Table 4

SDHR (95% CI) for the relationship between eCRF and breast cancer risk stratified by category of BMI and per cent body fat among postmenopausal women from the UK Biobank

ExposureUnadjusted† (N=17 840)Minimally adjusted†‡ (N=17 840)Fully adjusted†§ (N=17 840)Adjusted and restricted to protocol 1†§ (N=15 147)
BMI categoriesWithout OWOB (N=7332)With OWOB (N=10 508)Without OWOB (N=7332)With OWOB (N=10 508)Without OWOB (N=7332)With OWOB (N=10 508)Without OWOB (N=6518)With OWOB
(N=8629)
High fitness
 No¶RefRefRefRefRefRefRefRef
 Yes**0.91 (0.69 to 1.21)0.33 (0.11 to 1.02)0.94 (0.71 to 1.25)0.33 (0.10 to 1.01)0.94 (0.70 to 1.25)0.33 (0.11 to 1.01)0.89 (0.66 to 1.21)0.38 (0.12 to 1.18)
Quintiles of per cent body fat1st quintile (N=3737)2nd–5th quintile
(N=14 103)		1st quintile (N=3737)2nd–5th quintile
(N=14 103)		1st quintile (N=3737)2nd–5th quintile
(N=14 103)		Low body fat (N=3360)High body fat (N=11 787)
 High fitness
  No††RefRefRefRefRefRefRefRef
  Yes‡‡1.38 (0.85 to 2.25)0.66 (0.45 to 0.96)*1.40 (0.86 to 2.29)0.66 (0.45 to 0.95)*1.41 (0.86 to 2.32)0.65 (0.45 to 0.94)*1.33 (0.79 to 2.24)0.68 (0.46 to 1.00)

  • Fine and Gray regression analysis with other cancer, and all-cause mortality treated as competing risks was used to produce SDHRs. Age was used as a timescale to anchor findings to a clinically meaningful time point instead of arbitrary time point (ie, time since CRF assessment).

  • *Statistically significant (p<0.05).

  • †Sample sizes are based on the whole sample (ie, all those with an eCRF assessment using any of the three protocols).

  • ‡Estimates adjusted for ethnicity (white, black, other), smoke status (never, current, previous), alcohol use frequency (never, special occasions only, 1–3 times a month, once or twice a week, 3 or 4 times a week, daily) and physical activity (categorised as ≤50th percentile versus >50th percentile of MET minutes per week).

  • §Estimates are adjusted for all variables in the minimally adjusted model and education (secondary, postsecondary, missing), family history of breast cancer, age at menarche (<or≥13 years of age), number of live births (nulliparous, 1, 2, 3 or more children) and protocol (1, 2, 3) if relevant.

  • ¶Sample size in whole population is n=4018 and n=10 252 for without and with overweight/obesity, respectively.

  • **Sample size in whole population is n=3314 and n=256 for without and with overweight/obesity, respectively.

  • ††Sample size in whole population is n=1315 and n=12 955 for body fat within and greater than the first quintile, respectively.

  • ‡‡Sample size in whole population is n=2276 and n=1294 for body fat within and greater than the first quintile, respectively.

  • BMI, body mass index; eCRF, estimated cardiorespiratory fitness; MET, metabolic task equivalent; OWOB, overweight/obesity; SDHR, subdistribution HR.