Direct and Indirect Costs Among Caregivers of Patients With Major Depressive Disorder and Suicidal Ideation or Suicidal Attempt

Prim Care Companion CNS Disord 2021;23(4):20m02893

To cite: Kuvadia H, Beren IA, Starr HL, et al. Direct and indirect costs among caregivers of patients with major depressive disorder and suicidal ideation or suicidal attempt. Prim Care Companion CNS Disord. 2021;23(4):20m02893.
To share: https://doi.org/10.4088/PCC.20m02893

© Copyright 2021 Physicians Postgraduate Press, Inc.

^aJanssen Scientific Affairs, LLC, Titusville, New Jersey
^bIntegrated Data Analytics, Workpartners LLC, Cheyenne, Wyoming
^cRetrospective Research, Better Health Worldwide, Inc, Newfoundland, New Jersey
*Corresponding author: Richard Brook, MS, MBA, Retrospective Research, Better Health Worldwide, Inc, 18 Hirth Drive, Newfoundland, NJ 07435 (Rich@BH-WW.com).

Major depressive disorder (MDD) is a chronic psychiatric disease that affects approximately 7.1% of US adults (17.3 million individuals), with 80% experiencing difficulties with work, home, or social activities as a result of the condition. Symptoms of MDD may include suicidal ideation (SI) and suicide attempt (SA),^1,2 and MDD is the psychiatric diagnosis most often associated with suicide.³ Furthermore, more than 50% of those who attempt suicide have been reported to have MDD.^4,5 One prospective study⁶ of patients with MDD found that the incidence rate of SA was 21-fold higher during a major depressive episode versus when patients were in remission.

The economic burden of completed suicides and SAs is substantial. The national US cost of officially reported suicides and SA alone was estimated at $58.4 billion in 2013; adjustment for the underreporting of suicide events resulted in estimates that were nearly two-thirds higher ($93.5 billion).⁷ Of that $58.4 billion, 97.1% ($56.7 billion) is attributable to indirect costs related to productivity losses from premature death or injury.⁷

Research on the impact of MDD and SI/SA on patients is documented in the literature, but the impact on family members, who often serve as caregivers to patients, has not been well documented. Some studies^8,9 report caregiver stress or other conditions, including bereavement, resulting in caregivers themselves developing MDD. Self-reported data were used in several studies^9–12 to assess productivity and the quality of life of caregivers of patients with mental health conditions. Although SI/SA is prevalent—and has devastating consequences—among working-age adults, no information is available regarding the impact on caregivers of patients with MDD and SI/SA, including their health benefit costs and absenteeism. However, research has documented the impact on caregivers of spouses with late-life depression¹³ and epilepsy¹⁴ and caregivers of dependent children with pediatric asthma¹⁵ by using objective claims data, work absence data, and human resources data. The health care burden of caregivers is a potential liability for their employers, and, therefore, analysis of health care and absenteeism costs from the employer perspective is warranted.

This study documents the impact on caregivers of patients with MDD and SI/SA (hereafter, MDSI) or MDD alone in an employed population, focusing on caregiver health benefit costs and costs associated with caregivers’ work absences, using objectively measured data relevant to the payer perspective.

METHODS

Overview

This was a retrospective database study of US employees and eligible dependents with demographic and job-related information and data on employee absence payments and leaves by benefit type. Employees and their spouse-patients were identified for each cohort using adjudicated medical and prescription claims, and continuous eligibility was confirmed. A 20:1 matching procedure was used to increase sample size and lower standard errors, and 2-part statistical regression models were used for each outcome to address the non-normal distributions of the study outcomes. Because the data were de-identified and the study did not affect patient care, the study was exempt from institutional review board review and approval.

Data Source

Data for the study were extracted from the Workpartners Research Reference Database (RRDb) (formerly, Human Capital Management Systems or HCMS RRDb) during the period January 1, 2010, to July 31, 2019. The Workpartners database represents multiple geographically diverse, commercially insured US employers from all 50 states. Employers in the RRDb represent the retail, service, manufacturing, transportation, energy, technology, financial, and utility industries, and the database includes information on nearly 4 million employees, spouses, and eligible dependents. The database includes information on all individuals’ sex, age, and health plan claims (medical and prescription) from multiple insurers across the United States. The associated data are transactional in nature including information on individual medical services and prescription fills. The RRDb also contains employee self-reported race, job-related information (such as full- or part-time status, tenure [years with employer], job type, exempt or nonexempt status, and salary), region (work and home location), work absence cost based on actual payments, and absence time (based on days missed) for 4 benefit types: short-term disability, long-term disability, workers’ compensation, and sick leave. The database has been used in prior psychiatric^15–17 and caregiver research.^14,18

Data were de-identified to comply with the Health Insurance Portability and Accountability Act and the contractual obligations between the RRDb and their employer contributors, thus ensuring patient care was unaffected.

Sample Selection

Patients were identified by any primary, secondary, or tertiary claims containing International Classification of Diseases, 9th or 10th revision, Clinical Modification (ICD-9/ICD-10) codes for MDD alone or MDSI (Supplementary Table 1). ICD-9 codes were used for data prior to October 2014, when ICD-10 codes were implemented. Caregiver cohorts included in the current analysis included caregivers whose spouses or domestic partners have MDSI, MDD alone, or no claims for MDD or SI/SA (controls).

Patients and married or domestic partner caregivers were included in the extraction cohorts. Patients and caregivers in the same pair had the same index date. The MDSI cohort included patients identified with claims for self-harm, SI, or SA (within the pool of MDSI patients), and their index date is the date of the initial self-harm, SI, or SA. The MDD-alone cohort included patients with at least 1 MDD diagnosis (ICD-9 = 296.3x or ICD-10 = F33.xx) and no SI/SA diagnostic codes, with their index dates being the dates of their first MDD claims. The control cohort included patients and caregivers with no claims for MDD or SI/SA, and their index dates were matched to the MDSI cohort (Supplementary Figure 1).

A 6-month preindex baseline period and a 6-month postindex study period starting at the index date were required, as shown in Figure 1. All caregiver participants also had to meet the following criteria: caregivers were employed at some time from January 2010 to July 31, 2019; patients and their caregivers were required to be aged > 18 years at the beginning of the 6 months prior to the index date; and all pairs had to have continuous eligibility for their combined 12-month analysis period. Caregivers with MDD or SI/SA in the baseline or study period were excluded, and patients with certain specific psychiatric conditions were excluded (Supplementary Tables 2–5).

Matching

For further statistical validation, 20 pairs of patients and caregivers in the MDD-alone and control cohorts were propensity score matched to the MDSI cohort based on caregiver sex, age, region, tenure, and index year. The 3 matched cohorts were used in all comparisons throughout the study.

Descriptive Data

Descriptive information for all study participants was summarized over the baseline period and compared between cohorts. Means and standard errors of those descriptive variables were compared using t tests for continuous variables and χ² tests for discrete variables.

Study Outcomes

Study outcomes were evaluated for patients and caregivers. Patient outcomes included direct medical and prescription costs. Caregiver outcomes were components of health benefit costs, which included direct costs (eg, medical and prescription costs) as well as indirect costs that resulted from the caregiver’s workplace absences over the 6-month study (postindex) period. Indirect costs that resulted from absences were calculated based on payments for sick leave, short-term disability, long-term disability, and workers’ compensation. The workers’ compensation costs also included medical claims paid under the workers’ compensation benefit.

Statistical Methodology

Each of the outcomes was modeled separately by using 2-stage stepwise regression models (to account for non-normally distributed data and zero value outcomes) that controlled for age, tenure, marital status, caregiver race, caregiver job-related components (ie, exempt status, full- or part-time status, and salary), location, and caregiver Charlson Comorbidity Index (CCI) score.¹⁹ The first stage used logistic regression to model the likelihood of an outcome greater than zero—for example, those with disability claims versus those without disability claims. The second stage used generalized linear models on the portion of the population with a greater-than-zero outcome. The results of the generalized linear models were then combined with the results of the logistic models to reach an expected value of cost or days for all employees.

All costs were inflation adjusted to September 2019 US dollars by using various components of the Consumer Price Index (CPI).²⁰ Medical costs used the medical cost CPI, prescription costs used the prescription CPI, and all other components used the general CPI.

RESULTS

The identification and matching process identified 570 MDSI patient (mean age: 44.1 years, 67.5% female) and caregiver (mean age: 44.6 years, 32.3% female) pairs along with 11,400 matched MDD-alone pairs and 11,400 matched non-MDD (control) cohort pairs. Most of the differences in baseline demographics, job-related variables, index year, and region between study cohorts were not statistically significant (Table 1). However, there were some significant (P < .05) differences between cohorts. MDD-alone caregivers had a higher mean salary ($86,548) than both MDSI caregivers ($75,536) and control caregivers ($80,006), and more MDSI caregivers worked full-time (89.6%) than did MDD-alone caregivers (86.6%). Control caregivers were most often exempt (salaried) employees (47.3%) compared with MDD-alone caregivers (43.0%) and MDSI caregivers (35.4%), and control caregivers also had longer mean tenure (10.1 years with the employer) than MDD-alone caregivers (9.7 years). In the preindex (baseline) period (Table 1), patients in the MDSI and MDD-alone cohorts had higher mean CCI scores (0.46 and 0.39, respectively) than did patients in the control cohort (0.24). MDD-alone caregivers had a higher mean CCI score than did control caregivers (0.33 versus 0.26, respectively). Mean CCI scores for the MDSI caregivers and control caregivers were both 0.26.

Direct Health Benefit Costs

Caregiver total direct costs (Figure 2) in the MDSI and the MDD-alone cohorts ($6,983 and $5,973, respectively) were significantly (P < .0001) higher than among control caregivers ($4,890). The MDSI total direct costs were also significantly greater than those of MDD-alone caregivers (P < .05). Caregiver direct medical costs in the MDSI cohort ($5,131, SE: $294) were significantly (P < .001) higher than among control caregivers ($3,885, SE: $50) and nearly significantly higher than among MDD-alone caregivers ($4,548, SE: $58, P = .0513). MDD-alone caregiver costs were significantly higher (P < .001) than for controls. Caregiver prescription costs were significantly (P < .001) different for all cohorts and highest in the MDSI cohort ($1,852, SE: $127), next highest in the MDD-alone cohort ($1,425, SE: $22), and lowest in the control cohort ($1,005, SE: $16).

Although this study focused on caregivers, patients’ total direct costs (Figure 3) in the MDSI and MDD-alone cohorts ($25,365 and $9,034, respectively) were significantly (P < .0001) higher than those of control patients ($4,016). For patients with MDSI, total direct costs were also significantly greater than those of patients with MDD alone (P < .01). All of the medical costs were significantly different between cohorts (P < .0001), and mean costs were greatest among patients with MDSI ($23,510, SE: $1,458), next among patients with MDD alone ($7,490, SE: $104), and lowest among controls ($3,335, SE: $48). Prescription costs among patients with MDSI ($1,856, SE: $121) and those with MDD alone ($1,544, SE: $23) were significantly higher (P < .0001) than among controls ($682, SE $11), and significantly different from each other (P = .0116).

Indirect Costs

In total, the indirect costs (Figure 4) were highest for MDSI caregivers ($611), followed by those of caregivers of patients with MDD alone ($480) and controls ($427). Sick leave costs were higher among MDSI caregivers ($253, SE: $37) and MDD-alone caregivers ($218, SE: $8) than among control caregivers ($118, SE: $5). Short-term disability costs were highest among MDSI caregivers ($309, SE: $80), next highest among control caregivers ($133, SE: $11), and lowest among MDD-alone caregivers ($101, SE: $9). Long-term disability costs were higher among control caregivers ($73, SE: $34) than among MDSI caregivers ($1, SE: $1) and MDD-alone caregivers ($3, SE: $2). Workers’ compensation costs were significantly higher among MDD-alone caregivers ($157, SE: $16) than among MDSI caregivers ($48, SE: $27) and control caregivers ($103, SE: $11).

DISCUSSION

This is the first attempt, to our knowledge, to analyze and quantify caregivers’ economic burden of MDD in patients with and without SI/SA with regard to direct and indirect health care costs by using objectively measured data. Although the study does not establish causation, it demonstrates an association between patients’ conditions—in this case, MDSI—and their caregivers’ direct and indirect costs. It builds on prior work,¹³ showing the relationship between MDD caregiver status and the emotional health and quality of life burden. Uniquely, this current research adopts the employer perspective and details potential employer liabilities.

Family members who provide unpaid care have been described as a hidden patient group.²¹ The international literature²² suggests that caregiving similarly affects caregivers irrespective of geographic location or of the illness the care recipient is experiencing. A UK study²¹ reported that caregivers caring for family members with a range of illnesses experience worry, frustration, anger, and guilt and need professional support for family members’ conditions that include mental health problems²³ and schizophrenia.²⁴

For 6 of the 8 study outcomes, costs were highest for the MDSI cohort, followed by the MDD-alone group and controls, including direct medical and prescription costs for patients and caregivers, plus caregiver indirect costs for leaves due to sick leave and short-term disability. With regard to the other indirect components (indirect costs for leaves due to long-term disability and workers’ compensation), MDSI caregivers did not have the highest costs. Claims for those other components were rare in all 3 cohorts, with only 0.13% of the study population filing long-term disability claims, and 1.38% filing workers’ compensation claims.

Comparison With Other Relevant Caregiver Studies

Literature searches on caregiver studies identified few comparisons. Identified studies focused on caregivers of persons with late-life depression¹³ and of children with asthma,¹⁸ who may be cared for differently than a spouse. The research on caregivers of patients with epilepsy did not include a nonepilepsy control group, and it stratified caregivers based on whether patients were managed by monotherapy or adjunctive therapy with antiepileptic drugs.¹⁴ Other identified research used subjective data to assess the impact on caregivers.^11,24,25

The study findings likely underestimate the impact of MDD and SI/SA on caregiver direct and indirect costs. The continuous eligibility requirement for 6 months after the index date removes the impact of completed suicides and the impact of caregivers who leave the workforce. The exclusion of caregivers with diagnoses of MDD or SI/SA likely also resulted in an underestimation of caregiver direct and indirect costs. It is recognized that there is a higher rate of MDD among caregivers of patients with mental health disorders. Because family medical leaves and bereavement benefits vary from employer to employer or are reported inconsistently, they were excluded. Despite younger age being a risk factor for SA,⁶ the design excluded dependents and focused on spouses, as it was considered likely that care of a spouse differed from care of a dependent.

The strengths of this study include the breadth of the Workpartners RRDb, which contains many objectively measured components not subject to recall bias that are not available in other databases, such as salary, other job-related variables, work absences, and absence payments. The employers in the database represent various industries comprising commercial companies that provide benefits for their employees. In addition, the Workpartners RRDb differentiated the reasons for the caregiver absences. The study used matched cohorts designed to prevent the confounding of caregiver sex, age, region, tenure, and index year with the results. Plus, the 2-part regression methodology dealt with many of the challenges inherent in economic data—such as nonnormal distributions and large numbers of zeros—while controlling for many of the demographic, job-related, regional, and comorbidity factors. The regression methodology adjusted for differences between the cohorts, such as the higher mean salaries in the MDD and control cohorts compared with the MDSI cohort. The study also used baseline and study periods of the same 6-month duration, which may reduce the differences in identification of a comorbidity that manifests only rarely.

The current study has several limitations. Potential limitations of the administrative data include risk of clerical inaccuracies, recording bias secondary to financial incentives, temporal changes in billing codes, and lack of clinically relevant variables. The definition of SA was based on diagnosis codes and external cause-of-morbidity codes. Because the focus of the research was on caregivers, patient analysis was limited to overall direct medical and prescription cost components, not specific therapies used. The analysis was restricted to variables present in this database, and, therefore, other factors that were not measured may have confounded the observed relationships. The database did not contain information about the existence of other caregivers or the specific roles that employees played in the care of their spouses or domestic partners. Although the study controlled for job-related information (exempt status, full- or part-time status, and salary), it is unknown whether a caregiver’s employment was structured to provide the flexibility to care for a loved one. In addition, although patients were identified based on initial claims for entry into the cohorts, it is unknown whether they were previously identified with MDD or had prior SAs.

The administrative claims data were derived from employees with commercial health insurance and may not be generalizable to patients who do not have health insurance from their employers. Although the direct and indirect health care costs are higher in the MDSI and MDD-alone cohorts, it is plausible that such factors as industry sector, job level, and employer benefit coverage may influence caregiver health care usage to varying degrees. These questions could be explored in future studies.

Future research should also explore the impact of new therapies that might offer better management of MDD and SI/SA and potentially limit the effect of those conditions on both patients and caregivers.

CONCLUSIONS

Caregivers of patients with MDSI had significantly greater direct and indirect costs compared with caregivers of the MDD-alone and non-MDD control cohorts. The present analysis found that caregivers of patients with MDSI or MDD alone had approximately 43% and 22% higher direct health care costs, respectively, and used more sick days than did caregivers of patients who served as controls.

The economic impact of MDD and SI/SA on patients was found to be substantial. Quantification of those additional components of caregiver burden clarifies the severity of the impact of the conditions and the necessity for proper care and support of both patients and their caregivers.

Submitted: December 10, 2020; accepted February 19, 2021.
Published online: August 12, 2021.
Author contributions: Dr Kleinman and Mr Brook wrote the initial protocol, Mr Beren conducted the analysis under the direction of Dr Kleinman and Mr Brook. Mr Brook wrote the initial draft of the manuscript. Drs Starr and Dr Sheehan and Mr Kuvadia participated in the conception and design of the study. All authors participated in the interpretation of the results and revisions to and approval of the final version of the manuscript and have reported their contributions to the research.
Potential conflicts of interest (last 36 months): Mr Beren is an employee of Workpartners LLC and received funding through Better Health Worldwide, Inc. Dr Kleinman is a consultant to Workpartners LLC. Mr Brook is owner and president of Better Health Worldwide, Inc and received funding from Janssen Scientific Affairs, LLC. Drs Starr and Sheehan and Mr Kuvadia are employees of Janssen Scientific Affairs, LLC and hold stock in parent company Johnson & Johnson.
Funding/support: This research was supported by Janssen Scientific Affairs, LLC, Titusville, New Jersey.
Role of the sponsor: Janssen Scientific Affairs provided funding for this research. Although staff at Janssen Scientific Affairs reviewed the manuscript, final approval with regard to the decision to submit the manuscript was the sole decision of the authors.
Previous presentation: Poster presented virtually at the American Society of Consultant Pharmacists Annual Meeting, May 29–30, 2020 • Poster presented virtually at the American College of Clinical Pharmacy 2020 Poster Symposium, May 26–27, 2020.
Supplementary material: See accompanying pages.

Clinical Points

• Presence of major depressive disorder (MDD) with suicidal ideation/suicide attempt (MDSI) was associated with higher medical and prescription costs compared with the MDD alone and control groups.
• Caregivers of patients with MDSI or MDD alone had higher health care costs and absenteeism than control caregivers, and caregivers of patients with MDSI had higher costs than those caregivers of patients with MDD alone in most cases.
• Clinicians should recognize that MDD and suicidal ideation/suicide attempt impact not only the patients, but also their caregivers.