Table 3.

Two-part model: multiple logistic and multiple-linear regression analysis

Model	Independent variables	B coefficients	Standardised regression coefficients (β)	Significant p value	95% CI for B lower/upper
Multiple logistic regression Predicted variable: absence days	Constant	11.039		0.000	6.14	15.93
	Age	−0.065		0.013	−0.116	−0.014
	WAI	−0.203		0.000	−0.293	−0.113
Multiple linear regression Predicted variable: number of absence days	Constant	427.2*	–	0.000	317.32	537.08
	Disability pension†	−106.81*	−0.52	0.000	−141.60	−72.02
	WAI	−4.66*	−0.51	0.000	−6.13	−3.18
	Age	−0.498*	−0.07	0.429	−1.75	0.76
	Gender	−10.71*	−0.06	0.414	−36.82	15.40
	N° of diagnoses‡	10.24*	0.06	0.461	−17.45	37.93

The logistic regression has a Nagelkerke R=0.458, the Hosmer and Lemeshow test was not significant (p=0.09), the Omnibus test was very small (p=0.000).

For the multiple regression, the R² was 0.724, R² adjusted 0.7, the model is significant with p<0.001.

*Unstandardised regression coefficients (B).

†Disability pension (yes/no).

‡Number of diagnoses (up to 2/>2).