Introduction

Maria Crotty, Lynne C. Giles, Ian D. Cameron and Peter M. Brooks

Introduction

There is growing evidence that individuals with chronic diseases have difficulty securing employment (Petrides, Petermann, Henrichs, Petzoldt, Rolver, Schidlmeier, Webber, & Wendt, 1995), are more likely to drop out of the labour force (Doeglas, Suurmeijer, Krol, Sanderman, Van Leewuwen, & Van Rijswijkl, 1995; Mitchell, 1990; Yelin, Henke, & Epstein, 1987) and rarely have workplace adjustments made to assist them (Baanders, Andries, Rijken, & Dekker, 2001). With the ageing of the workforce the number of workers with chronic disease and disability will rise and the challenge to accommodate these groups in the workforce will increase. Of the chronic diseases the relationship between arthritis and work disability has probably had the most attention. National surveys examining arthritis have found that among people reporting symptomatic arthritis more than half will report some type of work disability (Dunlop, Manheim, Yeline, Song, & Chang, 2003). More than 40% of 8.78 million people aged 51-61 reported work disability in the US 1992 Health and Retirement Study (Yelin, 1995). A recent Australian review of the costs of arthritis used the National Health Survey to make estimates and found that arthritis is responsible for nearly 1.8 million days of reduced activity and about 213 000 days off work or school each year in Australia (Access Economics, 2001). The indirect costs including loss of earnings and lost production following premature retirement were triple the direct costs ($6.72 billion). Little however is known about how this compares to the impact of other disabling chronic diseases in Australia and little is known about the relationship of labour force participation with pain.

More work is needed to promote public policies, which are responsive to the needs of people with chronic diseases and more work is needed to promote an integration of clinical and social approaches. Secondary analyses of national data sets while constrained by a limited range of variables and the method of collection (which is often self report) give a picture of labour market issues.

In this analysis, using the 1998 Survey of Disability, Ageing and Carers (SDAC), we sought to examine the impact of musculoskeletal disease upon labour force participation and determine the effect of comorbidity and pain on workforce participation in a national sample of working age adults.

Methods

Data from the 1998 Survey of Disability, Ageing and Carers were used (Australian Bureau of Statistics, 1999b, 1999c). The subset of participants aged 15-64 years (n=25 217) were considered in this study. The Confidentialised Unit Record File distributed by the Australian Bureau of Statistics was used as the data source in these analyses (Australian Bureau of Statistics, 1999a).

A binary variable for labour force participation was created from the responses to questions concerning employment status and job seeking. Persons in full or part-time employment or looking for full or part-time work were defined as in the labour force. Persons not in the labour force included those who had retired, had family considerations, were full-time students or did not need or want to work, or were not participating in the labour force due to health and disability. Because our primary interest was in work disability, our conceptualisation of labour force participation focused on those respondents who were engaged in or seeking full or part-time employment, and those who were not in the labour force due to health and disability. Thus the usable sample comprised 19 496 persons.

We considered a range of possible predictors of labour force participation that were used by Badley and Wang (1998). Sociodemographic variables that we considered included age group, gender, education (currently attending school, incomplete secondary education, complete secondary education, missing), residence (capital city, other location), and family composition (couple only, couple with dependent children, couple with independent children, living alone, other).

Chronic conditions considered in these analyses included arthritis and musculoskeletal conditions, cancer, circulatory disorders, depression, diabetes, neurological conditions (excluding stroke), respiratory disease, hearing or vision (sensory) conditions, and stroke. The total number of chronic conditions was also calculated, and classified as 0, 1, 2 or at least 3 chronic conditions. We hypothesized that pain would be an additional independent factor influencing workforce participation (Badley & Wang, 1998). Participants' responses to a question concerning chronic pain were used to derive a binary variable indicating the presence or absence of pain.

Arthritis and other musculoskeletal diseases were examined in greater detail than other conditions because of their much higher prevalence and importance as an area for intervention (Australian Bureau of Statistics, 1999b). We considered a number of other chronic conditions to assess the effect of comorbidity with arthritis and musculoskeletal disease upon labour force participation. For arthritis, participants were classified as i) having no chronic conditions, ii) having arthritis and no other chronic conditions, iii) having chronic conditions but not arthritis, or iv) having both arthritis and other chronic conditions. A similar definition was applied for musculoskeletal disease excluding arthritis.

Analyses

All analyses were weighted, to take into account the unequal probability of selection for a person into the 1998 SDAC. The weights were derived from those supplied by the Australian Bureau of Statistics, and corrected to sum to 19 496.

Frequency distributions were initially tabulated for all variables. The prevalence of arthritis, musculoskeletal disease and chronic conditions among labour force participants and non-participants was plotted. The relationship between sociodemographic and health variables and labour force participation was examined using chi-square tests of association. Separate logistic regression analyses, controlling for the sociodemographic variables, were used to estimate the odds ratio (OR) of labour force participation associated with arthritis, musculoskeletal disease, and the number of chronic conditions respectively. The analyses were repeated including pain to assess its effect on the models.

Results

A total of 18 511 (94.9%) respondents were either engaged in or seeking full or part-time employment, while 985 (5.1%) were not in the labour force because of their health or disability. Table 1 summarizes labour force participation against the sociodemographic variables and pain. Chi-square tests of association showed that the relationship between labour force participation and each of these variables was significant (P<0.001), with the exception of gender, where a non-significant result (P=0.419) was found. Labour force participation generally declined with age, so that just over three quarters of the sample aged 60-64 were participants in the labour force, compared with almost 100 per cent of younger respondents. There was a strong relationship between non-participation and pain. More than one quarter of those with chronic pain were not participating in the labour force.

Table 1: Summary of labour force participation for sociodemographic variables and pain

in labour force

not in labour force

n

%b

n

%c

%d

Total

P-value

Age last birthday

<0.001

15-34 years

8 400

45.4

139

14.1

1.6

8 539

35-49 years

6 879

37.2

308

31.3

4.3

7 187

50-54 years

1 728

9.3

187

18.9

9.7

1 915

55-59 years

998

5.4

205

20.8

17.0

1 203

60-64 years

505

2.7

147

14.9

22.5

652

Gender

0.419

Male

10 372

56.0

539

54.7

4.9

10 911

Female

8 139

44.0

446

45.3

5.2

8 585

Education

<0.001

Did not complete secondary

10 090

54.5

805

81.7

7.4

10 895

Completed secondary

8 421

45.5

180

18.3

2.1

8 601

Residential location

<0.001

Capital city

12 218

66.0

585

59.4

4.6

12 803

Balance of state

6 293

34.0

400

40.6

6.0

6 693

Family Composition

<0.001

Couple only

3 554

19.2

254

25.8

6.7

3 808

Couple with dependent children

5 924

32.0

110

11.2

1.8

6 034

Couple with independent children

3 535

19.1

166

16.8

4.5

3 700

Lives alone

1 434

7.7

202

20.5

12.4

1 636

Other

4 064

22.0

253

25.7

5.9

4 318

Pain

<0.001

Absent

17 085

92.3

411

41.7

2.3

17 496

Present

1 426

7.7

574

58.3

28.7

2 000

a: reported not in labour force due to health or disability

b: % based on those in labour force

c: % based on those not in labour force

d: row-specific %

           

 

Table 2 shows 74.9% of the participants who were in the labour force had none of the chronic conditions we considered, whereas only 8.4% of the participants, who were not in the labour force for health or disability reasons, reported none of the chronic conditions. Of the participants not in the labour force, 40.7% were affected by a musculoskeletal disease (excluding arthritis) and at least one other condition, while 12.8% were affected by musculoskeletal disease alone. A total of 25.7% of the non-participants in the labour force had arthritis and at least once other condition, compared with only 2.5% of the labour force participants having this level of morbidity. Musculoskeletal disease and arthritis were the most commonly reported conditions.

Table 2: Summary of labour force participation for selected chronic conditions included in 1998 Survey of Disability, Ageing and Carers

in labour force

not in labour force due to disability

Condition

n

%b

n

%c

row %d

no conditions

13 863

74.9

83

8.4

0.6

Arthritis+/-chronic conditions

arthritis only

336

1.8

31

3.2

8.5

chronic conditions only

3 842

20.8

619

62.8

13.9

arthritis and chronic conditions

471

2.5

253

25.7

34.9

Cancer+/-chronic conditions

cancer only

40

0.2

11

1.1

21.6

chronic conditions only

4 577

24.7

858

87.0

15.8

cancer and chronic conditions

31

0.2

34

3.5

52.3

Circulatory+/-chronic conditions

circulatory only

454

2.5

19

1.9

4.0

chronic conditions only

3 787

20.5

626

63.5

14.2

circulatory and chronic conditions

408

2.2

258

26.2

38.7

Depression+/-chronic conditions

depression only

68

0.4

19

1.9

22.0

chronic conditions only

4 497

24.3

791

80.3

15.0

depression and chronic conditions

83

0.4

92

9.3

52.6

Diabetes+/-chronic conditions

diabetes only

111

0.6

5

0.5

4.2

chronic conditions only

4 433

23.9

810

82.2

15.5

diabetes and chronic conditions

104

0.6

88

8.9

45.7

Musculoskeletal (exc arthritis)+/-chronic conditions

musculoskeletal only

1 172

6.3

126

12.8

9.7

chronic conditions only

2 704

14.6

376

38.1

12.2

musculoskeletal and chronic conditions

772

4.2

401

40.7

34.2

Musculoskeletal (inc arthritis)+/-chronic conditions

musculoskeletal only

1 644

8.9

211

21.5

11.4

chronic conditions only

2 163

11.7

259

26.3

10.7

musculoskeletal and chronic conditions

841

4.5

432

43.8

33.9

Neurological+/-chronic conditions

neurological only

189

1.0

60

6.1

24.1

chronic conditions only

4 264

23.0

665

67.4

13.5

neurological and chronic conditions

195

1.1

178

18.1

47.7

Respiratory+/-chronic conditions

respiratory only

698

3.8

24

2.5

3.3

chronic conditions only

3 626

19.6

718

72.8

16.5

respiratory and chronic conditions

324

1.7

161

16.3

33.2

Sensory+/-chronic conditions

sensory only

358

1.9

6

0.6

1.7

chronic conditions only

3 945

21.3

749

76.0

16.0

sensory and chronic conditions

345

1.9

148

15.0

30.0

Stroke+/-chronic conditions

stroke only

16

0.1

3

0.3

17.6

chronic conditions only

4 594

24.8

847

86.0

15.6

stroke and chronic conditions

38

0.2

52

5.3

57.9

# Chronic conditions

0

14 430

78.0

155

15.7

1.1

1

3 233

17.5

348

35.3

9.7

2

693

3.7

283

28.8

29.0

3+

154

0.8

199

20.2

56.3

b: % based on those in labour force

c: % based on those not in labour force

d: row-specific %

           

 

Logistic regression models were fitted to ascertain the odds ratio of not being in the labour force associated with arthritis, musculoskeletal disease and the number of chronic conditions. Age group, education and family composition were included in all models. A model containing only age group, education, family composition and pain was also fitted, and this showed the OR of not participating in the labour force was 11.1 times greater (95% CI 9.6-12.9) for those participants with pain than those participants without pain.

The risk of not participating in the labour force was much greater for arthritis or musculoskeletal disease in conjunction with other chronic conditions than for arthritis or musculoskeletal disease alone. Table 3 shows that after controlling for pain, age group, education and family composition, the odds ratio for not participating in the labour force associated with arthritis plus other chronic conditions was 17.3 (95% CI 12.7-23.4). The odds ratio associated with musculoskeletal disease (excluding arthritis) plus other chronic conditions was similarly large (OR=22.6; 95% CI=17.0-30.0).

Table 3: Odds ratio of labour force non-participation for arthritis, musculoskeletal conditions and number of chronic conditions

Excluding pain from the model

Including pain in the model

ORa

95% CI

ORa

95% CI

Arthritis+/-chronic conditions

no conditions (referent category)

1.0

1.0

arthritis only

8.3

5.3 - 12.8

4.0

2.5 - 6.3

chronic conditions only

19.2

15.1 - 24.2

11.7

9.1 - 15.0

arthritis and chronic conditions

42.7

32.3 - 56.4

17.3

12.7 - 23.4

Musculoskeletal conditions

no conditions (referent category)

1.0

1.0

musculoskeletal only

14.5

10.9 - 19.4

6.9

5.0 - 9.4

chronic conditions only

15.0

11.7 - 19.2

11.2

8.6 - 14.4

musculoskeletal and chronic conditions

50.4

39.0 - 65.2

22.6

17.0 - 30.0

# Chronic conditions

0 conditions (referent category)

1.0

1.0

1 conditions

7.4

6.1 - 9.0

4.6

3.7 - 5.7

2 conditions

22.0

17.6 - 27.4

11.1

8.6 - 14.2

>3 conditions

62.8

47.5 - 83.1

26.9

19.7 - 36.7

a: All models adjusted for age group, education, family composition. Gender and residential location not significant covariates in any analysis so excluded from final models.