Table 3

Multinomial regression function: 'any OA'

N

666

Missing

49


NONE

ISO-PF

COMB

n = 212

n = 170

n = 284

aOR (95% CI)*

aOR (95% CI)*


Age (per year)

0.96 (0.93, 0.99)

1

1.04 (1.01, 1.07)

Female gender

2.87 (1.79, 4.59)

1

1.56 (0.99, 2.47)

BMI (per kg/m2)

0.94 (0.89, 0.99)

1

1.04 (0.99, 1.09)

Whole leg pain

2.12 (1.10, 4.08)

1

0.74 (0.37, 1.50)

Difficulty descending stairs

0.45 (0.27, 0.73)

1

1.15 (0.74, 1.80)

Intercondylar gap > 0 cm

1.38 (0.75, 2.57)

1

2.34 (1.32, 4.14)

Knee effusion†

0.82 (0.49, 1.35)

1

2.42 (1.55, 3.77)

Fixed flexion deformity

0.20 (0.04, 0.93)

1

2.31 (1.13, 4.77)

Coarse crepitus‡

0.64 (0.40, 1.02)

1

1.57 (1.03, 2.41)

Knee flexion ROM (per degree)

0.99 (0.96, 1.01)

1

0.96 (0.94, 0.98)


Nagelkerke's Pseudo R2

0.41

Pearson goodness-of-fit

P = 0.45

Pre-test probability

0.32

0.26

0.43


Correctly classified on 'balance of probabilities'

136 (64%)

36 (21%)

220 (77%)

Correctly classified on 'confident diagnosis' (probability ≥0.8)

18 (8%)

0

79 (28%)


* odds ratio with 95% confidence interval, adjusted for all other covariates in model. aOR is interpreted as the odds relative to the ISO-PF group (for example, compared to those people with ISO-PF, people with COMB have 2.4 times the odds of having knee effusion; compared with ISO-PF, those with no ROA have 0.45 times the odds of difficulty descending stairs, that is, lower risk).

†Mild/Moderate/Gross

‡Possible/definite

Peat et al. Arthritis Research & Therapy 2012 14:R63   doi:10.1186/ar3779

Open Data