Table 3

Regression models for fat mass

 Model Explained variance Model significance Unstandardised beta coefficient (95% CI), p value Likelihood-ratio test (compared with covariate) Covariate model R2=0.8531 F(13,1582)=706.71p<0.0001 OA type 2: (0.0 (−0.9 to 1.0))3: (−0.1 (−1.3 to 1.1))4: (−0.5 (−1.5 to 0.5))5: (−0.4 (−1.5 to 0.7))6: (−0.6 (−1.9 to 0.7))7: (−0.3 (−1.3 to 0.6)) Age (0.0 (0.0 to 0.1))Sex (4.9 (4.4 to 5.3))BMI (1.7 (1.7 to 1.7))CES-D-10 (0.0 (0.0 to 0.1))EDU (0.0 (−0.1 to 0.1))INC (0.1 (−0.1 to 0.3))SEQ (−0.1 (−0.2 to 0.1)) Covariates+NUTHC R2=0.8534 F(14,1581)=657.49p<0.0001 NUTHC (0.3 (0.0 to 0.6), p=0.06) p=0.06 Covariates+NUTFBR R2=0.8531 F(14,1581)=655.83p<0.0001 NUTFBR (−0.03 (−0.3 to 0.3), p=0.87) p=0.87 Covariates+NURSCR R2=0.8537 F(14,1581)=659.14p<0.0001 NUR SCR (−0.05 (−0.1 to 0.0), p=0.009) p=0.009 Covariates+NURCLS R2=0.8539 F(15,1580)=615.78p<0.0001 NUR CLS p=0.01 2: (0.6 (0.1 to 1.0)), p=0.009 3: (0.6 (0.1 to 1.1)), p=0.014 Covariates+PASE R2=0.8538 F(14,1581)=659.73p<0.0001 PA (−0.01 (0.0 to 0.0)), p=0.005 p=0.004
• Independent variables: frequency of high calorie snacks (NUTHC), frequency of high fibre cereal (NUTFBR), nutrition risk score (NURSCR), nutrition risk classification (NURCLS 2=moderate risk, NURCLS 3=high risk) and PASE (Physical Activity Scale for the Elderly). Covariates: OA type (1=hand, hip, knee; 2=knee; 3=hip and knee; 4=hip; 5=hand and knee; 6=hand and hip; 7=hand), age, sex, body mass index (BMI), depressive symptoms (CES-D-10), education level (EDU), income (INC), social inequality (SEQ). Bolded variables indicate those that were statistically significant (p<0.05).

• CES-D-10, 10-item Center for Epidemiological Studies Short Depression Scale ; OA, osteoarthritis; PA, physical activity.