Skip to main content

Table 2 Relative risk ratios (95% confidence intervals) for work impairment trajectory class membership from multinomial logit model and the relative importance in predicting trajectory class membership from dominance analysis

From: Digital self-management of hip and knee osteoarthritis and trajectories of work and activity impairments

Variable

VL–P vs. H–D

M–D vs. H–D

Dominance statisticsa

Rankingb

Female sex (male = ref)

1.11 (0.89, 1.38)

1.23 (1.01, 1.49)

0.0006

10

Age (24–50 years = ref)

1.0

1.0

0.0011

9

 51–60 years

0.94 (0.73, 1.23)

1.13 (0.90, 1.42)

 61–65 years

1.05 (0.78, 1.41)

1.32 (1.01, 1.71)

Hip as index joint (knee = ref)

0.82 (0.67, 0.99)

1.23 (1.04, 1.45)

0.0033

7

Education (Less than high school = ref)

1.0

1.0

0.0144

2

 High school

2.24 (1.37, 3.66)

1.23 (0.88, 1.74)

 College/university

4.96 (3.04, 8.11)

1.85 (1.31, 2.61)

Body mass index (< 25 = ref)

1.0

1.0

0.0127

3

 25–29

0.68 (0.54, 0.86)

0.92 (0.74, 1.14)

 30–34

0.48 (0.37, 0.64)

0.67 (0.52, 0.85)

 35 + 

0.39 (0.27, 0.59)

0.79 (0.58, 1.08)

Diabetes

1.19 (0.70, 2.01)

0.88 (0.57, 1.36)

0.0003

12

Lung diseases

1.18 (0.83, 1.68)

0.95 (0.70, 1.30)

0.0004

11

Balance troubles

1.22 (0.59, 2.49)

1.14 (0.64, 2.01)

0.0002

14

Rheumatoid arthritis

0.50 (0.28, 0.90)

0.70 (0.45, 1.09)

0.0015

8

Cardiovascular diseases

0.91 (0.51, 1.64)

1.12 (0.68, 1.85)

0.0002

13

Walking difficulties

0.44 (0.29, 0.68)

0.81 (0.58, 1.13)

0.0046

6

General health

1.24 (1.18, 1.31)

1.10 (1.05, 1.15)

0.0122

4

Pain

0.47 (0.44, 0.50)

0.61 (0.58, 0.64)

0.0869

1

Physical function

1.04 (1.02, 1.06)

1.01 (0.99, 1.03)

0.0062

5

McFadden pseudo-R2

0.145

  

MCC

0.271

  

CEN

0.581

  

PDI

0.600

  
  1. CEN Confusion entropy, H–D High–declining, VL–P Very low–persistent, MCC Matthews correlation coefficient, M–D Mild–declining, PDI Polytomous discrimination index
  2. aEach variable absolute contribution to McFadden pseudo-R2
  3. bThe relative importance of each variable in predicting trajectory class membership