Fig. 5
From: Are socio-economic inequalities in breast cancer survival explained by peri-diagnostic factors?
Excess hazard ratio of breast cancer death associated with increasing deprivationa: baseline and adjustedb in multivariable models fitted to imputed data. a) Women diagnosed 1 January 1988–31 December 2010 followed up to 31 December 2014, all tumours and early stages. b) Women diagnosed 1 January 2005–31 December 2010 followed up to 31 December 2014, all tumours. c) Women diagnosed 1 January 2005–31 December 2010 followed up to 31 December 2014, early stages. a Single unit increases derived from linear models are displayed with solid symbols. Where the effect was found to be non-linear, (e.g. (c)) numbers displayed within the symbols correspond to the deprivation quintile compared to the least deprived group (quintile 1). b Symbols are displayed only when the addition of the variable resulted in a significant improvement in the model fit (p < 0.05). Variable descriptions (see text for full coding), Baseline: Model adjusted for age and year of diagnosis only, Stage: Stage of disease at diagnosis, Alcohol: Drinking habits, BMI: Body Mass Index (kg/m2), Charlson: Charlson co-morbidity score, Consult.: Number of consultations for any reason, Br. Consult.: Number of consultations for breast symptom, Br. Sympt: Number of breast symptoms reported, Yrs prior diag: Number of days from last breast-related consultation to diagnosis, Referrals: Number of referrals for breast cancer, Time surg.: Number of days from diagnosis to major breast surgery