Increase in Work Productivity of Depressed Individuals With Improvement in Depressive Symptom Severity

In a comparison of disability-adjusted life years for 100 disorders, the World Health Organization’s Global Burden of Disease Study (1) found that major depressive disorder was the fourth leading cause of disability worldwide, accounting for 4.5% of total disability-adjusted life years in the year 2000. By 2020, major depression is projected to become the second leading cause of disability (2). Altogether, mental disorders accounted for 15.4% of the total disease burden on market economies as a result of lost productive time. Lost productive time refers to the sum of hours per week absent from work for a personal or family health-related reason (“absenteeism”) and the hour-equivalent of reduced performance attributable to health-related reasons on days at work (“presenteeism”). The total cost of lost productive time to U.S. employers has been estimated at $225.8 billion per year, the largest contributor (66%) being reduced work because of personal health conditions (3). Workers with depression had more total health-related lost productive time than those without depression (a mean of 5.6 hours/week compared with an expected 1.5 hours/week) (4). The 2-week prevalence of any depressive disorder in the U.S. workforce has been estimated at 9.4%, and 77.1% of individuals with depression report some lost productivity during the previous 2 weeks (4). Thus, depression in U.S. workers cost employers an estimated $44 billion a year in lost productivity, compared with $31 billion a year for those without depression (4). In addition, one study (5) found that major depression was associated with 8.7 days of absence and 18.2 days of lost productivity per year per person, at a cost of $4,426 per person annually. An 18-month observational study (6) found that depressed working patients had significantly greater work performance deficits than healthy comparison subjects. Even in depressed patients who experienced clinical improvement, job performance remained consistently worse than for nondepressed comparison subjects.

Depressive disorders in the workplace persist over time and have a major effect on workplace performance. Survey respondents with chronic depressive illness are seven times more likely to report decreased effectiveness in the workplace than those without depressive symptoms (7). The negative impact of depression on work is likely due in part to disrupted cognitive functioning (8, 9), including executive decision making. Unresolved depressive symptoms have been hypothesized to have a negative impact on work and psychosocial functioning because of cognitive dysfunction and a reduction in social interaction, which affects quality of life (10).

It is unclear which sociodemographic and clinical factors are associated with pretreatment impairment and treatment-related improvement in occupational functioning. This study was conducted to address the following questions:

The data for these analyses were gathered as part of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study (11, 12) from participants who were employed at study entry.

Method

Study Description

The STAR*D study has been described in detail elsewhere (11, 12). The study was designed to define prospectively which of several treatments are most effective for outpatients with nonpsychotic major depressive disorder who report an unsatisfactory clinical outcome to an initial treatment and, if necessary, subsequent treatments. The diagnosis of major depressive disorder was established clinically and confirmed using a DSM-IV checklist. Eligible and consenting patients were initially treated with the selective serotonin reuptake inhibitor citalopram. Those who had an adequate clinical response could enter a 12-month naturalistic follow-up phase, and those who did not could enter one or more subsequent randomized clinical trials.

Clinical research coordinators at each site worked closely with the participants and clinicians, functioned as study coordinators, and provided a liaison between clinical sites and regional centers. Research outcome assessors, who were blind to treatment assignment, collected symptom ratings in telephone interviews. Additional research outcomes were collected by telephone interview using an automated interactive voice response system (13).

Study Participants

Participants were outpatients 18–75 years of age with a diagnosis of nonpsychotic major depressive disorder who were recruited from 14 regional centers across the United States. Each regional center oversaw implementation of the protocol at two to four clinical sites that provided primary (N=18) or psychiatric specialty (N=23) care in either the public or private sector. To ensure recruitment of a representative sample, only self-declared outpatients seeking medical care (as opposed to symptomatic volunteers recruited through advertisements) were eligible for the study. Exclusion criteria were minimal: bipolar disorder, lifetime psychotic disorder or symptoms, a primary diagnosis of obsessive-compulsive disorder or an eating disorder, the presence of general medical conditions that contraindicated treatment with any of the medications used in the first two levels of the study, substance dependence requiring inpatient detoxification, or a history of nonresponse or intolerance (in the current depressive episode) to an adequate trial of any treatment used in the first two levels of the study. Patients were also excluded if they were breastfeeding, pregnant, or intending to conceive within the 9 months subsequent to study entry. Suicidal patients were eligible unless they required immediate inpatient hospitalization. The protocol was approved and monitored by the institutional review boards at the national coordinating, data coordinating, and regional centers and the relevant clinical sites. The Data Safety Monitoring Board of the National Institute of Mental Health approved and monitored the protocol. All potential risks and adverse events were explained to participants, who provided written informed consent before being assessed for study entry at a clinical site.

Assessment of Sociodemographic and Clinical Characteristics

Clinical research coordinators collected baseline measures, including sociodemographic information and severity of depression as assessed by the 17-item Hamilton Rating Scale for Depression (HAM-D) (14, 15), the 16-item Quick Inventory of Depressive Symptomatology–Clinician-Rated (QIDS-C), and the 16-item QIDS–Self-Report (QIDS-SR) (16–18). At baseline, research outcome assessors administered the HAM-D and the 30-item Inventory of Depressive Symptomatology–Clinician-Rated (17, 19). At study exit, the clinical research coordinators administered the QIDS-C and QIDS-SR, and research outcome assessors administered the HAM-D and the Inventory of Depressive Symptomatology.

Current burden due to general medical conditions was assessed at baseline using the 14-item Cumulative Illness Rating Scale (20, 21), which was completed by the clinician using a manual scoring guide (22). This scale gauges the severity or morbidity of general medical conditions relevant to different organ systems. Each condition is scored from 0 (no problem) to 4 (extremely severe/immediate treatment required/end organ failure/severe impairment in function). The scores are then summed across conditions to generate a total score. Participants also completed the self-report Psychiatric Diagnostic Screening Questionnaire (23, 24) at baseline, which was used to establish the presence or absence of concurrent axis I disorders.

Work Productivity and Health-Related Quality of Life Measures

The interactive voice response system was used to gather measures of work productivity, quality of life, and functioning at baseline and study exit. The six-item self-report Work Productivity and Activity Impairment scale (WPAI) (25) was used to measure the number of work hours missed in the past 7 days, the number of hours worked in the past 7 days, and impairment resulting from health conditions while working or performing usual daily activities. Higher WPAI scores indicate lower productivity. The construct validity and test-retest reliability of the self- and interviewer-administered WPAI–General Health (WPAI-GH) have been assessed by analyzing the extent to which the instrument correlates with symptom severity, global measures of work, and interference with regular activities, as well as with several domains of the Medical Outcomes Study 36-item Short Form Health Survey (26, 27).

The ability to work, to manage home and social affairs, and to form and maintain close relationships was assessed using the five-item self-report Work and Social Adjustment Scale (28); scores >20 suggest moderately severe functional impairment, and scores in the range of 10–20 are associated with significant functional impairment and less severe clinical symptoms. The 16-item Quality of Life Enjoyment and Satisfaction Questionnaire (29) was used to measure satisfaction and enjoyment in various domains of functioning: physical health, feelings, work, household duties, schoolwork/housework, leisure time activities, social relations, and general activities, with higher scores indicative of better quality of life. Perceived health and functioning were assessed with the 12-item Short Form Health Survey (30, 31).

Statistical Analysis

Of the six items on the WPAI, three (items 1, 2 and 5) were appropriate for this analysis. Item 1 (yes/no) determines current employment status; those who responded affirmatively were included in the analysis. Item 2 evaluates number of hours missed as a result of health problems (range, 0–80). Item 5 rates the degree to which current health problems affect productivity at work, using a 0–10 scale on which 0 indicates no impairment. The independent associations of scores on items 2 and 5 with baseline sociodemographic and clinical characteristics were assessed using exploratory stepwise regression models. Because the outcomes were not normally distributed, a log-normal model was used for number of days missed and a cumulative logit model for impairment score. Stepwise logistic regression models were used to assess work outcomes, which were operationalized as either a reduction of ≥50% in the number of hours missed or a reduction of ≥50% in WPAI impairment score. To be included in these analyses, data on hours missed and impairment, respectively, had to be present at baseline. An alpha level of 0.05 was used to test for statistical significance. For all outcomes, multiple imputation methods were used to impute missing data. For each outcome, 40 imputations were conducted based on the relative efficiency for multiple imputation.

Results

Of the 4,041 participants enrolled in STAR*D, 2,311 were employed, and of these, 1,928 reported impairment in work productivity or activity at baseline. Of these, 1,100 provided complete data regarding work productivity at exit from the initial treatment with citalopram (level 1) (see Figure S1 and Table S1 in the data supplement that accompanies the online edition of this article). Among the 1,928 who reported work impairment at baseline, the mean number of hours missed over the past 7 days was 7.0 (11.4), and the mean WPAI impairment score was 3.8 (SD=2.8). These measures were not highly correlated (r=0.22, p<0.0001).

Figure 1 summarizes past-week number of work hours missed at baseline and exit from level 1 treatment. At baseline, the proportion of participants reporting at least 1 hour of missed work was 56%, and at exit, the proportion decreased to 46%. Similarly, at the other extreme, while 22% reported missing more than 10 hours at baseline, less than 15% did so at exit. Figure 1 also summarizes participants’ subjective rating of past-week impaired productivity at work at baseline and exit on the WPAI impairment item. At baseline, 22% of the sample reported no impairment, and this proportion rose to 37% by the end of acute-phase treatment. To our surprise, rates of chronic depression were higher among those without baseline work impairment (41%) than those with baseline work impairment (17%), but with little difference in symptom severity.

Table 1 lists those baseline factors that were independently associated with missed work hours and impaired productivity. Baseline characteristics significantly associated with missing more hours of work included African American race, a history of suicide attempt, anxious features, an early age at onset of major depression, and greater symptom severity. Participants who were uninsured missed less work. Greater baseline work impairment was associated with having at least a college education, early onset of depression, and greater symptom severity. Less work impairment was associated with African American race, having less than a high school education, and a history of suicide attempt.

Table 2 list factors independently associated with improvements of ≥50% in hours missed and in WPAI impairment score at exit from the initial acute-phase treatment with citalopram (N=1,100). For the 884 of these participants who indicated impaired productivity at baseline, education was the only sociodemographic variable independently associated with improved productivity. Interestingly, participants with a college education were less likely to report improvement than those without a complete high school education. Clinically, a reduction in QIDS-IVR score (p<0.001) was associated with improved productivity; atypical mood features (p=0.02), melancholic features (p=0.002), and recurrent depression (p=0.005) were associated with decreased productivity. Among participants who reported having missed time from work at baseline (N=584), a greater reduction in QIDS-IVR score (p<0.001) and age (p=0.03) were the only factors independently associated with an improvement of ≥50% in hours missed.

An evaluation of differences between nonresponding (<50% improvement in QIDS-SR score and a QIDS-SR score >5), partially responding (≥50% improvement in QIDS-SR score and a QIDS-SR score >5), and fully responding or remitting (a QIDS-SR score ≤5) participants indicated that on average, work productivity improved with improvements in depression outcome (Table 3).

The WPAI results indicated that 357 patients reported missing time from work and/or a loss in productivity at work at level 2 baseline and exit. Of these, 183 (51%) reported having missed time from work and 277 (78%) reported decreased productivity as a result of their depression. The analysis of variance of level 1 partial responders (who received augmentation in level 2) and nonresponders (who received a medication switch in level 2) who reported work impairment at level 2 entrance and exit revealed no significant between-group differences in productivity at work and hours missed from work, by treatment group (medication switch options or medication augmentation options) or by response group (remission, partial response, nonremission), and no significant treatment group-by-response group interaction in productivity at work or hours missed from work. Table 4 lists results of the evaluation of the additional benefit associated with level 2 treatment response.

An examination of the pattern of treatment response by employment status at exit from levels 1 and 2 found that at level 1 exit the rate of remission was 38.8% for employed patients and 26.4% for those not employed, and at level 2 exit it was 36.9% for employed patients and 28.0% for those not employed. The response rate at level 1 exit was 49.1% for employed patients and 37.6% for those not employed, and at level 2 exit it was 34.1% for employed patients and 28.5% for those not employed.

Analyses utilizing multiple imputation methods were conducted on all the relevant outcome measures (percent change in number of hours missed, percent change in productivity impairment score, as well as improvements of ≥50% in number of hours missed and work productivity impairment) to assess the impact of missing data. The results remained largely unchanged, except that an association was observed between greater impairment in work productivity and higher levels of anxious depression (see Table 1).

Discussion

The results of this study show that at treatment baseline in a population of employed outpatients with major depression, several sociodemographic and clinical characteristics distinguish those who reported occupational impairment from those who did not. Of interest, participants with no insurance missed less work than those with insurance, be it private or public. Furthermore, participants with a college education reported the most occupational impairment, and those without a high school education reported the least. Clinically, prior to treatment, a greater burden of illness (history of attempted suicide, anxious features, greater duration of illness, and higher levels of depression severity) was found to be associated with a greater number of hours missed from work and/or more reported occupational impairment.

An examination of the level 1 exit data indicated that, again, education was associated with reported occupational impairment, this time with lower levels of education associated with greater subjective improvement in functioning. Also, participants with melancholic and recurrent depression reported significantly improved productivity at exit. Most importantly, decreases in depressive symptom severity were associated with significant improvements in occupational impairment, as manifest in a reduction in hours missed and improvements in productivity.

While past reports have highlighted the importance of work productivity impairment in depression, few large-scale clinical trials have obtained standardized longitudinal data that allowed for the analysis of objective markers of changes in workplace productivity and of how these changes are associated with what has previously been defined as successful treatment outcome (reduction in mood symptoms). Future trials would do well to include these measures, perhaps augmented with objective performance measures directly obtained from employers in order to assess the full spectrum of the impact of depression-related impairment and likely benefits from remission-focused treatments.

Consistent with our findings, there is evidence that the effective treatment of depression may enhance workplace performance (32), with recovery from depression associated not only with lower costs for primary care health services but also with less time missed from work because of illness (33). Zhang et al. (34) reported that depressed patients treated by a mental health specialist, compared with a nonspecialist, had less illness-related loss of earnings and higher medical costs, resulting in a net reduction of economic costs. Others have also found an association between reduction of chronic depressive symptom severity and improved workplace performance (35). However, evidence also suggests that even with remission of depressive symptoms, deficits remain in workplace performance. This indicates that while effective treatment of depression alone is critical, it is not sufficient (10) and that chronic illness management tools, such as peer support, use of care managers, and the like, might be useful.

An examination of response patterns indicated that participants with at least a partial response to level 1 treatment reported a significant reduction in hours missed and a significant decrease in occupational impairment. Interestingly, both partial responders and full responders or remitters showed significant benefit from treatment relative to nonresponders; however, no significant difference was seen between partial and full responders. This brings several important issues to light. After an acute-phase treatment with an antidepressant, patients who achieve remission of depressive symptoms may still manifest signs of the illness in the form of occupational impairment. This leads us to ask two questions: Are the remaining symptoms of impairment lingering or remitting in course? If remitting, how long is the response latency period from the time of depressive symptom remission? The results of this study indicate that patients with major depression who experience a reduction in depressive symptoms during the course of acute treatment with an antidepressant may also experience a significant improvement in functioning, but not necessarily a complete return to premorbid functioning. The duration of this response latency period in depressed remitters (the additional amount of time taken to return to normal occupational productivity after depressive symptoms have remitted) warrants additional investigation.

The level 1 nonremitters who were able to achieve remission of depressive symptoms in level 2 continued to have impairment at work, or at least impairment equivalent to that of level 2 nonresponders. The clinical message is that patients who have demonstrated some degree of treatment resistance appear to be more prone to lingering difficulties in occupational functioning despite remission of depressive symptoms.

The strengths of this study include its use of a large sample of treatment-seeking outpatients with major depression who were recruited from both primary care and psychiatric specialty care settings. These strengths contribute to the results’ generalizability to outpatients with major depression who seek treatment in typical clinical settings. Two basic limitations of the study are, first, that the measures of work and social functioning were self-reported and limited to the past 7 days and that a number of observations were missing (baseline and follow-up data were available for only 57% of participants) and, second, that the occupational impairment or success of these patients is unknown. Other limitations include an abbreviated study period, the lack of a nondepressed comparison group or normative community sample, and the need for more detailed analysis by type of employment. It is difficult to interpret the inverse relationship between higher levels of education and lower levels of improvement in occupational functioning. The type of job and related requirements may have influenced this outcome. Future studies may need to take this into consideration.

Conclusions

In this study, we identified a number of sociodemographic and clinical features that were associated with occupational impairment at baseline and at exit from acute treatment of major depression. This information may aid clinicians in determining which patients are most likely to experience occupational impairment in order to direct assistance accordingly. The general relationship found between depressive symptom severity and occupational impairment is consistent with the current disease model, which includes occupational impairment as part of the diagnostic criteria for the underlying disease, major depression (36). Participants who presented for the treatment of major depression typically had occupational impairment as part of their symptom presentation, and hence this symptom is also thought of as a treatment target. The clinical belief is often that the occupational dysfunction is due in part or in whole to the disease process and that if the underlying depressive pathology is treated, the manifest occupational impairment can also be expected to improve. Our results support this relationship in an initial antidepressant trial, as citalopram treatment reduced levels of both depressive symptoms and occupational impairment. While remission is the goal of treatment, even patients who experience a partial response to treatment in the form of a reduction in depressive symptom severity also experience an improvement in occupational functioning, as evidenced by a reduction in hours missed from work and decreased impairment while at work. However, this did not hold true for patients who experienced an initial treatment failure. Residual symptoms of occupational impairment remained, even after depressive symptoms were treated to remission with a second antidepressant agent. In such instances, additional treatment is indicated (37). These findings also highlight the importance of utilizing all possible treatment modalities (including nonpharmacological) as well as pursuing the goals of measurement-based care, as emphasized by our group (38, 39), and focus on maintenance of wellness. The implications of these findings should be further investigated to evaluate the financial impact on workplace productivity.