Hostname: page-component-78c5997874-s2hrs Total loading time: 0 Render date: 2024-11-10T06:50:40.634Z Has data issue: true hasContentIssue false

Pharmacotherapy guideline concordance for major depressive disorder and its link to functioning via symptom change

Published online by Cambridge University Press:  16 September 2024

Mason T. Breitzig
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
Fan He
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
Lan Kong
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
Guodong Liu
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA Department of Psychiatry and Behavioral Health, Penn State Health Milton S. Hershey Medical Center, Hershey, PA, USA
Daniel A. Waschbusch
Affiliation:
Department of Psychiatry and Behavioral Health, Penn State Health Milton S. Hershey Medical Center, Hershey, PA, USA
Jeff D. Yanosky
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
Duanping Liao
Affiliation:
Department of Public Health Sciences, Penn State College of Medicine, Hershey, PA, USA
Erika F.H. Saunders*
Affiliation:
Department of Psychiatry and Behavioral Health, Penn State Health Milton S. Hershey Medical Center, Hershey, PA, USA
*
Corresponding author: E. F. H. Saunders; Email: esaunders@pennstatehealth.psu.edu
Rights & Permissions [Opens in a new window]

Abstract

Introduction:

Alleviation of symptom severity for major depressive disorder (MDD) is known to be associated with a lagged improvement of functioning. Pharmacotherapy guidelines support algorithms for MDD treatment. However, it is currently unclear whether concordance with guidelines influences functional recovery. A guideline concordance algorithm (GCA-8) was used to measure this pathway in a naturalistic clinical setting.

Methods:

Data from 1403 adults (67% female, 84% non-Hispanic/Latino White, mean age of 43 years) with nonpsychotic MDD from the Penn State Psychiatry Clinical Assessment and Rating Evaluation System registry (visits from 02/01/2015 to 04/13/2021) were evaluated. Multivariable linear regression measured associations between GCA-8 and World Health Organization Disability Assessment Schedule 2.0 (WHODAS) scores. Mediation by MDD symptom severity using the Patient Health Questionnaire depression module (PHQ-9) was also evaluated.

Results:

This study found a statistically significant improvement in WHODAS scores (functioning) between baseline and final measures (−2 points, P < .001) within one year. A one standard deviation increase in the GCA-8 score was associated with a 0.48-point reduction in mean disability score (total effect; P = .02) with significant mediation by the change in MDD symptom severity (coefficient = −0.51, P < .001) and a nonsignificant natural direct effect of the GCA-8 independent of PHQ-9 change (coefficient = −0.02, P = .92).

Conclusions:

Higher pharmacotherapy guideline concordance is associated with better functioning for MDD patients; this association likely occurs through improvement in MDD symptom severity rather than directly.

Type
Research Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2024. Published by Cambridge University Press on behalf of Association for Clinical and Translational Science

Introduction

Globally, major depressive disorder (MDD) is recognized as a leading cause of disability and contributes, directly and indirectly, to detriments in work productivity [Reference Woo, Kim and Hwang1Reference Chow, Bowie, Morton, Lalovic, McInerney and Rizvi6]. It also impacts daily activities and social and family interaction [Reference Sheehan, Harnett-Sheehan and Raj7Reference Severe, Greden and Reddy11]. Prominent somatic MDD symptoms such as fatigue, low energy, disrupted sleep patterns, and change in appetite, as well as their overall severity, may contribute to an individual’s functional decline [Reference Chow, Bowie, Morton, Lalovic, McInerney and Rizvi6]. However, these symptoms often do not explain all of the variability in the functional impairment associated with MDD [Reference Chow, Bowie, Morton, Lalovic, McInerney and Rizvi6,Reference Hammer-Helmich, Haro and Jönsson8,Reference Kroenke, Spitzer and Williams12]. Regardless of its underlying causes, the detrimental impact of MDD on activities of daily living can be more severe than would be expected for those facing chronic physical disabilities such as diabetes [Reference Hammer-Helmich, Haro and Jönsson8,Reference Hays, Wells, Sherbourne, Rogers and Spritzer13]. More importantly, evidence shows that these functional impairments linger after remediation of the acute MDD episode [Reference Bothwell and Weissman14Reference Papakostas16]. The long-term nature of this impact raises essential questions about the applicability of current clinical guidelines and whether their degree of application (guideline concordance) influences functional recovery.

Individuals with MDD can endure functional impairment lasting 425 years after symptom remission [Reference Kennedy, Foy, Sherazi, McDonough and McKeon9,Reference Yang, Gao and Li10,Reference Rytsälä, Melartin, Leskelä, Lestelä-Mielonen, Sokero and Isometsä17]. Consequently, research and the resulting guidelines have made frequent calls to redefine “remission” and incorporate considerations of psychosocial functioning [Reference Severe, Greden and Reddy11,Reference Israel18,Reference Parikh, Quilty and Ravitz19]. However, given the multitude of robust symptom measures, such as the Patient Health Questionnaire depression module (PHQ-9), as well as a heightened interest in symptom remission engendered by outcomes-focused clinical systems (e.g., billing codes in the USA), symptom measures are collected far more frequently than functioning measures [Reference Yang, Gao and Li10]. Health systems in the USA are well-equipped to treat acute MDD symptoms for the 40%–66% with MDD who seek care but are less well-equipped to deal with protracted functional impairments [Reference Severe, Greden and Reddy11,Reference Rytsälä, Melartin, Leskelä, Lestelä-Mielonen, Sokero and Isometsä17,Reference Conradi, Ormel and de Jonge20].

Research increasingly supports measurement-based care, which promotes frequent assessment of MDD symptoms and functioning [Reference Fortney, Unützer and Wrenn21,Reference Morris and Trivedi22]. Unfortunately, clinical practices are less likely to routinely measure patients’ functioning due to the limited availability of comprehensive but practice-ready (i.e., quick/straightforward for patient use) psychometric measures supported by consensus [Reference Yang, Gao and Li10]. Additionally, functional remissions usually occur far beyond the typical one- to two-year treatment window if achieved [Reference Hammer-Helmich, Haro and Jönsson8,Reference Yang, Gao and Li10,Reference Severe, Greden and Reddy11,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23]. Consequently, longitudinal studies on functioning are not always feasible, and psychiatric research often does not track the return to total functional normalcy [Reference Hammer-Helmich, Haro and Jönsson8,Reference Yang, Gao and Li10,Reference Severe, Greden and Reddy11,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23].

Some studies focus on functioning, such as the European Prospective Epidemiological Research on Functioning Outcomes Related to Major depressive disorder (PERFORM) study [Reference Hammer-Helmich, Haro and Jönsson8,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23]. PERFORM was a two-year observational study of 1159 outpatients from France, Germany, Spain, Sweden, and the United Kingdom, showing that although the functional burden of MDD is intense and spans multiple domains, it is responsive to acute pharmacotherapy [Reference Hammer-Helmich, Haro and Jönsson8,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23]. However, this improvement tends to be less responsive than MDD symptom improvement and plateaus as symptom measures improve toward subsyndromal levels around the first two years of treatment [Reference Hammer-Helmich, Haro and Jönsson8,Reference Yang, Gao and Li10]. It has been posited that since subclinical symptoms and perceived residual cognitive symptoms are associated with the risk of recurrence of MDD, management of functional symptoms after clinical remission may be critical to preventing future episodes [Reference Yang, Gao and Li10].

The connection between symptom severity for MDD and functioning has been extensively studied. For instance, using longitudinal prospective data from 371 White patients with MDD in the National Institute of Mental Health Collaborative Depression Study, a study found evidence of a linear association between symptom severity and global functioning, emphasizing that higher levels of depressive symptoms may be associated with more severe functional impairments [Reference Judd, Akiskal and Zeller24].

This finding was later corroborated by a study of 193 Finnish patients, which reported that the two most significant predictors of functioning and social adjustment were the current severity of depression and the cumulative history of depression, each demonstrating significant inverse associations [Reference Rytsälä, Melartin, Leskelä, Lestelä-Mielonen, Sokero and Isometsä17]. Although studies demonstrate that treatment of acute episodes can rapidly alleviate some functional impairments, another study showed that the return to functional normalcy occurs more gradually and persists beyond MDD symptom remission, perhaps due to factors beyond depressive symptoms [Reference Yang, Gao and Li10,Reference Hirschfeld, Dunner and Keitner25].

Despite the considerable understanding of the link between symptom severity and functioning and the ability to alleviate functional impairments alongside depressive symptoms, it remains uncertain how specific treatment decisions impact patient functioning [Reference Judd, Akiskal and Zeller24,Reference Hirschfeld, Dunner and Keitner25]. Furthermore, it is unclear whether adherence to treatment guidelines (termed guideline concordance) influences such downstream outcomes.

As early as 1995, one study conducted simulations using real-world clinical data and found that guideline-concordant treatment significantly reduced daily role impairment and physical functioning [Reference Sturm and Wells26]. In a study of the MarketScan database spanning 1992 to 1996, another study similarly found that guideline-concordant practices improved outcomes, namely, extending the time to relapse [Reference Sood, Treglia, Obenchain, Dulisse, Melfi and Croghan27]. However, in the decades since, studies have rarely measured the potential associations between guideline concordance and patient functioning, focusing instead on symptoms [Reference Duhoux, Fournier and Menear28].

Our Penn State Psychiatry Clinical Assessment and Rating Evaluation System (PCARES) research team and others have developed more specific algorithms to study guideline concordance in practice and, most importantly, its connection to patient outcomes [Reference Yang, Su and Dong29Reference van Fenema, Giltay, van Noorden, van Hemert and Zitman34]. This study applied our previously described guideline concordance algorithm (GCA-8) and examined its association with the World Health Organization Disability Assessment Schedule (WHODAS 2.0), which measures self-reported functioning [Reference Breitzig, He and Kong32,Reference Ustun, Kostanjesek and Chatterji35]. Our aims were to (1) explore the association between guideline concordance and patient functioning using the GCA-8 and WHODAS and (2) determine whether that association, if any, was mediated by symptom severity as measured using the PHQ-9. We hypothesized that higher concordance with treatment guidelines, through a change (reduction) in MDD symptom severity, would result in better functioning among patients with MDD. Addressing these aims will help enhance the selection of existing treatment strategies and support the implementation of process measures for guideline concordance.

Methods

PCARES registry description

The Penn State Psychiatry Clinical Assessment and Rating Evaluation System (PCARES) registry includes a systematic clinical sample of 3556 individuals with mental illness and has been described previously [Reference Breitzig, He and Kong32,Reference Saunders, Brady and Mukherjee36]. The registry was considered a clinical quality improvement project and thus exempt from review by the Penn State College of Medicine Institutional Review Board (IRB). However, any research project using deidentified PCARES data, including this retrospective study, still necessitates approval by the PCARES steering committee and the Penn State College of Medicine IRB. Accordingly, this study was approved by the PCARES Steering Committee and IRB to use the PCARES data for this research (#00020184) and conducted according to the ethical principles of the Declaration of Helsinki.

Study participants

Eligible participants (1) were in the PCARES registry (n = 3556; individuals with significant cognitive impairment were excluded), (2) had an ICD-10 diagnosis for MDD within one year before and one year after their PCARES encounter (02/01/2015 to 04/13/2021), (3) were 18 years or older at baseline, (4) had at least one PHQ-9 and one World Health Organization Disability Assessment Schedule (WHODAS 2.0) measure (no minimum score required), and (5) had no diagnosis of bipolar disorder or psychosis within one year before baseline. Individuals with significant cognitive impairment were excluded from the registry. After applying the above criteria, we had 1403 eligible participants for whom we calculated a guideline concordance score (described below and previously) and 1241 for complete case analysis [Reference Breitzig, He and Kong32].

Guideline concordance algorithm (GCA-8)

Our novel pharmacotherapy guideline concordance algorithm (GCA-8) was informed by CANMAT guidelines and literature on guideline concordance and is described previously [Reference Breitzig, He and Kong32,Reference Duhoux, Fournier and Menear28,Reference Kennedy, Lam and McIntyre37]. Eight criteria generate the ordinal score: three focused on prescription sequence (e.g., prescribing a first-line drug before a second-line drug); three on dosing, duration, and modifications; one on drug-drug interactions; and one on visit frequency. According to the algorithm, a point was deducted from the baseline of eight for each failed criterion during the one-year treatment window [Reference Breitzig, He and Kong32].

PCARES PRO and EMR data

The PCARES patient-reported outcomes (PRO) battery included, among other questionnaires, the PHQ-9 to measure depression severity and somatic symptoms and the World Health Organization Disability Assessment Schedule (WHODAS 2.0) to measure disability and functioning [Reference Kroenke, Spitzer and Williams12,Reference Ustun, Kostanjesek and Chatterji35,Reference Tomitaka, Kawasaki and Ide38,Reference Andrews, Kemp, Sunderland, Von Korff and Ustun39]. For this study, all WHODAS 2.0 scores were tabulated using the 12-item format and the simple scoring strategy, ranging from 0 to 48 (directly comparable to the 12–60 range) as previously described [Reference Ustun, Kostanjesek and Chatterji35,Reference Andrews, Kemp, Sunderland, Von Korff and Ustun39Reference Sjonnesen, Bulloch, Williams, Lavorato and Patten41]. PHQ-9 and WHODAS questionnaires with two or more unanswered responses were excluded (described previously) [Reference Breitzig, He and Kong32]. Questionnaires with one missing value were addressed using mean-imputation (described previously) [Reference Breitzig, He and Kong32,Reference Ustun, Kostanjesek and Chatterji35,Reference Sjonnesen, Bulloch, Williams, Lavorato and Patten41,Reference Kroenke, Spitzer, Williams and Löwe42]. Briefly, the WHODAS 2.0 measures six domains of functioning, including cognition, mobility, self-care, getting along (social interaction), life activities (domestic responsibilities, leisure, work, and school), and participation (community activities) [Reference Ustun, Kostanjesek and Chatterji35]. Data from PROs were linked with electronic medical record (EMR) data detailing prescriptions and insurance type, among other practice-specific details described previously [Reference Breitzig, He and Kong32,Reference Saunders, Brady and Mukherjee36].

In this cross-sectional study (although nested in a retrospective cohort, many of the measures temporally overlap), we used (1) an arithmetic mean of all WHODAS and PHQ-9 scores in one year (except the baseline score; the median number of scores was 4 for each measure), (2) the final WHODAS score in one year (median occurrence at 265 days for both the WHODAS and the PHQ-9), (3) the standard deviation (SD) of WHODAS scores (including baseline) for those with at least three scores (measuring variability in treatment response), (4) and the change in each respective measure from baseline to the final measurement.

Statistical analyses

Associations between PCARES participants’ WHODAS 2.0 scores and the GCA-8 were evaluated with and without adjustment for the study covariates. Unadjusted analyses evaluated differences in median WHODAS scores across categorical variables (e.g., covariates) via Wilcoxon rank-sum tests and linear associations with continuous variables via simple linear regression (Table 1). Wilcoxon signed-rank test was used to evaluate the median change in WHODAS scores from baseline to the final measure to estimate the one-year improvement for PCARES. Multivariable linear regression analyses were conducted with adjustments for the confounders in Table 1. Finally, an exploratory cross-sectional mediation analysis was conducted for each outcome variable using the “causalmed” (causal mediation) procedure in SAS, which is based on a counterfactual framework [Reference Valeri and VanderWeele43,Reference Robins and Greenland44].

Table 1. Unadjusted associations between participants’ characteristics, guideline concordance, and other study measures and their baseline WHODAS score (n = 1403)

BMI = body mass index; GCA-8 = guideline concordance algorithm; IQR = interquartile range; MDD = major depressive disorder; PHQ-9 = Patient Health Questionnaire depression module; SD = standard deviation; SE = standard error; WHODAS = World Health Organization Disability Assessment Schedule.

a Testing the null hypothesis that the sample medians are equal (categorical variables) or testing the hypothesis that the regression coefficient in an unadjusted linear model is equal to 0.

b An ordinal measure of the overall burden of comorbidity (including presence or absence of congestive heart failure, stroke, coronary heart disease, diabetes, dyslipidemia, chronic kidney disease, thyroid disorders, hypertension, migraines, fibromyalgia, sleep disorders, anxiety, attention-deficit hyperactivity disorder, trauma, substance use disorder, and obsessive-compulsive disorder).

Results

Baseline characteristics of study participants

Unadjusted analyses of the associations between the participants’ characteristics and WHODAS scores for the 1403 participants with MDD in the sample (Table 1) revealed that participants’ degree of self-reported disability (as measured by the baseline WHODAS score) was significantly associated with age (all P < .001), where each additional 10 years was associated with a 0.73-point higher WHODAS score, 30 additional days of MDD episode duration (β = 0.26; P < .001), one-point increase in baseline PHQ-9 score (β = 0.97; P < .001), one-point increase in body mass index (BMI) (β = 0.17; P.001), one additional concurrent comorbidity (β = 1.26; P < .001), and one additional visit (β = 0.13; P < .001). Conversely, each additional point on the GCA-8 was associated with a significant decrease in a participant’s WHODAS score (β = −1.25; P < .001).

We also found that the mean (SD) WHODAS score at baseline for the PCARES sample was 15.7 (10.32). Non-Hispanic/Latino Black individuals had the highest median self-reported disability score at baseline 20/48 (Table 1), whereas non-Hispanic/Latino White individuals had the lowest baseline median score (P = .002). The median score for those with commercial insurance was 12, while the median for those with either Medicaid or Medicare was 18 and 19, respectively (P = .001). Divorced individuals had the highest median disability scores (19; P = .03), and those with recurrent MDD had significantly greater self-reported disability than those with nonrecurrent MDD (15 vs 13, P = .03). Finally, females reported slightly higher disability with median scores of 15 versus 14 (P = .06). There were no significant differences in self-reported disability between individuals who lived in rural versus urban municipalities (P = .08).

WHODAS score distribution at baseline and per the final measure

The distribution (Figure 1) of self-reported disability was assessed for the PCARES cohort at baseline for 1403 participants with MDD and at their last measure up to one year following baseline (n = 977). At baseline, 50% of the PCARES cohort reported having some functional impairment demonstrated by a median score of 14/48. The final measure showed a slight improvement, with the median disability rating reducing to a median score of 12/48. A Wilcoxon signed rank test found that the median change in WHODAS scores from baseline to the final measure was statistically significant (improved) for the 977 individuals who had both a baseline and last follow-up measure within one year of baseline (P < .001).

Figure 1. WHODAS scores in the PCARES cohort at baseline per their final WHODAS measurement. The baseline had an n = 1241 while the final score had an n = 977 due to needing at least two WHODAS measurements; the vertical line represents the 50th percentile score. PCARES = Penn State Psychiatry Clinical Assessment and Rating Evaluation System registry; WHODAS = World Health Organization Disability Assessment Schedule.

Adjusted associations between PHQ-9 change and functioning scores

Adjusted associations between PHQ-9 change and each of the three WHODAS outcome score variants (mean, final measure, and SD) were evaluated (Table 2). A one-point increase in PHQ-9 change was significantly associated with lower WHODAS scores. On average, a one-point increase in PHQ-9 change was associated with a 0.62-point decrease in mean WHODAS score (P < 0.001), a 0.86-point reduction in the final WHODAS score (P < .001), and a 0.09-point reduction in the SD of WHODAS scores (P < .001), as expected per literature.

Table 2. Adjusted associations between PHQ-9 change in one year and WHODAS scores among PCARES participants with complete data

The regression coefficient represents slope: mean change in outcome variable per one point increase the change of PHQ-9 score from baseline to the final measure taken for a given individual (the median occurrence of the final measure was at 265 days after baseline); some variables, but not the PHQ-9 change, were standardized before modeling by dividing the variable of interest by the SD of the given predictor as calculated using the entire cohort (n = 1403); the models were adjusted for age, average MDD episode duration, baseline PHQ-9 scores (standardized), baseline WHODAS scores (standardized), BMI, cohort, comorbidity burden (an ordinal score representing the presence or absence of ADHD, anxiety, CKD, CHF, CHD, diabetes, dyslipidemia, fibromyalgia, hypertension, migraines, sleep disorders, stroke, SUD, thyroid disorders, and trauma; observed range: 0–11), insurance type, marital status, visit frequency in 12 months, race and ethnicity, recurrent or nonrecurrent MDD diagnosis at baseline, rural/urban municipal classification, and patient-reported gender; the cohort variable represents the wave of recruitment into the PCARES registry based on the forms that were collected during that period, making it a temporal adjustment.

ADHD = attention-deficit hyperactivity disorder; CHF = congestive heart failure; CHD = coronary heart disease; CVD = cardiovascular disease; MDD = major depressive disorder; PHQ-9 = Patient Health Questionnaire depression module; PCARES = Psychiatry Clinical Assessment and Rating Evaluation System; SD = standard deviation; SUD = substance use disorder; WHODAS = World Health Organization Disability Assessment Schedule.

a The sample size was reduced since 883 individuals had the outcome, but only 958 had the PHQ-9 change variable; thus, the crossover sample size was 851 or less in the case of the WHODAS standard deviation analysis due to needing at least three WHODAS measures.

b Adjusted for the one-year score difference between the baseline WHODAS score and the final measure for a given individual (the median occurrence of the final measure was at 265 days after baseline) rather than the baseline measure.

Associations between guideline concordance and functioning

To evaluate the potential association between guideline concordance and functioning, a series of unadjusted models were constructed and assessed, each of which identified a significant association between the WHODAS measures and patients’ GCA-8 scores (Table 3), and then the models were adjusted for the covariates listed in Table 1. The adjusted analyses found that the associations between the GCA-8 scores and WHODAS scores remained statistically significant but were attenuated (Table 3). One SD increase in the GCA-8 score was associated with a 0.63-point decrease in the mean WHODAS score in one year (P = .004). In comparison, a one SD increase in baseline WHODAS score was associated with a 7.46 increase in the mean one-year WHODAS score (P < .001). Similarly, a one SD increase in GCA-8 was associated with a 0.83-point decrease in the final WHODAS score (P = .002). Finally, a one SD increase in GCA-8 was associated with a 0.29-point reduction in the intraindividual SD of WHODAS scores up to one year after the initial encounter (P = .002).

Table 3. Associations between standardized scores of guideline concordance and patient-reported functional disability scores

Regression coefficient represents slope; mean change in outcome variable per one SD increase in guideline concordance score; scores were standardized before modeling by dividing the variable of interest by the SD of the given predictor as calculated using the entire cohort (n = 1403); models also adjusted for age, average MDD episode duration, baseline PHQ-9 scores, BMI, cohort, comorbidity burden, race and ethnicity, insurance type, marital status, visit frequency in 12 months, recurrent or nonrecurrent MDD diagnosis at baseline, rural/urban classification, and patient-reported gender; the cohort variable represents the wave of recruitment into the PCARES registry based on the forms that were collected during that period, effectively making it a temporal adjustment. The adjusted model here does not contain the sample size-limiting variable representing PHQ-9 change; thus, the effect differs slightly from the total effect in Table 4.

BMI = body mass index; GCA-8 = guideline concordance algorithm; MDD = major depressive disorder; SD = standard deviation; SE = standard error; PCARES = Psychiatry Clinical Assessment and Rating Evaluation System; WHODAS = World Health Organization Disability Assessment Schedule.

a Adjusted for the one-year score difference between the baseline WHODAS score, the final measure for a given individual, rather than the baseline measure.

Table 4. Mediation analysis examining the pathway between guideline concordance score and WHODAS score via PHQ-9-measured symptom severity (n = 851)

The regression coefficient represents the mean change in the outcome variable per one standard deviation increase in the guideline concordance score; models also adjusted for age, average MDD episode duration, baseline PHQ-9 scores, BMI, cohort, comorbidity burden (described earlier), insurance type, marital status, visit frequency in 12 months, race and ethnicity, recurrent or nonrecurrent MDD diagnosis at baseline, rural/urban municipal classification, and patient-reported gender; the models examining the outcome of mean WHODAS and final WHODAS score in one year were also adjusted for the standardized baseline WHODAS score.

BMI = body mass index; GCA-8 = guideline concordance algorithm; MDD = major depressive disorder; PHQ-9 = Patient Health Questionnaire depression module; SE = standard error; WHODAS = World Health Organization Disability Assessment Schedule.

Cross-sectional mediation involving the GCA-8, symptom severity, and functioning

In a final analytic step, we evaluated the potential mediation of the association between guideline concordance and functioning by a change in symptom severity. An adjusted mediation analysis was conducted for each version of the WHODAS metric, with the change in PHQ-9 scores as the mediator between the GCA-8 and the WHODAS scores. For every SD increase in the GCA-8 score, the mean WHODAS score decreased by 0.53 points (P = .02; Table 4). This total effect was decomposed into a significant effect mediated by the change in the PHQ-9 score of −0.51 (P < .001) and a nonsignificant natural direct effect of −0.02 (P = .92), where the PHQ-9 change does not explain the influence of GCA-8. In this case, the mediator pathway could be entirely responsible for the association between a patient’s GCA-8 score and their mean WHODAS score in one year. While the GCA-8 score increases, it is possible that the difference between baseline PHQ-9 and final PHQ-9 also increases, which is associated with a lower mean WHODAS score.

For the final WHODAS score, a one SD increase in the GCA-8 score was associated with a 0.72-point decrease in patients’ final WHODAS score in one year (P = .005). However, just as above, the natural direct path was not statistically significant (P = .92), and the indirect path demonstrated that PHQ-9 change drove the total effect (P < .001). Lastly, the association with the one-year change in WHODAS was examined. A one SD increase for the GCA-8 was found to be associated with a 0.59 increase in WHODAS change (P = .03). The indirect mediated effect was again statistically significant (P < .001). In contrast, the direct, PHQ-9-independent GCA-8 effect was not statistically significant (P = .71).

Discussion

This study examined the association between guideline concordance (via GCA-8 scores) for MDD treatment and patient functioning and whether alleviating MDD symptoms could mediate this relationship. After adjusting for standard sociodemographic covariates, as well as practice-specific covariates (e.g., average MDD episode duration), the results demonstrated that (1) higher guideline concordance scores were associated with significantly lower WHODAS scores, and (2) this significant association could be explained by the change (improvement) in depression symptom severity over time. This study contributes to the literature by showing that the degree of guideline concordance for a given patient’s pharmacotherapy is essential, such that more guideline-concordant pharmacotherapy is associated with reduced symptom severity and, perhaps consequently, lower functional impairment.

Even though studies have established the association between symptom severity and functioning, including the seminal 12-month naturalistic Sequenced Treatment Alternatives to Relieve Depression (STAR*D), as reaffirmed here (Table 2), this study is among the first to report an association between guideline concordance and patient-reported functioning (Table 3) in a modern clinical cohort [Reference Yang, Gao and Li10,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23,Reference Judd, Akiskal and Zeller24,Reference Wang, Wang and Zhang45,Reference Trivedi, Rush and Wisniewski46]. In 1999, a study provided evidence of a linear association between guideline concordance and reduction in functional limitations [Reference Sturm and Wells26]. In 2006, the STAR*D study showed that remission is more likely for those with better baseline functioning (via measurement of physical and mental functioning with the 12-item Short Form Health Survey) [Reference Trivedi, Rush and Wisniewski46,Reference Rush, Trivedi and Wisniewski47]. However, per design, all patients received guideline-concordant care and, like many other studies, did not provide information on the role of guideline concordance [Reference Trivedi, Rush and Wisniewski46,Reference Rush, Trivedi and Wisniewski47].

Much later, a similar – but more functioning-focused – study in 2018, PERFORM, measured the “switch of antidepressant” and found that switching for the first time was associated with a significantly higher disability score on the Sheehan disability scale (SDS) [Reference Hammer-Helmich, Haro and Jönsson8]. Lastly, in 2020, another study showed that the time spent on medication was marginally associated with lower odds (95% CI: 0.83, 0.99) of having a social impairment [Reference Wang, Wang and Zhang45]. This study expands the literature by measuring more domains of concordance in a population receiving modern pharmacotherapies. More rigorous studies are needed to elucidate causality between guideline-concordant treatment and functioning.

To further describe the cross-sectional pathway between guideline concordance and WHODAS scores that we observed, this study also evaluated the change in PHQ-9 score as a mediator. We previously demonstrated that the GCA-8 score correlates with participants’ PHQ-9 scores and that avoiding one discordant event was associated with scoring approximately one point lower on PHQ-9 – a clinically relevant improvement [Reference Breitzig, He and Kong32]. Additionally, the results in Table 2 and existing evidence corroborate that a reduction in PHQ-9 scores via treatment is significantly associated with functional remediation [Reference Yang, Gao and Li10,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23,Reference Judd, Akiskal and Zeller24,Reference Wang, Wang and Zhang45]. This overlap could be due to changes in the somatic symptoms of fatigue, energy loss, sleep disruption, change in appetite, and overall symptom severity, measured in the PHQ-9, as mentioned above [Reference Chow, Bowie, Morton, Lalovic, McInerney and Rizvi6,Reference Kroenke, Spitzer and Williams12,Reference Kroenke, Spitzer, Williams and Löwe42]. Given this information, cross-sectional mediation models were constructed to determine whether the association between guideline concordance and functioning could be wholly or partly attributable to an improvement in symptom severity.

Results showed that PHQ-9 scores may fully mediate the association between GCA-8 and WHODAS scores (Table 4). Moreover, while the total effect of each analysis was statistically significant, all three of the PHQ-9-independent GCA-8 effects were nonsignificant (Table 4). In other words, lower symptom severity fully accounted for better functioning scores associated with higher guideline concordance. These results indicate that, among patients presenting for treatment of depression, a higher GCA-8 score (i.e., better concordance with treatment guidelines) may translate to significant improvement in both depression symptom severity and functioning [Reference Chow, Bowie, Morton, Lalovic, McInerney and Rizvi6,Reference Hammer-Helmich, Haro and Jönsson8,Reference Yang, Gao and Li10,Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23,Reference Judd, Akiskal and Zeller24]. Future longitudinal work should consider that some drugs and doses influence patients’ functioning through several mechanisms and that the treatment strategies supported by clinical guidelines can have a lesser negative impact on patient functioning, a more significant effect on improving functioning, or both. Guideline concordance may also directly impact aspects of functioning that are not measured by the WHODAS and may be independent or upstream of symptom severity, such as cognitive functioning [Reference Haro, Hammer-Helmich, Saragoussi, Ettrup, Larsen and Joan De Deu23].

In addition to these analyses, this study helps characterize disability in PCARES, which connects to other clinical populations with MDD. The data show that the patient-reported degree of disability significantly varied by most patient characteristics, including age, BMI, ethnicity, insurance type, and marital status. However, it did not differ significantly by rurality or patient-reported gender. Several studies find similar but inconsistent results for age, ethnicity, sex, and marital status [Reference Hammer-Helmich, Haro and Jönsson8,Reference da Silva Lima and de Almeida Fleck15,Reference Rytsälä, Melartin, Leskelä, Lestelä-Mielonen, Sokero and Isometsä17,Reference Andrews, Kemp, Sunderland, Von Korff and Ustun39,Reference Wang, Wang and Zhang45]. The results of this study also showed that non-Hispanic/Latino Black or Hispanic/Latino individuals had significantly greater median disability scores, often attributed to disparities in socioeconomic status, education, and health care [Reference Levine, Neidecker, Kiefe, Karve, Williams and Allison48,Reference Goyat, Vyas and Sambamoorthi49].

This study had several strengths, such as involving a large sample of clinically diverse (e.g., different comorbidities and insurance types) participants who received an array of pharmacotherapeutics in a natural clinical setting. An additional strength is our ability to evaluate patient outcomes in multiple ways due to having repeated measures of symptom severity (PHQ-9) and functioning (WHODAS). This study also had several limitations. Although we had some temporality in assessments where we evaluated the baseline measure of disability versus the final measure, the temporal overlap of the GCA-8 with our outcome measures precludes the establishment of causality. The PCARES sample includes those who seek treatment in a specific geographic region and, as such, has limited generalizability to the rest of the USA. There was no information on treatment adherence, psychotherapy, providers, tobacco smoking, treatment received out of network, or treatment received before 2015; such residual confounding may impact score distributions and further limit generalizability. As not all PCARES participants were at the same stage in their treatment, and not all patients received the same number of follow-up rating measures, the study may over- or underestimate the association between the GCA-8 and WHODAS. Loss to follow-up of those who ceased care due to improvement or nonresponse may also bias these associations toward or away from the null. This limitation is somewhat mitigated by looking at the associations only among those with multiple measures. Finally, diagnostic errors, lack of individual clinical context, and lack of information on treatment adherence or providers may lead to some outcome misclassification and mistakes in measuring guideline concordance.

The results of our study suggest that concordance with pharmacotherapy guidelines for MDD treatment is significantly associated with patient-reported functioning, such that higher treatment concordance (higher GCA-8 scores) is associated with better functioning scores. We showed that this relationship may be entirely due to a change in patients’ reported symptom severity, assuming the proposed causal pathway exists. We also demonstrated that greater guideline concordance is associated with better patient outcomes after adjustment for sociodemographic characteristics and clinical metrics. Future work should elucidate the causal relationship between guideline concordance and patient outcomes in real-world scenarios to improve process quality control and MDD treatment outcomes.

Acknowledgments

The authors would like to thank Dr Nancy Olsen, MD, of the Penn State College of Medicine, Hershey, USA, for her clinical insight and the PCARES participants, without whom this study would not be possible.

Author contributions

Mason Breitzig, Fan He, Duanping Liao, and Erika Saunders contributed to the conception and design of this study; Mason Breitzig, Fan He, Duanping Liao, and Lan Kong contributed to the analysis; Mason Breitzig drafted the manuscript, tables, and figures and interpreted the results with the help of all coauthors; all coauthors revised the manuscript and approved the final version.

Funding statement

None.

Competing interests

None.

Availability of data and materials

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

Footnotes

Poster presentation: Society of Epidemiologic Research Annual Conference June 2024, Austin, Texas.

Inclusion in a doctoral dissertation: This work is included as a chapter in Mason Breitzig’s doctoral dissertation. The approved work will be archived by the Penn State library.

References

Woo, JM, Kim, W, Hwang, TY, et al. Impact of depression on work productivity and its improvement after outpatient treatment with antidepressants. Value Health. 2011;14(4):475482. doi: 10.1016/j.jval.2010.11.006.Google Scholar
Kessler, RC. The costs of depression. Psych Clin North Am. 2012;35(1):114. doi: 10.1016/j.psc.2011.11.005.Google Scholar
Greenberg, PE, Fournier, AA, Sisitsky, T, et al. The economic burden of adults with major depressive disorder in the United States (2010 and 2018). Pharmacoeconomics. 2021;39(6):653665. doi: 10.1007/s40273-021-01019-4.Google Scholar
Birnbaum, HG, Kessler, RC, Kelley, D, Ben-Hamadi, R, Joish, VN, Greenberg, PE. Employer burden of mild, moderate, and severe major depressive disorder: mental health services utilization and costs, and work performance. Depress Anxiety. 2010;27(1):7889. doi: 10.1002/da.20580.Google Scholar
Kessler, RC, Aguilar-Gaxiola, S, Alonso, J, et al. The global burden of mental disorders: an update from the WHO world mental health (WMH) surveys. Epidemiol Psichiatr Soc. 2009;18(1):2333. doi: 10.1017/S1121189X00001421.Google Scholar
Chow, TK, Bowie, CR, Morton, M, Lalovic, A, McInerney, SJ, Rizvi, SJ. Contributors of functional impairment in major depressive disorder: a biopsychosocial approach. Curr Behav Neurosci Rep. 2022;9(2):5972. doi: 10.1007/s40473-022-00247-y.Google Scholar
Sheehan, DV, Harnett-Sheehan, K, Raj, BA. The measurement of disability. Int Clin Psychopharmacol. 1996;11(Supplement 3):8995. doi: 10.1097/00004850-199606003-00015.Google Scholar
Hammer-Helmich, L, Haro, JM, Jönsson, B, et al. Functional impairment in patients with major depressive disorder: the 2-year PERFORM study. Neuropsychiatr Dis Treat. 2018;14:239249. doi: 10.2147/NDT.S146098.Google Scholar
Kennedy, N, Foy, K, Sherazi, R, McDonough, M, McKeon, P. Long-term social functioning after depression treated by psychiatrists: a review. Bipolar Disord. 2007;9(1-2):2537. doi: 10.1111/j.1399-5618.2007.00326.x.Google Scholar
Yang, H, Gao, S, Li, J, et al. Remission of symptoms is not equal to functional recovery: psychosocial functioning impairment in major depression. Front Psychiatry. 2022;13:915689. doi: 10.3389/fpsyt.2022.915689.Google Scholar
Severe, J, Greden, JF, Reddy, P. Consequences of recurrence of major depressive disorder: is stopping effective antidepressant medications ever safe? Focus (Madison). 2020;18(2):120128. doi: 10.1176/appi.focus.20200008.Google Scholar
Kroenke, K, Spitzer, RL, Williams, JBW. The PHQ-9. J Gen Intern Med. 2001;16(9):606613. doi: 10.1046/j.1525-1497.2001.016009606.x.Google Scholar
Hays, RD, Wells, KB, Sherbourne, CD, Rogers, W, Spritzer, K. Functioning and well-being outcomes of patients with depression compared with chronic general medical illnesses. Arch Gen Psychiatry. 1995;52(1):1119. doi: 10.1001/archpsyc.1995.03950130011002.Google Scholar
Bothwell, S, Weissman, MM. Social impairments four years after an acute depressive episode. Am J Orthopsychiatry. 1977;47(2):231237. doi: 10.1111/j.1939-0025.1977.tb00978.x.Google Scholar
da Silva Lima, AFB, de Almeida Fleck, MP. Subsyndromal depression: an impact on quality of life? J Affect Disord. 2007;100(1-3):163169. doi: 10.1016/j.jad.2006.10.010.Google Scholar
Papakostas, GI. Major depressive disorder: psychosocial impairment and key considerations in functional improvement. Am J Manag Care. 2009;15(11 Suppl):S31621.Google Scholar
Rytsälä, HJ, Melartin, TK, Leskelä, US, Lestelä-Mielonen, PS, Sokero, TP, Isometsä, ET. Determinants of functional disability and social adjustment in major depressive disorder. J Nerv Ment Dis. 2006;194(8):570576. doi: 10.1097/01.nmd.0000230394.21345.c4.Google Scholar
Israel, JA. Remission in depression: definition and initial treatment approaches. J Psychopharmacol. 2006;20(3_suppl):510. doi: 10.1177/1359786806064306.Google Scholar
Parikh, SV, Quilty, LC, Ravitz, P, et al. Canadian network for mood and anxiety treatments (CANMAT) 2016 clinical guidelines for the management of adults with major depressive disorder: section 2. Psychological treatments. Can J Psychiatr. 2016;61(9):524539. doi: 10.1177/0706743716659418.Google Scholar
Conradi, HJ, Ormel, J, de Jonge, P. Presence of individual (residual) symptoms during depressive episodes and periods of remission: a 3-year prospective study. Psychol Med. 2011;41(6):11651174. doi: 10.1017/S0033291710001911.Google Scholar
Fortney, JC, Unützer, J, Wrenn, G, et al. A tipping point for measurement-based care. Psychiatric Serv. 2017;68(2):179188. doi: 10.1176/appi.ps.201500439.Google Scholar
Morris, DW, Trivedi, MH. Measurement-based care for unipolar depression. Curr Psychiatry Rep. 2011;13(6):446458. doi: 10.1007/s11920-011-0237-8.Google Scholar
Haro, JM, Hammer-Helmich, L, Saragoussi, D, Ettrup, A, Larsen, KG., Joan De Deu, SS. Patient-reported depression severity and cognitive symptoms as determinants of functioning in patients with major depressive disorder: a secondary analysis of the 2-year prospective PERFORM study. Neuropsychiatr Dis Treat. 2019;15:23132323. doi: 10.2147/NDT.S206825.Google Scholar
Judd, LL, Akiskal, HS, Zeller, PJ, et al. Psychosocial disability during the long-term course of unipolar major depressive disorder. Arch Gen Psychiatry. 2000;57(4):375. doi: 10.1001/archpsyc.57.4.375.Google Scholar
Hirschfeld, RMA, Dunner, DL, Keitner, G, et al. Does psychosocial functioning improve independent of depressive symptoms? a comparison of nefazodone, psychotherapy, and their combination. Biol Psychiatry. 2002;51(2):123133. doi: 10.1016/S0006-3223(01)01291-4.Google Scholar
Sturm, R, Wells, KB. Health policy implications of the RAND medical outcomes study: improving the value of depression treatment. Behav Healthc Tomorrow. 1996;5(5):6366.Google Scholar
Sood, N, Treglia, M, Obenchain, RL, Dulisse, B, Melfi, CA, Croghan, TW. Determinants of antidepressant treatment outcome. Am J Manag Care. 2000;6(12):13271336.Google Scholar
Duhoux, A, Fournier, L, Menear, M. Quality indicators for depression treatment in primary care: a systematic literature review. Curr Psychiatry Rev. 2011;7(2):104137. doi: 10.2174/157340011796391166.Google Scholar
Yang, L, Su, Y, Dong, S, et al. Concordance of the treatment patterns for major depressive disorders between the Canadian network for mood and anxiety treatments (CANMAT) algorithm and real-world practice in China. Front Pharmacol. 2022;13:954973. doi: 10.3389/fphar.2022.954973.Google Scholar
Renes, JW, Maciejewski, DF, Regeer, EJ, Hoogendoorn, AW, Nolen, WA, Kupka, RW. Guideline concordance and outcome in long-term naturalistic treatment of bipolar disorder - a one-year longitudinal study using latent change models. J Affect Disord. 2021;283:395401. doi: 10.1016/j.jad.2020.12.106.Google Scholar
Van fenema, E, Van der wee, NJA, Bauer, M, Witte, CJ, Zitman, FG. Assessing adherence to guidelines for common mental disorders in routine clinical practice. Int J Qual Health Care. 2012;24(1):7279. doi: 10.1093/INTQHC/MZR076.Google Scholar
Breitzig, MT, He, F, Kong, L, et al. Novel quality control metric for the pharmacotherapy of major depressive disorder: measuring guideline concordance and its impact on symptom severity. J Clin Psychiatry. 2024;85:23m14916. doi: 10.4088/JCP.23m14916.Google Scholar
van Fenema, EM, van der Wee, NJA, Giltay, EJ, den Hollander-Gijsman, ME, Zitman, FG. Vitality predicts level of guideline-concordant care in routine treatment of mood, anxiety and somatoform disorders. J Eval Clin Pract. 2012;18(2):441448. doi: 10.1111/j.1365-2753.2010.01593.x.Google Scholar
van Fenema, E, Giltay, E, van Noorden, M, van Hemert, A, Zitman, F. Assessing adherence to guidelines with administrative data in psychiatric outpatients. J Eval Clin Pract. 2017;23(1):513. doi: 10.1111/jep.12414.Google Scholar
Ustun, TB, Kostanjesek, N, Chatterji, SRJ, World Health Organization. Measuring Health and Disability: Manual for WHO Disability Assessment Schedule WHODAS 2.0. Geneva, Switzerland: World Health Organization; 2010. https://apps.who.int/iris/handle/10665/43974 Google Scholar
Saunders, EFH, Brady, M, Mukherjee, D, et al. Gender differences in transdiagnostic domains and function of adults measured by DSM-5 assessment scales at the first clinical visit: a cohort study. BMC Psychiatry. 2023;23(1):709. doi: 10.1186/s12888-023-05207-8.Google Scholar
Kennedy, SH, Lam, RW, McIntyre, RS, et al. Canadian network for mood and anxiety treatments (CANMAT) 2016 clinical guidelines for the management of adults with major depressive disorder. Can J Psychiatr. 2016;61(9):540560. doi: 10.1177/0706743716659417.Google Scholar
Tomitaka, S, Kawasaki, Y, Ide, K, et al. Distributional patterns of item responses and total scores on the PHQ-9 in the general population: data from the national health and nutrition examination survey. BMC Psychiatry. 2018;18(1). doi: 10.1186/s12888-018-1696-9.Google Scholar
Andrews, G, Kemp, A, Sunderland, M, Von Korff, M, Ustun, TB. Normative data for the 12 item WHO disability assessment schedule 2.0. Ross JS, ed. PLoS One. 2009;4(12):e8343. doi: 10.1371/journal.pone.0008343.Google Scholar
Axelsson, E, Lindsäter, E, Ljótsson, B, Andersson, E, Hedman-Lagerlöf, E. The 12-item self-report world health organization disability assessment schedule (WHODAS) 2.0 administered via the internet to individuals with anxiety and stress disorders: a psychometric investigation based on data from two clinical trials. JMIR Ment Health. 2017;4(4):e58. doi: 10.2196/MENTAL.7497.Google Scholar
Sjonnesen, K, Bulloch, AGM, Williams, J, Lavorato, D, Patten, SB. Characterization of disability in Canadians with mental disorders using an abbreviated version of a DSM-5 emerging measure: the 12-item WHO disability assessment schedule (WHODAS) 2.0. Can J Psychiatr. 2016;61(4):227235. doi: 10.1177/0706743716632514.Google Scholar
Kroenke, K, Spitzer, RL, Williams, JBW, Löwe, B. The patient health questionnaire somatic, anxiety, and depressive symptom scales: a systematic review. Gen Hosp Psychiatry. 2010;32(4):345359. doi: 10.1016/j.genhosppsych.2010.03.006.Google Scholar
Valeri, L, VanderWeele, TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods. 2013;18(2):137150. doi: 10.1037/a0031034.Google Scholar
Robins, JM, Greenland, S. Identifiability and exchangeability for direct and indirect effects. Epidemiology. 1992;3(2):143155. doi: 10.1097/00001648-199203000-00013.Google Scholar
Wang, Y, Wang, G, Zhang, N, et al. Association between residual symptoms and social functioning in patients with depression. Compr Psychiatry. 2020;98:152164. doi: 10.1016/j.comppsych.2020.152164.Google Scholar
Trivedi, MH, Rush, AJ, Wisniewski, SR, et al. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. Am J Psychiatr. 2006;163(1):2840. doi: 10.1176/appi.ajp.163.1.28.Google Scholar
Rush, AJ, Trivedi, MH, Wisniewski, SR, et al. Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. Am J Psychiatr. 2006;163(11):19051917. doi: 10.1176/ajp.2006.163.11.1905.Google Scholar
Levine, DA, Neidecker, MV, Kiefe, CI, Karve, S, Williams, LS, Allison, JJ. Racial/ethnic disparities in access to physician care and medications among US stroke survivors. Neurology. 2011;76(1):5361. doi: 10.1212/WNL.0b013e318203e952.Google Scholar
Goyat, R, Vyas, A, Sambamoorthi, U. Racial/Ethnic disparities in disability prevalence. J Racial Ethn Health Disparities. 2016;3(4):635645. doi: 10.1007/s40615-015-0182-z.Google Scholar
Figure 0

Table 1. Unadjusted associations between participants’ characteristics, guideline concordance, and other study measures and their baseline WHODAS score (n = 1403)

Figure 1

Figure 1. WHODAS scores in the PCARES cohort at baseline per their final WHODAS measurement. The baseline had an n = 1241 while the final score had an n = 977 due to needing at least two WHODAS measurements; the vertical line represents the 50th percentile score. PCARES = Penn State Psychiatry Clinical Assessment and Rating Evaluation System registry; WHODAS = World Health Organization Disability Assessment Schedule.

Figure 2

Table 2. Adjusted associations between PHQ-9 change in one year and WHODAS scores among PCARES participants with complete data

Figure 3

Table 3. Associations between standardized scores of guideline concordance and patient-reported functional disability scores

Figure 4

Table 4. Mediation analysis examining the pathway between guideline concordance score and WHODAS score via PHQ-9-measured symptom severity (n = 851)