http://informahealthcare.com/pgm ISSN: 0032-5481 (print), 1941-9260 (electronic) Postgrad Med, 2015; 127(5): 455–462 DOI: 10.1080/00325481.2014.994468
ORIGINAL RESEARCH
United States comparative costs and absenteeism of diabetic ophthalmic conditions Richard A. Brook1, Nathan L. Kleinman2, Sunil Patel3, Jim E. Smeeding4, Ian A. Beren2 & Adam Turpcu5 The JeSTARx Group, Newfoundland, NJ, USA, 2HCMS Group, Cheyenne, WY, USA, 3The Retina Research Group, Abilene, TX, USA, 4The JeSTARx Group, Dallas, TX, USA, and 5Genentech, Inc., South San Francisco, CA, USA
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
1
Abstract
Keywords:
Objective: This retrospective cohort study examined the impact of diabetic macular edema (DME), diabetic retinopathy (DR), or diabetes on annual health benefit costs and absenteeism in US employees. Methods: Claims data from 2001 to 2012 was extracted from the Human Capital Management Services Group Research Reference Database on annual direct/indirect health benefit costs and absences for employees aged ‡ 18 years. Employees with DME, DR, or diabetes were identified by International Classification of Diseases, Ninth Revision, Clinical Modification codes. Employees were divided into two groups, drivers or nondrivers, and examined in separate analyses. For drivers and nondrivers, the DME, DR, and diabetes cohorts were compared with their respective control groups (without diabetes). Two-part regression models controlled for demographics and job-related characteristics. Results: A total of 39,702 driver and 426,549 nondriver employees were identified as having ‡ 1 year’s continuous health plan enrollment. Direct medical costs for drivers with DME, DR, or diabetes were $6470, $8021, and $5102, respectively (>2.8 times higher and statistically significant compared with driver controls). Nondrivers with DME and DR incurred significantly higher sick leave and short-term disability costs compared with the nondrivers with diabetes and nondriver controls. In drivers with DME, the majority of days of absence were for short- and long-term disability (12.41 and 11.43 days, respectively). In drivers with DR, the majority of days of absence were for short-term disability (10.70 days). In nondrivers with DME and nondrivers with DR, the majority of days of absence were for sick leave (5.74 and 4.93 days, respectively) and short-term disability (5.08 and 4.93 days, respectively). Conclusion: DME and DR are associated with substantial direct medical cost and absenteeism in this real-world sample of medically insured employees. This research highlights the negative impact of DME and DR on annual costs and absenteeism and may assist employers in assessing the impact of these conditions on employees.
Cost studies, macular edema, outcomes studies, retinopathy
Introduction Diabetes mellitus affects ~ 8.3% (25.8 million) of the US population, of whom an estimated 7.0 million are undiagnosed [1]. Diabetic retinopathy (DR) is the main cause of new blindness cases in adults aged 20 to 74 years in the United States [1]. Diabetic macular edema (DME), which can occur at any stage of DR, is diagnosed when retinal thickening involves or approaches the central macula [2,3]. Consistent with the increasing prevalence of diabetes in the United States [4], the number of adults with diagnosed diabetes who reported vision impairment (defined as visual difficulty despite using corrective lenses) increased from 2.7 to 4.0 million from 1997 to 2011 [5]. DR was reported in 4.2 million (28.5%) people with diabetes aged ‡ 40 years in 2005 to 2008. Of this group, 655,000 (4.4%) had advanced DR and were consequently at risk of severe vision loss [1,6]. Increasing prevalence of diabetes in younger adults (aged
History Received 11 September 2014 Accepted 21 November 2014 Published online 31 December 2014
20–44 years) [1] suggests that these trends are potentially relevant to employers [7]. Diabetes is associated with a significant economic burden in the United States because of increased health resource use [8,9] and loss of productivity [7,10,11]. The American Diabetes Association reported that, in 2012, the total estimated direct medical costs and costs related to reduced productivity associated with diagnosed diabetes were $176 billion and $69 billion, respectively [9]. Moreover, medical expenditures in individuals with diagnosed diabetes were found to be ~ 2.3 times higher than for individuals without diabetes [9]. However, the contribution of retinopathy to the cost and burden of diabetes is unknown. Certain aspects of the lifestyle of commercial drivers, such as the sedentary nature of the occupation and limited healthy food choices on the road, are associated with increased risk of developing diabetes [12,13]. The National Survey of Long-Haul Truck Driver Health and Injury,
Correspondence: Richard A. Brook, MS, MBA, Retrospective Analysis, The JeSTARx Group, 18 Hirth Drive, Newfoundland, NJ 07435-1710, USA. Tel.: +1 973 208 8621. Fax: +1 973 954 2968. E-mail: [email protected] 2015 Informa UK Ltd.
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
456
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
conducted by the National Institute for Occupational Safety and Health, found that the self-reported prevalence of diabetes was 14.4% among US long-haul truck drivers but was 6.7% in working adults (based on the 2010 National Health Interview Survey) [14]. Commercial drivers may be particularly susceptible to develop DR and DME, as job conditions may make diabetes management difficult, and they may be unwilling to start insulin therapy owing to regulations surrounding commercial vehicle operation and insulin use [15]. However, visual acuity (Snellen chart) of at least 20/40 or better in each eye (with or without correction) is required to qualify for a commercial driver’s license and to maintain employment [16,17]. A literature search conducted in early 2013 using the keywords diabetic retinopathy, diabetic macular edema, drivers, costs, and absenteeism failed to identify any publications. Consequently, this study may be the first to examine the impact of DR and DME on employer costs in a cohort of commercial drivers in the United States. The present study was designed to quantify and compare the burden of DME and DR in US employees, including commercial drivers for whom good vision is required to maintain employment [17]. This research compared annual absence days, annual health benefit costs, and the likelihood of using health benefits in employees with DME, DR, or diabetes versus controls without diabetes.
Materials and methods Study design This retrospective analysis utilized the Human Capital Management Services Group Research Reference Database (RRDb; Cheyenne, WY, USA). Employees were sourced from > 20 geographically dispersed US employers. Demographic, payroll, health insurance, health-related work absence, and workers’ compensation data were maintained in the RRDb for all employees. This database, containing > 1.7 million deidentified individuals employed at some time between 1 January 2001 and 30 June 2012, has been used in previous research and is a unique data source that classifies employees by job type allowing for the identification of drivers, and records actual absence costs and time by type of benefit [18,19]. Employees identified as commercial drivers were designated as “drivers” in this analysis. All other employees, with jobs unrelated to commercial driving, were identified as “nondrivers.” These two populations were examined in separate parallel analyses; this study was not designed to directly compare drivers and nondrivers. Based on International Classification of Diseases, Ninth Revision, Clinical Table 1. ICD-9-CM codes for DME, DR, and diabetes. Condition
ICD-9-CM code [15] (description)
DME
362.07 (DME) 362.53 (Cystoid macular degeneration) 362.83 (Retinal edema) 362.0x (DR) 250.xx
DR Diabetes
Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification.
Modification (ICD-9-CM) codes (Table 1) [20], drivers and nondrivers were further divided into the following groups: employees with DME, employees with DR but without DME, employees with diabetes without DME or DR, and controls. The control group for the drivers consisted of drivers without DME, DR, or diabetes. The control group for the nondrivers consisted of nondrivers without DME, DR, or diabetes. For driver and nondriver analyses, the DME, DR, diabetes, and control cohorts were compared. The index date for the DME and DR cohorts was the first claim date for the respective condition. For the diabetes and control cohorts, the index date was the average index dates of the DME and DR cohorts by company. All included individuals were required to have continuous employment and health plan enrollment for a 12-month study period after their index date, during which costs, days of absence, and likelihood outcomes were measured. Only individuals with the required eligibility and benefit were included in the analyses for that benefit. All subjects were assigned to a region of the country based on the first digit of their ZIP code. Outcomes of this study included days of absence, costs, and likelihood of using health benefits. Days of absence were categorized as sick leave, short- and long-term disability, and workers’ compensation (the workers’ compensation variable also included medical claims specifically paid under the individual employee’s workers’ compensation benefit). Annual health benefit cost components were derived from a number of different benefits provided by employers, including direct components (medical/healthcare costs, prescription costs) and indirect components (payments made to the employee for sick leave, short- and long-term disability, and workers’ compensation). The likelihood of using health benefits was defined as the percentage of each driver and nondriver population using each direct and indirect benefit type during the study period. All annual direct and indirect cost outcomes were inflation-adjusted to June 2012 US dollars by the appropriate component of the Bureau of Labor Statistics Consumer Price Index (CPI) [21]. The medical care services CPI and prescription drugs CPI were used to adjust medical cost components and prescription drug components, respectively, and the allconsumer-goods CPI was utilized for all other cost elements. Statistical analysis Nonregression-based statistical tests were used to compare the demographics of cohorts for the analyses of drivers and nondrivers. Statistically significant differences in continuous demographic characteristics and between dichotomous demographic variables were determined using Tukey-Kramer ttests (p < 0.05) and Marascuilo adjusted c2 tests, respectively. Two-part regression modeling was used to estimate each outcome. First, logistic regression was used to model the likelihood of an outcome > 0. Generalized linear models were then used to estimate costs or days of absence for the portion of the population with an outcome > 0. Independent variables, including demographics, job-related information, Charlson Comorbidity Index (CCI), and US region, were used to control for confounding factors between cohorts. Sick leave
DME, DR, and diabetes: impact on US employees
DOI: 10.1080/00325481.2014.994468
457
All employees n = 903,255
Employees with 12 months of continuous employment and health plan enrollment n = 466,251
Nondrivers n = 426,549
Drivers n = 39,702
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Drivers (with diabetes) n = 1320
Nondrivers (with diabetes) n = 20,618
Drivers controls (without diabetes) n = 38,382
Nondrivers controls (without diabetes) n = 405,931
DME n = 538
DME n = 26
DR (without DME) n = 1441
DR (without DME) n = 103
Diabetes (without DME or DR) n = 18,639
Diabetes (without DME or DR) n = 1191
Figure 1. Flow chart showing patient identification. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
models also controlled for employer, to account for variability between employers for this benefit type. All models utilized stepwise selection methods to determine the variables included in the final model. Model outcomes were considered significant at p < 0.05. All models were generated using SAS version 9.2 for Microsoft Windows (SAS Institute, Inc., Cary, NC, USA).
Results Descriptive statistics Overall, 466,251 employees with 12 months of continuous employment and health plan enrollment were identified from 903,255 employees who had some health plan enrollment (Figure 1). Descriptive statistics for the 39,702 commercial drivers and 426,549 nondrivers are shown in Table 2. In the driver analysis, the DME, DR, and diabetes cohorts were, on average, 6 years older compared with the driver controls. In the analysis of nondrivers, the DME, DR, and diabetes cohorts were 8 years older or more compared with the nondriver controls (all comparisons, p < 0.05). In both the driver and nondriver analyses, the DME, DR, and diabetes cohorts had significantly higher CCI scores versus their respective control cohorts (p < 0.05). However, the driver and nondriver DR cohorts had higher CCI score versus their respective DME cohorts (p < 0.05 for drivers; p < 0.05 for nondrivers). Driver and nondriver employees were distributed throughout the United States. Drivers were more concentrated in the eastern half of the United States, whereas nondrivers were more concentrated in the western region of the United States.
Annual absence days In the analysis of drivers, total annual days of absence were greatest in the drivers with DME (27.0 days), followed by the drivers with DR (16.1 days), drivers with diabetes (15.0 days), and driver controls (8.2 days) (Figure 2A). Most employee absenteeism among the drivers with DME and drivers with DR cohorts was in the short-term disability category (12.41 and 10.70 days, respectively). Also, compared with controls, drivers with diabetes had more sick leave days (p < 0.0001) and short- and long-term disability days (p < 0.0001 and p = 0.003, respectively) but had fewer workers’ compensation days (p = 0.003). Differences in absence categories among the drivers with DME were not significantly different compared with drivers with DR, drivers with diabetes, or driver controls. Drivers with DR were absent from work for more short-term disability days versus controls (p = 0.0054), and for fewer long-term disability days compared with the drivers with diabetes (p = 0.0024). No significant differences in total annual days of absence were reported between the drivers with DME and drivers with DR for all absence categories. In the analysis of nondrivers, total annual days of absence were greatest in the nondrivers with DME (13.8 days), followed by the nondrivers with DR (12.7 days), nondrivers with diabetes (8.9 days), and nondriver controls (5.4 days) (Figure 2B). The majority of days of absence in the nondrivers with DME and nondrivers with DR were for sick leave (5.74 and 4.93 days, respectively) and short-term disability (5.08 and 4.93 days, respectively). Additionally, the number of sick leave days and short-term disability days were significantly higher for nondrivers with DME and
458
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
Table 2. Descriptive statistics for the driver and nondriver populations. Drivers DME (n = 26)
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Variable Age, index Dx date, y Tenure, index Dx date, y Female, % Marital status, % Married Not married Missing Race/ethnicity, % White Black Hispanic Other Missing Exempt, % Annual salary, $d Full time, % Charlson Comorbidity Index score
Mean
DR (n = 103) SE
a
Mean a,b
Diabetes (n = 1191) SE
Mean a
Control (n = 38,382) SE
Mean
SE
46.87 5.97b,c 3.8a
1.72 1.12 3.8
48.93 11.40a 3.9a
0.93 0.89 1.9
46.53 10.78a 5.1a
0.25 0.23 0.6
39.05 8.41 17.0
0.04 0.03 0.2
65.4 30.8 3.8
9.5 9.2 3.8
66.0 16.5a 17.5a
4.7 3.7 3.8
65.7a 26.5a 7.7a
1.4 1.3 0.8
59.8 37.6 2.6
0.3 0.2 0.1
61.5 23.1 15.4 0.0a,b 0.0a,b 0.0a 35,463 96.2a 2.42a,b,c
9.7 8.4 7.2 0.0 0.0 0.0 1383 3.8 0.23
38.8a 39.8a 16.5 2.9 1.9 0.0a 36,569 96.1a,b 3.26a,b
4.8 4.8 3.7 1.7 1.4 0.0 690 1.9 0.12
51.5a 31.6a 14.2 2.0 0.8 0.3 37,164 89.3a 1.58a
1.4 1.3 1.0 0.4 0.3 0.1 211 0.9 0.03
62.8 21.6 12.7 2.5 0.4 0.5 35,016 78.2 0.09
0.2 0.2 0.2 0.1 0.0 0.0 244 0.2 0.0
Nondrivers DME (n = 538)
DR (n = 1441)
Diabetes (n = 18,639)
Control (n = 405,931)
Variable
Mean
SE
Mean
SE
Mean
SE
Mean
SE
Age, index Dx date, ye Tenure, index Dx date, y Female, % Marital status, % Married Not married Missing Race/ethnicity, % White Black Hispanic Other Missing Exempt, % Annual salary, $f Full time, % Charlson Comorbidity Index score
50.43a,b 11.74a 46.7
0.41 0.46 2.2
49.14a 12.20a 46.5
0.27 0.28 1.3
48.90a 12.22a 48.7
0.07 0.07 0.4
40.83 8.87 48.3
0.02 0.01 0.1
36.8 24.2a,b 39.0a,b
2.1 1.8 2.1
38.0b 27.6a,b 34.5a,b
1.3 1.2 1.3
42.1a 31.4a 26.4a
0.4 0.3 0.3
36.4 33.0 30.5
0.1 0.1 0.1
33.1 13.2 8.2 3.0 42.6 27.7a 50,160 97.8a,b 2.89a,b,c
2.0 1.5 1.2 0.7 2.1 1.9 1360 0.6 0.07
31.8 11.5 11.4a,b 3.5 41.9a 30.7a 54,680 97.4a,b 3.41a,b
1.2 0.8 0.8 0.5 1.3 1.2 933 0.4 0.03
33.3a 13.7a 7.5a 3.9 41.6a 29.3a 52,557a 95.9a 1.72a
0.3 0.3 0.2 0.1 0.4 0.3 250 0.1 0.01
35.0 9.3 5.4 3.6 46.7 36.2 56,972 93.1 0.15
0.1 0.0 0.0 0.0 0.1 0.1 143 0.0 0.0
p < 0.05 versus control cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. b p < 0.05 versus diabetes cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. c p < 0.05 versus DR cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. d Twelve drivers in the control cohort were missing salary information. e Three nondrivers in the diabetes cohort and 15 employees in the control cohort were missing age values. f There were 6, 25, 433, and 12,942 nondriver employees in the DME, DR, diabetes, and control cohorts, respectively, who were missing salary information. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy; Dx = Diagnosis; SE = Standard error. a
nondrivers with DR compared with those of nondrivers with diabetes and nondriver controls. Nondrivers with DR missed fewer days in the category of workers’ compensation compared with the nondrivers with diabetes and nondriver controls (both p < 0.0001). Nondrivers with diabetes had a greater number of days of absence compared with nondriver controls for all absence categories (p < 0.006). No significant differences were reported in any absence category between the nondrivers with DME and nondrivers with DR.
Annual health benefit costs Total annual health benefit costs in drivers with DME, DR, or diabetes were $12,511, $12,163, and $8785, respectively. In the driver controls, the total annual health benefit cost was $4149 (Figure 3A). Drivers with DME, DR, or diabetes had significantly higher direct medical costs (medical and prescription drug) compared with driver controls (all comparisons, p < 0.02). In drivers with DR, significantly lower longterm disability costs were reported versus the drivers with
DME, DR, and diabetes: impact on US employees
DOI: 10.1080/00325481.2014.994468
Drivers
a
b
Nondrivers Workers’ compensation Long-term disability Short-term disability Sick leave
30
30 0.77
25
25 20
459
20
11.43
15
2.94
Days
Days
0.66* 1.40†
10.70†
1.94 6.10‡
5.08‡,§
1.09
5.74*,‡
DME (27.04)
1.79
1.67
DR (16.08)
Diabetes (14.98)
Control (8.16)
4.93*
0.31 0.30 1.81
3.31‡
4.93*,‡
4.37‡
3.03
DR (12.65)
Diabetes (8.90)
Control (5.44)
0 DME (13.77)
Driver cohort (total absence days)
Nondriver cohort (total absence days)
Figure 2. Annual days of absence are shown for (a) drivers and (b) nondrivers. Within each analysis only: *p < 0.01 versus diabetes cohort; † p < 0.01 versus control cohort; ‡p < 0.0001 versus control cohort; §p < 0.05 versus diabetes cohort; kp < 0.0001 versus diabetes cohort; ¶ p < 0.05 versus control cohort. The n values for the driver and nondriver categories are shown in Table 3. Copyright 2013 American Diabetes Association. From Diabetes,Vol. 62, Suppl. 1; 2013. Modified by permission of the American Diabetes Association. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
diabetes (p = 0.0002) and driver controls (p = 0.0227). Workers’ compensation costs also were lower among the drivers with DR compared with the drivers with diabetes and driver controls (p = 0.0050 and p < 0.0001, respectively). However, for short-term disability costs, drivers with DR were significantly more expensive than the driver controls (p = 0.0037), and the difference in costs approached significance compared with the drivers with diabetes (p = 0.0548). For the drivers with DME, indirect costs did not differ significantly from the drivers with DR, drivers with diabetes, or driver controls, and Drivers
b Workers’ compensation Long-term disability Short-term disability Sick leave Prescription Medical
$20,000 $18,000 $740 $687 $1086 $389
$16,000 $14,000 $12,000 $10,000 $8000
$4000
$14,000
$75§,¶ $240† $760‡,§ $924§,II
$3733§,¶ $3338§,¶ $229* $66* $483§ $830§
$12,000 $905* $214* $497§ $348§
$8021§,II $5102§
$10,000 $8000
$1266 $57 $267 $277 $490
$1720§
$2000
$18,000 $16,000
$3139*
$6470†
Nondrivers $119II $136 $705*,II $1009‡,§
$20,000
$415‡,§ $20†,‡ $970* $299
$2438§
$6000
no significant differences were reported in any cost category between drivers with DME or drivers with DR. For nondrivers, total annual health benefit costs were $17,433, $19,280, $10,926, and $5258 for the nondrivers with DME, nondrivers with DR, nondrivers with diabetes, and nondrivers controls, respectively (Figure 3B). All cost components in the nondrivers with diabetes cohort were significantly higher than the nondriver controls (p < 0.001). Sick leave and short-term disability costs were significantly higher among the nondrivers with DME and nondrivers with DR cohorts versus
Costs
a
Costs
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
0
2.12‡
0.42† 0.78†
,‡
5
3.45 2.42
0.10‡,II 2.69¶
10
12.41
5
0.15 2.80
5.37†
10
15
$6000
$13,547§,¶
$2502§
$4000
$6815§
$2000
$3326
$1792 $0
$189 $29 $259 $606 $849
$12,125§,¶
$0 DME ($12,511)
DR ($12,163)
Diabetes ($8785)
Driver cohort (total costs)
Control ($4149)
DME ($17,433)
DR ($19,280)
Diabetes ($10,926)
Control ($5258)
Nondriver cohort (total costs)
Figure 3. Annual direct and indirect health benefit costs are shown for (a) drivers and (b) nondrivers. Within each analysis only: *p < 0.01 versus control cohort; †p < 0.05 versus control cohort; ‡p < 0.01 versus diabetes cohort; §p < 0.0001 versus control cohort; kp < 0.05 versus diabetes cohort; ¶ p < 0.0001 versus diabetes cohort. The n values for the driver and nondriver categories are shown in Table 3. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
460
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
Table 3. Likelihood of using health benefits during the 12 months following the index date for drivers and nondrivers. DME Employee cost category
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Drivers Medical Prescription drug Sick leave Short-term disability Long-term disability Workers’ compensation Nondrivers Medical Prescription drug Sick leave Short-term disability Long-term disability Workers’ compensation
DR
Diabetes
Control
n
Likelihood, %
n
Likelihood, %
n
Likelihood, %
n
Likelihood, %
26 26 26 25 25 26
100.0a 100.0a 69.6 11.4 5.8a 12.5
103 103 99 97 96 103
95.0b 93.2b 50.3 16.8a 1.6 8.7
1191 1191 1159 1067 1032 1185
97.5b 96.6b 59.7b 10.9b 1.8b 10.9
38,382 38,382 37,792 30,738 28,470 38,332
65.2 65.4 52.4 6.8 0.5 11.4
538 538 243 289 319 479
99.8b 97.0b 77.7b,d,e 13.4b,c 0.6f 1.5
1441 1441 635 811 964 1277
100.0b,c 97.9b,c 70.0b 11.3b,c 0.7b,c 2.0
18,639 18,639 7115 10,013 12,437 16,655
99.6b 96.9b 66.9b 9.1b 0.3b 2.6a
405,931 405,931 156,101 230,080 274,749 356,291
82.2 76.5 60.6 5.0 0.1 2.2
p < 0.01 versus control cohort. p < 0.0001 versus control cohort. c p < 0.05 versus diabetes cohort. d p < 0.05 versus DR cohort. e p < 0.001 versus diabetes cohort. f p < 0.05 versus control cohort. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy. a
b
the nondrivers with diabetes and nondriver controls (all comparisons, p < 0.05). No significant differences in any annual cost component were reported for comparisons between nondrivers with DME or DR. Workers’ compensation costs were significantly lower among the nondrivers with DME versus the nondrivers with diabetes (p = 0.0165), and in the nondrivers with DR cohort versus the nondrivers with diabetes and nondriver controls (both p < 0.0001). Long-term disability costs were significantly higher among the nondrivers with DR versus nondriver controls (p = 0.0218), but not compared with the nondrivers with diabetes (p = 0.0609). Likelihood of using health benefits In the driver population, the likelihood of using direct and indirect benefits was similar between the drivers with DME and drivers with DR, and for both versus the drivers with diabetes cohort (Table 3). Drivers with DR, DME, or diabetes were significantly more likely to use direct medical components (medical and prescription drug) compared with driver controls (all comparisons, p < 0.0003). The likelihood of using the long-term disability benefit was significantly higher among the drivers with DME versus driver controls (p = 0.0003) and, of all four cohorts, drivers with DME had the highest likelihood of using this benefit (drivers with: DME, 5.8%; DR, 1.6%; diabetes, 1.8%). Nondrivers with DR were more likely to use medical, prescription drug, and short- and long-term disability compared with the nondrivers with diabetes and nondriver controls (p < 0.04), and more likely to use sick leave compared with the nondriver control (p < 0.0001; Table 3). Nondrivers with DME were more likely to use the sick leave benefit compared with nondrivers with DR (p = 0.0241), with diabetes (p = 0.0004), or with controls (p < 0.0001). Compared with nondriver controls, the nondriver DME, DR, and diabetes cohorts were each more likely to use individual direct and indirect benefits (p < 0.04), with the exception of workers’
compensation, which was significantly higher in only the nondrivers with diabetes versus nondriver controls (p < 0.003).
Discussion This analysis demonstrates the significant impact of DME and DR on employee direct and indirect costs across a range of benefits, as well as the detrimental impact on employee absenteeism using actual, not proxied, data (as used in previous studies) [22]. In general, drivers and nondrivers with DME or DR were more expensive to manage, missed more work days, and were more likely to use health benefits than drivers and nondrivers with diabetes or their respective control groups. In separate analyses of drivers and nondriver employees, this study found that differences in days of absence, costs, and likelihood of using health benefits did not differ significantly between the DR and DME cohorts. This may be owing to the relatively small size of the DME and DR cohorts and lack of severity measures within the database. It is also possible that more severely affected patients with DME exit the workforce entirely and rely on government-sponsored sources of benefits (Medicare, Medicaid, or federal disability programs) instead of employer-sponsored benefits. The economic burden of diabetes and diabetes-related vision disorders has been previously reported [22,23]. A 2008 claims analysis by Lee et al. reported that DR employees had higher mean annual total costs compared with non-DR employees (who had diabetes but no DR) [22]. Employees with DME had 74.8% higher mean costs versus employees without DME. Indirect costs due to absenteeism were also higher in employees with DR compared with nonDR employees ($1640 vs. $1218, difference = $422; p < 0.0001). It is important to note that Lee et al. estimated absenteeism by assuming that 1 day of hospitalization resulted in the loss of 1 day of work and an outpatient visit resulted in the loss of one-half day of work [22].
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
DOI: 10.1080/00325481.2014.994468
Unlike clinical trials, which enroll a well-defined patient population, this retrospective cohort study allows for the examination of the impact of diabetic eye disease on costs and days of absence in a real-world population. However, this study has several limitations. In both the driver and nondriver analyses, the mean ages of the DME, DR, and diabetes cohorts were significantly higher compared with their respective control cohorts. Since age is approximately normally distributed, we calculated that ~95% of each study cohort was within two standard deviations of the cohort mean. The large control group sample sizes provided sufficient overlap in the distribution of ages between groups, and adjusting for age in the multivariate model is a sufficient control measure. DR and DME have been shown to worsen over time as more blood vessels are involved/affected [24]. However, as with most claims databases, data on retinal disease stage, severity, or duration is not captured. The present study was not designed to compare driver with nondriver populations. Although all available data were included in the regression analysis, the driver and nondriver populations were modeled separately. Therefore, variances and standard errors were based on each individual study population, and demographic differences or other factors were not controlled for between driver and nondriver populations. This study was limited to reporting direct and indirect costs for employees with DME, DR, or diabetes and did not assess the impact of these conditions on caregivers (i.e., employees with spouses and/or other family members with DME or DR). Since individuals were identified based on ICD-9-CM codes, control cohorts may have included individuals with undiagnosed diabetes. Some observations may be attributed to the limited sample size of the DR and DME disease cohorts within the driver population. For example, the CCI data revealed that the drivers with DR ranked higher than the drivers with DME; this observation is likely to be a result of the small sample size of drivers with DME. In the present study, we observed similar trends in costs, absences, and likelihood variables in employees with DME or DR within the driver and nondriver analyses. Given that the incidence of diabetic complications is directly related to poor glycemic control [25], these findings may suggest that uncontrolled diabetes indirectly leads to increased costs to employers. Employee wellness programs can provide health coaching, with some larger companies (> 750 employees) offering health risk assessment or screening [26]. In 2012, employee absenteeism accounted for $5 billion of the $69 billion attributed to diabetes-related reduced productivity in 2012 (25 million workdays) [9]; thus, the implementation of such plans may be of financial benefit to employers. This may be particularly necessary for drivers with DME as their annual absences totaled ~ 30 days.
Conclusion In this real-world sample of medically insured employees, DME and DR were associated with substantial direct medical costs and absenteeism. These data provide employers
DME, DR, and diabetes: impact on US employees
461
with important information necessary to assess the impact of DME and DR on employees.
Acknowledgments The authors thank Suzanne Douthwaite, PhD, and Rebecca Jarvis, PhD, of Envision Scientific Solutions for assistance with manuscript preparation, which was supported by Genentech, Inc.
Declaration of interest This study was sponsored by Genentech, Inc. Adam Turpcu is an employee of and owns stock in Genentech, Inc. Richard A. Brook, Jim E. Smeeding, Nathan L. Kleinman, and Ian A. Beren received consulting fees from Genentech, Inc. through their respective employers. Nathan L. Kleinman and Ian A. Beren were employees of HCMS Group, Cheyenne, WY, United States, at the time the study was conducted. Sunil Patel received grants to his institution from Alcon, Allergan, Alimera, Genentech, Inc., Pfizer, Ophthotech, and Regeneron; consulting fees and fees for participation in review activities or honorarium from Genentech, Inc.; consulting fees from Allergan and Ophthotech; fees for participation in speakers bureaus from Alcon; and owns stock/ stock options in Ophthotech.
References [1] Centers for Disease Control and Prevention. National diabetes fact sheet: national estimates and general information on diabetes and prediabetes in the United States. 2011. http://www.cdc. gov/diabetes/pubs/pdf/ndfs_2011.pdf. Accessed 7 February 2014. [2] Bandello F, Battaglia Parodi M, Lanzetta P, Loewenstein A, Massin P, Menchini F, et al. Diabetic macular edema. Dev Ophthalmol 2010;47:73–110. [3] Fenwick EK, Pesudovs K, Khadka J, Dirani M, Rees G, Wong TY, et al. The impact of diabetic retinopathy on quality of life: qualitative findings from an item bank development project. Qual Life Res 2012;21:1771–82. [4] Centers for Disease Control and Prevention. Increasing prevalence of diagnosed diabetes–United States and Puerto Rico, 1995-2010. MMWR Morb Mortal Wkly Rep 2012;61:918–21. [5] Centers for Disease Control and Prevention, National Center for Health Statistics, Health Interview Survey. Number (in millions) of adults aged 18 years or older with diagnosed diabetes reporting visual impairment, United States, 1997–2011. http://www.cdc.gov/diabetes/statistics/visual/fig1.htm. Accessed 7 February 2014. [6] Zhang X, Saaddine JB, Chou CF, Cotch MF, Cheng YJ, Geiss LS, et al. Prevalence of diabetic retinopathy in the United States, 2005-2008. JAMA 2010;304:649–56. [7] Tunceli K, Bradley CJ, Nerenz D, Williams LK, Pladevall M, Elston Lafata J. The impact of diabetes on employment and work productivity. Diabetes Care 2005;28:2662–7. [8] American Diabetes Association. Economic costs of diabetes in the US in 2007. Diabetes Care 2008;31:596–615. [9] Yang W DT, Halder P, Gallo P, Kowal SL, Hogan PF. Economic costs of diabetes in the U.S. in 2012. Diabetes Care 2013;36:1033–46. [10] Lavigne JE, Phelps CE, Mushlin A, Lednar WM. Reductions in individual work productivity associated with type 2 diabetes mellitus. Pharmacoeconomics 2003;21:1123–34. [11] Ng YC, Jacobs P, Johnson JA. Productivity losses associated with diabetes in the US. Diabetes Care 2001;24:257–61. [12] Izadi N, Malek M, Aminian O, Saraei M. Medical risk factors of diabetes mellitus among professional drivers. J Diabetes Metab Disord 2013;12:23. [13] Wilmot EG, Edwardson CL, Achana FA, Davies MJ, Gorely T, Gray LJ, et al. Sedentary time in adults and the association with
462
[14]
[15] [16]
[17]
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
[18] [19]
R. A. Brook et al. diabetes, cardiovascular disease and death: systematic review and meta-analysis. Diabetologia 2012;55:2895–905. Sieber WK, Robinson CF, Birdsey J, Chen GX, Hitchcock EM, Lincoln JE, et al. Obesity and other risk factors: the National Survey of U.S. Long-Haul Truck Driver Health and Injury. Am J Ind Med 2014;57:615–26. Soule JB, Egede LE. Challenges of managing diabetes in commercial truck drivers. Clin Diabetes 2007;25:110–11. Federal Motor Carrier Safety Administration, US Department of Transportation. Medical examination report for commercial driver fitness determination. http://www.fmcsa.dot.gov/regulations/medical/ medical-examination-report-commercial-driver-fitness-determination. Accessed on 18 November 2014. Federal Motor Carrier Safety Administration, US Department of Transportation. 49 CFR Ch. III (10-1-11 edition). Subpart E – physical qualifications and examinations. 2011. http://www.gpo. gov/fdsys/pkg/CFR-2011-title49-vol5/pdf/CFR-2011-title49-vol5sec391-41.pdf. Accessed 7 February 2014. Brook RA, Kleinman NL, Rajagopalan K. Employee costs before and after treatment initiation for bipolar disorder. Am J Manag Care 2007;13:17–86. Brook RA, Kleinman NL, Su J, Corey-Lisle PK, Iloeje UH. Absenteeism and productivity among employees being treated for hepatitis C. Am J Manag Care 2011;17:657–64.
Postgrad Med, 2015; 127(5):455–462
[20] Centers for Disease Control and Prevention. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). http://www.cdc.gov/nchs/icd/icd9cm.htm. Accessed 7 February 2014. [21] US Department of Labor, Bureau of Labor Statistics. CPI detailed report – data for June 2012. http://www.bls.gov/cpi/cpid1206.pdf. Accessed 7 February 2014. [22] Lee LJ, Yu AP, Cahill KE, Oglesby AK, Tang J, Qiu Y, et al. Direct and indirect costs among employees with diabetic retinopathy in the United States. Curr Med Res Opin 2008;24:1549–59. [23] Rein DB, Zhang P, Wirth KE, Lee PP, Hoerger TJ, McCall N, et al. The economic burden of major adult visual disorders in the United States. Arch Ophthalmol 2006;124:1754–60. [24] National Eye Institute, US Department of Health and Human Services. Diabetic retinopathy. What you should know. https:// www.nei.nih.gov/sites/default/files/health-pdfs/diabeticretino.pdf. Accessed 7 February 2014. [25] Klein R, Klein BE. Relation of glycemic control to diabetic complications and health outcomes. Diabetes Care 1998;21: C39–43. [26] Schmittdiel JA, Brown SD, Neugebauer R, Adams SR, Adams AS, Wiley D, et al. Health-plan and employer-based wellness programs to reduce diabetes risk: the Kaiser Permanente Northern California NEXT-D Study. Prev Chronic Dis 2013;10:E15.
ORIGINAL RESEARCH
United States comparative costs and absenteeism of diabetic ophthalmic conditions Richard A. Brook1, Nathan L. Kleinman2, Sunil Patel3, Jim E. Smeeding4, Ian A. Beren2 & Adam Turpcu5 The JeSTARx Group, Newfoundland, NJ, USA, 2HCMS Group, Cheyenne, WY, USA, 3The Retina Research Group, Abilene, TX, USA, 4The JeSTARx Group, Dallas, TX, USA, and 5Genentech, Inc., South San Francisco, CA, USA
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
1
Abstract
Keywords:
Objective: This retrospective cohort study examined the impact of diabetic macular edema (DME), diabetic retinopathy (DR), or diabetes on annual health benefit costs and absenteeism in US employees. Methods: Claims data from 2001 to 2012 was extracted from the Human Capital Management Services Group Research Reference Database on annual direct/indirect health benefit costs and absences for employees aged ‡ 18 years. Employees with DME, DR, or diabetes were identified by International Classification of Diseases, Ninth Revision, Clinical Modification codes. Employees were divided into two groups, drivers or nondrivers, and examined in separate analyses. For drivers and nondrivers, the DME, DR, and diabetes cohorts were compared with their respective control groups (without diabetes). Two-part regression models controlled for demographics and job-related characteristics. Results: A total of 39,702 driver and 426,549 nondriver employees were identified as having ‡ 1 year’s continuous health plan enrollment. Direct medical costs for drivers with DME, DR, or diabetes were $6470, $8021, and $5102, respectively (>2.8 times higher and statistically significant compared with driver controls). Nondrivers with DME and DR incurred significantly higher sick leave and short-term disability costs compared with the nondrivers with diabetes and nondriver controls. In drivers with DME, the majority of days of absence were for short- and long-term disability (12.41 and 11.43 days, respectively). In drivers with DR, the majority of days of absence were for short-term disability (10.70 days). In nondrivers with DME and nondrivers with DR, the majority of days of absence were for sick leave (5.74 and 4.93 days, respectively) and short-term disability (5.08 and 4.93 days, respectively). Conclusion: DME and DR are associated with substantial direct medical cost and absenteeism in this real-world sample of medically insured employees. This research highlights the negative impact of DME and DR on annual costs and absenteeism and may assist employers in assessing the impact of these conditions on employees.
Cost studies, macular edema, outcomes studies, retinopathy
Introduction Diabetes mellitus affects ~ 8.3% (25.8 million) of the US population, of whom an estimated 7.0 million are undiagnosed [1]. Diabetic retinopathy (DR) is the main cause of new blindness cases in adults aged 20 to 74 years in the United States [1]. Diabetic macular edema (DME), which can occur at any stage of DR, is diagnosed when retinal thickening involves or approaches the central macula [2,3]. Consistent with the increasing prevalence of diabetes in the United States [4], the number of adults with diagnosed diabetes who reported vision impairment (defined as visual difficulty despite using corrective lenses) increased from 2.7 to 4.0 million from 1997 to 2011 [5]. DR was reported in 4.2 million (28.5%) people with diabetes aged ‡ 40 years in 2005 to 2008. Of this group, 655,000 (4.4%) had advanced DR and were consequently at risk of severe vision loss [1,6]. Increasing prevalence of diabetes in younger adults (aged
History Received 11 September 2014 Accepted 21 November 2014 Published online 31 December 2014
20–44 years) [1] suggests that these trends are potentially relevant to employers [7]. Diabetes is associated with a significant economic burden in the United States because of increased health resource use [8,9] and loss of productivity [7,10,11]. The American Diabetes Association reported that, in 2012, the total estimated direct medical costs and costs related to reduced productivity associated with diagnosed diabetes were $176 billion and $69 billion, respectively [9]. Moreover, medical expenditures in individuals with diagnosed diabetes were found to be ~ 2.3 times higher than for individuals without diabetes [9]. However, the contribution of retinopathy to the cost and burden of diabetes is unknown. Certain aspects of the lifestyle of commercial drivers, such as the sedentary nature of the occupation and limited healthy food choices on the road, are associated with increased risk of developing diabetes [12,13]. The National Survey of Long-Haul Truck Driver Health and Injury,
Correspondence: Richard A. Brook, MS, MBA, Retrospective Analysis, The JeSTARx Group, 18 Hirth Drive, Newfoundland, NJ 07435-1710, USA. Tel.: +1 973 208 8621. Fax: +1 973 954 2968. E-mail: [email protected] 2015 Informa UK Ltd.
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
456
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
conducted by the National Institute for Occupational Safety and Health, found that the self-reported prevalence of diabetes was 14.4% among US long-haul truck drivers but was 6.7% in working adults (based on the 2010 National Health Interview Survey) [14]. Commercial drivers may be particularly susceptible to develop DR and DME, as job conditions may make diabetes management difficult, and they may be unwilling to start insulin therapy owing to regulations surrounding commercial vehicle operation and insulin use [15]. However, visual acuity (Snellen chart) of at least 20/40 or better in each eye (with or without correction) is required to qualify for a commercial driver’s license and to maintain employment [16,17]. A literature search conducted in early 2013 using the keywords diabetic retinopathy, diabetic macular edema, drivers, costs, and absenteeism failed to identify any publications. Consequently, this study may be the first to examine the impact of DR and DME on employer costs in a cohort of commercial drivers in the United States. The present study was designed to quantify and compare the burden of DME and DR in US employees, including commercial drivers for whom good vision is required to maintain employment [17]. This research compared annual absence days, annual health benefit costs, and the likelihood of using health benefits in employees with DME, DR, or diabetes versus controls without diabetes.
Materials and methods Study design This retrospective analysis utilized the Human Capital Management Services Group Research Reference Database (RRDb; Cheyenne, WY, USA). Employees were sourced from > 20 geographically dispersed US employers. Demographic, payroll, health insurance, health-related work absence, and workers’ compensation data were maintained in the RRDb for all employees. This database, containing > 1.7 million deidentified individuals employed at some time between 1 January 2001 and 30 June 2012, has been used in previous research and is a unique data source that classifies employees by job type allowing for the identification of drivers, and records actual absence costs and time by type of benefit [18,19]. Employees identified as commercial drivers were designated as “drivers” in this analysis. All other employees, with jobs unrelated to commercial driving, were identified as “nondrivers.” These two populations were examined in separate parallel analyses; this study was not designed to directly compare drivers and nondrivers. Based on International Classification of Diseases, Ninth Revision, Clinical Table 1. ICD-9-CM codes for DME, DR, and diabetes. Condition
ICD-9-CM code [15] (description)
DME
362.07 (DME) 362.53 (Cystoid macular degeneration) 362.83 (Retinal edema) 362.0x (DR) 250.xx
DR Diabetes
Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification.
Modification (ICD-9-CM) codes (Table 1) [20], drivers and nondrivers were further divided into the following groups: employees with DME, employees with DR but without DME, employees with diabetes without DME or DR, and controls. The control group for the drivers consisted of drivers without DME, DR, or diabetes. The control group for the nondrivers consisted of nondrivers without DME, DR, or diabetes. For driver and nondriver analyses, the DME, DR, diabetes, and control cohorts were compared. The index date for the DME and DR cohorts was the first claim date for the respective condition. For the diabetes and control cohorts, the index date was the average index dates of the DME and DR cohorts by company. All included individuals were required to have continuous employment and health plan enrollment for a 12-month study period after their index date, during which costs, days of absence, and likelihood outcomes were measured. Only individuals with the required eligibility and benefit were included in the analyses for that benefit. All subjects were assigned to a region of the country based on the first digit of their ZIP code. Outcomes of this study included days of absence, costs, and likelihood of using health benefits. Days of absence were categorized as sick leave, short- and long-term disability, and workers’ compensation (the workers’ compensation variable also included medical claims specifically paid under the individual employee’s workers’ compensation benefit). Annual health benefit cost components were derived from a number of different benefits provided by employers, including direct components (medical/healthcare costs, prescription costs) and indirect components (payments made to the employee for sick leave, short- and long-term disability, and workers’ compensation). The likelihood of using health benefits was defined as the percentage of each driver and nondriver population using each direct and indirect benefit type during the study period. All annual direct and indirect cost outcomes were inflation-adjusted to June 2012 US dollars by the appropriate component of the Bureau of Labor Statistics Consumer Price Index (CPI) [21]. The medical care services CPI and prescription drugs CPI were used to adjust medical cost components and prescription drug components, respectively, and the allconsumer-goods CPI was utilized for all other cost elements. Statistical analysis Nonregression-based statistical tests were used to compare the demographics of cohorts for the analyses of drivers and nondrivers. Statistically significant differences in continuous demographic characteristics and between dichotomous demographic variables were determined using Tukey-Kramer ttests (p < 0.05) and Marascuilo adjusted c2 tests, respectively. Two-part regression modeling was used to estimate each outcome. First, logistic regression was used to model the likelihood of an outcome > 0. Generalized linear models were then used to estimate costs or days of absence for the portion of the population with an outcome > 0. Independent variables, including demographics, job-related information, Charlson Comorbidity Index (CCI), and US region, were used to control for confounding factors between cohorts. Sick leave
DME, DR, and diabetes: impact on US employees
DOI: 10.1080/00325481.2014.994468
457
All employees n = 903,255
Employees with 12 months of continuous employment and health plan enrollment n = 466,251
Nondrivers n = 426,549
Drivers n = 39,702
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Drivers (with diabetes) n = 1320
Nondrivers (with diabetes) n = 20,618
Drivers controls (without diabetes) n = 38,382
Nondrivers controls (without diabetes) n = 405,931
DME n = 538
DME n = 26
DR (without DME) n = 1441
DR (without DME) n = 103
Diabetes (without DME or DR) n = 18,639
Diabetes (without DME or DR) n = 1191
Figure 1. Flow chart showing patient identification. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
models also controlled for employer, to account for variability between employers for this benefit type. All models utilized stepwise selection methods to determine the variables included in the final model. Model outcomes were considered significant at p < 0.05. All models were generated using SAS version 9.2 for Microsoft Windows (SAS Institute, Inc., Cary, NC, USA).
Results Descriptive statistics Overall, 466,251 employees with 12 months of continuous employment and health plan enrollment were identified from 903,255 employees who had some health plan enrollment (Figure 1). Descriptive statistics for the 39,702 commercial drivers and 426,549 nondrivers are shown in Table 2. In the driver analysis, the DME, DR, and diabetes cohorts were, on average, 6 years older compared with the driver controls. In the analysis of nondrivers, the DME, DR, and diabetes cohorts were 8 years older or more compared with the nondriver controls (all comparisons, p < 0.05). In both the driver and nondriver analyses, the DME, DR, and diabetes cohorts had significantly higher CCI scores versus their respective control cohorts (p < 0.05). However, the driver and nondriver DR cohorts had higher CCI score versus their respective DME cohorts (p < 0.05 for drivers; p < 0.05 for nondrivers). Driver and nondriver employees were distributed throughout the United States. Drivers were more concentrated in the eastern half of the United States, whereas nondrivers were more concentrated in the western region of the United States.
Annual absence days In the analysis of drivers, total annual days of absence were greatest in the drivers with DME (27.0 days), followed by the drivers with DR (16.1 days), drivers with diabetes (15.0 days), and driver controls (8.2 days) (Figure 2A). Most employee absenteeism among the drivers with DME and drivers with DR cohorts was in the short-term disability category (12.41 and 10.70 days, respectively). Also, compared with controls, drivers with diabetes had more sick leave days (p < 0.0001) and short- and long-term disability days (p < 0.0001 and p = 0.003, respectively) but had fewer workers’ compensation days (p = 0.003). Differences in absence categories among the drivers with DME were not significantly different compared with drivers with DR, drivers with diabetes, or driver controls. Drivers with DR were absent from work for more short-term disability days versus controls (p = 0.0054), and for fewer long-term disability days compared with the drivers with diabetes (p = 0.0024). No significant differences in total annual days of absence were reported between the drivers with DME and drivers with DR for all absence categories. In the analysis of nondrivers, total annual days of absence were greatest in the nondrivers with DME (13.8 days), followed by the nondrivers with DR (12.7 days), nondrivers with diabetes (8.9 days), and nondriver controls (5.4 days) (Figure 2B). The majority of days of absence in the nondrivers with DME and nondrivers with DR were for sick leave (5.74 and 4.93 days, respectively) and short-term disability (5.08 and 4.93 days, respectively). Additionally, the number of sick leave days and short-term disability days were significantly higher for nondrivers with DME and
458
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
Table 2. Descriptive statistics for the driver and nondriver populations. Drivers DME (n = 26)
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Variable Age, index Dx date, y Tenure, index Dx date, y Female, % Marital status, % Married Not married Missing Race/ethnicity, % White Black Hispanic Other Missing Exempt, % Annual salary, $d Full time, % Charlson Comorbidity Index score
Mean
DR (n = 103) SE
a
Mean a,b
Diabetes (n = 1191) SE
Mean a
Control (n = 38,382) SE
Mean
SE
46.87 5.97b,c 3.8a
1.72 1.12 3.8
48.93 11.40a 3.9a
0.93 0.89 1.9
46.53 10.78a 5.1a
0.25 0.23 0.6
39.05 8.41 17.0
0.04 0.03 0.2
65.4 30.8 3.8
9.5 9.2 3.8
66.0 16.5a 17.5a
4.7 3.7 3.8
65.7a 26.5a 7.7a
1.4 1.3 0.8
59.8 37.6 2.6
0.3 0.2 0.1
61.5 23.1 15.4 0.0a,b 0.0a,b 0.0a 35,463 96.2a 2.42a,b,c
9.7 8.4 7.2 0.0 0.0 0.0 1383 3.8 0.23
38.8a 39.8a 16.5 2.9 1.9 0.0a 36,569 96.1a,b 3.26a,b
4.8 4.8 3.7 1.7 1.4 0.0 690 1.9 0.12
51.5a 31.6a 14.2 2.0 0.8 0.3 37,164 89.3a 1.58a
1.4 1.3 1.0 0.4 0.3 0.1 211 0.9 0.03
62.8 21.6 12.7 2.5 0.4 0.5 35,016 78.2 0.09
0.2 0.2 0.2 0.1 0.0 0.0 244 0.2 0.0
Nondrivers DME (n = 538)
DR (n = 1441)
Diabetes (n = 18,639)
Control (n = 405,931)
Variable
Mean
SE
Mean
SE
Mean
SE
Mean
SE
Age, index Dx date, ye Tenure, index Dx date, y Female, % Marital status, % Married Not married Missing Race/ethnicity, % White Black Hispanic Other Missing Exempt, % Annual salary, $f Full time, % Charlson Comorbidity Index score
50.43a,b 11.74a 46.7
0.41 0.46 2.2
49.14a 12.20a 46.5
0.27 0.28 1.3
48.90a 12.22a 48.7
0.07 0.07 0.4
40.83 8.87 48.3
0.02 0.01 0.1
36.8 24.2a,b 39.0a,b
2.1 1.8 2.1
38.0b 27.6a,b 34.5a,b
1.3 1.2 1.3
42.1a 31.4a 26.4a
0.4 0.3 0.3
36.4 33.0 30.5
0.1 0.1 0.1
33.1 13.2 8.2 3.0 42.6 27.7a 50,160 97.8a,b 2.89a,b,c
2.0 1.5 1.2 0.7 2.1 1.9 1360 0.6 0.07
31.8 11.5 11.4a,b 3.5 41.9a 30.7a 54,680 97.4a,b 3.41a,b
1.2 0.8 0.8 0.5 1.3 1.2 933 0.4 0.03
33.3a 13.7a 7.5a 3.9 41.6a 29.3a 52,557a 95.9a 1.72a
0.3 0.3 0.2 0.1 0.4 0.3 250 0.1 0.01
35.0 9.3 5.4 3.6 46.7 36.2 56,972 93.1 0.15
0.1 0.0 0.0 0.0 0.1 0.1 143 0.0 0.0
p < 0.05 versus control cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. b p < 0.05 versus diabetes cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. c p < 0.05 versus DR cohort, adjusting for multiple comparisons between cohorts using the Tukey-Kramer procedure for continuous variables or the Marascuilo procedure for binary variables. d Twelve drivers in the control cohort were missing salary information. e Three nondrivers in the diabetes cohort and 15 employees in the control cohort were missing age values. f There were 6, 25, 433, and 12,942 nondriver employees in the DME, DR, diabetes, and control cohorts, respectively, who were missing salary information. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy; Dx = Diagnosis; SE = Standard error. a
nondrivers with DR compared with those of nondrivers with diabetes and nondriver controls. Nondrivers with DR missed fewer days in the category of workers’ compensation compared with the nondrivers with diabetes and nondriver controls (both p < 0.0001). Nondrivers with diabetes had a greater number of days of absence compared with nondriver controls for all absence categories (p < 0.006). No significant differences were reported in any absence category between the nondrivers with DME and nondrivers with DR.
Annual health benefit costs Total annual health benefit costs in drivers with DME, DR, or diabetes were $12,511, $12,163, and $8785, respectively. In the driver controls, the total annual health benefit cost was $4149 (Figure 3A). Drivers with DME, DR, or diabetes had significantly higher direct medical costs (medical and prescription drug) compared with driver controls (all comparisons, p < 0.02). In drivers with DR, significantly lower longterm disability costs were reported versus the drivers with
DME, DR, and diabetes: impact on US employees
DOI: 10.1080/00325481.2014.994468
Drivers
a
b
Nondrivers Workers’ compensation Long-term disability Short-term disability Sick leave
30
30 0.77
25
25 20
459
20
11.43
15
2.94
Days
Days
0.66* 1.40†
10.70†
1.94 6.10‡
5.08‡,§
1.09
5.74*,‡
DME (27.04)
1.79
1.67
DR (16.08)
Diabetes (14.98)
Control (8.16)
4.93*
0.31 0.30 1.81
3.31‡
4.93*,‡
4.37‡
3.03
DR (12.65)
Diabetes (8.90)
Control (5.44)
0 DME (13.77)
Driver cohort (total absence days)
Nondriver cohort (total absence days)
Figure 2. Annual days of absence are shown for (a) drivers and (b) nondrivers. Within each analysis only: *p < 0.01 versus diabetes cohort; † p < 0.01 versus control cohort; ‡p < 0.0001 versus control cohort; §p < 0.05 versus diabetes cohort; kp < 0.0001 versus diabetes cohort; ¶ p < 0.05 versus control cohort. The n values for the driver and nondriver categories are shown in Table 3. Copyright 2013 American Diabetes Association. From Diabetes,Vol. 62, Suppl. 1; 2013. Modified by permission of the American Diabetes Association. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
diabetes (p = 0.0002) and driver controls (p = 0.0227). Workers’ compensation costs also were lower among the drivers with DR compared with the drivers with diabetes and driver controls (p = 0.0050 and p < 0.0001, respectively). However, for short-term disability costs, drivers with DR were significantly more expensive than the driver controls (p = 0.0037), and the difference in costs approached significance compared with the drivers with diabetes (p = 0.0548). For the drivers with DME, indirect costs did not differ significantly from the drivers with DR, drivers with diabetes, or driver controls, and Drivers
b Workers’ compensation Long-term disability Short-term disability Sick leave Prescription Medical
$20,000 $18,000 $740 $687 $1086 $389
$16,000 $14,000 $12,000 $10,000 $8000
$4000
$14,000
$75§,¶ $240† $760‡,§ $924§,II
$3733§,¶ $3338§,¶ $229* $66* $483§ $830§
$12,000 $905* $214* $497§ $348§
$8021§,II $5102§
$10,000 $8000
$1266 $57 $267 $277 $490
$1720§
$2000
$18,000 $16,000
$3139*
$6470†
Nondrivers $119II $136 $705*,II $1009‡,§
$20,000
$415‡,§ $20†,‡ $970* $299
$2438§
$6000
no significant differences were reported in any cost category between drivers with DME or drivers with DR. For nondrivers, total annual health benefit costs were $17,433, $19,280, $10,926, and $5258 for the nondrivers with DME, nondrivers with DR, nondrivers with diabetes, and nondrivers controls, respectively (Figure 3B). All cost components in the nondrivers with diabetes cohort were significantly higher than the nondriver controls (p < 0.001). Sick leave and short-term disability costs were significantly higher among the nondrivers with DME and nondrivers with DR cohorts versus
Costs
a
Costs
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
0
2.12‡
0.42† 0.78†
,‡
5
3.45 2.42
0.10‡,II 2.69¶
10
12.41
5
0.15 2.80
5.37†
10
15
$6000
$13,547§,¶
$2502§
$4000
$6815§
$2000
$3326
$1792 $0
$189 $29 $259 $606 $849
$12,125§,¶
$0 DME ($12,511)
DR ($12,163)
Diabetes ($8785)
Driver cohort (total costs)
Control ($4149)
DME ($17,433)
DR ($19,280)
Diabetes ($10,926)
Control ($5258)
Nondriver cohort (total costs)
Figure 3. Annual direct and indirect health benefit costs are shown for (a) drivers and (b) nondrivers. Within each analysis only: *p < 0.01 versus control cohort; †p < 0.05 versus control cohort; ‡p < 0.01 versus diabetes cohort; §p < 0.0001 versus control cohort; kp < 0.05 versus diabetes cohort; ¶ p < 0.0001 versus diabetes cohort. The n values for the driver and nondriver categories are shown in Table 3. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy.
460
R. A. Brook et al.
Postgrad Med, 2015; 127(5):455–462
Table 3. Likelihood of using health benefits during the 12 months following the index date for drivers and nondrivers. DME Employee cost category
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
Drivers Medical Prescription drug Sick leave Short-term disability Long-term disability Workers’ compensation Nondrivers Medical Prescription drug Sick leave Short-term disability Long-term disability Workers’ compensation
DR
Diabetes
Control
n
Likelihood, %
n
Likelihood, %
n
Likelihood, %
n
Likelihood, %
26 26 26 25 25 26
100.0a 100.0a 69.6 11.4 5.8a 12.5
103 103 99 97 96 103
95.0b 93.2b 50.3 16.8a 1.6 8.7
1191 1191 1159 1067 1032 1185
97.5b 96.6b 59.7b 10.9b 1.8b 10.9
38,382 38,382 37,792 30,738 28,470 38,332
65.2 65.4 52.4 6.8 0.5 11.4
538 538 243 289 319 479
99.8b 97.0b 77.7b,d,e 13.4b,c 0.6f 1.5
1441 1441 635 811 964 1277
100.0b,c 97.9b,c 70.0b 11.3b,c 0.7b,c 2.0
18,639 18,639 7115 10,013 12,437 16,655
99.6b 96.9b 66.9b 9.1b 0.3b 2.6a
405,931 405,931 156,101 230,080 274,749 356,291
82.2 76.5 60.6 5.0 0.1 2.2
p < 0.01 versus control cohort. p < 0.0001 versus control cohort. c p < 0.05 versus diabetes cohort. d p < 0.05 versus DR cohort. e p < 0.001 versus diabetes cohort. f p < 0.05 versus control cohort. Abbreviations: DME = Diabetic macular edema; DR = Diabetic retinopathy. a
b
the nondrivers with diabetes and nondriver controls (all comparisons, p < 0.05). No significant differences in any annual cost component were reported for comparisons between nondrivers with DME or DR. Workers’ compensation costs were significantly lower among the nondrivers with DME versus the nondrivers with diabetes (p = 0.0165), and in the nondrivers with DR cohort versus the nondrivers with diabetes and nondriver controls (both p < 0.0001). Long-term disability costs were significantly higher among the nondrivers with DR versus nondriver controls (p = 0.0218), but not compared with the nondrivers with diabetes (p = 0.0609). Likelihood of using health benefits In the driver population, the likelihood of using direct and indirect benefits was similar between the drivers with DME and drivers with DR, and for both versus the drivers with diabetes cohort (Table 3). Drivers with DR, DME, or diabetes were significantly more likely to use direct medical components (medical and prescription drug) compared with driver controls (all comparisons, p < 0.0003). The likelihood of using the long-term disability benefit was significantly higher among the drivers with DME versus driver controls (p = 0.0003) and, of all four cohorts, drivers with DME had the highest likelihood of using this benefit (drivers with: DME, 5.8%; DR, 1.6%; diabetes, 1.8%). Nondrivers with DR were more likely to use medical, prescription drug, and short- and long-term disability compared with the nondrivers with diabetes and nondriver controls (p < 0.04), and more likely to use sick leave compared with the nondriver control (p < 0.0001; Table 3). Nondrivers with DME were more likely to use the sick leave benefit compared with nondrivers with DR (p = 0.0241), with diabetes (p = 0.0004), or with controls (p < 0.0001). Compared with nondriver controls, the nondriver DME, DR, and diabetes cohorts were each more likely to use individual direct and indirect benefits (p < 0.04), with the exception of workers’
compensation, which was significantly higher in only the nondrivers with diabetes versus nondriver controls (p < 0.003).
Discussion This analysis demonstrates the significant impact of DME and DR on employee direct and indirect costs across a range of benefits, as well as the detrimental impact on employee absenteeism using actual, not proxied, data (as used in previous studies) [22]. In general, drivers and nondrivers with DME or DR were more expensive to manage, missed more work days, and were more likely to use health benefits than drivers and nondrivers with diabetes or their respective control groups. In separate analyses of drivers and nondriver employees, this study found that differences in days of absence, costs, and likelihood of using health benefits did not differ significantly between the DR and DME cohorts. This may be owing to the relatively small size of the DME and DR cohorts and lack of severity measures within the database. It is also possible that more severely affected patients with DME exit the workforce entirely and rely on government-sponsored sources of benefits (Medicare, Medicaid, or federal disability programs) instead of employer-sponsored benefits. The economic burden of diabetes and diabetes-related vision disorders has been previously reported [22,23]. A 2008 claims analysis by Lee et al. reported that DR employees had higher mean annual total costs compared with non-DR employees (who had diabetes but no DR) [22]. Employees with DME had 74.8% higher mean costs versus employees without DME. Indirect costs due to absenteeism were also higher in employees with DR compared with nonDR employees ($1640 vs. $1218, difference = $422; p < 0.0001). It is important to note that Lee et al. estimated absenteeism by assuming that 1 day of hospitalization resulted in the loss of 1 day of work and an outpatient visit resulted in the loss of one-half day of work [22].
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
DOI: 10.1080/00325481.2014.994468
Unlike clinical trials, which enroll a well-defined patient population, this retrospective cohort study allows for the examination of the impact of diabetic eye disease on costs and days of absence in a real-world population. However, this study has several limitations. In both the driver and nondriver analyses, the mean ages of the DME, DR, and diabetes cohorts were significantly higher compared with their respective control cohorts. Since age is approximately normally distributed, we calculated that ~95% of each study cohort was within two standard deviations of the cohort mean. The large control group sample sizes provided sufficient overlap in the distribution of ages between groups, and adjusting for age in the multivariate model is a sufficient control measure. DR and DME have been shown to worsen over time as more blood vessels are involved/affected [24]. However, as with most claims databases, data on retinal disease stage, severity, or duration is not captured. The present study was not designed to compare driver with nondriver populations. Although all available data were included in the regression analysis, the driver and nondriver populations were modeled separately. Therefore, variances and standard errors were based on each individual study population, and demographic differences or other factors were not controlled for between driver and nondriver populations. This study was limited to reporting direct and indirect costs for employees with DME, DR, or diabetes and did not assess the impact of these conditions on caregivers (i.e., employees with spouses and/or other family members with DME or DR). Since individuals were identified based on ICD-9-CM codes, control cohorts may have included individuals with undiagnosed diabetes. Some observations may be attributed to the limited sample size of the DR and DME disease cohorts within the driver population. For example, the CCI data revealed that the drivers with DR ranked higher than the drivers with DME; this observation is likely to be a result of the small sample size of drivers with DME. In the present study, we observed similar trends in costs, absences, and likelihood variables in employees with DME or DR within the driver and nondriver analyses. Given that the incidence of diabetic complications is directly related to poor glycemic control [25], these findings may suggest that uncontrolled diabetes indirectly leads to increased costs to employers. Employee wellness programs can provide health coaching, with some larger companies (> 750 employees) offering health risk assessment or screening [26]. In 2012, employee absenteeism accounted for $5 billion of the $69 billion attributed to diabetes-related reduced productivity in 2012 (25 million workdays) [9]; thus, the implementation of such plans may be of financial benefit to employers. This may be particularly necessary for drivers with DME as their annual absences totaled ~ 30 days.
Conclusion In this real-world sample of medically insured employees, DME and DR were associated with substantial direct medical costs and absenteeism. These data provide employers
DME, DR, and diabetes: impact on US employees
461
with important information necessary to assess the impact of DME and DR on employees.
Acknowledgments The authors thank Suzanne Douthwaite, PhD, and Rebecca Jarvis, PhD, of Envision Scientific Solutions for assistance with manuscript preparation, which was supported by Genentech, Inc.
Declaration of interest This study was sponsored by Genentech, Inc. Adam Turpcu is an employee of and owns stock in Genentech, Inc. Richard A. Brook, Jim E. Smeeding, Nathan L. Kleinman, and Ian A. Beren received consulting fees from Genentech, Inc. through their respective employers. Nathan L. Kleinman and Ian A. Beren were employees of HCMS Group, Cheyenne, WY, United States, at the time the study was conducted. Sunil Patel received grants to his institution from Alcon, Allergan, Alimera, Genentech, Inc., Pfizer, Ophthotech, and Regeneron; consulting fees and fees for participation in review activities or honorarium from Genentech, Inc.; consulting fees from Allergan and Ophthotech; fees for participation in speakers bureaus from Alcon; and owns stock/ stock options in Ophthotech.
References [1] Centers for Disease Control and Prevention. National diabetes fact sheet: national estimates and general information on diabetes and prediabetes in the United States. 2011. http://www.cdc. gov/diabetes/pubs/pdf/ndfs_2011.pdf. Accessed 7 February 2014. [2] Bandello F, Battaglia Parodi M, Lanzetta P, Loewenstein A, Massin P, Menchini F, et al. Diabetic macular edema. Dev Ophthalmol 2010;47:73–110. [3] Fenwick EK, Pesudovs K, Khadka J, Dirani M, Rees G, Wong TY, et al. The impact of diabetic retinopathy on quality of life: qualitative findings from an item bank development project. Qual Life Res 2012;21:1771–82. [4] Centers for Disease Control and Prevention. Increasing prevalence of diagnosed diabetes–United States and Puerto Rico, 1995-2010. MMWR Morb Mortal Wkly Rep 2012;61:918–21. [5] Centers for Disease Control and Prevention, National Center for Health Statistics, Health Interview Survey. Number (in millions) of adults aged 18 years or older with diagnosed diabetes reporting visual impairment, United States, 1997–2011. http://www.cdc.gov/diabetes/statistics/visual/fig1.htm. Accessed 7 February 2014. [6] Zhang X, Saaddine JB, Chou CF, Cotch MF, Cheng YJ, Geiss LS, et al. Prevalence of diabetic retinopathy in the United States, 2005-2008. JAMA 2010;304:649–56. [7] Tunceli K, Bradley CJ, Nerenz D, Williams LK, Pladevall M, Elston Lafata J. The impact of diabetes on employment and work productivity. Diabetes Care 2005;28:2662–7. [8] American Diabetes Association. Economic costs of diabetes in the US in 2007. Diabetes Care 2008;31:596–615. [9] Yang W DT, Halder P, Gallo P, Kowal SL, Hogan PF. Economic costs of diabetes in the U.S. in 2012. Diabetes Care 2013;36:1033–46. [10] Lavigne JE, Phelps CE, Mushlin A, Lednar WM. Reductions in individual work productivity associated with type 2 diabetes mellitus. Pharmacoeconomics 2003;21:1123–34. [11] Ng YC, Jacobs P, Johnson JA. Productivity losses associated with diabetes in the US. Diabetes Care 2001;24:257–61. [12] Izadi N, Malek M, Aminian O, Saraei M. Medical risk factors of diabetes mellitus among professional drivers. J Diabetes Metab Disord 2013;12:23. [13] Wilmot EG, Edwardson CL, Achana FA, Davies MJ, Gorely T, Gray LJ, et al. Sedentary time in adults and the association with
462
[14]
[15] [16]
[17]
Postgraduate Medicine Downloaded from informahealthcare.com by Emory University on 08/25/15 For personal use only.
[18] [19]
R. A. Brook et al. diabetes, cardiovascular disease and death: systematic review and meta-analysis. Diabetologia 2012;55:2895–905. Sieber WK, Robinson CF, Birdsey J, Chen GX, Hitchcock EM, Lincoln JE, et al. Obesity and other risk factors: the National Survey of U.S. Long-Haul Truck Driver Health and Injury. Am J Ind Med 2014;57:615–26. Soule JB, Egede LE. Challenges of managing diabetes in commercial truck drivers. Clin Diabetes 2007;25:110–11. Federal Motor Carrier Safety Administration, US Department of Transportation. Medical examination report for commercial driver fitness determination. http://www.fmcsa.dot.gov/regulations/medical/ medical-examination-report-commercial-driver-fitness-determination. Accessed on 18 November 2014. Federal Motor Carrier Safety Administration, US Department of Transportation. 49 CFR Ch. III (10-1-11 edition). Subpart E – physical qualifications and examinations. 2011. http://www.gpo. gov/fdsys/pkg/CFR-2011-title49-vol5/pdf/CFR-2011-title49-vol5sec391-41.pdf. Accessed 7 February 2014. Brook RA, Kleinman NL, Rajagopalan K. Employee costs before and after treatment initiation for bipolar disorder. Am J Manag Care 2007;13:17–86. Brook RA, Kleinman NL, Su J, Corey-Lisle PK, Iloeje UH. Absenteeism and productivity among employees being treated for hepatitis C. Am J Manag Care 2011;17:657–64.
Postgrad Med, 2015; 127(5):455–462
[20] Centers for Disease Control and Prevention. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). http://www.cdc.gov/nchs/icd/icd9cm.htm. Accessed 7 February 2014. [21] US Department of Labor, Bureau of Labor Statistics. CPI detailed report – data for June 2012. http://www.bls.gov/cpi/cpid1206.pdf. Accessed 7 February 2014. [22] Lee LJ, Yu AP, Cahill KE, Oglesby AK, Tang J, Qiu Y, et al. Direct and indirect costs among employees with diabetic retinopathy in the United States. Curr Med Res Opin 2008;24:1549–59. [23] Rein DB, Zhang P, Wirth KE, Lee PP, Hoerger TJ, McCall N, et al. The economic burden of major adult visual disorders in the United States. Arch Ophthalmol 2006;124:1754–60. [24] National Eye Institute, US Department of Health and Human Services. Diabetic retinopathy. What you should know. https:// www.nei.nih.gov/sites/default/files/health-pdfs/diabeticretino.pdf. Accessed 7 February 2014. [25] Klein R, Klein BE. Relation of glycemic control to diabetic complications and health outcomes. Diabetes Care 1998;21: C39–43. [26] Schmittdiel JA, Brown SD, Neugebauer R, Adams SR, Adams AS, Wiley D, et al. Health-plan and employer-based wellness programs to reduce diabetes risk: the Kaiser Permanente Northern California NEXT-D Study. Prev Chronic Dis 2013;10:E15.
