Student
3
(N=21)		 6
(N=11)		 9
(N=7)		 12
(N=5)		 15+
(N=5)		 3
(N=23)		 6
(N=13)		 9
(N=6)		 12
(N=2)
Trauma Outcomes
Improved Trauma 13 (61.9%) 7 (63.6%) 7 (100%) 3 (60.0%) 3 (60.0%) 21 (91.3%) 11 (84.6%) 4 (66.7%) 2 (100%)
No change in Trauma 3 (14.3%) 1 (9.1%) 0 (0%) 1 (20.0%) 0 (0%) 1 (4.3%) 0 (0%) 2 (33.3%) 0 (0%)
Worsened Trauma 5 (23.8%) 3 (27.3%) 0 (0%) 1 (20.0%) 2 (40.0%) 1 (4.3%) 2 (15.4%) 0 (0%) 0 (0%)

References

1.
Ekers D, Webster L, Van Straten A, Cuijpers P, Richards D, Gilbody S. Behavioural Activation for Depression; An Update of Meta-Analysis of Effectiveness and Sub Group Analysis. Aleman A, ed. PLoS ONE. 2014;9(6):e100100. doi:10.1371/journal.pone.0100100
2.
Hsieh HF, Shannon SE. Three Approaches to Qualitative Content Analysis. Qualitative Health Research. 2005;15(9):1277-1288. doi:10.1177/1049732305276687
3.
Karliner LS, Jacobs EA, Chen AH, Mutha S. Do Professional Interpreters Improve Clinical Care for Patients with Limited English Proficiency? A Systematic Review of the Literature. Health Services Research. 2007;42(2):727-754. doi:10.1111/j.1475-6773.2006.00629.x
4.
Muskett C. Trauma-informed care in inpatient mental health settings: A review of the literature. International Journal of Mental Health Nursing. 2013;23(1):51-59. doi:10.1111/inm.12012
5.
Öst LG, Karlstedt A, Widén S. The Effects of Cognitive Behavior Therapy Delivered by Students in a Psychologist Training Program: An Effectiveness Study. Behavior Therapy. 2012;43(1):160-173. doi:10.1016/j.beth.2011.05.001