Socioeconomic Predictors of Treatment Outcome in Adults with Major Depressive Disorder: An Analysis of the CO-MED Trial

Abstract

Objective:

We sought to examine the impact of socioeconomic variables on outcomes in major depressive disorder while controlling for treatment access and level of care.

Methods:

Using data from CO-MED, a prospective clinical trial of 665 adults with major depressive disorder randomized to 3 pharmacotherapeutic strategies, we developed Bayesian hierarchical models of treatment trajectories for the change in the Quick Inventory of Depressive Symptomatology-Self-Report ratings. Posterior tail probabilities were used to evaluate the effect of education, income, race and employment.

Results:

Controlling for sex, age and treatment type, not having a college education, being unemployed or being non-white were associated with slower and decreased improvement. At week 12, not having a college degree reduced response by 10% (p=0.045), being unemployed by 7% (p=0.007) and being non-white by 11% (p<0.001). Treatment response was significantly related to income; having income at the 25th percentile of the income distribution decreased improvement by approximately 5% compared to those with income at the 75th percentile (p=0.018).

Conclusions:

Even within a short-term, randomized trial, socioeconomic factors play a critical role in acute treatment response.

Introduction

Major depressive disorder inflicts substantial morbidity and mortality across the lifespan. It increases the risk of suicide, increases health utilization, and, in patients under the age of 50, contributes to more disability adjusted life years lost than diabetes, stroke, heart disease and hypertension. While many patients respond to first-line treatments (i.e., antidepressant mediations and psychotherapy), accumulating data suggest that the socioeconomically disadvantaged—including those in poverty, unemployed, with lower income and educational attainment—have worse mental health outcomes compared to other socioeconomic strata¹.

Historically, research has focused on the hypothesis that socioeconomic status affects mental health outcomes through its impact on access to healthcare, provider availability, cost of treatment and engagement with the healthcare system^2,3. These studies suggest that the socioeconomic environment of patients with major depressive disorder directly affects treatment outcome. However, this hypothesis is difficult to assess through examination of observational data since variation in access, provider availability, cost and engagement are generally difficult to measure and are also often correlated with other important factors, such as financial barriers, health insurance generosity, income, and quality of care that can confound findings^4,5. Randomized controlled trials allow socioeconomic factors to be examined while controlling for related variables such as access and quality of care⁶.

The Combining Medications to Enhance Depression Outcomes (CO-MED; www.co-med.org) study provides a vessel in which socioeconomic variables can be examined across a large heterogeneous sample. Patients (n=665) in CO-MED received one of 3 antidepressant treatments (escitalopram + placebo, venlafaxine + mirtazapine or escitalopram + bupropion) for 12 weeks. The primary outcomes have been described previously; however, the impact of socioeconomic factors—other than race and ethnicity—on treatment outcome has received limited attention^7,8 in prospective, randomized controlled trials. Further, the trajectory of treatment response with regard to these features has received almost no attention. All CO-MED patients had access to the same treatments, which were provided free of charge and by the same providers working in Academic health centers, thus potentially eliminating any differential effects of financial barriers, health insurance generosity and provider quality.

With this in mind, we employed Bayesian hierarchical models^9,10 to examine the impact of educational achievement, income, employment status and race on treatment response (trajectory) in adults with major depressive disorder, while controlling for clinical features. We hypothesized, based on accumulating data from the socioeconomic disparities’ literature, that income, unemployment, lack of a college degree, and being a member of a minority would be separately associated with less improvement in depressive symptoms even after eliminating access and quality of care effects.

Methods

As previously described,¹¹ outpatients with nonpsychotic chronic or recurrent major depressive disorder were recruited from six primary and nine psychiatric care sites and randomized to escitalopram plus a placebo, escitalopram plus bupropion, or venlafaxine plus mirtazapine (1:1:1 ratio). The inclusion criteria were broad and exclusion criteria were minimal. Patients had to be in the index depressive episode for ≥2 months and had a score ≥16 on the 17-Item Hamilton Rating Scale for Depression (HRSD-17).¹¹ A list of exclusion criteria is available at www.co-med.org. The protocol and all consent and study procedures were approved by the institutional review boards at the National Coordinating Center (University of Texas Southwestern Medical Center) and all clinical sites. The study was conducted from March, 2008 to April, 2014. Treatment visits for the acute phase used in this study were at baseline and weeks 4, 8 and 12, and medication dose was adjusted in response to the 16-item Quick Inventory of Depressive Symptomatology–Clinician-rated (QIDS) based on the CO-MED Operations Manual (available at www.co-med.org).

The individual trajectory of improvement (i.e., week 0, 4, 8, 12) in depressive symptoms (QIDS-SR₁₆) was modeled using an individual logarithmic trends “random effects” coefficients Bayesian hierarchical model.¹² The multi-level Bayesian hierarchical model has several advantages over the standard MMRM approach – most notably allowing exact posterior inference, and explicit modeling and estimation of both observed and unobserved heterogeneity – and provides improved inference and precision.^9,10,12 Several functional forms for the trend (linear, time indicator fixed effects, log-linear, linear-log, double-log) were compared with AIC, BIC and predictive performance and, as in a number of previous studies of psychopharmacological treatment efficacy, the linear-log trend model is overall the best parsimonious specification.¹² Model specification details are provided in the Online Supplement.

Results

Demographic and clinical characteristics, including race, ethnicity and income are reported in Supplemental Table 1 available online and have been previously described⁷. Briefly, patients with college education had similar baseline depression symptom severity compared to those who did not (mean QIDS = 15.28 vs. 15.61, p=0.956). Similarly, age (42.30 vs. 43.69, p=0.937), and level of anxiety (mean HAMD = 7.86 vs. 8.35, p=0.886) were similar for those with and without a college education. Those who were employed had similar baseline depression symptom severity (15.07 vs. 15.92, p=0.888), age (41.40 vs. 45.25, p=0.830), level of anxiety (8.05 vs. 8.38, p=0.920) compared to those who did not.

In the whole sample (N=569 available for estimation after patient visits with missing values omitted), when controlling for age, sex, baseline symptom severity and the presence of anxiety, the mean log trend coefficient provides significant evidence of overall improvement (β=−0.489±0.018, CrI: −0.455 to −0.524, p<0.001) (Supplemental Table 2). Note that a negative coefficient estimate indicates a declining QIDS and so improvement over time. Not having a college education (β=0.410±0.205, CrI: 0.009 to 0.811, p=0.045), being unemployed (β=0.490±0.182, CrI: 0.133 to 0.848, p=0.007) and non-white (β=0.753±0.193, CrI: 0.374 to 1.131, p<0.001) all have positive trend coefficients and so are associated with slower and decreased improvement (Figure 1, Supplemental Table 2). Additionally, treatment improvement was significantly related to (standardized) income (β=−0.225±0.095, CrI: −0.038 to −0.412, p=0.019), with lower income being associated with less improvement (Figure 1, Supplemental Table 2).

Figure 1: The Impact of Social Determinants on Treatment Outcome in the Combining Medications to Enhance Depression Outcomes (CO-MED) Trial.

QIDS, Quick Inventory of Depressive Symptoms

The trajectory of improvement is shown for patients with and without college education in the first row (A and B) and for employed and unemployed patients in the second row (C and D) and for whites and non-whites in the third row (E and F). Average trajectories for the groups are shown in the left column (A, C, E) and the individual trends for each patient are shown in the right column (B, D, F). The relative influence of specific variables on response trajectory coefficients are shown in the radar plot (G). Panel H compares the outcome trajectory for two patient groups, Patients A: white, employed, with a college degree and with income at the 75^th percentile, to Patients B: nonwhite, unemployed, without a college degree and with income at the 25^th percentile.

Being unemployed results in 6.58% less improvement at end-point (week 12), not having a college degree leads to 9.57% less improvement at end-point, being Nonwhite leads to 11.26% less improvement at end-point and having income at the 25^th percentile of the income distribution results in 4.82% less improvement than having income at the 75^th percentile. Since these characteristics are likely associated with each other (e.g., people with lower income are less likely to have college education), it is worthwhile to examine the impact of multiple negative influences. Panel H of Supplemental Figure 1 compares the outcome trajectory for two patient groups, Patients A: white, employed, with a college degree and with income at the 75^th percentile, to Patients B: nonwhite, unemployed, without a college degree and with income at the 25^th percentile. This resulted in a 25.85% difference in improvement at week 12 (visit 4) between the two groups.

Discussion

Social determinants of health outcomes have been well described in medicine and recent calls encourage clinicians to attend to access to health care, affordability of treatment, language barriers and other constraints.¹³ However, while acknowledging that these social inequality-driven structural factors collectively produce disproportionate outcomes for some individuals, most prior work has focused on their naturalistic effects. Observing these effects in clinical trials that control for several of these factors--namely access to healthcare, cost of treatment, provider availability, and engagement with the healthcare system—has important clinical implications and implications for researchers and policymakers. Our findings reveal significant effects of employment, education, race and income on treatment response that are independent of access to care, cost of medication, insurance coverage, and quality of care. How much a patient with major depressive disorder improves within a clinical trial has long been thought to relate to the medication being administered, as well as its pharmacokinetics and pharmacodynamics. In some trials, clinical factors (e.g., baseline symptom severity, comorbidity) impact response. The present results suggest that a significant amount of variation in treatment response relates to socioeconomic characteristics. These findings have implications for public policy, clinical trial design and individualized approaches to treatment.

From a public policy perspective, approaches to treating depression are frequently siloed and rarely consider the patient’s socioeconomic milieu¹⁴. Our findings raise the possibility that attending to socioeconomic factors could enhance outcomes for a substantial number of patients and may be independent of the specific psychopharmacologic treatment. Historically, the effects of socioeconomic adversity on treatment outcome have been attributed to decreases in access to care and to the quality of care received. However, the influence of quality and access are generally controlled for in the context of a randomized controlled trial, highlighting that socioeconomic differences in outcomes can arise from other factors related to the patient’s environment.

As interventions—whether psychosocial or pharmacologic—are evaluated in randomized controlled trials, variation in response is generally attributed to differences in the efficacy of the interventions being evaluated. However, these findings suggest that substantial variation results from socioeconomic factors. These factors are infrequently assessed in many clinical trials and are rarely considered in outcome models. The analyses of most clinical trials have included clinical characteristics (e.g., baseline symptom severity, co-morbidity) and demographic characteristics (e.g., age, sex) rather than socioeconomic factors. Traditional analyses of these trials often assume that socioeconomic variables are ‘controlled for’ given that there is universal access to care and that other barriers to treatment are removed by the structure of the RCT. Thus, not only measuring but incorporating these factors into analyses of RCTs could enhance our ability to detect treatment-placebo differences or differences among active treatments. Additionally, consideration of socioeconomic factors could decrease the likelihood of abandoning effective treatments because of socioeconomically driven noise and could further enhance our understanding of patient-specific trajectories of improvement and help create targeted therapies.

While very few studies have examined socioeconomic factors that influence improvement in adults with depressive disorders in the context of a prospective, randomized trial, there are several limitations that warrant additional discussion. First, many of the socioeconomic factors likely represent proxies for other features that are difficult to measure. For example, patients with less education experienced slower improvement and less improvement; however, this could relate to multiple factors¹⁵. Those with less education may attribute or express symptoms and improvement differently compared to those who have had the opportunity to pursue more education. Further, these individuals with less education, even when employed, could also face lower job security and more stressful job conditions that could attenuate their improvement. Second, our results are based on a deterministic logarithmic trend model which imposes a more restrictive functional form for the trajectory of response but improves our precision. Third, we were limited in our ability to examine all moderators of treatment response and, despite the impressive sample size of CO-MED, had a limited number of observations over time.

In summary, this study demonstrates the impact of social determinants on treatment outcomes in depression. Future clinical trials should measure these socioeconomic factors and incorporate them analytically. Additionally, to optimize outcomes for adults with depression, we must leverage socioeconomic variables as potential predictors to inform treatment selection, monitoring, and delivery.

HIGHLIGHTS.

• Controlling for healthcare access, provider availability, and engagement with the healthcare system, socioeconomic factors represent statistically significant determinants of mental health outcome in a randomized controlled trial of pharmacotherapy for major depressive disorder.

• Not having a college education, being unemployed and non-white were associated with slower and decreased improvement. Significantly greater improvement was observed in wealthier patients.