Objective: To evaluate the relationship between antidepressant medication adherence and short-term disability (STD) in an employed population.
Study Design: Retrospective observational cohort study of 2112 employees with a new episode of treatment with an antidepressant medication (selective serotonin reuptake inhibitors or serotonin norepinephrine reuptake inhibitors).
Methods: Both Health Plan Employer Data and Information Set (HEDIS) acute and continuation treatment guidelines were applied to categorize patients' medication adherence. STD events were followed for 365 days after the date that an initial antidepressant medication prescription was filled. The association between STD and adherence was analyzed with multiple logistic regression models, adjusting for demographic and other confounding factors.
Results: A total of 1301 employees (61.6% of 2112) adhered to acute phase treatment, and 966 (45.7% of 2112) remained adherent to continuation phase treatment. After adjusting for sociodemographic factors, employees nonadherent with acute treatment were 38.7% more likely to have STD claims than adherent employees (odds ratio [OR] = 1.387; 95% confidence interval [CI] = 1.025, 1.876; = .0339); and employees nonadherent with continuation treatment were 46.1% more likely to have STD claims than adherent employees (OR = 1.461; 95% CI = 1.071, 1.993; = .0167).
Conclusions: A higher incidence of STD was associated with antidepressant medication nonadherence in both acute and continuation treatment phases. Employers may save indirect costs by providing assistance to encourage employees to adhere to their antidepressant medication treatment.
(Am J Manag Care. 2007;13:105-112)
Depressive disorders are among the most common of psychological disorders, affecting around 35 million adults in the United States each year, with a lifetime prevalence of 16.2%.1 The National Comorbidity Survey found that 59% of the US adults with lifetime prevalence of major depressive disorder were unable to work on average 35 days in the past year.1 The associated economic burden of depressive disorders is estimated at $81.5 billion annually, with much of that total stemming from losses in worker productivity and high relapse rates.2 Other researchers estimated that $32 billion in lost productive work time (ie, time absent from work and time at work unable to perform the job) is attributable to depression.3
The work-related impacts of depression are significant, and employers need to be aware of the potential costs of the disease.3-9 Lerner and colleagues found that depressed employees had significantly more absenteeism and presenteeism compared with employees who had rheumatoid arthritis.10 Wang et al found that depression was significantly associated with decrements in both task focus and productivity equal to about 2.3 days absent per month of being depressed.11
Measuring adherence to medication regimens has value in that nonadherence may not only prolong the illness episode of the patient but also contribute to the overall cost impact of depressive disorders in the workplace. Thompson et al investigated the costs of depression care by pattern of antidepressant use and found the highest total medical costs for the patients who switched or augmented their medication, or discontinued their treatment early.12 Those who adhered to their treatment for 3 months had the lowest total medical costs. Similarly, Revicki et al reported that patients receiving the minimum recommended levels of antidepressant therapy for 3 months showed lower total healthcare costs compared with patients receiving less treatment.13 Furthermore, when patients discontinued their medication early, they were more likely to experience a depression recurrence than patients who continued their medication for 3 months.14
The issue of depression management and workplace productivity and absenteeism was addressed by Rost et al.15 Primary care physician practices were assigned to provide usual or enhanced depression management. Their patients were followed for 2 years. The patients receiving enhanced care reported 6.1% greater productivity and 22.8% less absenteeism over 2 years compared with patients receiving usual care. Absenteeism and productivity savings among consistently employed individuals were $2500 per employee per year.15 Results obtained by Claxton et al show that absenteeism drops after treatment for depression.16
The previous studies demonstrate that when patients adhere to recommended treatment guidelines, they have a lower risk of depression recurrence and lower total medical costs. The current study examines short-term disability (STD) absences associated with employees who do and do not adhere to depression medication treatment guidelines from the Health Plan Employer Data and Information Set (HEDIS), published annually by the National Committee for Quality Assurance.17 The purpose of this study was to determine the relationship between adherence to depression treatment and mental health-related STD events.
This retrospective, observational study design had a 6-month preindex period, an identification period (from June 29, 2000, through January 1, 2004), and a 1-year follow-up period. A unique index date was determined for each subject, which was the first antidepressant medication fill date during that identification time period. The preindex period also was unique to each subject and was the 180 days before the index date. Similarly, the follow-up period was the 365 days after the index date. The earliest possible preindex period began on January 1, 2000, and the latest possible follow-up period ended on December 31, 2004.
The study population was from a large financial services corporation headquartered in the Midwest with employees in more than 25 states. At the start of the study, the company employed approximately 100 000 people. However, 72 000 people were continuously employed from 2000 to 2004; about 70% of the population was female and the average employee age was 38 years. The majority (87%) of employees reported Caucasian ethnicity. Approximately one quarter of employees had prescription medication coverage from a single pharmacy benefit management company. This research was approved by the University of Michigan's institutional review board.
These de-identified files from 2000 to 2004 were combined to compile the study dataset: company personnel files, pharmacy claims records, and STD claims. The STD absences were managed internally by a dedicated professional staff under the direction of the firm's corporate medical director.18
The first step in identifying the study population was to select the cohort of employees with a new episode of treatment with an antidepressant medication (selective serotonin reuptake inhibitors or serotonin norepinephrine reuptake inhibitors) during the identification period. A new treatment was defined as being free of any antidepressants during the 180 days of the preindex period. The antidepressant medications dispensed by the pharmacy benefit plan included citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline, venlafaxine, bupropion, mirtazapine, nefazodone, and trazodone.
Of the 182 292 employees aged 18 years and over who were employed for any continuous 1-year period from 2000 to 2004, 51 700 (28.4%) participated in the company's pharmacy benefit plan in any 1 year during the study period. Of these employees, 6568 employees (12.7% of pharmacy subscribers) filed a claim for 1 of the study medications sometime during 2000-2004. A total of 4456 employees were eliminated due to the presence of 1 or more of the following exclusion criteria: use of antipsychotic medication in the preindex or follow-up period (n = 298), use of other antidepressants in the preindex period (n = 371), having a preindex period less than 180 days (n = 2120), having a follow-up period less than 365 days (n = 1320), or employment termination during follow-up (n = 347). The majority of those excluded were for reasons connected to insufficient tenure. The final study population of 2112 employees was 4.1% of pharmacy benefit plan enrollees and 32.2% of the 6568 medication claimants.
Treatment Adherence Criteria
Effective depression treatment includes both an acute and a continuation phase of treatment. The sum of the days' supply of all prescriptions in a time period was calculated for each patient. We followed the HEDIS treatment adherence criteria, which are characterized as follows19: The acute treatment phase lasts 114 days, during which an employee needs to fill a sufficient number of antidepressant prescriptions to provide medication for at least 84 days. Medication gaps due to washout when changing medication or refill gaps can total a maximum of 30 days during the 114-day period. The continuation phase of treatment lasts 231 days, during which an employee needs to fill a sufficient number of antidepressant prescriptions to provide medication for at least 180 days. Again, medication gaps due to washout or refills can total a maximum of 51 days during the 231-day period. Employees who were nonadherent during the 3-month acute phase were automatically considered nonadherent during the 6-month continuation phase. According to these adherence guidelines, employees were classified as either adherent or nonadherent for both the acute phase and continuation phase of treatment.
Short-term Disability Absences
International Classification of Diseases, Ninth Revision
The primary outcome measure in this study was the incidence and duration of STD absence days. By examining the primary and secondary diagnostic codes, each STD episode was further identified as representing any depression disorder ( codes 296.2x, 296.3x, 300.4x, or 311.xx) and/or anxiety disorder (codes 293.89, 300.00, 300.01, 300.02, 300.21, 300.22, 300.23, 300.3x, 308.3x, or 309.81). The number of absence days in the preindex and follow-up time periods was summed separately for each employee.
Group differences in adherent and nonadherent patients regarding sociodemographics, clinical characteristics, and STD claims were examined by using ?2 tests and tests. Multiple logistic regression models were used to study the association with adherence while controlling for baseline characteristics. Another set of multiple logistic regressions modeled the odds of incurring any STD claims in the follow-up period. The full model contained sociodemographic factors (age, sex, employment exempt status, and ethnicity); the presence of any STD claims in the preindex period; the presence of comorbidities based on the Chronic Disease Score20; the number of unique prescription fills during the preindex period; and the indicator variable for adherence. The Chronic Disease Score is a summary measure of the comorbid conditions among participants assessed by their prescription medication combination during the preindex period. We also tested potential interactions between age and sex with other factors, but only main effects were reported in final models. Odds ratios were estimated for all variables in the models. To examine a trend analysis of 3 levels of adherence (none, acute phase only, and continuation phase), multiple logistic regression was used, controlling for baseline characteristics. All analyses were conducted using SAS 9.0 software (SAS Institute Inc, Cary, NC).
Adherence in the Acute Phase of Treatment
Table 1 summarizes baseline characteristics of the study population by adherence in the acute treatment phase. A total of 2112 subjects met study eligibility criteria. Of this group, 1301 (61.6%) were adherent during the acute treatment phase. All subjects (N = 2112) were compared with those who were nonadherent (n = 811) and those who were adherent (n = 1301) to the treatment guidelines in the acute phase. Multiple logistic regression analysis determined whether nonadherent and adherent employees were different from each other. Adherent employees were significantly more likely to be older (< .0001), female (= .0065), and Caucasian (< .0001). A greater percentage of adherent employees had exempt status compared with nonadherent employees (< .0115). There were no significant differences in the Chronic Disease Score and the number of unique prescription fills between the adherent and nonadherent employees.
These employee groups were compared with respect to their STD absences. Table 2 shows that there were no significant differences in STD absences in the preindex period between the adherent and nonadherent employees.
In the follow-up period, a significantly smaller percentage of adherent employees (8.8%) had any STD episode compared with the nonadherent employees (12.7%). Very small percentages of adherent (1.5%) and nonadherent (2.8%) employees had a depression-or anxiety-related STD event in the follow-up time period. In the multiple logistic regression model, this was not a significant difference (= .5044). More nonadherent employees than adherent employees had multiple STD episodes (2.3% vs 1.1%; = .0377). The duration of STD events during the follow-up period was not significantly different for adherent and nonadherent employees. The average STD duration was calculated for all employees in the group and also solely for STD claimants; in both cases, nonadherent and adherent employees were similar.
Adherence in the Continuation Phase of Treatment
The demographics of employees classified as adherent and nonadherent during the continuation phase of treatment are presented in Table 3. Similar to the acute treatment phase, employees adherent during the continuation treatment phase were more likely than nonadherent employees to be female, older, and Caucasian, and to have exempt status.
As can be seen in Table 4, adherent employees had fewer STD events in the follow-up time period (8.4% vs 12.0%; = .0167). Similarly, adherent employees had fewer depression/anxiety STD events in the follow-up time period (1.3% vs 2.5%), although the difference was not significant in the multiple logistic regression model (= .4997). In the continuation phase, more nonadherent employees than adherent employees had multiple episodes (2.1% vs 0.9%; = .0392). The difference in the duration of STD events between all continuation phase adherent (3.6 days) and nonadherent (5.5 days) employees in the follow-up time period was only marginally significant (= .0646). The duration of STD events for claimants was not significantly different between adherent and nonadherent employees.
Odds of Filing Any STD Claim
The multiple logistic regression analysis was used to study the odds of filing any STD claim in the follow-up time period, adjusted for baseline characteristics in the preindex period. Employees who were nonadherent during the acute phase of treatment were 1.387 times more likely to file any STD claim compared with adherent employees (95% confidence interval [CI] = 1.025, 1.876; = .0339). Employees who were nonadherent during the continuation phase of treatment were 1.461 times more likely to file any STD claim compared with adherent employees (95% CI = 1.071, 1.993; = .0167).
Potential Indirect Savings Related to Treatment Adherence
The continuation phase adherence results were used to calculate the potential savings associated with adherence. The 1146 employees who were nonadherent to the continuation phase treatment guidelines had an STD event incidence rate of 12.0% (n = 137) in the 1-year follow-up time period. This rate is 46.1% higher than that for the adherent employees (based on the 1.461 odds ratio after adjusting for confounding factors). Therefore, the STD event rate for this group of employees could have been 8.2% (12.0% divided by 1.461), which translates to 94 employees with any STD incidence. The potential number of STD cases saved would be 43 (137 employees -94 employees). The average duration of an STD claim is 46 workdays per year, so the potential STD days saved could be 1978 workdays (46 Ã— 43). Assuming a salary and benefits cost of $200 per employee per workday, the total potential indirect savings in this study population (N = 2112) would be $395 600 (1978 days Ã— $200). The average savings for 1 employee taking antidepressant medication is $187 per year ($395 600/2112). About 12% of employees participating in the pharmacy benefit plan fill at least 1 antidepressant medication prescription each year. In a company with 70 000 employees, that translates to 8400 people taking any antidepressant. If all employees adhered to the treatment guidelines, there is a potential savings in STD absences of $1.57 million ($187 Ã— 8400 employees).
This retrospective, observational study investigated whether adherence to HEDIS antidepressant medication treatment guidelines was associated with lower STD absences in an employed population. Of the 2112 study subjects, 61.6% were adherent in the acute phase and 45.7% were adherent for the continuation phase. There were significant demographic differences among adherent and nonadherent employees in both phases of treatment adherence. Sex, age, ethnicity, and exempt status were included in the models to control for these initial differences. These adherence rates are similar to those found in other studies. A large managed care database was analyzed to determine that 43% of patients were adherent to depression medication treatment guidelines.21 A study of a large database found a continuation phase adherence rate of 44%.22
In the 365-day follow-up period, STD absences were compared for adherent and nonadherent employee populations. Employees who were adherent during the acute phase of treatment had significantly fewer STD events and also significantly fewer depression/anxiety-related STD events compared with those who were not adherent. The employees who remained adherent during the continuation phase of treatment also had significantly fewer STD events compared with the employees who were nonadherent.
To further analyze these results, we divided the 1146 employees who did not adhere to the continuation phase of treatment according to whether they were or were not adherent in the acute phase. This gave us 3 groups of employees for comparison: those who did not adhere to the treatment guidelines in either the acute or the continuation phase (n = 811), those who were adherent in the acute phase but not in the continuation phase (n = 335), and those who were adherent in the continuation phase (n = 966).
The Figure shows that 12.7% of the nonadherent employees, 10.1% of the employees adherent only in the acute phase, and 8.4% of the employees adherent in the continuation phase had an STD absence in the follow-up period. Multiple logistic regression analysis of this trend, controlling for demographic factors, found it to be significant (= .0461 for the overall trend). Although there was not a significant difference between those adherent in the acute treatment phase only (n = 335) and those who remained adherent for the continuation phase (n = 966), the overall trend for the likelihood of STD absences to decrease with longer antidepressant medication adherence was significant.
Returning to the standard comparison groups used throughout the study, the multiple logistic regression analysis results demonstrated that employees who were nonadherent in the acute phase were 1.387 times (38.7%) more likely to file an STD claim during the follow-up period compared with the employees who were adherent in the acute phase ( = .0339). Similarly, employees who were nonadherent to the continuation phase guidelines were 1.461 times (46.1%) more likely to file an STD claim during follow-up compared with the adherent employees (= .0167).
Using these odds and examining the employees who remained adherent through the continuation phase versus the employees who did not, the potential savings in STD benefits for the study group would be $395 600. Extrapolating this savings to the potential group of depressed employees in this corporation, the savings could reach $1 573 000 if all employees remained adherent to treatment guidelines.
Langleib and Kahn underscore the fact that many corporations do not yet understand the high productivity cost of mental health issues among their employees.23 They reason that it is crucial to provide quality mental healthcare benefits to help employees and to moderate costs, particularly because it has been shown that those who receive appropriate care for their anxiety or depression have less disability and greater productivity.
This retrospective, observational study design was susceptible to bias in data selection and analysis, and confounding variables may have gone unrecognized. Although these results may show associations among variables, causal relationships cannot be established. Furthermore, as this study was based solely on pharmacy claims, we could not determine the effect on adherence of the severity of the cases, nor could we determine the influence of behavioral, economic, or provider factors.
Although employees were followed for 1 year after receiving their prescription for an antidepressant, longer-term study is warranted. Given that the guidelines established by the Agency for Healthcare Policy and Research and the American Psychiatric Association recommend treatment with an antidepressant for at least 4 to 9 months after symptom resolution,19,24 the current study does not examine the impact of maximized treatment. It may be that significantly better outcomes occur with even longer treatment adherence.
This study demonstrates that employees who remain adherent through the continuation treatment phase of the HEDIS antidepressant medication guidelines are 46% less likely to experience an STD event during the following year compared with employees who are not adherent. In this study of 2112 antidepressant users, about $400 000 in lost STD workdays could have been saved had the employees maintained adherence. Other studies have shown that when patients do not continue their depression medication treatment for at least 3 months, they are more likely to have higher healthcare costs, recurring depression episodes, and more absenteeism from work.12,13,19,25,26
This study has important implications for employers, as well as pharmacy and medical benefit plan design. Clearly, employers that provide encouragement and support for patient adherence to treatment guidelines may limit the burden of STD absences. This encouragement may take the form of lowered copayments, deductibles, or tier status of medications for chronic conditions to decrease economic barriers to medication adherence. Also, reinforcing antidepressant medication compliance via education and awareness would seem to be an essential component that could lead to quantifiable cost justification.27 Decreases in recidivism due to disability and more successful return to work outcomes should more than offset the costs of these educational interventions. In any case, studies such as this one demonstrate to employers the importance of medication compliance by their overall strategy in managing the total burden of cost for medical conditions.
We wish to thank Kristina S. Yu-Isenberg, PhD, RPh, and Michael C. Sokol, MD, of GlaxoSmithKline for their invaluable assistance and contributions to this study and manuscript.
Author Affiliations: From the Department of Environmental and Occupational Sciences, University of Illinois at Chicago, Chicago, Ill (WNB); the University of Michigan Health Management Research Center, Ann Arbor, Mich (CYC, ABS, DWE); and DePaul University, Chicago, Ill (DJC).
This study was funded by GlaxoSmithKline.
Dee W. Edington, PhD, Health Management Research Center, University of Michigan, 1027 E. Huron St, Ann Arbor, MI 48104-1688. E-mail: email@example.com.