Therapist-Driven Telephone Call Outreach to Improve Depression Remission in a Federally Qualified Health Center

This study took place during 2020–2022 at a federally qualified health center (FQHC) located in the Midwest. Of the 50,067 patients served in 2022, about 21,054 (42.1%) were uninsured, 22,829 (45.6%) were at or below the federal poverty level, and 38,970 (77.8%) were members of racial-ethnic minority groups, primarily Hispanic or Latino (6). Spanish was identified as the preferred language by 29,028 (58.0%) of the patients. The center’s 1-year depression remission rates were 7%, 7%, and 12% in 2018, 2019, and 2020, respectively. Depression remission at 12 months is defined by the Health Resources & Services Administration (HRSA) as a Patient Health Questionnaire–9 (PHQ-9) score <5 at 12 months±60 days after an initial depression diagnosis (6). The study protocol was reviewed by the Creighton University Institutional Review Board and given a quality improvement waiver (reference 2002591).

Patients included in this study were adults (ages ≥19 years) who received a diagnosis of major depression or dysthymia; had a PHQ-9 score >9 between November 1, 2020, and October 31, 2021; and did not have a recorded PHQ-9 score at a follow-up 1 year±60 days later. Patients with a concurrent diagnosis of bipolar disorder, personality disorder, schizophrenia, psychotic disorder, or pervasive developmental disorder were excluded. Patients were also excluded if they were receiving hospice or palliative services, were residents of permanent nursing homes, or died during the period of the intervention. Previous or current depression treatment was not an exclusion criterion.

Starting in January 2021, the health center’s electronic health record (EHR) system was updated so that it could automatically add depression to the medical diagnosis list in the patient chart and enable the generation of a list of patients due for 1-year follow-up PHQ-9 screenings. The list of patients was divided monthly among 10–13 therapists who telephoned patients from June 2021 to May 2022 to perform PHQ-9 screenings and recommend treatment based on the updated score (a procedure flowchart is available in Figure S1 of the online supplement to this column). Lists were separated according to the patient’s preferred language. Bilingual therapists interviewed Spanish-speaking patients by using the Spanish version of the PHQ-9. Patients who received a positive score for suicidal ideation on the PHQ-9 were further screened with the Columbia–Suicide Severity Rating Scale (C-SSRS). Recommended treatments included warm handoff, cold handoff, and visit with a medical provider. No additional treatment was recommended if a PHQ-9 score was <5. A warm handoff was defined as a therapy session conducted via telephone, and a cold handoff was a session scheduled for a later date. Patients were offered a warm handoff or a cold handoff if the updated PHQ-9 score was ≥5 or if the screening indicated suicidal ideation. Along with the offered therapy session, patients were recommended to schedule an appointment with their primary care physician if their PHQ-9 score was ≥10. If patients had an updated PHQ-9 score of ≥20 or were determined to be at high risk on the C-SSRS, an electronic consultation message to the psychiatric team was initiated. The therapists attempted to call up to three times and left a voicemail if they were able to. If the patient returned the call, a behavioral health receptionist conducted the screening and offered treatment based on the aforementioned score criteria. Patients with a PHQ-9 score of ≥10 were advised to seek additional care from a primary care physician, and follow-up scores and visits were recorded for 6 months after the call with the therapist.

One year after implementation of the intervention, a semistructured focus group was conducted with all 13 therapists by using an interview guide (Table S1 in the online supplement) to understand their experiences with the intervention. Interviews were audio-recorded and later transcribed by one author (R.L.) or with transcription software (Otter.ai). We performed a basic qualitative analysis to identify frequently mentioned themes. The results of this analysis were later presented to interview participants to solicit their feedback about identified themes. This member check was performed by two authors (S.K., R.L.).

Descriptive rates for clinically relevant improvement (defined as a decrease in PHQ-9 score of ≥50%) and depression remission, intervention rates among patients experiencing continued depression symptoms, and year-end institutional remission rates were estimated as frequencies and percentages. Statistical significance of 1-year changes in PHQ-9 scores was assessed with paired-sample t tests. Six-month treatment reengagement among patients experiencing continued depression symptoms was evaluated by using a linear mixed model with random intercepts to account for the correlation of repeated observations by individual patients. Additionally, a fixed-effect predictor for treatment type and an interaction effect of treatment type with time were estimated to evaluate the difference in change over time (up to 6 months after the initial assessment).

During the 12-month intervention window, 795 patients met the inclusion criteria, 502 (63%) of whom received at least one contact attempt (i.e., a telephone call). Of the patients with a contact attempt, 241 (48%) were contacted, 225 (93%) of whom completed a PHQ-9 questionnaire, and of 150 patients who did not have remission in depression symptoms, 136 (91%) agreed to a therapy session. The full intervention results are shown in Figure S2 in the online supplement.

Before evaluating PHQ-9 changes among patients who were contacted for follow-up, we compared mean baseline index scores between patients with follow-up and those lost to follow-up. On average, no significant differences in baseline PHQ-9 scores were observed between patients lost to 1-year follow-up (mean±SD score=16.6±4.5) and those who were reached (mean score=16.1±4.4), mitigating the concern that the observed 1-year change in PHQ-9 score was due to nonresponse bias (given that 261 [52%] of the 502 patients were not reached).

On average, the PHQ-9 score dropped by 7 points over the 1-year study period (95% CI=5.94–7.91, p<0.001). Among the 225 participants with both baseline and follow-up PHQ-9 scores, scores decreased for 187 (83%) participants, did not change for 7 (3%) participants, and increased for 31 (14%) participants. Notably, 105 (47%) participants had a score reduction of ≥50%, indicating clinically significant depression responses, and 75 (33%) met the criteria for remission (PHQ-9 score <5) at the 1-year follow-up.

Of 88 patients who participated in the 1-year survey and who had a PHQ-9 score of ≥10 (i.e., patients who were recommended for treatment), 63 (72%) attended at least one additional clinic (medical or behavioral health) appointment over the next 6 months that addressed depression. For these patients, the median number of additional follow-up visits was two (interquartile range 1–3, maximum=10). Evaluating the change in PHQ-9 score over time among the patients with high initial PHQ-9 scores, we observed a statistically significant decrease of 0.72 points in PHQ-9 score per month. No significant interaction was observed among treatment types (warm handoff, cold handoff, or PHQ-9 assessment only), suggesting similar rates of change in PHQ-9 score for these treatment types over the follow-up period (Figure S3 in the online supplement).

Clinic-level 1-year depression remission rates, defined as above, increased to 14% (161 of 1,115 patients had a PHQ-9 score <5) in 2021 and to 15% (190 of 1,241 patients) in 2022. This metric is calculated as a year-end measure, but this quality improvement project spanned the middle of 2021 and 2022. A qualitative analysis of the focus group found that the therapists perceived cold-calling as unhelpful. Future implementation should consider notifying patients of the telephone call via a prior text message and use of a designated, recognizable clinic telephone number. Furthermore, therapists noted positive feedback and gratitude from patients who had been contacted and that the intervention was useful for following up with patients who typically become lost to follow-up. (Table S2 in the online supplement provides additional details and representative quotes.)

This project may have contributed to the health care center’s improvement in depression remission scores of 14%–15% in 2021 and 2022, compared with the 3 years before the intervention (7%–12% in 2018–2020 [6]). The 2021 and 2022 percentages of remission surpassed the national 2022 average of 13.6% reported by the HRSA’s National Health Center Program Uniform Data System Awardee Data, which uses the same measures as this study, and the 11% (2021) national average among FQHCs (6). Specifically, and by design, this improvement primarily reflected an increase in data collection, an essential first step in health care improvement. The intervention led to the inclusion of data from an additional 241 patients in the 2021–2022 depression remission data set by gathering PHQ-9 information via telephone (see Figure S4 in the online supplement). It is noteworthy and encouraging that depression symptoms were in remission for 33% of the patients contacted via telephone. The telephone call also presented an opportunity for treatment. Patients whose symptoms were not in remission were offered a therapy session, and 91% of these patients agreed to participate either on the same day or a few days later. These findings complement those of past studies about patient preferences of Latinos in public-sector primary care settings, which found that patients preferred counseling or counseling plus medication to medication alone, availability of telephone sessions, and receiving their mental health treatment in primary care (7).

Findings were mixed regarding implementation feasibility. This intervention was an additional responsibility on top of therapists’ already busy schedules, and assigned lists were not fully completed—the therapist team attempted to contact only 63% of patients on the EHR list. Lack of time and need for clinic productivity were listed as barriers to the outreach intervention. Therapists were encouraged that if the telephone call led to a warm handoff or cold handoff, the call would be counted toward their productivity. At the end of the project, the therapists expressed positive regard for the intervention, emphasizing that patients felt cared for because of the telephone call.

This study had some limitations. Its design, including the lack of a comparison group, precludes causal statements. Social desirability bias may have contributed to the observed improvements in PHQ-9 scores, but this concern is partially mitigated by the fact that patients responded via telephone rather than in person. This study had no measures of comorbid conditions, treatments, caregivers’ opinions, or psychosocial stressors but instead relied on the PHQ-9 as the only measure of depression severity. Individual patient characteristics such as age, race and ethnicity, income, insurance status, and primary language spoken were not analyzed because this was a quality improvement project that focused on process.

The findings of this quality improvement project and of other studies of telephone-based follow-up (4, 5) suggest that patient outreach interventions can be effective. Our results indicate that these interventions improve depression remission and treatment reengagement among adults in a primary care population with a high proportion of non-White, Spanish-speaking, and uninsured patients. Our findings contribute to the growing literature supporting the use of patient outreach to improve depression remission, inform measurement-based care, and increase follow-up care. These findings support the need for further research to assess the effects of implementation and continued use of therapist outreach in primary care clinics in which the patient population comprises underserved individuals.