Table 4.

Multivariable analysis of CRRM.

	Odds ratio (95% CI)	p value
AJ to cancer (years)	0.96 (0.90–1.02)	0.170
Step-change	9.61 (3.54–26.14)	< 0.001
Change in gradient post-AJ (per year)*	0.86 (0.69–1.08)	0.189
Age at breast cancer (per decade)	0.30 (0.20–0.43)	< 0.001
Estimated lifetime risk (%)		< 0.001
< 30%	–	–
30–39%	2.26 (1.08–4.75)	0.031
40+%	5.90 (2.97–11.73)	< 0.001

Results are from a multivariable binary logistic regression model. All factors in Table 1 were initially considered for inclusion in the model, with forwards stepwise approach used to select those that were independently associated with CRRM. Factors selected by the stepwise procedure were then entered into a model alongside the factors from Table 2. Bold p values are significant at p < 0.05. *Represents the difference between the gradients in the pre- vs. post-AJ periods. Re-parameterisation of the model found the gradient in the post-AJ period to have an odds ratio of 0.83 per year (95% CI 0.67–1.02, p = 0.076).