Virtual Collaborative Care Versus Specialty Psychiatry Treatment for Depression or Anxiety

Virtual Collaborative Care Versus Specialty Psychiatry Treatment for Depression or Anxiety

Approximately 1 in 5 adults in the United States experience a lifetime depressive disorder,^1,2 leading to significant disability, disability-related burden,³ and $326.2 billion in annual costs.⁴ Depression impacts outcomes of many prevalent, chronic medical conditions including diabetes and hypertension, along with quality of life, functioning, and overall health status.⁵ Depressive disorders impart high risk for suicide,^6,7 a leading cause of death across age groups in the United States.⁸ Similarly, anxiety disorders impact 31% of the US population over their lifetimes⁹ and are associated with significant functional impairment⁹ and poor outcomes.¹⁰ Timely, evidence-based care with psychotherapy and/or medications has been shown to improve outcomes for both disorders.^11–13 Yet, approximately half of patients do not receive adequate depression or anxiety care,¹⁴ suggesting that strategies to improve patient connection with evidence-based care are needed.

One evidence-based care model associated with reduced depressive or anxiety symptoms, improved quality of life, and sustained remission is collaborative care.^15,16 Collaborative care involves a multidisciplinary team focused on population-based treatment using measurement-based care, treatment to target, accurate diagnosis, and patient-centered care.¹⁷ Randomized controlled trials and meta-analyses have demonstrated that collaborative care is superior to noncollaborative care models in primary care for depression and anxiety.^15,18–21 Yet, collaborative care is not widely implemented, and the majority of patients with depression or anxiety are referred to specialty psychiatry, which is associated with limited referral connection and treatment delays.^22–24 Such delays are associated with poorer patient outcomes,²⁵ 43%–52% higher ambulatory costs,^26–29 and >$210 billion lost earnings/year.⁴

To date, outcomes for depression or anxiety after treatment in specialty psychiatry compared to collaborative care are unclear. Further, there is a relative lack of data regarding real-world collaborative care implementation outcomes,^30–36 which would facilitate collaborative care implementation, spread, and real-world practice. Finally, previous studies were conducted prior to the COVID-19 pandemic when telehealth was considered an alternative strategy for care delivery.³⁷ As such, research examining whether collaborative care can be successfully implemented in a virtual care first setting is needed.³⁸

Here, we compare depressive or anxiety symptoms between patients treated in a virtual collaborative care program versus virtual specialty psychiatry treatment. We hypothesized that patients enrolling in collaborative care would have improved depressive or anxiety symptom scores at 6 months comparable to patients in specialty psychiatry.

METHODS

Design Overview

This was a retrospective observational study with target trial emulation methods (Supplementary Table 1 outlines the specified and emulated target trial study designs).³⁹ This study followed the Strengthening the Reporting of Observational Studies in Epidemiology(STROBE) guidelines for observational studies with the target trial framework⁴⁰ and was approved by the Kaiser Permanente Northern California (KPNC) Institutional Review Board. Research was conducted in accordance with the principles of the Declaration of Helsinki. All variables were derived from KPNC’s electronic health record (EHR) system.

Setting/Participants

KPNC is a large, integrated health care system serving >4.5 million patients (36% of the regional population) through commercial, Medicare, Medicaid, and insurance exchange plans. Patients represent the ethnic and socioeconomic diversity of surrounding/ statewide population.⁴¹ Eligible patients had depression or anxiety diagnoses (Supplementary Table 2: International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM], codes) between April 1, 2020, and May 31, 2021; were aged ≥18 years; had Patient Health Questionnaire (PHQ)-9 or Generalized Anxiety Disorder (GAD)-7 Questionnaire score of ≥5 and <15; and had ≥1 year of continuous KPNC membership prior to the index date. Exclusion criteria included suicide risk (answer of ≥1 on question 9 of the PHQ-9 or a suicide-related diagnostic code [R45.85, T14.91, T36.xx3(4)-T50.xx2(4), T51.xx2(4)-T65.xx2(4), T71.xx2(4), X71-X83]⁴²) ≤30 days prior to eligibility screening. Additional exclusion criterion for the specialty psychiatry group was history of collaborative care program care. Multivariate analysis was limited to patients with PHQ-9/GAD-7 ≥10 and primary depression or anxiety diagnoses (not adjustment disorders) prior to baseline.

Interventions and Assignment

Study procedures and protocol were previously published.⁴³ Consistent with target trial emulation, patients undergoing eligibility screening were eligible for either intervention and came from the same source population (Supplementary Table 1). We identified patients referred by primary care or self-referred to screening eligibility (index visit) where symptoms were assessed, and patients were assigned to treatment based on shared decision making and program access.

Collaborative care. This intervention followed collaborative care principles including validated tool use, accountable care and population management–aligned evidence-based treatments, systematic follow-up, care team and patient communication and coordination, and program oversight.^17,43 Novel program components included integration of the care manager role by therapists into every visit, medication management by program pharmacists, integration of registry clinical outcome tracking into the EHR and visit note, and treatment to target guided by workflows.⁴³ The entire program was designed to be virtual prior to the COVID-19 pandemic.⁴³ At the first treatment (baseline) appointment, the therapist remeasured depressive or anxiety symptoms, performed treatment goal setting, and scheduled follow-up appointments. Treatment consisted of problem-solving therapy weekly or every other week. If a patient’s condition worsened or did not improve after 2–4 weeks, therapists offer pharmacist medication consultation who could discuss, prescribe, and titrate antidepressants by clinical protocol supplemented with case conference and psychiatrist consultation. Measurement-based symptom tracking was performed at each appointment; outreach was attempted 6 months post baseline. All team members and referring providers received ongoing education in care models and workflows.

Specialty psychiatry. At the baseline appointment, the therapist remeasures symptom scores, performs treatment goal setting, and schedules follow-up therapy or medication management appointments at the discretion of the therapist and patient. Medication management was performed by psychiatrists or clinical pharmacists when deemed necessary by a patient or therapist. Measurement based symptom tracking was performed through PHQ-9 or GAD-7 at baseline and each appointment.

Outcomes

Mean difference in continuous PHQ-9 scores for depression or GAD-7 scores for anxiety was the primary treatment outcome.

Depressive symptoms. Depressive symptoms were measured using the PHQ-9 (9 questions reflecting the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition [DSM-5], criteria as “0” [not at all] to “3”[nearly every day]; scores range from 0 to 27). The PHQ 9 is validated for depressive symptom screening and follow up with high sensitivity (>88%) and specificity (>88%) for major depression. PHQ-9 scores of 5, 10, 15, and 20 represent mild, moderate, moderately severe, and severe depressive symptoms.⁴⁴

Anxiety symptoms. Anxiety symptoms were measured using the GAD-7 Questionnaire (7 questions reflecting DSM-5 criteria for GAD with responses ranging from “0” [not at all] to “3” [nearly every day]; scores range from 0 to 21). The GAD-7 is validated for screening and follow-up with a sensitivity of 89% and specificity of 82%.⁴⁵

Symptom response and remission. Symptom response was defined as a ≥50% reduction in symptoms,⁴⁶ while remission was defined as a PHQ-9 score of <5 for depression and a GAD-7 score of <3 for anxiety.^44,45

Covariates

Covariates evaluated at the index date included age, sex, race, ethnicity, and neighborhood deprivation index (NDI, a geocoded measure of socioeconomic status with categories based on the tertiles of the overall cohort; higher scores indicate more deprivation). Body mass index (BMI) was evaluated at the last measurement in the year prior to the index date; the Charlson comorbidity index (CCI) was calculated from data in the year prior to the index date. A positive history of clinic encounters containing psychiatric diagnoses in the year prior to the index date was defined as >3 in-person, telephone, virtual, inpatient, institutional, rehab, or nursing facility clinic encounters containing psychiatric diagnoses using a clinician-curated list of ICD-10-CM codes for psychiatry, substance use, and lifestyle. Patient antidepressant use in the year prior to the index date and baseline symptom score were also included. Covariates evaluated during the follow-up period included antidepressant drug use during the follow-up (defined as 1 day to up to 6 months after baseline).

Statistical Analysis

A significance level (type I error rate) of 0.05 was set for all tests. Clinical and demographic characteristics were summarized using means and proportions. Using the full cohort, we used bivariate analysis to examine treatment assignment dependency on patient characteristics. We identified gender, race, age, CCI, NDI, screening symptom score, history of clinic encounters containing psychiatric diagnoses in the year prior to the index date, antidepressant therapy in the year prior to the index date, and time from eligibility screening to baseline symptom assessment as significant for both depression and anxiety and BMI for depression. Using these results, we conducted a propensity score weight analysis using inverse probability of treatment weighting (IPTW, assuming every patient in the population would be offered the treatment) to create a variable weight used in all subsequent analyses to emulate balance achieved via randomization for known covariates. A weighted linear mixed-effects model including patients with a baseline PHQ-9/GAD-7 score of ≥10 and a primary depression or anxiety diagnosis (not adjustment disorders) evaluated the outcome of mean PHQ-9 difference for patients with depressive disorders or mean GAD-7 difference for patients with anxiety disorders, adjusting for baseline scores. For patients without follow-up symptom measures, we imputed baseline symptom value assuming a conservative null effect (average mean difference of zero). Models included treatment group, continuous time, interaction between treatment group and time, and covariates as fixed effects and participants as random effects. Models were fit using the restricted maximum likelihood method assuming a spatial power covariance structure due to unequally spaced measurements, allowing all available responses to be used and handling data missing at random. Fitted models were used to estimate the adjusted mean difference (AMD) in PHQ-9 or GAD-7 scores at time points of interest. Unadjusted bivariate analysis estimated mean scores by day to illustrate trends in follow-up scores to compare interventions. All analyses were performed using SAS 9.4 (Cary, North Carolina).

RESULTS

Patient Characteristics

There were N = 10,380 patients (15% in virtual collaborative care; 85% in virtual specialty psychiatry) with depressive disorders and N = 2,935 (19% in collaborative care; 81% in specialty psychiatry) with anxiety disorders. For the multivariate analysis including only patients with a baseline PHQ-9/GAD-7 score of ≥10 and a depression or anxiety diagnosis, there were N = 4,563 patients (14% in collaborative care; 86% in specialty psychiatry) with depressive disorders and N = 1,805 (20% in collaborative care; 80% in specialty psychiatry) with anxiety disorders (Figure 1). During the study period, 99.5% of specialty psychiatry appointments were virtual. For both interventions and diagnoses, most patients were <40 years old, female, and non-Latinx white (Table 1). There were significant differences by intervention in these characteristics and baseline BMI, CCI, and NDI. Clinical characteristics differed between interventions, including prior year history of clinic encounters with a psychiatric diagnosis (P < .0001 for both diagnoses), time of eligibility screening to baseline symptom assessment (P = .002 for both diagnoses), and antidepressant use in the year prior to the eligibility screening (P < .0001 for both diagnoses). Baseline PHQ-9 scores were 9.1 ± 3.5 for collaborative care and 10.0 ± 4.2 for specialty psychiatry (P < .0001); baseline GAD-7 scores were 10.6 ± 2.9 for collaborative care and 11.2 ± 3.1 for specialty psychiatry (P < .0001).

Response to Treatment Over Time for Depressive Disorders

For patients with major depressive disorder, patients in both interventions showed significant improvements in depressive symptoms compared to baseline (Figure 2A and 2B), with patients in collaborative care showing an AMD of −9.0 (95% CI, −9.7 to −8.4; Table 2) and those in specialty psychiatry showing an AMD of −5.0 (95% CI, −5.6 to −4.5) at 6 months; collaborative care showed significantly greater improvement compared to specialty psychiatry (P < .0001). Estimates at the mean treatment time (72 days) showed similar significant improvements. In an analysis including patients with only an adjustment disorder diagnosis, depressive symptoms significantly improved in both interventions at 6 months (AMD =−9.1, 95% CI, −9.8 to −8.3 in collaborative care; AMD =−5.6, 95% CI, −6.3 to −4.8 in specialty psychiatry) with collaborative care having significantly higher improvements (P < .0001).

There was a significant time-by-treatment interaction in collaborative care vs specialty psychiatry (P < .0001; Table 3). Lower baseline depressive symptoms, time from baseline to treatment start, antidepressant use during treatment, and older age were significantly associated with depressive symptom reduction during follow-up (P < .0001 except age 40–64 years; P = .042). Antidepressant use in the year prior to screening (P = .011) and a history of clinic encounters containing psychiatric diagnoses in the year prior (P < .0001) were all associated with worsening depressive symptoms during the follow-up. Post hoc examination of unadjusted rates of depression response and remission using the full cohort (N = 10,380) at the last follow-up measure revealed that 55% of patients had treatment response and 43% had depression remission in virtual collaborative care vs 36% response and 27% remission in virtual specialty psychiatry (Supplementary Table 3).

Response to Treatment Over Time for Anxiety Disorders

For patients with primary anxiety disorders, patients in both interventions showed significant anxiety symptom improvement from baseline (Figure 2C and 2D), with virtual collaborative care showing an AMD of −5.4 (95% CI, −6.2 to −4.7; Table 2) and specialty psychiatry showing an AMD of −2.8 (95% CI, −3.6 to −2.1) at 6 months. Patients in collaborative care had a significantly greater improvement compared to those in specialty psychiatry (P < .0001). Estimates at the mean treatment time (49 days) revealed that patients in virtual collaborative care (AMD = −2.6, 95% CI, −3.2 to −2.1) and specialty psychiatry (AMD = −1.3, 95% CI, −1.8 to −0.7) had significant improvement in anxiety symptoms, with collaborative care having significantly greater improvements compared to specialty psychiatry (P < .0001). In an analysis including patients with adjustment disorder diagnoses, anxiety symptoms significantly improved in both groups (AMD = −3.4, 95% CI, −6.5 to −0.4 in collaborative care; AMD = −3.8, 95% CI, −6.5 to −0.4 in specialty psychiatry).

There was a significant time-by-treatment interaction in collaborative care vs specialty psychiatry (P < .0001; Table 3). Lower baseline anxiety symptoms (P < .0001) and time from baseline to treatment (P < .0001) were significantly associated with symptom reduction at follow-up, while a high NDI (score of 2 [P = .043]) was significantly associated with higher follow-up symptoms compared to baseline. Post hoc examination of unadjusted response and remission rates using the full cohort (N = 2,935) at last measure revealed that 14% of collaborative care patients had treatment response and 0.3% had anxiety remission vs 10% response and 0.1% remission in specialty psychiatry (Supplementary Table 3).

DISCUSSION

These results show that a virtual collaborative care program for depression with novel programmatic features⁴³ was associated with significant improvements durable up to 6 months after treatment initiation compared to specialty psychiatry. On average, patients moved from moderate symptoms to not meeting criteria for a depressive disorder (Table 2). Similarly, patients with anxiety showed clinical improvements during collaborative care treatment at least as effective as specialty psychiatry (Table 2). Our results extend previous work comparing collaborative care for depression to enhanced primary care.⁴⁷ In addition to demonstrating the effectiveness of a virtual collaborative care program compared to specialty psychiatry in a large population, we demonstrate that novel programmatic elements, including manager role integration into each clinical contact and utilization of pharmacist prescribers,⁴³ are efficacious.

Collaborative care had a large effect size (Cohen d = −2.23; 95% CI, −2.46 to −2.00) for major depressive disorder; previous efficacy trials in smaller populations showed small effect sizes for collaborative care.^47,48

Differences in baseline population or programmatic elements may contribute to these findings. Many collaborative care models focus on treating older adults.¹⁵ The mean age of this study population was 37 ± 15 years for anxiety and 41 ± 16 years for depression, yet it demonstrated comparable, positive outcomes to collaborative care in older populations, suggesting that the care model extends well to younger populations. Patient outcomes in specialty psychiatric care were comparable to previous literature,^49,50 with the average depression severity significantly improving from moderate to mild at 6 months.

Specialty psychiatry in an integrated health care system may share similarities to collaborative care compared to specialty psychiatry in other systems. As such, the finding that patients in collaborative care showed larger reductions in depression or anxiety symptoms compared to specialty psychiatry and a 159% and 153% improvement in remission and response for depression and a 300% and 140% improvement in remission and response for anxiety (Supplementary Table 3) likely reflects programmatic features of collaborative care. The implemented model had routine symptom collection (measurement-based care), scheduled follow-up, structured therapy content, time limited therapy, algorithms for addressing worsening symptoms (action plan), and symptom-based program graduation.⁴³ Such model components have been associated with collaborative care program outcomes^15,38 and may have impacted treatment effectiveness compared to specialty psychiatry, where treatment was individually determined by providers and patients.

Clinical variables predicting depressive symptom improvement included baseline PHQ-9, time from baseline to treatment, age, and antidepressant use during treatment. Variables predicting worsening depressive symptoms included a history of clinic encounters containing psychiatric diagnoses, or antidepressant use in the year prior to screening eligibility. This aligns with previous literature, suggesting that individuals closer to their first depressive episode with lower baseline depressive symptoms, or lacking comorbidities, have a greater likelihood of treatment response,⁵¹ and lower socioeconomic status is associated with poorer depression outcomes.⁵²

Older age predicted higher treatment response, which is counterintuitive, given the literature suggesting that treatment resistance incidence increases with age.⁵³ However, most patients in this cohort had no psychiatric history or comorbidities, suggesting that they may be presenting for their first depressive episode later in life.

For anxiety, lower baseline anxiety symptoms and shorter time between baseline and treatment were significantly associated with anxiety symptom reduction, while a high NDI score was significantly associated with worsening anxiety symptoms. Similar to depression, baseline anxiety scores have been associated with anxiety outcomes,⁵⁴ although direct evidence regarding primary delays in anxiety screening to treatment impacting outcomes is limited. Previously, there have been reports of data trends suggesting an association between income and anxiety outcomes⁵⁵; data from this study support this and warrant further investigation.

While this study provides some of the first real-world data regarding virtual collaborative care treatment outcomes for a diverse population in an integrated health setting vs specialty psychiatry treatment, it has limitations. Factors influencing decision making regarding intervention enrollment could impact study results. To address this limitation, we performed IPTW, which only accounts for measured covariates. A randomized controlled trial is needed to control for unmeasured covariates. This study took place in an integrated health system with insured patients. As such, some findings may not generalize to populations without similar insurance or access to care.

These results support the effectiveness of virtual collaborative care to treat mild-to-moderate depressive and anxiety symptoms. Depression is one of the most common mental health conditions in the United States,¹ yet most patients with depression do not achieve treatment response or reach Healthcare Effectiveness Data and Information Set (HEDIS) care quality goals of a 50% symptom reduction.^46,56,57 Collaborative care is a powerful model to help patients obtain timely access to high-quality, evidence-based mental health care, leading to improved outcomes.¹⁵ These results suggest that collaborative care models may have value toward improving specialty psychiatric care and can help define how scarce psychiatric resources should be deployed.

Article Information

Published Online: September 4, 2024. https://doi.org/10.4088/JCP.24m15332
© 2024 Physicians Postgraduate Press, Inc.
Submitted: March 5, 2024; accepted June 4, 2024.
To Cite: Ridout KK, Alavi M, Lee C, et al. Virtual collaborative care versus specialty psychiatry treatment for depression or anxiety. J Clin Psychiatry. 2024;85(3):24m15332.
Author Affiliations: The Permanente Medical Group, Kaiser Permanente Northern California, Oakland, California (K. Ridout, Fazzolari, S. Ridout, Koshy, Awsare); Division of Research, Kaiser Permanente Northern California, Oakland, California (K. Ridout, Alavi, Lee, Weisner, Iturralde); Kaiser Foundation Hospitals, Kaiser Permanente Northern California, Oakland, California (Harris).
Corresponding Author: Kathryn K. Ridout, MD, PhD, Kaiser Permanente Northern California, 401 Bicentennial Way, Santa Rosa, CA 95403 (Kathryn.Erickson-Ridout@KP.org).
Author Contributions: Concept and design: (K. Ridout, S. Ridout, Awsare, Weisner, Iturralde). Acquisition, analysis, or interpretation of data: (K. Ridout, Alavi, S. Ridout, Lee, Iturralde). Drafting of the manuscript: (K. Ridout, Alavi, S. Ridout, Iturralde). Critical revision of the manuscript for important intellectual content: (K. Ridout, Koshy, Awsare, Harris, Weisner, Fazzolari, Iturralde). Statistical analysis: (K. Ridout, S. Ridout, Alavi, Lee). Funding acquisition: (K. Ridout). Administrative, technical, or material support: (K. Ridout, S. Ridout, Awsare, Harris, Fazzolari). Supervision: (Awsare, Koshy, Iturralde).
Relevant Financial Relationships: The authors report no financial or other relationship relevant to the subject of this article.
Funding/Support: This project was supported by competitive, peer-reviewed funding in The Permanente Medical Group (TPMG) Delivery Science and Applied Research program (Oakland, CA). Dr K. Ridout’s time was supported by The Permanente Medical Group’s Physician Researcher Program.
Role of the Sponsor/Funder: The supporters had no role in the design, analysis, interpretation, or publication of this study.
Author Data Access: Drs K. Ridout and Iturralde and Ms Alavi had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
ORCID: Kathryn Ridout: https://orcid.org/0000-0003-2101-8165; Esti Iturralde: https://orcid.org/0000-0001-6837-4998; Catherine Lee: https://orcid.org/0000-0003-0008-9052; Brooke Harris: https://orcid.org/0000-0001-9166-052X
Supplementary Material: Available at Psychiatrist.com.

Clinical Points

• Collaborative care for depression or anxiety is an evidence-based treatment framework to improve depressive or anxiety symptoms, but real-world evidence compared to specialty psychiatry treatment is lacking.
• Collaborative care is an effective treatment paradigm for depressive and anxiety symptom reduction compared to specialty psychiatry referral and treatment.

The post Virtual Collaborative Care Versus Specialty Psychiatry Treatment for Depression or Anxiety appeared first on Psychiatrist.com.