Risk of cancer in regular and low meat-eaters, fish-eaters, and vegetarians: a prospective analysis of UK Biobank participants

Watling, Cody Z.; Schmidt, Julie A.; Dunneram, Yashvee; Tong, Tammy Y. N.; Kelly, Rebecca K.; Knuppel, Anika; Travis, Ruth C.; Key, Timothy J.; Perez-Cornago, Aurora

doi:10.1186/s12916-022-02256-w

Table 2 Summary of estimated direct effect, indirect effect, and total effect using potential mediators for the association of diet groups in comparison to regular meat-eaters and risk of all cancer, colorectal cancer, postmenopausal breast cancer, and prostate cancer risk

From: Risk of cancer in regular and low meat-eaters, fish-eaters, and vegetarians: a prospective analysis of UK Biobank participants

	Potential mediators (hazard ratio; 95% CI)
All cancer	Mediation through BMI	Mediation through IGF-I^a	Mediation through free testosterone^a
Low meat-eaters versus regular meat-eaters	Mediation through BMI (n = 450,412)
Total effect	0.99 (0.96–1.00)
Natural indirect effect	0.99 (0.98–1.00)
Natural direct effect	0.99 (0.95–1.00)
	Mediation through BMI (n = 256,727)
Fish-eaters versus regular meat-eaters
Total effect	0.90 (0.83–0.97)
Natural indirect effect	0.99 (0.90–1.01)
Natural direct effect	0.90 (0.83–0.98)
	Mediation through BMI (n = 254,709)
Vegetarians versus regular meat-eaters
Total effect	0.86 (0.78–0.96)
Natural indirect effect	0.94 (0.81–1.08)
Natural direct effect	0.92 (0.77–1.09)
Colorectal cancer	Mediation through BMI (n = 450,412)
Low meat-eaters versus regular meat-eaters
Total effect	0.91 (0.85–0.97)
Natural indirect effect	1.00 (0.98–1.02)
Natural direct effect	0.91 (0.86–0.97)
Postmenopausal breast cancer	Mediation through BMI (n = 111,574)	Mediation through IGF-I (n = 103,853)	Mediation through free testosterone (n = 93,662)
Vegetarians versus regular meat-eaters
Total effect	0.82 (0.68–0.99)	0.86 (0.71–1.05)	0.86 (0.71–1.05)
Natural indirect effect	0.83 (0.63–1.08)	0.91 (0.73–1.15)	1.06 (0.76–1.38)
Natural direct effect	0.99 (0.79–1.23)	0.94 (0.70–1.21)	0.81 (0.62–1.05)
Prostate cancer^b		Mediation through IGF-I (n = 126,538)	Mediation through free testosterone (n = 116,087)
Vegetarians versus regular meat-eaters
Total effect		0.71 (0.56–0.92)	0.71 (0.56–0.92)
Natural indirect effect		1.10 (0.77–1.56)	0.99 (0.67–1.48)
Natural direct effect		0.64 (0.50–1.01)	0.71 (0.51- 1.01)
Fish-eaters versus regular meat-eaters			Mediation through free testosterone (n = 116,186)
Total effect			0.80 (0.65–0.99)
Natural indirect effect			0.95 (0.70–1.29)
Natural direct effect			0.86 (0.56–1.32)

All models used age as the underlying time variable and are stratified by sex (for only all cancer and colorectal cancer) and age groups at recruitment, and adjusted for region of recruitment, height, physical activity, Townsend deprivation index, education, employment status, smoking status, alcohol consumption, ethnicity, diabetes status, and body mass index (except when it was considered a potential mediator). For all cancer and colorectal cancer, models are further adjusted for menopausal hormone therapy use and menopausal status. Colorectal cancer models are adjusted for non-steroid anti-inflammatory drug use. For prostate cancer, models are further adjusted for marital status. For breast cancer, models are further adjusted for menopausal hormone therapy use, age at menarche, and age at first birth/ parity. Full details for each covariate are provided in the statistical analysis section in the main text.
Mediation analyses restricted to significant associations between diet-cancer in the main analyses (Fig. 1) and if there was a significant difference in biomarker concentrations between diet group (Additional File 1 Table S7).
Natural indirect effect represents the estimated association of diet group and cancer outcome through the potential mediator
Natural direct effect represents the estimated association of diet group and cancer outcome not through the potential mediator
Models exclude participants with missing values for mediator(s)
^aModels are adjusted for BMI
^bBMI not assessed as a mediator with total prostate cancer risk. Association of IGF-I and free testosterone presented as both hormones have been associated with prostate cancer risk. IGF-I concentrations not assessed for fish-eaters as no difference in concentrations in comparison to regular meat-eaters was observed.
Abbreviations: BMI body mass index, CI confidence intervals, IGF-I insulin like growth factor-I

Back to article page

ISSN: 1741-7015

Contact us

Submission enquiries: bmcmedicineeditorial@biomedcentral.com
General enquiries: info@biomedcentral.com

BMC Medicine

Contact us