Abstract
Objectives. We evaluated the risk of work disability (sick leave and disability pension) before and after diabetes diagnosis relative to individuals without diabetes during the same time period, as well as the trajectory of work disability around the diagnosis.
Methods. This Swedish population-based cohort study with register data included 14 428 individuals with incident diabetes in 2006 and 39 702 individuals without diabetes during 2003 to 2009.
Results. Work disability was substantially higher among people with diabetes (overall mean = 95 days per year over the 7 years, SD = 143) than among those without diabetes (mean = 35 days, SD = 95). The risk of work disability was slightly higher after diabetes diagnosis than before and compared with the risk of those without diabetes. The trajectory of work disability was already increasing before diagnosis, increased even more at the time of diagnosis, and leveled off after diagnosis. Individual sociodemographic characteristics and comorbid conditions contributed both to the risk and to the trajectory of work disability.
Conclusions. Although diabetes has an independent effect on work disability, sex, age, education, and comorbid conditions play a significant role.
The burden of diabetes ranks in the top 10 as measured by disability-adjusted life-years, and it is projected to increase.1–4 In addition to the individual burden, diabetes has also a vast societal and economic significance, as reported in studies from Scandinavia5,6 as well as the United States.4,7 Studies have shown that diabetes is associated with a 1.3 to 3 times higher risk of sick leave and disability pension and lower labor force participation.8 These studies have largely been based on self-reports of either diabetes status or work disability, and studies based on longitudinal data or more objective measures are needed.8
The risk of work disability (sick leave and disability pension) may already be higher before diabetes diagnosis.9 Prediagnosis morbidity, and subsequent work disability, is plausible because diabetes-related symptoms, such as fatigue, are likely to be most severe right before diagnosis. The prediabetic period, characterized by impaired glucose regulation and elevated blood glucose level not reaching the values needed for diabetes diagnosis, can last for more than 10 years and often remains undetected but may still manifest itself in a variety of symptoms and micro- and macrovascular complications.10
Although people with diabetes have been shown to be heterogeneous in terms of disease prognosis,10 only a few previous studies have focused on such differences and on possible high-risk subgroups. The risk of not working was found to be higher for men than women with self-reported diabetes compared with individuals of the same sex without diabetes.11 Socioeconomic status is a strong predictor of onset of and of poorer prognosis for diabetes.12–18 Comorbid conditions are also a likely contributor to diabetes prognosis: Public-sector employees with self-reported diabetes had a risk of sick leave twice that of those who reported no chronic disease, but comorbid noncardiovascular diseases explained 55% of that excess risk.19
To address the gaps in previous research, we examined (1) the prevalence and risk of work disability (sick leave and disability pension) after diabetes diagnosis compared with before diagnosis and compared with a population sample without diabetes during an observation window of 7 years around the diabetes diagnosis; (2) the trajectories of work disability before, during, and after the diagnosis of diabetes; and (3) the factors that contributed to both risk of work disability after relative to years before diabetes diagnosis and to the trajectories of work disability. Such factors include sex, education, and comorbid conditions.
METHODS
This study was based on the nationwide research database Insurance Medicine All-Sweden20–22 from which we drew individuals living in Sweden on December 31, 2005, who had been living in Sweden since December 31, 2002 (n = 4 123 104). We restricted the study population to working-aged individuals (25–59 years) to avoid inclusion of full-time students and to homogenize those with diabetes to individuals more likely to have type 2 diabetes. From those individuals, we identified people with diabetes and a reference group without diabetes.
Our diabetes population was those individuals in the cohort who were diagnosed with diabetes in 2006 and who had no indication of diabetes in 2003 to 2005 (n = 14 428). The population without diabetes was a 1% random sample of the whole cohort living in Sweden on December 31, 2005, without any indication of diabetes in 2003 to 2009 (n = 39 702). This sample was used as a reference group on work disability. Those who died or emigrated during the follow-up were excluded from the analyses from death or emigration year onward. The analyses were based on a 7-year observation window: from 3 years before diagnosis (2003–2005) to year of diagnosis (2006) and 3 years after diagnosis (2007–2009). The study setting is further illustrated in Figure 1.
FIGURE 1—
Study setting: Insurance Medicine All-Sweden, 2003–2009.
Note. ATC = Anatomical Therapeutic Chemical; CVD = cardiovascular disease; ICD-10 = International Classification of Diseases, 10th Revision.23
Data for each individual were obtained from nationwide Swedish registers and linked through the personal identity number unique to each resident of Sweden. We used the following registers:
Statistics Sweden: Longitudinal Integration Database for Health Insurance and Labor Market Studies, regarding sex, age, education, family situation, place of birth (Sweden or not Sweden), type of living area (large city, medium-sized municipality, small municipality), geographic region (8 categories) on December 31, 2005, and year of emigration.
National Board of Health and Welfare: (1) International Classification of Diseases, 10th Revision (ICD-10)23–coded diagnosis-specific data on hospitalizations or specialized outpatient care from January 1, 2003, through December 31, 2009; (2) from the Swedish Prescribed Drug Register, medication purchases from July 1, 2005, through December 31, 2009 (no data were available before July 2005); and (3) from the death register, date of death from January 1, 2006, through December 31, 2009.
National Social Insurance Agency: annual net number of days with sick leave and disability pension benefits paid by the Social Insurance Agency from January 1, 2003, through December 31, 2009.
Measures
Indication of incident diabetes was based on having at least 1 of the following: (1) insulin or other diabetes medication purchase (Anatomical Therapeutic Chemical24 code A10) from January 1 through December 31, 2006, and none from July 1 through December 31, 2005; (2) in- or outpatient hospitalization record with ICD-10, code E10, E11, E12, E13, or E14 during January 1 through December 31, 2006, and none from January 1, 2003, through December 31, 2005. Specific information on type 1 and type 2 diabetes was not available. However, the onset of type 2 diabetes is typically in middle age,25 and the onset of type 1 diabetes is typically in childhood or adolescence.26 Because the minimum age at diabetes onset in our study sample was 25 years, most of the cases were likely to be type 2. The individuals in the reference group were chosen from among those who had no indication of diabetes in 2003 to 2009, based on the previously mentioned data on medication or in- and outpatient care.
We defined work disability as annual sick leave or disability pension days, as indicated by benefits received from the Social Insurance Agency, from January 1, 2003, through December 31, 2009. Employees usually get sick pay from their employer during the first 14 days of a sick-leave spell; thus, we did not include those days. All people with income from work or unemployment benefits who have a reduced work capacity as a result of disease or injury can get sick leave benefits. Also, those with no previous income can be granted disability pension if their work capacity is permanently reduced as a result of disease or injury.
Sociodemographic covariates, all measured on December 31, 2005 (i.e., before the diagnosis of diabetes) were age, sex, education (compulsory school, high school, university), family situation (married or cohabitating without children; married or cohabitating with children; single, divorced, separated, or widowed without children; single, divorced, separated, or widowed with children), place of birth (born in Sweden or not born in Sweden), type of living area (large city, medium-sized municipality, small municipality), and geographic region of the place of residence (Stockholm County, East Middle Sweden, South Sweden, North Middle Sweden, Middle Norrland, Småland and islands, West Sweden, Upper Norrland).
Comorbid conditions concurrent with diabetes diagnosis (January 1, 2006–December 31, 2006) were depression, other psychiatric disorders, cardiovascular disease, hypertension, musculoskeletal disorders, and asthma. These conditions are common causes of work disability in Sweden,27 and depression,28,29 cardiovascular disease,30 musculoskeletal disorders,31,32 and to some extent asthma33 have been linked with diabetes. Indication of these conditions was based on having either a medication purchase record entry or in- or outpatient hospitalization record entry, or both. The identification of these disorders and data sources are further described in Table A (available as a supplement to this article at http://www.ajph.org).
Statistical Analyses
We calculated the annual mean work disability days for those with incident diabetes and for the general population sample without diabetes during the whole 7-year observation window. To examine the annual prevalence of work disability and to plot the trajectory of work disability in relation to diabetes diagnosis, we calculated unadjusted means per year and sociodemographics-adjusted least square means of annual work disability days.
To examine the risk of work disability during the 3 years after relative to the 3 years before diabetes diagnosis, we applied a repeated-measures Poisson regression analysis using the generalized estimating equations method with exchangeable correlation structure.34 This method takes into account the intraindividual correlation between measurements.
To express the risk of work disability during the years after the diabetes diagnosis in relation to the years before the diagnosis, we calculated rate ratios (RR) and their 95% confidence intervals (CIs) by contrasting the 3-year window after diagnosis (years 1–3) with the 3-year window before the diagnosis (years −3 to −1) adjusting for sex and age at diagnosis. We then examined the role of contributing factors—that is, whether sex, education, or comorbid diseases modified the risk of work disability in time—by entering interaction terms (contributing factor × year) into the regression models adjusted for other contributing factors. Year was specified as a class (categorical) variable in the analyses.
We used Poisson regression analysis using the generalized estimating equations method with exchangeable correlation structure to examine the trajectories of work disability. To examine the shape of trajectories (i.e., slopes) in work disability before, during, and after diabetes diagnosis, we distinguished among 3 periods: prediagnosis (years −3 and −2), diagnosis (years −1 to 1), and postdiagnosis period (years 2 and 3). To evaluate the differences in the shape of the trajectory or slope between periods, we entered the interaction term (year [continuous variable] × period) into the regression model.35,36 We then examined the role of contributing factors—that is, whether sex, education, or comorbid diseases modified the shape of the trajectory (slope)—by stratifying the regression analyses by these factors. To express the difference in the shape of the trajectory or slope of work disability, we calculated the RRs per 2 years in work disability within the prediagnosis, diagnosis, and postdiagnosis periods. We also adjusted the models for the contributing sociodemographic factors and comorbid conditions. We performed all statistical analyses with SAS version 9.4 (SAS Institute, Cary, NC).
RESULTS
Of the 4 123 104 individuals in the nationwide cohort, 14 428 individuals had incident diabetes in 2006. This resulted in an incidence rate of 3.5 per 1000 people in 2006 in Sweden. The descriptive statistics of those with incident diabetes compared with the sample without diabetes are shown in Table 1. A higher rate of those with incident diabetes were men, had a lower level of education, were childless, were somewhat older, were not born in Sweden, and were more likely to have comorbid chronic diseases than those without diabetes. Differences in geographic area and type of living area between those with and without diabetes were small. The overall mean of annual work disability days was 95 (SD = 143) among those with incident diabetes and 35 (SD = 95) among those without diabetes over the 7-year study period. A total of 4% of those with incident diabetes in 2006 died during 2006 to 2010. The corresponding death rate for those without diabetes was 2%.
TABLE 1—
Descriptive Statistics for People With Incident Diabetes and the Population-Based Sample Without Diabetes: Insurance Medicine All-Sweden, 2003–2009
| Individuals With Diabetes Onset in 2006 (n = 14 428) |
Population-Based Sample Without Diabetes 2003–2009 (n = 39 702) |
|||||
| Characteristic | % or Mean (SD) | Annual Work Disability Days During 2003–2009, Mean (SD) | No. | % or Mean (SD) | Annual Work Disability Days During 2003–2009, Mean (SD) |
No. |
| All | 100 | 94.9 (143.1) | 100 | 35.2 (95.0) | ||
| Sexa | ||||||
| Men | 58 | 84.7 (139.3) | 50 | 27.8 (86.6) | ||
| Women | 42 | 108.8 (147.0) | 50 | 42.7 (102.1) | ||
| Age, yb | 48.5 (8.8) | 42.0 (9.9) | ||||
| Educational levela | ||||||
| Low/medium | 77 | 105.3 (147.8) | 64 | 44.9 (106.4) | ||
| High | 23 | 56.9 (116.2) | 36 | 17.5 (65.7) | ||
| Type of living areac | ||||||
| Large city | 36 | 92.3 (142.9) | 37 | 30.8 (89.6) | ||
| Medium-sized city | 34 | 94.5 (142.5) | 35 | 35.9 (95.6) | ||
| Small city or village | 30 | 98.5 (143.8) | 27 | 40.8 (101.1) | ||
| Geographic regionc | ||||||
| Stockholm county | 22 | 91.1 (142.7) | 22 | 29.3 (87.5) | ||
| East middle Sweden | 17 | 95.6 (142.8) | 16 | 38.6 (99.4) | ||
| South Sweden | 15 | 93.0 (143.1) | 14 | 34.6 (94.7) | ||
| North middle Sweden | 10 | 96.6 (142.8) | 9 | 37.7 (96.7) | ||
| Middle Norrland | 5 | 95.3 (142.2) | 4 | 43.3 (103.6) | ||
| Småland and islands | 8 | 87.8 (137.7) | 9 | 31.8 (89.6) | ||
| West Sweden | 17 | 100.9 (145.6) | 20 | 36.9 (97.4) | ||
| Upper Norrland | 6 | 100.5 (145.3) | 6 | 40.5 (99.6) | ||
| Family situationc | ||||||
| Married or cohabiting without children | 21 | 97.0 (143.6) | 13 | 48.2 (108.1) | ||
| Married or cohabiting with children | 35 | 73.2 (130.1) | 45 | 22.9 (75.5) | ||
| Single without children | 37 | 112.5 (150.5) | 35 | 42.3 (105.2) | ||
| Single with children | 8 | 102.4 (147.8) | 8 | 52.1 (112.5) | ||
| Birth countrya | ||||||
| Sweden | 74 | 87.2 (137.6) | 86 | 32.2 (90.4) | ||
| Other | 26 | 116.8 (155.7) | 14 | 53.9 (117.6) | ||
| Comorbid depressive disorder (2006)a | ||||||
| Yes | 18 | 185.5 (159.5) | 9 | 122.9 (150.3) | ||
| No | 82 | 75.2 (131.2) | 91 | 26.4 (82.5) | ||
| Comorbid other psychiatric disordera | ||||||
| Yes | 23 | 189.3 (158.3) | 11 | 128.3 (153.1) | ||
| No | 77 | 67.6 (125.8) | 89 | 24.3 (78.6) | ||
| Comorbid cardiovascular disease (2006)a | ||||||
| Yes | 18 | 136.6 (155.5) | 7 | 69.3 (126.7) | ||
| No | 82 | 85.9 (138.7) | 93 | 32.8 (91.8) | ||
| Comorbid hypertension (2006)a | ||||||
| Yes | 11 | 123.0 (152.9) | 1 | 86.6 (136.5) | ||
| No | 89 | 91.2 (141.4) | 99 | 34.8 (94.5) | ||
| Comorbid musculoskeletal disordera | ||||||
| Yes | 9 | 155.8 (156.7) | 5 | 91.7 (136.7) | ||
| No | 91 | 89.1 (140.4) | 95 | 32.5 (91.5) | ||
| Comorbid asthma (2006)a | ||||||
| Yes | 3 | 169.1 (156.9) | 1 | 97.4 (146.3) | ||
| No | 97 | 92.8 (142.1) | 99 | 34.9 (94.5) | ||
| Work disability daysc | ||||||
| 2003 | 82.8 (135.5) | 14 428 | 33.2 (91.0) | 39 702 | ||
| 2004 | 86.9 (139.1) | 14 428 | 34.6 (93.4) | 39 702 | ||
| 2005 | 91.4 (141.3) | 14 428 | 35.8 (95.1) | 39 702 | ||
| 2006 | 101.6 (143.7) | 14 266 | 36.6 (96.3) | 39 450 | ||
| 2007 | 102.1 (146.6) | 14 098 | 37.3 (97.4) | 39 238 | ||
| 2008 | 101.0 (147.5) | 13 953 | 37.1 (97.6) | 39 056 | ||
| 2009 | 98.7 (146.4) | 13 833 | 35.7 (96.3) | 38 889 | ||
Note. All differences between those with and without diabetes are statistically significant (χ2 or t test, P < .001).
P for difference (t test) < .001.
P for difference (Pearson r) < .001.
P for difference (analysis of variance) < .001.
Risk of Work Disability After vs Before Diabetes Diagnosis
Figure 2 displays the annual means of work disability days among those with incident diabetes in 2006 and those without diabetes during the same time period. Although the level of work disability was substantially higher among those with incident diabetes during the whole follow-up period, the work disability days increased toward the year of diabetes diagnosis and leveled off after the diagnosis year. We observed no corresponding increase among individuals without diabetes.
FIGURE 2—
Annual means of work disability days in relation to diabetes diagnosis in 2006: Insurance Medicine All-Sweden, 2003–2009.
Note. Dotted lines indicate means adjusted for sex, age, education, family situation, birth country, type of living area, and geographic region. Error bars indicate 95% confidence intervals.
The results regarding the risk of work disability during the observation window of 3 years before and 3 years after diabetes diagnosis, and the factors that potentially modified that risk, are presented in Table B (available as a supplement to this article at http://www.ajph.org). Among those with incident diabetes in 2006, the age- and sex-adjusted risk of work disability during the years after diabetes diagnosis (2007–2009) was 1.18 times (95% CI = 1.16, 1.20) that in the years before the diagnosis (2003–2005). Among those without diabetes, the age- and sex-adjusted risk of work disability during 2007–2009 did not increase as much (RR = 1.07; 95% CI = 1.05, 1.08) relative to 2003–2005. Among those with incident diabetes, the risk of work disability was higher during the years after diagnosis in relation to the years before diagnosis among men (RR = 1.22; 95% CI = 1.20, 1.25) than among women (RR = 1.14; 95% CI = 1.11, 1.16) and among those who were highly educated (RR = 1.24; 95% CI = 1.19, 1.30) than among those with a lower level of education (RR = 1.17; 95% CI = 1.15, 1.18). The risk of work disability was higher among those with diabetes and cardiovascular disease (RR = 1.23; 95% CI = 1.19, 1.26) than among those without diabetes but with cardiovascular disease (RR = 1.11; 95% CI = 1.07, 1.15) during the years after diabetes diagnosis in relation to the years before (supplementary Table B).
Trajectories of Work Disability Among Individuals With Diabetes
We then examined the trajectories, that is, the slopes of relative risk of work disability, among those with diabetes in 3 periods: prediagnosis, diagnosis, and postdiagnosis. After adjustments for age, sex, education, family situation, birth country, type of living area, geographic region, and comorbid conditions, the shape of the work disability trajectory was different at the time of diabetes diagnosis (RR2 years = 1.13; 95% CI = 1.12, 1.15) compared with prediagnosis (RR2 years = 1.10; 95% CI = 1.08, 1.12) or postdiagnosis (RR2 years = 0.96; 95% CI = 0.95, 0.98) among people with diabetes. In the population sample without diabetes, the shape of the work disability trajectory was different in 2003 to 2004 (corresponding to prediagnosis period; RR2 years = 1.09; 95% CI = 1.06, 1.12), in 2005 to 2007 (corresponding to diagnosis period; RR2 years = 1.04; 95% CI = 1.02, 1.05), and in 2008 to 2009 (corresponding to postdiagnosis period; RR2 years = 0.94; 95% CI = 0.92, 0.96; Table 2).
TABLE 2—
Differences in the Shape of Work Disability Trajectories (Slopes) Within 3 Periods Around Diabetes Onset Among Individuals With Diabetes, Stratified by Sex, Education, and Comorbid Conditions: Insurance Medicine All-Sweden; 2003–2009
| Characteristic | Days per Year, Mean (SD) | RR2 ya (95% CI) |
| Without diabetes (n = 275 755) | ||
| Prediagnosis | 33.9 (92.2) | 1.09 (1.06, 1.12) |
| Diagnosis | 36.5 (96.3) | 1.04 (1.02, 1.05) |
| Postdiagnosis | 36.4 (96.9) | 0.94 (0.92, 0.96) |
| With diabetes (n = 98 341) | ||
| Prediagnosis | 84.9 (137.3) | 1.10 (1.08, 1.12) |
| Diagnosis | 98.3 (144.0) | 1.13 (1.12, 1.15) |
| Postdiagnosis | 99.9 (147.0) | 0.96 (0.95, 0.98) |
| Men (n = 57 734) | ||
| Prediagnosis | 74.3 (132.2) | 1.09 (1.06, 1.13) |
| Diagnosis | 88.0 (140.2) | 1.16 (1.14, 1.18) |
| Postdiagnosis | 90.6 (144.3) | 0.99 (0.97, 1.01) |
| Women (n = 41 700) | ||
| Prediagnosis | 99.5 (142.8) | 1.11 (1.07, 1.14) |
| Diagnosis | 112.7 (147.9) | 1.10 (1.08, 1.12) |
| Postdiagnosis | 112.6 (149.6) | 0.94 (0.92, 0.96) |
| Low or intermediate education (n = 77 274) | ||
| Prediagnosis | 95.0 (142.7) | 1.09 (1.07, 1.12) |
| Diagnosis | 108.8 (148.5) | 1.12 (1.11, 1.14) |
| Postdiagnosis | 110.6 (151.3) | 0.96 (0.95, 0.98) |
| High education (n = 23 253) | ||
| Prediagnosis | 48.9 (108.4) | 1.16 (1.08, 1.23) |
| Diagnosis | 60.4 (118.1) | 1.20 (1.15, 1.25) |
| Postdiagnosis | 59.7 (120.7) | 0.97 (0.93, 1.02) |
| Musculoskeletal (n = 8 604) | ||
| Prediagnosis | 130.8 (153.1) | 1.17 (1.11, 1.24) |
| Diagnosis | 167.1 (155.5) | 1.16 (1.12, 1.21) |
| Postdiagnosis | 164.3 (159.3) | 0.95 (0.91, 0.99) |
| Cardiovascular (n = 17 474) | ||
| Prediagnosis | 119.1 (151.2) | 1.09 (1.05, 1.13) |
| Diagnosis | 142.5 (154.8) | 1.20 (1.16, 1.23) |
| Postdiagnosis | 145.9 (159.4) | 0.96 (0.94, 0.99) |
| Hypertension (n = 11 284) | ||
| Prediagnosis | 101.3 (144.8) | 1.14 (1.07, 1.20) |
| Diagnosis | 128.8 (152.4) | 1.29 (1.24, 1.35) |
| Postdiagnosis | 137.0 (159.4) | 0.95 (0.91, 0.99) |
| Asthma (n = 2 623) | ||
| Prediagnosis | 151.0 (155.0) | 1.03 (0.95, 1.11) |
| Diagnosis | 173.3 (155.9) | 1.14 (1.08, 1.21) |
| Postdiagnosis | 182.1 (158.9) | 0.96 (0.91, 1.03) |
| Depression (n = 17 743) | ||
| Prediagnosis | 162.8 (158.6) | 1.15 (1.11, 1.19) |
| Diagnosis | 194.4 (157.8) | 1.13 (1.10, 1.15) |
| Postdiagnosis | 195.6 (160.6) | 0.94 (0.92, 0.97) |
| Other psychiatric (n = 22 273) | ||
| Prediagnosis | 167.1 (158.5) | 1.11 (1.08, 1.14) |
| Diagnosis | 197.5 (156.2) | 1.13 (1.11, 1.15) |
| Postdiagnosis | 200.6 (159.0) | 0.96 (0.94, 0.98) |
Note. CI = confidence interval; RR = rate ratio; RR2 y = RRs per 2 years. The sample sizes are the number of person-observations during the follow-up. P < .002 for interaction term period × continuous time in all models.
RR2 y describe the shape of the work disability trajectory (slope) in each period (prediagnosis, diagnosis, and postdiagnosis). Models are adjusted for age, sex, education, family situation, birth country, type of living area, geographic region, comorbid depression, other psychiatric disorder, cardiovascular disease, hypertension, musculoskeletal disorder, and asthma in 2006.
The adjusted models stratified by sociodemographic factors and comorbid diseases among people with diabetes are shown in Table 2. Among men, the shape of the work disability trajectory was different at diagnosis (RR2 years = 1.16; 95% CI = 1.14, 1.18) from prediagnosis (RR2 years = 1.09; 95% CI = 1.06, 1.13) or postdiagnosis (RR2 years = 0.99; 95% CI = 0.97, 1.01). In women, the shape of the work disability trajectory was similar (increasing) at prediagnosis (RR2 years = 1.11; 95% CI = 1.07, 1.14) and at diagnosis (RR2 years = 1.10; 95% CI = 1.08, 1.12), but different (decreasing) after diagnosis (RR2 years = 0.94; 95% CI = 0.92, 0.96).
Among those with comorbid hypertension, the shape of the work disability trajectory was different at diagnosis (RR2 years = 1.29; 95% CI = 1.24, 1.35) from prediagnosis (RR2 years = 1.14; 95% CI = 1.07, 1.20) or postdiagnosis (RR2 years = 0.95; 95% CI = 0.91, 0.99). The trajectories of those with comorbid cardiovascular disease were similar to the trajectories of those with hypertension (Table 2).
Those with comorbid psychiatric disorders, asthma, and musculoskeletal disorders had the highest levels of work disability. Among those with comorbid musculoskeletal disorders, the shape of the work disability trajectory was similar at prediagnosis (RR2 years = 1.17; 95% CI = 1.11, 1.24) and at diagnosis (RR2 years = 1.16; 95% CI = 1.12, 1.21), but different (decreasing) at postdiagnosis (RR2 years = 0.95; 95% CI = 0.91, 0.99). The work disability trajectories of those with comorbid depression and other psychiatric disorders were similar to the trajectories of those with musculoskeletal disorders (Table 2).
DISCUSSION
In this study among the population of Sweden, we used nationwide register data with repeated measures of work disability days to assess the trajectories of work disability before and after incident diabetes. We demonstrated that, in general, the annual work disability rates were substantially higher among those with diabetes (on average, 95 days per year) than in the comparison group, a random sample of the general Swedish population with no indication of diabetes (35 days per year). Those with incident diabetes were more likely to have other risk factors of work disability as well, including older age and a lower level of education and comorbid conditions, than those without diabetes.
A previous review reported 5 to 18 days of absenteeism per year for people with diabetes and 3 to 9 days for those without diabetes.8 The higher level of absence associated with diabetes corresponds with our findings. However, the numbers are not comparable because our study also included disability pension days.
The age- and sex-adjusted risk of work disability was higher (RR = 1.18 vs 1.07) in the years after the diabetes diagnosis relative to the years before diagnosis and compared with the risk of those without diabetes. This result adds to the evidence that incident diabetes has an independent effect on future work disability,7,8 but that comorbid conditions explain that effect to a large degree.7,19,37
Among people with diabetes, the trajectory (slope) in relative risk of work disability had already started to increase before diabetes diagnosis. The slope increased even more sharply during the year of diagnosis and leveled off thereafter. Among people without diabetes, the slope was similar to that of those with diabetes during the prediagnosis period but did not increase as much during diagnosis period as among those with diabetes (RR2 years = 1.13 vs 1.04). Men and women had a different trajectory of work disability: In men, the trajectory was more gently sloping during prediagnosis than during diagnosis. In women, the slope was similar (increasing) at prediagnosis and at diagnosis, indicating that although the overall level of work disability was higher among women during the follow-up, the increase in the relative risk at the time of diabetes diagnosis was steeper among men. This finding may indicate that prediabetic symptoms and associated work disability differ between men and women so that women might be more likely than men to consult physicians and be absent as a result of illness at prediagnosis stage for prediabetic symptoms.
Comorbid conditions also contributed differentially to the trajectories of work disability among people with diabetes. Among those with comorbid musculoskeletal or psychiatric disorders, the slope of work disability was similar (gently increasing) from before diagnosis to during diagnosis of diabetes and leveled off after diagnosis. Among those with comorbid hypertension or cardiovascular disease, the trajectory of work disability was sharply sloping during the diagnosis period, indicating that the increase in the relative risk of work disability at diagnosis was steeper among those with comorbid cardiovascular disease or hypertension. Thus, the prediabetes period seems to be different with different comorbid conditions: having preexisting cardiovascular problems or hypertension may further increase the risk of work disability with incident diabetes. However, with all comorbid conditions, the overall level of work disability was high even before diabetes diagnosis.
The major strength of the study is the large population-based prospective cohort data with reliable register measures of high coverage and specificity.38 We thus avoided the recall and response biases associated with self-report data. We were also able to follow work disability across a 7-year period centered on diabetes diagnosis. There was no loss to follow-up, so we also avoided selection bias.
The limitations include the relatively short follow-up after diabetes diagnosis (3 years). Future studies should have longer follow-up periods, which would allow studies to be carried out on the prognosis and causes and diagnoses of work disability after diabetes diagnosis in more detail. We also did not have data on sick leave episodes lasting less than 14 days for those who were employed (i.e., for most of the study population), which can be considered as a limitation of this study.
Other register data also had some limitations: first, our indication of diabetes was based on purchases of diabetes medication or having had in- or outpatient care attributable to diabetes. Although these purchases and hospitalizations have an exact date, it is not the same as the date of diabetes diagnosis. However, new purchase of medication or treatment contact is likely to be temporally close to the day of diagnosis.
We may have missed some individuals with diabetes, because we had data on prescribed drug purchases only from July 2005 onward. However, in Sweden one is not allowed to buy prescribed medication for more than 3 months at a time. Thus, most individuals should have done that at least once during the 6-month period. Second, the in- and outpatient records for 2003 to 2005 were from specialized health care, and part of diabetes care might be managed in primary health care. Most diabetes patients, however, should have been identified through the medication register.
A limitation regarding the role of comorbid conditions is that people with chronic diseases, such as diabetes, cardiovascular disease, hypertension, asthma, and musculoskeletal or psychiatric disorders consume more health care and are therefore more likely to become diagnosed with additional disorders. This may lead to overrepresentation of comorbid disorders among people with diabetes and subsequent overestimation of the contribution of comorbid conditions to the association between diabetes and work disability. Finally, our register data did not include any information about diabetes severity or the control of diabetes (e.g., blood sugar or glucose concentration measurements).
Knowledge about the trajectories of work disability among people with diabetes is beneficial for occupational and other health care professionals for the prevention of work disability. This information can help to identify subgroups at high risk for work disability and highlights that the risk of work disability is already increased before the diagnosis of diabetes. Along with diabetes, which has a small but independent effect on the risk of work disability, people with diabetes have accumulated risk factors of work disability, such as lower socioeconomic status and comorbid conditions. The trajectory of work disability in diabetes was characterized by an increase during the prediagnosis and diagnosis periods, which leveled off during the postdiagnosis period. The risk of work disability was increased already at the prediabetes period, especially among women and those with comorbid noncardiovascular conditions, whereas among men and those with comorbid cardiovascular conditions, the risk was increased at the diagnosis period.
Acknowledgments
J. Ervasti received funding from the Academy of Finland (265174) and the Finnish Work Environment Fund (114260). M. Virtanen received funding from the Academy of Finland (258598). T. Lallukka had funding from the Association for Promotion of Occupational Health. P. Tinghög, L. Kjeldgård, and K. Alexanderson had funding from the Swedish Research Council of Health, Working Life and Welfare (2007-1762). E. Mittendorfer-Rutz had funding from the Swedish Research Council (522-2010-2683).
Human Participant Protection
The study was approved by the Regional Ethical Review Board of Stockholm, Sweden.
References
- 1.Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract. 2010;87(1):4–14. doi: 10.1016/j.diabres.2009.10.007. [DOI] [PubMed] [Google Scholar]
- 2.Whiting DR, Guariguata L, Weil C, Shaw J. IDF diabetes atlas: global estimates of the prevalence of diabetes for 2011 and 2030. Diabetes Res Clin Pract. 2011;94(3):311–321. doi: 10.1016/j.diabres.2011.10.029. [DOI] [PubMed] [Google Scholar]
- 3.Institute for Health Metrics and Evaluation. GBD arrow diagram. Available at: http://vizhub.healthdata.org/irank/arrow.php. Accessed October 17, 2014.
- 4.Caspersen CJ, Thomas GD, Boseman LA, Beckles GLA, Albright AL. Aging, diabetes, and the public health system in the United States. Am J Public Health. 2012;102(8):1482–1497. doi: 10.2105/AJPH.2011.300616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bolin K, Gip C, Mörk A-C, Lindgren B. Diabetes, healthcare cost and loss of productivity in Sweden 1987 and 2005—a register-based approach. Diabet Med. 2009;26(9):928–934. doi: 10.1111/j.1464-5491.2009.02786.x. [DOI] [PubMed] [Google Scholar]
- 6.Solli O, Jenssen T, Kristiansen IS. Diabetes: cost of illness in Norway. BMC Endocr Disord. 2010;10:15. doi: 10.1186/1472-6823-10-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.American Diabetes Association. Economic costs of diabetes in the US in 2012. Diabetes Care. 2013;36(4):1033–1046. doi: 10.2337/dc12-2625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Breton M-C, Guénette L, Amiche MA, Kayibanda J-F, Grégoire J-P, Moisan J. Burden of diabetes on the ability to work. A systematic review. Diabetes Care. 2013;36(3):740–749. doi: 10.2337/dc12-0354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Dray-Spira R, Herquelot E, Bonenfant S, Guéguen A, Melchior M. Impact of diabetes mellitus onset on sickness absence from work—a 15-year follow-up of the GAZEL occupational cohort study. Diabet Med. 2013;30(5):549–556. doi: 10.1111/dme.12076. [DOI] [PubMed] [Google Scholar]
- 10.Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet. 2012;379(9833):2279–2290. doi: 10.1016/S0140-6736(12)60283-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.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(11):2662–2667. doi: 10.2337/diacare.28.11.2662. [DOI] [PubMed] [Google Scholar]
- 12.Brown AF, Ettner SL, Piette J et al. Socioeconomic position and health among persons with diabetes mellitus: a conceptual framework and review of the literature. Epidemiol Rev. 2004;26(1):63–77. doi: 10.1093/epirev/mxh002. [DOI] [PubMed] [Google Scholar]
- 13.Ricci-Cabello I, Ruiz-Perez I, Olry de Labry-Lima A, Marquez-Calderon S. Do social inequalities exist in terms of the prevention, diagnosis, treatment, control and monitoring of diabetes? A systematic review. Health Soc Care Community. 2010;18(6):572–587. doi: 10.1111/j.1365-2524.2010.00960.x. [DOI] [PubMed] [Google Scholar]
- 14.Agardh E, Allebeck P, Hallqvist J, Moradi T, Sidorchuk A. Type 2 diabetes incidence and socio-economic position: a systematic review and meta-analysis. Int J Epidemiol. 2011;40(3):804–818. doi: 10.1093/ije/dyr029. [DOI] [PubMed] [Google Scholar]
- 15.Espelt A, Arriola L, Borrell C, Larrañaga I, Sandín M, Escolar-Pujolar A. Socioeconomic position and type 2 diabetes mellitus in Europe 1999-2009. Curr Diabetes Rev. 2011;7(3):148–158. doi: 10.2174/157339911795843131. [DOI] [PubMed] [Google Scholar]
- 16.Gaskin DJ, Thorpe RJ, Jr, McGinty EE et al. Disparities in diabetes: the nexus of race, poverty, and place. Am J Public Health. 2014;104(11):2147–2155. doi: 10.2105/AJPH.2013.301420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Dray-Spira R, Gary-Webb TL, Brancati FL. Educational disparities in mortality among adults with diabetes in the US. Diabetes Care. 2010;33(6):1200–1205. doi: 10.2337/dc09-2094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Saydah SH, Imperatore G, Beckles GL. Socioeconomic status and mortality: contribution of health care access and psychological distress among US adults with diagnosed diabetes. Diabetes Care. 2013;36(1):49–55. doi: 10.2337/dc11-1864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kivimäki M, Vahtera J, Pentti J, Virtanen M, Elovainio M, Hemingway H. Increased sick absence in diabetic employees: what is the role of co-morbid conditions? Diabet Med. 2007;24(9):1043–1048. doi: 10.1111/j.1464-5491.2007.02216.x. [DOI] [PubMed] [Google Scholar]
- 20.Jansson C, Alexanderson K. Sickness absence due to musculoskeletal diagnoses and risk of diagnosis-specific disability pension: a nationwide Swedish prospective cohort study. Pain. 2013;154(6):933–941. doi: 10.1016/j.pain.2013.03.001. [DOI] [PubMed] [Google Scholar]
- 21.Mittendorfer-Rutz E, Kjeldgård L, Runeson B et al. Sickness absence due to specific mental diagnoses and all-cause and cause-specific mortality: a cohort study of 4.9 million inhabitants of Sweden. PLoS ONE. 2012;7(9):e45788. doi: 10.1371/journal.pone.0045788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Tinghög P, Hillert J, Kjeldgård L, Wiberg M, Glaser A, Alexanderson K. High prevalence of sickness absence and disability pension among multiple sclerosis patients: a nationwide population-based study. Mult Scler. 2013;19(14):1923–1930. doi: 10.1177/1352458513488234. [DOI] [PubMed] [Google Scholar]
- 23.International Classification of Diseases, 10th Revision. Geneva, Switzerland: World Health Organization; 1992. [Google Scholar]
- 24.World Health Organization. Anatomical Therapeutic Chemical code index. Available at: http://www.whocc.no/atc_ddd_index. Accessed March 16, 2015.
- 25.Koopman RJ, Mainous AG, Diaz VA, Geesey ME. Changes in age at diagnosis of type 2 diabetes mellitus in the United States, 1988 to 2000. Ann Fam Med. 2005;3(1):60–63. doi: 10.1370/afm.214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nokoff N, Rewers M. Pathogenesis of type 1 diabetes: lessons from natural history studies of high-risk individuals. Ann N Y Acad Sci. 2013;1281:1–15. doi: 10.1111/nyas.12021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Swedish Social Insurance Agency. Social insurance in figures 2014. Available at: https://www.forsakringskassan.se/wps/wcm/connect/3107947a-6e12-4dcc-b1a6-e952a289ea95/sfis+2014-e.pdf?MOD=AJPERES. Accessed January 26, 2015.
- 28.Mezuk B, Eaton WW, Albrecht S, Hill Golden S. Depression and type 2 diabetes over the lifespan. Diabetes Care. 2008;31(12):2383–2390. doi: 10.2337/dc08-0985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tabák AG, Akbaraly TN, Batty GD, Kivimaki M. Depression and type 2 diabetes: a causal association? Lancet Diabetes Endocrinol. 2014;2(3):236–245. doi: 10.1016/S2213-8587(13)70139-6. [DOI] [PubMed] [Google Scholar]
- 30.American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2010;33(suppl 1):S62–S69. doi: 10.2337/dc10-S062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Silva MB, Skare TL. Musculoskeletal disorders in diabetes mellitus. Rev Bras Reumatol. 2012 doi: 10.1590/s0482-50042012000400010. 52():601–609. [DOI] [PubMed] [Google Scholar]
- 32.Larkin ME, Barnie A, Braffett BH et al. Musculoskeletal complications in type 1 diabetes. Diabetes Care. 2014;37(7):1863–1869. doi: 10.2337/dc13-2361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Mueller NT, Koh W-P, Odegaard AO, Gross MD, Yuan J-M, Pereira MA. Asthma and the risk of type 2 diabetes in the Singapore Chinese Health Study. Diabetes Res Clin Pract. 2013;99(2):192–199. doi: 10.1016/j.diabres.2012.11.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lipsitz SR, Kim K, Zhao L. Analysis of repeated categorical data using generalized estimating equations. Stat Med. 1994;13(11):1149–1163. doi: 10.1002/sim.4780131106. [DOI] [PubMed] [Google Scholar]
- 35.Oksanen T, Vahtera J, Westerlund H et al. Is retirement beneficial for mental health? Antidepressant use before and after retirement. Epidemiology. 2011;22(4):553–559. doi: 10.1097/EDE.0b013e31821c41bd. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol. 2005;162(3):199–200. doi: 10.1093/aje/kwi188. [DOI] [PubMed] [Google Scholar]
- 37.Von Korff M, Katon W, Lin EHB et al. Work disability among individuals with diabetes. Diabetes Care. 2005;28(6):1326–1332. doi: 10.2337/diacare.28.6.1326. [DOI] [PubMed] [Google Scholar]
- 38.Ludvigsson JF, Andersson E, Ekbom A et al. External review and validation of the Swedish national inpatient register. BMC Public Health. 2011;11:450. doi: 10.1186/1471-2458-11-450. [DOI] [PMC free article] [PubMed] [Google Scholar]