A Review of the Clinical, Economic, and Societal Burden of Treatment-Resistant Depression: 1996–2013

Almost 50% of the U.S. population has experienced at least one psychiatric disorder in their lifetime (1). The lifetime prevalence of major depressive disorder is reported to be as high as 17%, and the 12-month prevalence is 5%−9% (2–4). The World Health Organization ranks major depressive disorder among the diseases that are most debilitating to society, in part because of its association with increased utilization of health care resources, diminished quality of life, and indirect personal and societal costs (5,6).

More than 50% of patients with major depressive disorder do not reach remission with an initial treatment; of those, 30%−50% also do not respond (4,5,7–9). The designation “treatment resistant” is used to describe patients who do not respond to antidepressant therapy after one or more adequate trials (9–11) (duration of at least six weeks and use of appropriate dosages [12–15]). A response to treatment is commonly measured as a ≥50% decrease in baseline scores on the Hamilton Rating Scale for Depression (HAM-D) (7). The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial studied the effectiveness of different treatments for major depressive disorder among patients who did not become symptom free after one or more treatments. Seventy-two percent of patients did not experience remission after treatment with citalopram, the initial medication used in this study; with each subsequent treatment, failure to remit increased from 79% to 84% to 93% (8,16–19).

Patients with treatment-resistant depression contribute a disproportionately high burden of illness compared with patients who respond to treatment. Several studies have documented a subset of outcomes related to treatment-resistant depression, such as societal costs and effects on quality of life, labor force participation, or medical resource utilization, but none has provided a comprehensive, systematic, and rigorous review of the overall burden.

The aim of this systematic review was to assess the aggregate burden of treatment-resistant depression in the United States by compiling the data available from published studies on clinical, societal, and economic outcomes associated with treatment-resistant depression. The review assessed characteristics and comorbid conditions, costs of treatment, mortality, quality of life, severity of symptoms, response rates to subsequent treatments, and adverse events among patients with treatment-resistant depression. This review also investigated factors that contribute most to the societal burden of treatment-resistant depression and identified potential areas for improvement.

Methods

We used PubMed to search MEDLINE for articles published between January 1996 and May 2011 by using terms related to treatment-resistant depression, outcomes, economics, and society. A supplemental search of PubMed using additional terms was conducted for articles published between May 2011 and August 2013. We also conducted a search of the Tufts Cost-Effectiveness Analysis Registry (20) for articles published from 1996 through 2013 and reviewed the reference lists of articles identified by the searches related to costs of treatment. [A complete list of search terms and a full description of the study methods are available online as a data supplement to this article.]

Two researchers screened each abstract of the retrieved articles. The articles were included if the abstract referred to primary data collection as a part of the design; if endpoints of the study were pertinent to clinical, societal, or economic outcomes; and if the study enrolled adults ages 18 and older. We defined treatment resistance as failure to respond to one or more adequate trials of drug therapy. This definition was intended to conform to accepted criteria of treatment resistance while including heterogeneous definitions. The evidence grade for each study was assessed by using the quality index developed by the Mental Disorders and Illicit Drug Use Expert Group (21). Quality index scores range from 0 to 17, with higher scores indicating that more complete reporting and higher quality methods were used.

We included articles that reported results from the STAR*D trial, even though its definition of treatment resistance as “failure to remit” is more inclusive than the one used in this review, which included patients who failed to respond. However, as the largest and longest study evaluating depression treatment, the STAR*D study provides valuable information for comparing responses to treatment, despite differences in definitions of treatment resistance.

Summary statistics were performed by using Stata, version 9.2, and were weighted by sample size. Unless stated otherwise, results are reported for the treatment-resistant population.

Results

Literature search and study characteristics

The original search identified 442 articles; 62 were included in this review. [The complete list of included articles is available in the online data supplement.] The mean±SD study duration was 7.75±1.75 years. The quality index score varied from 6 to 18 (mean±SD=13±3). Sample sizes in the studies varied from six to 24,415 patients (median=42; Table 1) and, together, enrolled 59,462 patients. Thirteen studies summarized data on outpatients, eight on inpatients, and six on both populations; 22 articles did not specify a population but likely included predominantly outpatients.

Baseline patient characteristics

Patients’ baseline characteristics varied by study (Table 1). The mean age was 46.7, and illness duration was 4.4 years. Women represented 71% and non-Hispanic whites represented 89% of the study populations. Patients had 3.8 prior depressive episodes (range .8–7.2). On average, patients had not responded to 4.7 drug trials (range 1–10) and 2.1 drug classes (Table 2).

Symptom severity

The scales used to assess symptom severity varied greatly. Commonly used scales included the HAM-D, the Montgomery-Asberg Depression Scale (MADRS), the Clinical Global Impression–Severity (CGI-S) scale, and the Quick Inventory of Depressive Symptomatology–Clinician Rated (QIDS-C). The average scores at baseline on the HAM-D-17, the MADRS, the CGI-S, and the QIDS-C indicated that symptom severity was within or close to the scoring range used to classify conditions as severe or markedly ill (Table 2) (22–32). At the end of the studies, severity scores had improved by an average of 35%±8% (improvements of 12.8±4.1, or 42%, on the HAM-D-17; 23.2±7.7, or 27%, on the MADRS; and 3.0±1.4, or 36%, on the CGI-S). Scores at the end of the studies were not reported for the QIDS-C.

Comorbid conditions

Comorbid conditions were relatively common among patients with treatment-resistant depression, as indicated in Table 1. Comorbid conditions included joint, limb, or back pain (73%); hypertension (67%); and dyslipidemia (61%) (33,34). Some psychiatric conditions, such as malaise or fatigue, anxiety, and personality disorder, were more prevalent among patients with treatment-resistant versus treatment-responsive depression (15,33–37). Suicidal ideation was reported for 15%±8% of patients with treatment-resistant depression, 6% of patients with treatment-responsive depression, and 1% of the general population (33,38,39). Approximately 17%±6% of patients with treatment-resistant depression had a history of suicide attempt, with an average of 1.1±.2 prior suicide attempts each (8,15,33,39–44).

Mortality rates (deaths per 1,000 patient-years) were similar among patients with treatment-resistant (46.2) and treatment-responsive (46.8) depression in a Medicare population and were about 4% lower than the rate for individuals in the general population (48.2). (33) No articles summarized associations between treatment-resistant depression and other outcomes, such as crime rates, incarceration, social services’ utilization, or caregivers’ quality of life.

Response and remission rates

Therapeutic options for treatment-resistant depression consisted of deep brain stimulation, transcranial magnetic stimulation, transcranial direct-current stimulation, vagus nerve stimulation, group psychoeducation, cognitive therapy, and a variety of drugs, most commonly lamotrigine, lithium, olanzapine, and venlafaxine. Sixteen studies reported rates of remission and response (Table 3) (8,9,17,38,43–54). Response rates for different treatment groups varied between 0% and 80%, for an average of 36%±1%. Remission rates varied from 8% to 80%, for an average of 20%±1%. The lowest average response and remission rates were reported in the STAR*D trial: scores on the QIDS–Self-Report–16 indicated that only 15% and 10%, respectively, had responded to treatment or were in remission (8). One study found cognitive therapy to be effective; patients averaged a significant 9-point decline in the Beck Depression Inventory score, and 26% had a “sustained recovery” at 26 weeks (55).

Medication-related adverse events

Twenty-seven studies reported on the incidence of adverse events among patients with treatment-resistant depression. Rush and others (56) found that 53% of patients experienced at least one adverse event, of which 81% were mild or moderate. The most frequent mild or moderate drug-related adverse events were decreased sexual desire (33%), orgasmic dysfunction (26%), and blurred vision (15%). The most frequent severe drug-related adverse events were dissociative reactions (13%), ataxia (13%), mixed states (dysphoric mania or agitated depression) (10%), and tremor and nausea (10%). The most frequent procedure-related adverse events were swollen eye (55%), headache (38%), and erythema (36%); no severe procedure-related adverse events were reported. Only four studies that reported adverse events had a placebo arm. Adverse events that occurred in more than 5% of the placebo groups were blood pressure change (40%), headache (15%), dissociative reaction (in a study utilizing ketamine) (13%), and manic reaction (13%) (10,48,54,57).

Quality of life and costs

Quality-of-life data were taken from published models that used data sources ranging from reviews of prior literature to original trials (58–63). The studies measured quality of life with a continuous scale, with 0 indicating death and 1 indicating perfect health. Average baseline scores were .552±.120 for patients with major depressive disorder (58–61), .826±.065 for patients in remission (59–63), .673±.031 for patients who responded to therapy without remission (61,62), and .417±.126 for patients who did not respond to therapy (61–63). Adverse events caused a further loss of .01 to .12 quality-of-life units (59).

Five studies provided detailed information about annual per-patient costs for treatment-resistant depression, one from Medicare and four from databases of national employers and private payer claims (Table 4) (33,36,54,64,65). Annual claims for visits to a medical facility were relatively common among patients with treatment-resistant (N=28.3 claims) versus treatment-responsive depression (N=15.1 claims) (64). Fifty-two percent of patients with treatment-resistant depression were hospitalized over their lifetimes (39).

Among private payers, the mean annual direct health care costs per patient for management of treatment-resistant depression ($13,196) were $5,481 higher than for management of treatment-responsive depression ($7,715) (Table 4) (36,64,65). These comparisons were from the same studies, so observed differences were not an artifact of sample heterogeneity. Annual direct medical costs for persons in the general population were $3,997 (64). Total costs in productivity per patient-year were $4,048 higher among patients with treatment-resistant depression ($6,924) than among patients with treatment-responsive depression ($2,876) (36,64) (Table 4). For the general population, annual productivity costs were $1,098 (64). The total annual direct and indirect costs per patient-year were $20,120 for patients with treatment-resistant depression and $10,592 for an age-matched population with treatment-responsive depression (Table 4) (36,64,65). These costs were $5,095 in the general population (64).

Annual direct medical costs among Medicare patients were $20,736 for patients with treatment-resistant depression, $14,098 for patients with treatment-responsive depression, and $10,380 in the general population of Medicare patients (33). These differences in cost in the Medicare population reinforce the considerable health care benefit that could be obtained from more effectively treating major depressive disorder.

Discussion

Our review of literature shows that among the many burdens of patients with treatment-resistant depression, they experience only a 20% probability of achieving remission in the course of treatment, a 17% prevalence of prior suicide attempts, substantially lower quality-of-life scores than patients whose depression remits, and direct and indirect annual costs among private payers that are $9,529 higher than those of patients with treatment-responsive depression. Patients with treatment-resistant depression had 28.3 annual general medical visits, more than three times as many as the general population average of 8.7, whereas patients with treatment-responsive depression had less than twice as many annual general medical visits as the general population (64).

Although response and remission rates for patients with treatment-resistant depression varied widely among studies, average improvements on the CGI-S (36%), the HAM-D-17 (42%), and the MADRS (27%) were similar. The STAR*D study found particularly low rates of response (15%) and remission (10%), possibly because this study consisted of patients who did not respond to up to three well-monitored treatments, whereas many other studies used nonresponse to just one or two naturalistic treatments as their criterion for treatment resistance.

Among adults in the United States, the 12-month prevalence of major depressive disorder in 2012 was 6.7% (3), or 16 million individuals. Assuming that 12% of these patients had treatment-resistant depression, which was the lowest estimate reported by studies in this review (64), a total of 1.9 million adults in the United States had treatment-resistant depression in 2012. If annual costs of treatment per patient-year for private payers totaled $20,120, as estimated, the total costs for treatment of this population in 2012 was $38 billion. Adding in the annual cost of treatment of the 14 million patients with treatment-responsive depression ($10,592 each; $148 billion total), the societal burden of major depressive disorder for the United States in 2012 was $188 billion. By comparison, the societal cost for cancer was $131 billion (66), and in 2007, the societal cost for diabetes in 2012 dollars was $173 billion (67).

Corey-Lisle and colleagues (64) stated that treatment resistance among patients with depression might actually be as high as 20%. When this higher estimate was used, the estimated societal burden of major depressive disorder in the United States reaches $200 billion per year and the burden of treatment-resistant depression alone reaches $64 billion per year. A prior study estimated that the societal cost of major depressive disorder was $124 billion per year, in 2012 dollars (2). This estimate may be lower than our estimate of $188–$200 billion because of temporal changes in unit service costs, for example, increasing wages, and the possible exclusion of costs for patient visits that were not reported as being primarily for depression (36,64).

The burden of major depression can also be calculated as an increase in costs above those of the general population. If 12% of patients with depression are resistant to therapy, the costs of depression above general population costs are $29 billion for patients with treatment-resistant depression, $77 billion for patients with treatment-responsive depression, and $106 billion for all major depression. If 20% of depressed patients are resistant to therapy, the estimated costs are $48 billion for patients with treatment-resistant depression, $70 billion for patients with treatment-responsive depression, and $118 billion for all major depression. Conversely, if 70% more patients respond to antidepressant therapy, the costs for the 3.6% of patients who remain treatment resistant would decrease to $8.6 billion, and costs for patients with treatment-responsive depression would be $84.3 billion. The aggregate $93 billion annual cost of major depression above general population costs would save $13 billion, a 12% reduction in the annual cost of treating all patients with major depressive disorder.

Gillum and colleagues (68) recently compared 29 common conditions and diseases in terms of incidence, prevalence, and disability-adjusted life years. Depression ranked first in populationwide burden by disability-adjusted life years. The next four most impactful conditions or disorders were injuries, ischemic heart disease, alcohol abuse, and chronic obstructive pulmonary disorder. The low quality-of-life scores (.417 on a scale of 0 to 1) reported by patients with treatment-resistant depression who did not respond to therapy fell within the range of scores reported for metastatic cancer, chronic moderate-to-severe pain, or acquired blindness (20). Although these quality-of-life measures may be confounded by patients’ current mood and depressive symptoms (69), the estimates amount to one million quality-adjusted years lost due to treatment-resistant depression in the United States. Treatment-responsive depression accounts for an additional loss of more than 1.5 million quality-adjusted years.

The small numbers and heterogeneity that characterized the study populations and treatments described in the articles summarized here may limit the general applicability of some findings. For example, only a few studies included information about comorbid disorders and adverse events. Our summaries of the five studies that reported costs per patient-year for depression treatment and our extrapolation of the data to determine nationwide costs can be combined with results of future studies to more accurately calculate the monetary impact of treatment-resistant depression. Other limitations of this analysis included the heterogeneity of methods and the degree to which complete findings were reported, the relatively short periods of patient follow-up, and a focus primarily on clinical endpoints during follow-up.

The burden of treatment-resistant depression was likely underestimated because there is limited published research, or none at all, on the incidence of crime rates, incarceration, and use of social services among persons with depression and on costs and quality-of-life burden for family members and caregivers. The burden might also have been underestimated because we excluded studies with adolescent populations, which have similar prevalence rates for depression as adults but lower treatment rates (70–72). Among patients with treatment-resistant depression, adolescents experienced a greater burden with regard to suicide attempts compared with adults, and their rate of substance use disorders (54%) was higher than the current (3%) and lifetime (18%) rates for adults (73).

In response to these challenges, personalized medicine technologies are finding increasing application in reducing the burden of major depression. One such approach, pharmacogenomics, matches the pharmacokinetic and pharmacodynamic properties of antidepressant medications to the genetic profile of individual patients. Pharmacogenomic test results can identify variability in psychiatric drug response and, when applied clinically, can increase antidepressant responses by 70% (74–78). Identifying the right medication and dose for patients will shorten patients’ treatment odyssey, thereby decreasing the proportion that become treatment resistant. The resulting savings would diminish annual health expenditures and increase productivity. Psychiatric pharmacogenomics and other innovations may also improve the classification of major depressive disorder subtypes; define more individualized and earlier, potentially prodromal treatments; and reduce adverse event rates (79,80).

Conclusions

The studies reviewed here reveal that the personal and societal burdens of depression are disproportionately greater among patients who do not respond to antidepressant therapies. Contributory factors to this extra burden include more unsuccessful drug trials; more comorbid disorders, such as malaise, fatigue, and anxiety; and increases in suicidal ideation. The far greater personal financial burden of treatment resistance is likely to compound anxiety, depression, and comorbid disorders as financial hardships mount.

Despite multiple treatment options available to clinicians, treatment resistance remains highly prevalent and exacts a substantial toll on patients’ quality of life and on society. The burden of treatment-resistant depression is on a par with or is greater than that of other chronic conditions such as cancer and diabetes, yet depression ranks only 15th among conditions or disorders that receive National Institutes of Health research funds (68).

Improvements in technology are likely to diminish treatment resistance and mitigate its extra costs. These include pharmacogenomic tests that incorporate additional DNA variants and DNA methylation status; new medications that are based on nonmonoaminergic approaches, such as N-methyl-D-aspartate receptor antagonism; and refinements in transcranial magnetic stimulation.

Disparities in care also need to be addressed so that innovations can achieve a broad societal impact. Certain racial or ethnic groups and unemployment are associated with underdetection of mental illness and inadequate mental health care access or quality (81–84). Solutions include more social programs, better patient education, quality improvement programs, and increased resources for indigent care clinics (81–84). Consistent processes for assessing regulatory and reimbursement implications of new therapeutic and diagnostic practices are also needed (79). Expanded treatment alternatives, augmented with pharmacogenomic decision support, have the potential to improve outcomes and help patients retain productive lives. Along with decreases in health care costs and increased global productivity, there is much to achieve—and expect—by decreasing the individual and societal burdens of treatment-resistant depression.