Table 1.

Logistic regression models evaluating the association of post-diagnosis anxiety and depression symptoms with the odds of having an unhealthy lifestyle at the end of follow-up (either 2006 or 2010)

	Stratified Analyses
	Anxiety		Depression
	Healthy	Unhealthy	Healthy	Unhealthy
	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)
Sample size	53	94	76	128
(number of women with unhealthy lifestyle at follow-up)	(21)	(81)	(26)	(109)
Odds Ratio	2.42	5.30a	0.36	0.32a
(95% CI)	(0.69, 8.50)	(0.81, 34.82)	(0.05, 2.55)	(0.10, 1.04)
	Interaction: p=0.73		Interaction: p=0.81
	Anxiety		Depression
	≤10 years of FU	> 10 years of FU	≤10 years of FU	> 10 years of FU
	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)	(Higher vs. lower symptoms)
Sample size	102	45	92	112
(number of women with unhealthy lifestyle at follow-up)	(71)	(31)	(61)	(74)
Odds Ratio	6.03*	0.80	1.44	0.35a
(95% CI)	(1.39, 26.17)	(0.22, 2.98)	(0.28, 7.37)	(0.10, 1.24)
	Interaction: p=0.05		Interaction: p=0.08

Notes. Sensitivity analyses using only the 2010 lifestyle score as the outcome led to highly similar estimates but with larger confidence intervals compared to results obtained with the 2006–2010 assessments combined. To ensure more stable estimates and make best use of the available sample, only the latter results are presented. Models adjusted for age at diagnosis (continuous), year of diagnosis (continuous), time between cancer diagnosis and completion of the anxiety or depression questionnaire (continuous), cancer stage (advanced [stages III–IV] vs. non-advanced [stages 0–II]), tumor location (proximal colon vs. distal colon and rectal tumors), marital status (in a relationship vs. not), education level (registered nurses vs. university degree), and pre-diagnosis lifestyle score (ranges from 0 to 5). CI=confidence intervals, FU=follow-up;

^ap≤0.10,

^*p≤0.05,

^**p≤0.01,

^***p≤0.001.