The Impact of Gout on Work Absence and Productivity

Authors


Nathan L. Kleinman, HCMS Group, 1800 Carey Avenue, Suite 300, Cheyenne, WY 82001, USA. E-mail: Nathan_Kleinman@hcmsgroup.com

ABSTRACT

Objective:  The goal of this analysis was to evaluate the impact of gout, a painful inflammatory arthritis condition, on an employed population's health-related work absence and objectively measured productivity output.

Methods:  Payroll, demographic, medical, pharmaceutical, sick leave, short- and long-term disability, and workers' compensation data were collected from multiple large employers with employees widely dispersed across the United States. Data were collected during the time period of 2001 to 2004 from approximately 300,000 employees. Objective productivity output data were also available for a subset of employees (captured electronically in the form of units of work processed per person). T-tests and chi-square tests were used to compare demographic data. Two-stage multivariate regression models were used to compare annual work absence and at-work productivity between employees with and without gout, while controlling for group differences indemographics, salary, other work-related variables, and comorbidities (using the Charlson Comorbidity Index).

Results:  The annual prevalence of gout was 4.7 per 1000 employees from 2001 to 2004. Employees with gout had 4.56 more annual absence days for all categories of health-related work absence than those without gout. Objective productivity (units of work processed) results were only available for a small subsample of employees (86 with gout and 27,472 without gout). Employees with gout processed 3.51% fewer units per hour worked and 2.38% fewer units per year than employees without gout (nonsignificant at P = 0.49 and P = 0.78, respectively).

Conclusion:  This study suggests that gout has a substantial impact on work absence and may also negatively impact productivity.

Introduction

Gout is now the most common form of inflammatory joint disease in men aged 40 years and older [1], and is thought to affect approximately 5.1 million US adults [2]. Prevalence of gout is thought to be increasing [3–8], in part, because of an increase in overall longevity, as well as an increase in multiple contributory risk factors, including obesity, alcohol consumption, high purine intake, hypertension, increased prevalence of end-stage renal failure, and use of certain predisposing medications such as aspirin and thiazide diuretics [9].

Patients with gout present, particularly in the acute and recurrent phases, with rapid onset of severe pain, accompanied by swelling and erythema in the affected joint. The involved joint is usually monoarticular (90%) and found in the lower extremity [10,11]. Pain is considered the primary symptom, but treatment of the condition is aimed at treating both pain and inflammation [12]. Even with treatment, it usually takes 5 to 7 days for symptoms to resolve [11,12]. Patients generally enter an asymptomatic phase between episodes, and with time if left untreated, attacks are more likely to become polyarticular, more frequent, and of longer duration [10]. Ultimately, and particularly when hyperuricemia is poorly controlled, a patient may develop an advanced, chronic picture, with continuous pain and symptoms from chronic arthropathy [13]. This advanced picture may also include the comorbid conditions of urolithiasis, urate nephropathy, and/or tophi [14].

As the incidence and prevalence of gout appear to be increasing, the effect of this disease on employers in terms of health-related work absence and productivity requires evaluation. Studies have confirmed that patients with musculoskeletal disease reduce work hours, take breaks, or may even be forced to change jobs, in efforts referred to as “self-management” or “behavioral coping” to deal with pain and disability [15–17]. Stewart et al. have evaluated the effect of pain from arthritis on work absence and reduced performance using data collected between 2001 and 2002 using the American Productivity Audit, a national survey of the US workforce [18]. They found that approximately 2% of subjects suffered a loss in productivity due to pain from arthritis, with a mean loss inproductive time (either from actual absence or presence at work with a reduction in performance) of 5.2 hours per week. Unfortunately, this study does not differentiate these results for inflammatory versus noninflammatory arthritis, conditions that have very different clinical presentations and long-term consequences [19]. In addition, the analysis only examined the effect of pain on these variables, and not the effect of the overall condition. Furthermore, it does not examine objective work output data when estimating productivity loss. An additional study on arthritis and associated joint disorders (AJD) in an employed population found employees with AJD produced 4% fewer annual work units (P < 0.05), had higher annual absence costs ($84), and more short- and long-term disability than employees without AJD [20].

The goal of this analysis was to evaluate the impact of gout, an inflammatory arthritis condition, on an employed population's health-related work absence and productivity output.

Materials and Methods

Data Source

A retrospective study was utilized using two comparison cohorts selected from the Human Capital Management Services Research Reference Database (HCMS RRDb): a cohort of employees with gout, and a cohort of employees without gout. This database of approximately 300,000 employees is compiled from several large national US employers, and represents retail, service, manufacturing, and financial industries. The HCMS RRDb includes demographic and claims data from 2001 to 2004 for direct medical care, prescription drugs, absence, disability, and workers' compensation. There are also data on employee-specific at-work productivity for a subset of the overall population. This subset of employees performs work that can be measured in units of work processed per hour. The demographic data include age, tenure (years with current employer), sex, exempt status (indicating a salaried versus hourly pay schedule), marital status, annual salary, full- or part-time status, and geographic region (defined by the first digit of the employee's ZIP code). The HCMS RRDb contains all medical claims for subjects allowing an analysis of comorbid conditions using Agency for Healthcare and Research Quality's 17 major diagnostic categories and 261 specific categories [21]. This database has been used previously for this type of research [22].

The cohort of employees with gout was selected from the HCMS RRDb if they had an International Classification of Diseases-9 (ICD-9) code 274.xx for the diagnosis (primary, secondary, or tertiary) during the calendar years of 2001 to 2003. The subject's index date was defined as the date the employee was first associated with the gout ICD-9 diagnosis.

Employees without gout, the control cohort, were selected from the subjects that did not have a primary, secondary, or a tertiary ICD-9 code for gout (274.xx). The index date for this cohort was assigned based on the average index date of the employees with gout.

Employees from both cohorts were required to have continuous eligibility for medical and prescription benefits for at least 1 year after the index date. Annual outcomes data for each analysis were measured for the year immediately after the index date for both cohorts.

Statistical Analysis

The mean values for the demographic data of the two cohorts were calculated and compared using t-tests or chi-square tests. These data included age (at index date), sex, annual salary, tenure (years with current employer as of the index date), exempt status, full- or part-time status, and marital status. Demographic data on race were not available for all records, and were only used to help control for possible confounding factors in regression modeling. Demographic data were also available and compared for region, as defined by the first number of the subject's ZIP code. All t-test comparisons were considered statistically significant when P ≤ 0.05.

Regression models were used to compare health-related work absence and lost at-work productivity between the two cohorts. The regression models controlled for the confounding factors of age, sex, tenure, marital status, race, exempt status, full- or part-time status, salary, and region. In addition, the models were controlled for comorbidities using a Charlson Comorbidity Index score [23].

Variables modeled in the work absence analyses included lost days from work due to sick leave (SL), short-term disability (STD), long-term disability (LTD), and workers' compensation indemnity (WC). A two-stage regression methodology was utilized for each of these dependent variables as these data were highly skewed, had nonconstant variance, and had many observations in which values were zero [24,25]. Logistic regression was used in the first stage model to predict the likelihood of absence during the year after the index date (dependent variable was 1 if the employee had more than 0 absence days and was 0 otherwise). In the second stage, a generalized linear model (with a gamma distribution and log link) was used to estimate average annual lost days for those employees with work absences. The results were then combined to yield estimates of annual lost days for all employees in the cohort. Only employees eligible for a specific work absence benefit were included in regression models for that benefit.

Productivity data for subjects used in this analysis were only available for a subset of the HCMS RRDb (those with values provided for number of units processed per hour). These included 86 employees with gout and 27,632 employees without gout. The analysis allowed for examination of productivity while at work (hourly productivity) and for a 12-month period (total annual productivity). The at-work productivity analyses were performed using only the second-stage regression modeling described above. Subjects with hourly or annual productivity values greater than 4 standard deviations from the mean were removed (0 employees with gout and 160 employees without gout were removed from the hourly analysis, and 0 employees with gout and 117 employees without gout were removed from the annual analysis) [22].

SAS System for Windows version 9.1 (SAS Institute, Inc., Cary, NC, USA) was used to generate all statistical analysis.

Results

The analysis identified 1171 employees with gout and 247,867 employees without gout yielding a prevalence of approximately 4.7 per 1000 employees. The means and confidence intervals for demographic data for both cohorts are given in Table 1. All t-test and chi-square test comparison results between both cohorts were found to be significant at P ≤ 0.05 level. Employees with gout were older than those without gout (5.5 years) and 56.5% more likely to be male. They were more likely to be Caucasian, and to be married. Employees with gout had higher salaries and longer tenure, were more likely to have exempt status, and were more likely to be employed full time. It was also found that employees with gout were more likely to live in the Northeast region of the United States (ZIP codes starting with 0 or 1). They were less likely to live in ZIP code regions starting with 3 (Florida, Alabama, Mississippi, Tennessee, and Georgia), ZIP code regions starting with 4 (Kentucky, Indiana, Ohio, and Michigan), and ZIP code regions starting with 9 (Alaska, Hawaii, Washington, Oregon, and California) (P ≤ 0.05).

Table 1.  Demographic statistics for employees with and without gout (during the year after the subject's index date*)
VariableEmployees with goutEmployees without gout
nMeanLower 95% CIUpper 95% CInMeanLower 95% CIUpper 95% CI
  • *

    For employees with gout, the index date is the date of the first gout diagnosis (ICD-9 274.xx) in the study period. For employees without gout, the index date is the average index date from the group of employees with gout.

  • Differences evaluated using t-tests or chi-square tests and found to be significant (P ≤ 0.05).

  • Values given in years.

  • §

    Exempt status (indicating a salaried vs. hourly pay schedule).

  • CI, confidence interval; n, sample size.

Age (at index date)†,‡1,17145.9145.4046.43247,84940.4140.3740.45
Male sex (%)1,17185.082.987.0247,86754.354.154.5
Annual salary ($)1,14561,36157,62265,100244,39750,31449,94250,686
Tenure (at index date)†,‡1,17112.7912.2313.36247,8679.739.709.76
Exempt (%)†,§1,17136.033.238.7247,85929.529.329.7
Full-time (%)1,17194.493.095.7247,86786.686.586.8
Race736   170,951   
 White (%) 71.768.575.0 65.465.165.6
 Black (%) 15.512.918.1 19.619.419.8
 Hispanic (%) 5.84.17.5 9.79.59.8
Married (%)1,08766.163.268.9225,03756.656.456.8
ZIP code first digit =1,171   247,867   
0 (%) 17.915.720.1 12.212.112.3
1 (%) 17.014.819.1 14.514.414.7
2 (%) 13.311.415.3 13.613.513.7
3 (%) 17.114.919.2 21.521.321.6
4 (%) 3.82.74.8 5.15.05.2
5 (%) 0.40.10.8 0.70.70.7
6 (%) 2.81.93.8 3.13.03.2
7 (%) 13.111.115.0 12.912.713.0
8 (%) 5.03.86.3 4.54.54.6
9 (%) 9.67.911.3 11.811.712.0

Figure 1 displays the annual lost days per employee with gout versus the employee without gout. Employees with gout had 2.78 more days of SL than employees without gout (P < 0.0001). They had 3.03 more days of STD absence (P = 0.0003). Employees without gout had 1.45 more days of absence associated with LTD than did employees with gout (P < 0.0001). Differences in lost days associated with WC were not statistically significant. WC days were nonsignificantly greater for employees with gout (0.2 more days). The total number of absence days for all categories of disability for employees with gout was 14.39 days versus 9.83 days for employees without gout, or a 4.56-day difference.

Figure 1.

Comparison of annual lost days per employee. Lost days were calculated for employees eligible for each specific benefit for each regression model for that benefit, and were adjusted controlling for age, tenure, sex, marital status, race, exempt status, full- or part-time status, region, and Charlson Comorbidity Index. For employees with gout, the index date is the date of the first gout diagnosis (ICD-9 274.xx) in the study period. For employees without gout, the index date is the average index date based on the cohort of employees with gout. Sample sizes of eligible employees with gout and without gout were: 600 and 123,461 employees for sick leave; 484 and 102,234 for short-term disability; 822 and 177,477 for long-term disability; and 1085 and 224,723 for workers' compensation, respectively.

Table 2 displays the objectively measured productivity comparison of units processed per hour worked and units processed per year adjusted for age, tenure, sex, marital status, race, exempt status, full- or part-time status, salary, region, and Charlson Comorbidity Index. Employees with gout averaged 3.51% fewer units processed per hour worked than those without gout, but this difference was found to be nonsignificant (P = 0.4939). The difference in units processed per year (2.38% fewer units for employees with gout) was also nonsignificant (P = 0.7758).

Table 2.  Objective productivity comparison of units of work processed per hour and per year*,†
Adjusted units processedEmployees with goutEmployees without goutΔ in meansP-value
MeanLower 95% CIUpper 95% CIMeanLower 95% CIUpper 95% CI
  • *

    For employees with gout, the index date is the date of the first gout diagnosis (ICD-9 274.xx) in the study period. For employees without gout, the index date is the average index date from the cohort of employees with gout.

  • The subsets were analyzed (restricted to those employees with productivity data) using regression models adjusted for age, tenure, sex, marital status, race, exempt status, full- or part-time status, salary, region, and Charlson Comorbidity Index. Outliers (>4 standard deviations) were removed.

  • Significant at P ≤ 0.05.

  • CI, confidence interval.

Per hourn = 86  n = 27,472    
17.8516.0319.6718.5018.3918.600.650.4939
Per yearn = 86  n = 27,515    
27,38223,63531,32928,04927,82928,268566.40.7758

Discussion

This analysis, using a comprehensive source of claims data including metrics for health-related work absence and at-work productivity, adds to the limited literature of the impact of gout on employers in the United States. The results found that the overall prevalence of gout in this employed population was 4.7 per 1000 employees. This is consistent with other physician-derived prevalence rates that have been reported in the literature [4,7,26,27]. Physician-reported cases of gout are lower than self-reported cases, with current self-reported prevalence in the United States documented at 22.4 per 1000 adults aged 45 to 64 years [4]. The lower rate of cases reported by physicians may lead to an underestimation of the work absence and productivity-loss in this study, as employees with gout who self-treat and are subsequently not reported in this database may not only have frequent absences from work, but also be less productive because of the condition.

The results of this analysis also revealed that in this population of employed workers, employees with gout were older and more likely to be male, results that are consistent with current literature [1,28]. More importantly, from the perspective of the employer, employees with gout in this population had higher salaries, longer tenure, and were more likely to have exempt status and full-time status. These figures are important for several reasons. As suggested above, the data may suggest that there is an element of self-management that is occurring in the gout cohort, as has been previously documented in patients with arthritis [29]. Employees with gout may be unable to maintain positions that require strenuous activity, positions that are often paid on an hourly basis or are more likely to include overtime status. They are therefore less likely to choose positions that include manual labor, managing their disease by the amount of activity they perform. This is also suggested by the small sample size of subjects that are included in the subset of employees with gout that are employed in positions that allow for measurement of productivity in units per hour, positions that often include manual labor. Of equal interest to the employer are the data which show that gout employees have higher salaries than those without gout. Recent studies designed to capture variations in ranges of salary indicate that with increasing salary (especially at levels ≥$50,000), lost productivity, as measured in percentage of lost productive time in hours, may poorly reflect true lost productive time cost in dollars [18]. Additionally, differences in gout prevalence were found for different regions in the United States. Gout prevalence was relatively higher in the Northeast and relatively lower in the Southeast, Midwest, and West Coast regions. These differences may be due to lifestyle and factors as such as alcohol intake that varies throughout the country [30–32].

Not surprisingly, SL and STD absences are significantly greater for employees with gout. Absence due to workers' compensation was statistically even between the two cohorts, indicating that gout may not increase an employee's likelihood of being injured on the job. LTD absences, however, were higher for employees without gout than for employees with gout. One explanation may be that LTD claims are rare, yet result in a large number of lost days from work. The results of the analysis revealed that over all health-related work absence benefits, the cohort of employees with gout lost almost 5 more days of work on average annually than did employees without gout. Using this incremental value and the study's prevalence rate of 4.7 per 1000 employees, an annual total of additional lost days in persons with gout can be conservatively projected across the US civilian labor force (a total of 144.9 million persons in 2002) [33] at approximately 3.105 million additional lost days. Another way to undertake this analysis is to calculate the days lost by using values for the population insured by an employer in the United States [34,35]. In 2002, approximately 64% of adults aged 18 to 64 years, or an estimated 114.9 million population, fell into this insured category. Using this population number, the overall incremental lost days in persons with gout is estimated at 2.463 million days.

The productivity comparison per hour and per year between employees with gout and without gout, over a small subset of the database that included these data, did not reveal significant differences between the two groups. As noted above, this is a group of employees who are employed in positions that generally require more strenuous activity, and the small sample size may reflect self-selection of employees with severe gout away from such positions. Only 7.3% of employees with gout fell into this subset of employees with productivity data (86 persons out of a total of 1171) versus 11.1% of employees without gout (27,515 persons out of a total of 247,849). Upon examination of these results, it may be hypothesized that the population that is still capable of this type of more strenuously physical employment is in an earlier stage of the disease, and suffers less from pain and has less work impairment.

Limitations of this analysis include the fact that the study was not designed to evaluate some unit of measure of productivity for “white collar” workers with gout. With increasing pain and severity of gout, disability increases. While the study was able to capture health-related work absence for most employees, it was only able to capture objectively measured work output data for a subset of “blue collar” workers. Additional research is needed to better measure the loss in work productivity due to the disease for employees that are engaged in “white collar” positions to more completely capture lost productivity.

As noted above, another limitation that leads to potential underestimation of both work absence and lost productivity is the fact that the cohort with gout was restricted to cases that were diagnosed with ICD-9 codes during the study time frame. As self-reported cases of gout are substantially higher than physician-derived cases, there is a possibility that cases of gout were not recorded because of misdiagnosis, the lack of recording the diagnosis on the claims record, or the possibility that a patient with gout simply did not seek medical attention [36]. All of these possibilities lead to underestimation of the impact of gout on both work absence and lost productivity. Similarly, because patients with gout who have been treated may have fewer flares and pain than untreated patients, another form of underestimation may result from not taking gout treatment into account in the analysis.

Some limitations arise from the data used. First, the absenteeism data (sick leave, short- and long-term disability, and workers' compensation) were not available for some employees. This is due to the fact that some employees are not eligible for all benefits, and a small number of employers have not tracked or supplied these data. Second, the HCMS RRDb primarily includes working-age persons, whereas the prevalence of gout may be higher in older, nonworking populations. Nevertheless, the focus of the current analysis is impact of gout on employee absence and productivity, and the HCMS RRDb is consistent with the 2004 US Employed Civilian Labor Force (N = 139.2 million) in terms of age and sex proportions [37]. Nevertheless, the HCMS database has a lower proportion of white employees (65.4% vs. 82.8%), and higher proportions of black (19.6% vs. 10.7%) and full-time employees (86.6% vs. 72.0%). Furthermore, the median salary in the current study was $42,044 compared with the US median salary of $33,176 [38]. The differences may be due to the employers in the database being large and nationally based. Lastly, claims data are collected for reimbursement, not for research purposes. As a result, these findings need to be replicated in a clinical study.

Further research is needed to determine the differences in lost days and productivity in early stages of gout versus more chronic, severe stages. Research is also needed to determine how much of work absence and lost productivity is due to the pain accompanying the condition versus the disability that accompanies more chronic disease. This would include the contribution of associated comorbidities of progressive gout, such as urolithiasis, urate nephropathy, and/or tophi. Finally, the contribution to work absence and lost productivity of comorbidities commonly associated with gout such as those found with metabolic syndrome (obesity, hyperlipidemia, insulin resistance, and hypertension), and/or the actual development of type 2 diabetes or cardiovascular disease should be evaluated.

The results of this type of detailed research allow for employers to begin targeting specific treatment and educational programs to not only improve the employee's condition, but also improve work-place outcome metrics. The current indications for treatment of gout include that the patient has had: 1) at least two attacks of gout (regardless of serum uric acid level); 2) tophi; 3) uric acid lithiasis; or 4) a serum uric acid greater than 10 mg/dL. The goal is to reduce the serum uric acid to less than 6 mg/dL [10,14]. Current treatment for controlling hyperuricemia includes allopurinol, a xanthine-oxidase inhibitor. Compliance with allopurinol has been shown to be low [39], and the use is associated with side effects in up to 5% of patients, including gastrointestinal symptoms, bone marrow suppression, and hypersensitivity syndrome [14]. Development of more effective treatment for hyperuricemia may improve clinical and economic outcomes and potentially reduce the number of gout flares.

In conclusion, gout is a chronic, progressive disease that, when untreated, may lead to permanent joint and bone destruction, with resulting disability. This study adds to the limited literature of the effect of gout on the employed population and suggests that there is a substantial impact of this illness on health-related work absence and objectively measured productive output. Furthermore, it suggests that further research is required to develop new and innovative solutions to improve the status of the employee with gout.

The authors would like to thank Suzanne Novak, MD, PhD, for her invaluable contributions to early drafts of this article. Funding for this project was provided by TAP Pharmaceutical Products, Inc., Lake Forest, IL, USA.

Source of financial support: TAP Pharmaceutical Products, Inc., Lake Forrest, IL, USA.