Skip to main content

Table 2 Univariate associations and multivariable regression coefficients of the predictors in the final model of the development cohort

From: Long-term sickness absence in a working population: development and validation of a risk prediction model in a large Dutch prospective cohort

  

Univariate

Multivariable

Predictor

Level

OR (95%-CI)a

OR

Coefficientb

Female gender

 

1.26 (1.05–1.51)

1.10

0.09

Age, per year

 

1.02 (1.00–1.03)

1.00

0.007

Educational levelc

(ref: low)

Medium

1.00 (0.84–1.19)

0.88

−0.13

High

0.75 (0.62–0.91)

0.83

−0.19

SF-12 physical healthd (ref: 1st quartile, poorest health)

2nd quartile

0.97 (0.80–1.18)

0.55

−0.59

3rd quartile

0.51 (0.41–0.65)

0.42

−0.87

4th quartile

0.46 (0.36–0.59)

0.41

−0.90

Physically fite

 

0.77 (0.64–0.94)

0.80

−0.22

Physical job loadf (ref: 1st-3rd quartile, less demanding)

4th quartile

1.63 (1.35–1.98)

1.33

0.29

Knowledge and skills match the job (ref: bad/mediocre)

Reasonable/good

0.43 (0.30–0.59)

0.62

−0.48

Major life events previous year (ref: none)

1

1.10 (0.92–1.33)

1.12

0.11

≥2

1.62 (1.32–2.00)

1.43

0.35

Work abilityg (ref: good)

Average

1.40 (1.14–1.71)

1.10

0.10

Poor

4.70 (3.50–6.30)

2.28

0.82

Sickness absence days previous year (ref: none)

1–5

0.90 (0.73–1.12)

1.39

0.33

6–10

2.25 (1.76–2.87)

2.53

0.93

11–27

4.68 (3.77–5.80)

3.84

1.35

Self-employed

 

0.49 (0.33–0.73)

0.57

−0.57

Intercept

−2.55

  1. aPooled Odds Ratio (95% confidence interval) from the m = 30 multiple imputed datasets
  2. bPooled regression coefficients and intercept from the m = 30 multiple imputed datasets. An individuals predicted probability can be computed using the logistic regression formula P (LTSA) = 1/(1 + exp.(−LP), in which ‘exp’ denotes e-raised-to-the-power-of. The LP is the linear predictor, i.e. the linear sum of all predictor values multiplied by their regression coefficients, or − 2.55 + 0.09*gender (female = 1) + 0.007*age (years) -0.13*education (medium education = 1) -0.19*education (high education = 1) -0.59*physical health (2nd quartile = 1) -0.87*physical health (3rd quartile = 1) -0.90*physical health (4th quartile = 1) -0.22*physically fit (yes = 1) + 0.29*physical job load (4th quartile = 1) -0.48*knowledge (reasonable/good = 1) + 0.11*major life events (one event = 1) + 0.35*major life events (two or more = 1) + 0.10*work ability (average = 1) + 0.82*work ability (poor = 1) + 0.33*sickness absence (1–5 days = 1) + 0.93*sickness absence (6–10 days = 1) + 1.35*sickness absence (11–27 days = 1) -0.57*employment status (self-employed = 1)
  3. cLow: lower general secondary educational, preparatory secondary vocational education. Medium: intermediate vocational training, higher general secondary education, pre-university education. High: higher vocational education, university education
  4. dWeighted summary score (range 0–100) assessing physical health using 6 items of the 12-Item Short-Form Health Survey. Higher scores indicating better perceived physical health. 1st quartile < 46.1, 2nd quartile = 46.1–54.1, 3rd quartile = 54.2–56.5, 4th quartile > = 56.6
  5. eIntensive physical exercise ≥3 days per week for ≥20 min
  6. fAverage of five items (range: 1 = never, 5 = always) from the Dutch Musculoskeletal Questionnaire [22]. 1st-3rd quartile < 2.4, 4th quartile > = 2.4
  7. gMeasured with the first item of the Work Ability Index (WAI) [23]. Good = 8–10, average = 6/7, poor = 0–5
Back to article page

BMC Public Health

ISSN: 1471-2458

Contact us