Table 2.

Crude and adjusted odds ratios for factors associated with later stage at diagnosis derived from ordinal logistic regression (unknown stage excluded) N = 4362

			Crude			Adjusted#
Variables		OR	95 % CI	p-value	OR	95 % CI	p-value
Site	Colon	1.00			1.00
	Rectum	0.82	0.73-0.91	<0.001	0.81	0.72-0.91	<0.001
Age group	50-59 yrs	1.00			1.00
	60-69 yrs	0.82	0.70-0.95	<0.008	0.76	0.66-0.89	<0.001
	70-79 yrs	0.88	0.77-1.02	0.087	0.77	0.67-0.89	0.001
Sex	Females	1.00			1.00
	Males	0.97	0.87-1.08	0.562	0.96	0.86-1.07	0.436
SES	Lowest (quintile)	1.00			1.00
	Low	0.89	0.75-1.06	0.208	0.90	0.75-1.08	0.248
	Mid	0.99	0.83-1.18	0.909	0.96	0.79-1.15	0.636
	Low-high	0.93	0.78-1.10	0.395	0.90	0.75-1.08	0.268
	Highest	1.08	0.91-1.26	0.438	1.03	0.86-1.23	0.756
Residence	Inner urban	1.00			1.00
	Outer urban	0.85	0.70-1.05	0.080	0.89	0.74-1.07	0.248
	Rural	1.02	0.88-1.20	0.717	0.99	0.83-1.16	0.812
	Remote	0.92	0.73-1.16	0.473	0.99	0.77-1.26	0.912
Private Insurance	No	1.00			1.00
	Yes	0.88	0.76-0.94	0.002	0.84	0.75-0.95	0.003
Co-morbidities	None	1.00			1.00
	One (not severe)	1.25	1.09-1.44	0.002	1.25	1.09-1.44	0.002
	Multiple or severe	1.48	1.25-1.76	<0.001	1.50	1.26-1.79	<0.001
Diagnosis year	2003-2008 (cont)	0.97	0.94-1.00	0.024	0.97	0.94-1.00	0.027

#Ordinal logistic regression model adjusted for all factors simultaneously – excludes missing stage (n = 328)

Ordinal logistic regression is an extension of logistic regression that incorporates the ordinal nature of the dependent variable (in this case stage at diagnosis). It can be used for modelling a dependant variable that has more than two ordered categories. The method is analogous to a series of binary models predicting the following combinations of binary stage groupings (e.g. stage A v stage B + C + D, A + B v C + D and A + B + C v D)