The Economic Burden of Adults With Major Depressive Disorder in the United States (2005 and 2010)

See our Focus Collection of J Clin Psychiatry articles on healthcare economics.

The Economic Burden of Adults With Major Depressive Disorder in the United States (2005 and 2010)

Paul E. Greenberg, MS, MA; Andree-Anne Fournier, MA; Tammy Sisitsky, MA; Crystal T. Pike, MBA; and Ronald C. Kessler, PhD

Depression is among the most burdensome disorders worldwide, giving rise to considerable adverse effects on activities of daily living for extended periods of time.1,2 In the United States, it is a leading cause of disability for people aged 15-44 years, resulting in almost 400 million disability days per year, substantially more than most other physical and mental conditions.3,4 The economic burden of depression, including major depressive disorder (MDD), bipolar disorder, and dysthymia, was estimated at $83.1 billion in 2000 in the United States.5 This total was composed of $26.1 billion in direct medical costs, $5.4 billion in suicide-related mortality costs, and $51.5 billion in indirect workplace costs (absenteeism from work and presenteeism while at work).5 Several subsequent studies have quantified specific components of the cost of depression, such as workplace costs or outpatient treatment costs.6-8 Other studies have analyzed the economic burden of closely related conditions, such as treatment-resistant depression.9 However, no study has attempted to quantify the overall economic burden of MDD. In addition, although recent literature highlights the importance of comorbidities among people with depression,10,11 prior estimates focused on the cost of depression alone, with no attention to the added economic burden from physical and psychiatric comorbidities.5,12

The objectives of the current study are to examine (1) excess costs incurred by adults with compared to those without MDD having otherwise similar profiles, (2) the portion of these costs attributable to MDD itself as opposed to comorbid conditions, and (3) the relative importance of the different components of the overall economic burden of MDD, with attention to direct costs, suicide-related costs, and workplace costs. In doing so, we update previous research with attention to any changes that occurred between 2005 and 2010. In the absence of a single database from which to draw, we rely on findings from the literature to supplement original results.

METHOD

Key elements of the method are described below. Our framework for evaluating the economic burden of adults with MDD is also summarized in Supplementary eTable 1.

Prevalence Data

Prevalence rates by gender, age, employment, and treatment status relied on the 2005 and 2010 updates to the National Survey on Drug Use and Health (NSDUH), a national probability sample of the adult US civilian, noninstitutionalized population.13,14 In the NSDUH depression module, adult respondents were asked questions adapted from the National Comorbidity Survey Replication,15,16 which was based on World Health Organization World Mental Health Survey Initiative version of the Composite International Diagnostic Interview. On the basis of these responses, DSM-IV criteria17 were used to identify people with a past-year major depressive episode (MDE), defined as symptoms occurring for a period of 2 weeks or longer over the past 12 months among those with lifetime MDE.18

Cost Data: Sample and Control Group

Individuals aged 18-64 years with diagnosed MDD in 2005 or 2010 (study years) were selected from the OptumHealth Reporting and Insights administrative claims database.19,20 This private insurance database includes over 16 million beneficiaries (ie, employees, spouses, and dependents) from 69 large, self-insured US companies. Patients with MDD were included for analysis if they had at least 2 ICD-9-CM claims for MDD—296.2 (single episode) or 296.3 (recurrent episode)—occurring on different dates during 1 of the 2 study years. To ensure that a complete claims history was available, each patient was required to have continuous health care eligibility during the study period. Patients with MDD having health maintenance organization, capitated, or Medicare coverage were excluded from the analysis because payment information may be incomplete for these patients. Controls with no diagnosis of MDD and no prescription for antidepressant or antipsychotic/antimanic drugs during the study years were selected using similar criteria.

• The cost of major depressive disorder in the United States is known to be substantial, but little attention has been given to the added economic burden from physical and psychiatric comorbidities.
• Comorbid conditions account for the largest portion of the growing economic burden of major depressive disorder and should be considered in the treatment of the disease.

For each study year, patients with MDD were matched 1-to-1 to controls using a combination of direct characteristic matching and propensity score analysis. Specifically, patients were matched directly by age, gender, region, insurance type, employment status, relationship to primary beneficiary, and Charlson Comorbidity Index21,22 and by propensity score using a caliper of 0.25 standard deviations of the propensity score. The propensity score was calculated using a logistic regression with controls for the most frequently observed general physical comorbidities that were found to be statistically significantly different at baseline between MDD patients and controls and that were not known to be MDD related.23-25 For example, we controlled for hypertension, which has no reported association with MDD, but did not control for back pain since we wanted to capture the incremental costs of this condition, which has been documented as MDD related (despite the ambiguity of the causal direction). Detailed comparisons of characteristics of MDD patients and controls before and after patient matching in both study years are presented in Supplementary eTables 2-5.

Direct Costs Estimation

Average costs per patient, including medical services and prescription drug costs, were calculated in the 2 study years for both MDD and control patients. Three categories of costs were estimated: (1) MDD costs, including costs that occurred on the same day and in the same location as a medical claim with an MDD diagnosis, as well as pharmaceutical costs for antidepressant and antipsychotic/antimanic drugs; (2) other depression costs, including medical costs that occurred on the same day and in the same location as a medical claim with a diagnosis for another type of depression but not MDD specifically, as well as pharmaceutical costs for antianxiety and anticonvulsant drugs; and (3) nondepression costs, including all costs not captured in the first 2 categories. Incremental costs were calculated in each study year by subtracting average costs of controls from those of MDD patients. While the first cost category was the basis for estimating the direct costs of MDD, attention to all 3 categories taken together yielded an estimate of the direct costs of individuals with MDD.

Direct costs were estimated according to employment and MDD treatment status: (1) employed and treated—costs estimated from claims data; (2) employed and not treated—MDD costs set equal to 0, and non-MDD costs (the second and third cost categories above) set equal to those incurred by employed and treated patients; and (3) not employed (treated or not treated)—costs assumed to be 1.7 times those found in the employed population based on the ratio of health care costs incurred by MDD patients in a Medicaid population compared with a privately insured population.26 (See Supplementary eTable 6 for detailed calculation of ratios used to infer missing cost categories.) Since the direct costs of MDD for people aged ≥ 65 years could not be observed in these claims data, they were assumed equal to those calculated for individuals aged 50-64 years. Societal direct costs were extrapolated by multiplying NSDUH estimates of the number of people with MDD by the direct cost estimates per patient for each of these 3 categories, stratified by age and gender.

Suicide-Related Costs Estimation

Suicide-related costs were estimated using the human capital method, incorporating the conservative assumption that household services had no human capital value.5,12 The total number of suicides by age and gender cohort in 2005 and 2010 was obtained from the Centers for Disease Control.27 On the basis of prior literature, we attributed 50% of suicides to MDD in our cost model.28-30 In addition, the present value of lifetime earnings was estimated based on mortality rates and life expectancies from the National Vital Statistics Report31,32 together with data from the Bureau of Labor Statistics.33,34 To express future earnings in present value terms, we applied a 3% discount rate.35

Workplace Costs Estimation

Workplace costs were estimated by assessing the value of missed days of work (absenteeism) and reduced productivity while at work (presenteeism) of individuals with MDD. By following the same approach as that described above for direct costs, workplace costs were estimated for the 3 categories: MDD costs, other depression costs, and nondepression costs. We estimated 3 categories of absenteeism costs: (1) injury/illness, (2) discretionary time off, and (3) disability. Absenteeism due to injury and illness was imputed for the employed and treated subgroup based on OptumHealth data, with outpatient visits on workdays counting as half a day missed and inpatient or emergency department visits on workdays counting as full days missed. The second absenteeism cost category was based on NSDUH, which reported the number of workdays missed “because the respondent didn’ t want to be there.” The added number of days missed by individuals with MDD compared with controls was attributed to the effects of MDD. Costs associated with these 2 categories were estimated for each patient based on the cumulative number of workdays missed (combining claims and NSDUH information) multiplied by that employee’s daily wage (from claims data). The third category of absenteeism costs, disability, was estimated directly from the claims data, which included duration and costs of short- and long-term disability for employed and treated beneficiaries. Absenteeism costs in the employed and not-treated group were assumed to be 48% of those incurred by the employed and treated group based on the workdays missed reported by both groups (from NSDUH). Presenteeism costs were assessed at 6.1 times the cost of absenteeism due to injury and illness, following previous MDD literature estimates.7 (See Supplementary eTable 6 for detailed calculation of ratios used to infer missing cost categories.) Finally, societal workplace costs were extrapolated using the approach described above for direct costs.

RESULTS

Prevalence, Employment, and Treatment Rates

Between 2005 and 2010, MDD prevalence rose from 13.8 million to 15.4 million adults (Table 1A). This growth was unevenly distributed by age, with the ≥ 50-year age group increasing fastest and the other age groups experiencing minor growth or slight decline (Figure 1). The overall change in prevalence was partly the result of the US adult population growing from 216 million to 228 million and partly due to an MDD prevalence rate increase from 6.4% to 6.8%.

Worsening economic conditions after the 2008 downturn took a particularly heavy toll, with 0.3 million fewer persons with MDD employed full-time, 0.3 million more employed part-time, and 1.6 million more not employed at all (ie, unemployed or not looking for work). Whereas the full-time employment rate was 8.7 percentage points lower for persons with than those without MDD in 2005, this gap widened to 10.3 percentage points in 2010 (Table 1B). Furthermore, full-time employment rates in the MDD group decreased 6.9 percentage points during this period, from 47.2% to 40.3%. This represented both a lower starting level and a steeper decline compared with the non-MDD group, which experienced a 5.3 percentage point reduction from 55.9% to 50.6%. Compared with the non-MDD group, the MDD group was also more frequently unemployed or not looking for work. Taken together, following the economic downturn, the “not-employed” rate increased by 6.2 percentage points among those with MDD compared with 3.8 percentage points within the non-MDD group.

There were 1.5 million more people treated for MDD in 2010 than in 2005, with the treatment rate increasing from 52.2% to 56.2% in that period (Table 1C). While the rate of treatment increased by 6.1 percentage points among the full-time employed, less pronounced rates of increase were observed among the part-time employed and the not-employed groups (1.2 and 0.8 percentage points, respectively).

MDD Costs: 2010 Versus 2005

The incremental economic burden of individuals with MDD was $173.2 billion in 2005 and $210.5 billion in 2010, an increase of 21.5% over this period (Table 2A). A large portion of this increase was attributable to higher direct medical and presenteeism costs. The composition of total incremental costs was stable over time, with 48%-50% attributable to workplace costs, 45%-47% to direct costs, and 5% to suicide-related costs. Of these total incremental costs, only 38% was attributable to MDD itself as opposed to comorbid conditions (Table 2B).

Direct costs. Direct costs incurred by individuals with MDD totaled $77.5 billion in 2005, rising 27.5% to $98.9 billion in 2010 (Table 2A). The majority of the costs were for outpatient and inpatient medical services representing 15%-18% and 9%-10% of total costs in each study year, respectively. In addition, pharmacy costs accounted for 15% of the total in 2005 and 13% in 2010. Furthermore, direct costs attributable to individuals with MDD aged ≥ 50 years increased the fastest, the costs incurred by 18- to 25- and 26- to 34-year-old patients grew less rapidly, while those stemming from the 35- to 49-year age group remained flat (Figure 1). (See Supplementary eTable 7 for additional results on incremental direct costs of individuals with MDD by employment status, treatment status, and age group in 2005 and 2010.)

For employed and treated patients, incremental direct costs of health care services were $5,707 per MDD patient in 2005, increasing by 5% in 2010 to $5,988 (Table 3). In both study years, the costs of services that were directly attributable to MDD accounted for only 40% of the incremental direct costs, with MDD drug costs declining over time as generics became more widely used.36 An additional 10%-11% was due to depression other than MDD, with another 48%-51% stemming from other conditions (eg, hospitalization or physician office visits for mental illnesses, including anxiety, adjustment disorder, and posttraumatic stress disorder, as well as a range of non-mental health services). This was mostly composed of various manifestations of pain, including disc and back disorders, abdominal pain, and chest pain, as well as sleep disorders, and migraines (Table 3).37,38 (See Supplementary eTable 8 for additional results on incremental cost per patient for non-mental health medical services in 2010.)

Suicide-related costs. The adult suicide rate was 14.2 per 100,000 in 2005 and 15.9 per 100,000 in 2010,27 with 50% of those suicides attributed to MDD. Suicide-related costs for the MDD group were estimated at $9.4 billion in 2005, 5% of the overall economic burden, rising by 2.7% to $9.7 billion in 2010 (Table 2). (See Supplementary eTable 9 for the number of suicides and a detailed calculation of lifetime earnings lost due to MDD in 2005 and 2010.)

Workplace costs. Presenteeism accounted for approximately 3 quarters of workplace costs and represented 37% of the overall economic burden of individuals with MDD (Table 2). In each study year, the equivalent of approximately 32 incremental workdays was lost due to presenteeism by the average individual with MDD (data not shown). Total presenteeism costs were estimated at $64.7 billion in 2005, and rose 21.5% over the ensuing 5 years to $78.7 billion in 2010. In contrast, incremental costs associated with absenteeism grew only 8.4% over the period 2005 to 2010, from $21.5 billion to $23.3 billion. This is composed of 21.5% growth in absenteeism due to injury/illness costs, 19.4% growth in discretionary absenteeism costs, and 24.6% decrease in disability costs. (See Supplementary eTable 10 for detailed results on the incremental workplace costs of individuals with MDD in 2005 and 2010.)

DISCUSSION

The purpose of this study was to estimate the economic burden of people with MDD; assess the cost share attributable to MDD versus comorbid conditions; understand the relative contributions of direct costs, suicide-related costs, and workplace costs; and analyze changes in these cost categories between 2005 and 2010. The emphasis on understanding the role of comorbid conditions dates back to our earlier cost-of-illness study,5 which concluded that “future research will incorporate additional costs associated with depression sufferers, including the excess costs of their coexisting psychiatric and medical conditions.”5(p1465)

The current study adds to our understanding of MDD as a source of significant economic burden. We estimated the incremental cost of people with MDD at $173.2 billion in 2005 and $210.5 billion in 2010, with 45%-47% attributable to direct costs, 5% to suicide-related costs, and 48%-50% to workplace costs. But MDD direct costs (including both medical and pharmaceutical services directly related to MDD treatment itself) accounted for only 12%-13% of this incremental burden of people with MDD, totaling $21.6 billion in 2005 and $27.7 billion in 2010. In fact, for every dollar spent on MDD direct costs in 2010, an additional $1.90 was spent on MDD-related indirect costs (ie, suicide-related, workplace), and another $4.70 was spent on direct and workplace comorbidity costs incurred by persons with MDD (Figure 2).

Individuals with MDD thereby experience a wide range of disease burdens, much of which is not usually categorized as directly related to the MDD treatment itself. The treatment rate of MDD remained relatively low in 2005 and 2010 (in the 50% range), likely contributing to the increasing burden of the disease. Thus, to the extent there is success in raising treatment rates, we would expect a shift from costs of people with the disease (ie, reduced cost of comorbid conditions) to costs associated with MDD (ie, increased treatment costs). The extent to which these competing forces would offset one another in dollar terms is a topic for further research. For example, whereas a condition like back pain may well be exacerbated by the presence of depression and therefore could potentially be alleviated by its successful treatment, the same is unlikely to be true for a condition like hypertension or cancer. Recent studies39-41 suggest that collaborative care approaches that are mindful of these different pathways could allow for cost savings in the long term.

Understanding the source of cost changes between 2005 and 2010 reported in this study also merits further attention. The 27.5% increase in direct costs for people with MDD can be broken into 2 approximately equal-sized components: change in number of MDD cases, accounting for 12.8 percentage points of the increase in direct cost (of which 5.8 percentage points are due to growth in the adult US population, and 7.0 percentage points are attributable to an increase in the MDD prevalence rate) (Table 1); and change in cost per case, accounting for an additional 14.8 percentage points of the increase in direct costs.

Further research on the underlying drivers of these observed changes is warranted.42 Indeed, the increase in the incremental direct cost per employed and treated patient with MDD reported in Table 3 was more modest among the employed and treated patients with MDD, at 5%. This was most likely driven by the changing mix of employed patients with MDD in 2010 compared with 2005. That is, the worsening economy most likely led to a less severely depressed employed group by 2010, which would be an offsetting factor affecting the cost change for this group.

Through the business cycle, labor force attachment is far more volatile for people with MDD compared to those without. On the one hand, involuntary unemployment could spur more MDD, and on the other hand, the symptoms of illness could diminish employment prospects.43 In economically robust times, persons with MDD tend to be highly employable, but when economic conditions worsen, they are disproportionately adversely affected, especially those aged 50 years and older. Furthermore, during economic downturns, the buffer of part-time work is not as widely available or accessed by the MDD group. Thus, the complex interplay of work status, MDD symptoms, and MDD treatment warrants continued study. Shining a bright light on the relative impact of these different contributors to direct cost changes, and understanding how they have moved historically, would offer insight into the available levers that could be brought to bear in most effectively managing resource utilization in this context over time.

There are several limitations of this study that are noteworthy. First, in the absence of a single data source to evaluate the economic burden of MDD, we relied on both original as well as literature-based estimates. Our ability to overlay literature-based estimates on our original analyses depends on the underlying consistency of these sources. However, it is comforting that many of these estimates are based on representative samples of persons with MDD drawn from the same or similar time periods and that the results do not appear to be highly sensitive to changes in these estimates. That is, we performed a sensitivity analysis with respect to the 3 parameters that are drawn from estimates in the literature (ie, the direct costs of MDD subjects who are not employed, the absenteeism costs of MDD subjects who are employed but not treated, and the presenteeism costs of MDD subjects who are employed) and found that increasing (or decreasing) these parameters by 10% results in an increase (or decrease) in the overall costs of MDD of 6% and in the costs of people with MDD of 5%. Second, presenteeism is estimated based on the relationship between presenteeism and absenteeism costs in 2002. To the extent this ratio changed through the business cycle, our estimate may not properly capture the workplace dynamic. Also, by 2010 many more jobs did not require a physical presence in the workplace, which could have changed the relationship between absenteeism and presenteeism relative to an earlier era. Third, our data do not allow costs for patients aged 65 years and older to be estimated directly, and so we have imputed these costs based on the cohort aged 50-64 years. In addition, our data do not allow for analysis of beneficiaries covered under certain types of managed care plans. This raises the possibility that our extrapolation does not fully reflect the experiences of these individuals. Finally, there are no severity markers in the data on which this study is based. Further research is needed on the sources of the economic burden of depression by severity of illness.

Taken together, the economic burden of adults with MDD is large and has grown over time. Several factors most likely contributed to the increase in costs between 2005 and 2010, including growth in the US population, increase in MDD prevalence, increase in treatment cost per individual with MDD, changes in employment and treatment rates, as well as changes in the composition and quality of MDD treatment services. Future research should focus on the relative importance of these different factors and analyze further the comorbidities associated with MDD that account for the largest portion of the total economic burden of the disease.

Author affiliations: Harvard Medical School (Dr Kessler) and Analysis Group, Inc (Mr Greenberg and Mss Fournier, Sisitsky, and Pike), Boston, Massachusetts.

Potential conflicts of interest: In the past 12 months, Dr Kessler has served as a consultant for Hoffmann-La Roche and Johnson & Johnson Wellness and Prevention; has served on advisory boards for Mensante, Johnson & Johnson Services, Lake Nona Life Project, and US Preventive Medicine; and owns 25% share in DataStat. Mr Greenberg and Mss Fournier, Sisitsky, and Pike have no financial conflicts of interest to disclose.

Funding/support: None reported.

Acknowledgments: The authors are grateful for the substantial research assistance provided by Elizabeth Chertavian, BA; Michael Kaminsky, BS; and Ngoc Pham, BA, of Analysis Group, Inc. The authors also thank Ana Bozas, PhD, of Analysis Group, for editing assistance. Dr Bozas, Mss Chertavian and Pham, and Mr Kaminsky have no additional financial conflict of interests to disclose.

Supplementary material: See accompanying pages.

REFERENCES

1. Ustün TB, Ayuso-Mateos JL, Chatterji S, et al. Global burden of depressive disorders in the year 2000. Br J Psychiatry. 2004;184(5):386-392. PubMed doi:10.1192/bjp.184.5.386

2. Bruffaerts R, Vilagut G, Demyttenaere K, et al. Role of common mental and physical disorders in partial disability around the world. Br J Psychiatry. 2012;200(6):454-461. PubMed doi:10.1192/bjp.bp.111.097519

3. World Health Organization. The Global Burden of Disease: 2004 Update. Geneva, Switzerland: WHO Press, World Health Organization; 2008:160. http://www.who.int/healthinfo/global_burden_disease/2004_report_update/en/. Accessed February 2, 2014.

4. Merikangas KR, Ames M, Cui L, et al. The impact of comorbidity of mental and physical conditions on role disability in the US adult household population. Arch Gen Psychiatry. 2007;64(10):1180-1188. PubMed doi:10.1001/archpsyc.64.10.1180

5. Greenberg PE, Kessler RC, Birnbaum HG, et al. The economic burden of depression in the United States: how did it change between 1990 and 2000? J Clin Psychiatry. 2003;64(12):1465-1475. PubMed doi:10.4088/JCP.v64n1211

6. Marcus SC, Olfson M. National trends in the treatment for depression from 1998 to 2007. Arch Gen Psychiatry. 2010;67(12):1265-1273. PubMed doi:10.1001/archgenpsychiatry.2010.151

7. Stewart WF, Ricci JA, Chee E, et al. Cost of lost productive work time among US workers with depression. JAMA. 2003;289(23):3135-3144. PubMed doi:10.1001/jama.289.23.3135

8. Birnbaum HG, Kessler RC, Kelley D, et al. Employer burden of mild, moderate, and severe major depressive disorder: mental health services utilization and costs, and work performance. Depress Anxiety. 2010;27(1):78-89. PubMed doi:10.1002/da.20580

9. Mrazek DA, Hornberger JC, Altar CA, et al. A review of the clinical, economic, and societal burden of treatment-resistant depression: 1996-2013. Psychiatr Serv. 2014;65(8):977-987. PubMed doi:10.1176/appi.ps.201300059

10. Gadermann AM, Alonso J, Vilagut G, et al. Comorbidity and disease burden in the National Comorbidity Survey Replication (NCS-R). Depress Anxiety. 2012;29(9):797-806. PubMed doi:10.1002/da.21924

11. Kessler RC. The costs of depression. Psychiatr Clin North Am. 2012;35(1):1-14. PubMed doi:10.1016/j.psc.2011.11.005

12. Greenberg PE, Stiglin LE, Finkelstein SN, et al. The economic burden of depression in 1990. J Clin Psychiatry. 1993;54(11):405-418. PubMed

13. United States Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. Office of Applied Studies. National Survey on Drug Use and Health, 2005. Ann Arbor, MI: Inter-university Consortium for Political and Social Research (ICPSR) [distributor]; 2005. http://www.icpsr.umich.edu/icpsrweb/SAMHDA/studies/4596/version/4 . Accessed December 12, 2014.

14. United States Department of Health and Human Services. Substance Abuse and Mental Health Services Administration. Center for Behavioral Health Statistics and Quality, United States Department of Health and Human Services. Substance Abuse and Mental Health Services Administration. Center for Behavioral Health Statistics and Quality. National Survey on Drug Use and Health, 2010. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2011:1-869. http://www.icpsr.umich.edu/icpsrweb/SAMHDA/studies/32722/version/. Accessed December 12, 2014.

15. Kessler RC, Berglund P, Chiu WT, et al. The US National Comorbidity Survey Replication (NCS-R): design and field procedures. Int J Methods Psychiatr Res. 2004;13(2):69-92. PubMed doi:10.1002/mpr.167

16. Kessler RC, Merikangas KR. The National Comorbidity Survey Replication (NCS-R): background and aims. Int J Methods Psychiatr Res. 2004;13(2):60-68. PubMed doi:10.1002/mpr.166

17. American Psychological Association, American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. Arlington, VA: American Psychiatric Publishing; 1994.

18. Hedden S, Gfroerer J, Barker P, et al. Comparison of NSDUH mental health data and methods with other data sources. Rockville, MD: Center for Behavioral Health Statistics and Quality (formerly the Office of Applied Studies), Substance Abuse and Mental Health Services Administration (SAMHSA); 2012:1-16. www.optum.com/content/dam/optum/resources/productSheets/
5302_Data_Assets_Chart_Sheet_ISPOR.pdf. Accessed December 12, 2014.

19. Optum Inc. US Core Data Assets. 2014. http://www.optum.com/content/dam/optum/resources/productSheets/
5302_Data_Assets_Chart_Sheet_ISPOR.pdf. Accessed December 12, 2014.

20. Optum Inc. Life Sciences Data Assets. 2014. https://www.optum.com/life-sciences/develop-evidence/data-assets.html Accessed December 12, 2014.

21. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130-1139. PubMed doi:10.1097/01.mlr.0000182534.19832.83

22. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. PubMed doi:10.1016/0021-9681(87)90171-8

23. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8, pt 2):757-763. PubMed doi:10.7326/0003-4819-127-8_Part_2-199710151-00064

24. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70(1):41-55. doi:10.1093/biomet/70.1.41

25. Elixhauser A, Steiner C, Harris DR, et al. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. PubMed doi:10.1097/00005650-199801000-00004

26. Ivanova JI, Birnbaum HG, Chen L, et al. Cost of post-traumatic stress disorder vs major depressive disorder among patients covered by medicaid or private insurance. Am J Manag Care. 2011;17(8):e314-e323. PubMed

27. Centers for Disease Control and Prevention. National Center for Injury Prevention and Control. Web-based Injury Statistics Query and Reporting System (WISQARS). 2014. http://www.cdc.gov/injury/wisqars/index.html. Accessed December 12, 2014.

28. Blumenthal SJ, Kupfer DJ. Suicide Over the Life Cycle: Risk Factors, Assessment, and Treatment of Suicidal Patients. Washington, DC: American Psychiatric Pub; 1990:135.

29. Nock MK, Kessler RC. Prevalence of and risk factors for suicide attempts versus suicide gestures: analysis of the National Comorbidity Survey. J Abnorm Psychol. 2006;115(3):616-623. PubMed doi:10.1037/0021-843X.115.3.616

30. Karch DL, Barker L, Strine TW. Race/ethnicity, substance abuse, and mental illness among suicide victims in 13 US states: 2004 data from the National Violent Death Reporting System. Inj Prev. 2006;12(suppl 2):ii22-ii27. PubMed doi:10.1136/ip.2006.013557

31. Kung H-C, Hoyert DL, Xu J, et al. Deaths: final data for 2005. Natl Vital Stat Rep. 2008;56(10):1-120. PubMed

32. Murphy SL, Xu J, Kochanek KD, et al. National Vital Statistics Reports Deaths: Preliminary Data for 2010. 2012;60(4). http://www.cdc.gov/nchs/data/nvsr/nvsr60/nvsr60_04.pdf. Updated January 11, 2012. Accessed December 12, 2014.

33. US Department of Labor, US Bureau of Labor Statistics. Highlights of Women’s Earnings in 2010. Washington, DC: US Department of Labor; 2011:91. http://www.bls.gov/cps/cpswom2010.pdf. Accessed December 12, 2014.

34. US Department of Labor, US Bureau of Labor Statistics. Highlights of Women’s Earnings in 2005. Washington, DC: US Department of Labor; 2006:40. http://www.bls.gov/cps/cpswom2005.pdf. Accessed December 12, 2014.

35. Max W, Rice DP, Sung H-Y, et al; UC San Francisco. Valuing human life : estimating the present value of lifetime earnings, 2000. Cent Tob Control Res Educ. 2004. http://escholarship.org/uc/item/82d0550k. Accessed December 12, 2014.

36. Ventimiglia J, Kalali AH. Generic penetration in the retail antidepressant market. Psychiatry (Edgmont). 2010;7(6):9-11. PubMed

37. Demyttenaere K, Bonnewyn A, Bruffaerts R, et al. Comorbid painful physical symptoms and depression: prevalence, work loss, and help seeking. J Affect Disord. 2006;92(2-3):185-193. PubMed doi:10.1016/j.jad.2006.01.007

38. Trivedi MH. The link between depression and physical symptoms. Prim Care Companion J Clin Psychiatry. 2004;6(suppl 1):12-16. PubMed

39. Druss BG, Walker ER. Mental disorders and medical comorbidity. Synth Proj Res Synth Rep. 2011;(21):1-26. PubMed

40. Katon WJ, Russo JE, Von Korff M, et al. Long-term effects on medical costs of improving depression outcomes in patients with depression and diabetes. Diabetes Care. 2008;31(6):1155-1159. PubMed doi:10.2337/dc08-0032

41. Unutzer J, Katon WJ, Fan M-Y, et al. Long-term cost effects of collaborative care for late-life depression. Am J Manag Care. 2008;14(2):95-100. PubMed

42. Starr M, Dominiak L, Aizcorbe A. Decomposing growth in spending finds annual cost of treatment contributed most to spending growth, 1980-2006. Health Aff (Millwood). 2014;33(5):823-831. PubMed doi:10.1377/hlthaff.2013.0656

43. Jefferis BJ, Nazareth I, Marston L, et al. Associations between unemployment and major depressive disorder: evidence from an international, prospective study (the predict cohort). Soc Sci Med. 2011;73(11):1627-1634. PubMed doi:10.1016/j.socscimed.2011.09.029