Fig. 1

Outcomes by race/ethnicity and FEV₁ impairment defined using white and/or black normal values, in terms of (a) breathlessness, and (b) mortality. Breathlessness probability was analyzed using multinomial logistic regression, and mortality using Cox proportional hazards regression. Impaired lung function was defined as a forced expiratory volume in one second (FEV₁) < lower limit of normal (LLN) using GLI-2012 predicted normal values for white and black people, respectively [5]. Groups were categorized by race/ethnicity and FEV₁ impairment according to different race-specific prediction equations as: ‘White Normal’ (white race/ethnicity with FEV₁ ≥ LLN_white); ‘Black Normal’ (black race/ethnicity with FEV₁ ≥ LLN_white); ‘Black Abnormal (White Reference)’ (black race/ethnicity with FEV₁ < LLN_white but ≥ LLN_black); ‘Black Abnormal (Black Reference)’ (black race/ethnicity and FEV₁ < LLN_white and < LLN_black); and ‘White Abnormal’ (white race/ethnicity and FEV₁ < LLN_white). The main finding is that black people who were categorized as having a normal FEV₁ using LLN_black but not using LLN_white had increased breathlessness prevalence and mortality compared with people categorized as normal using reference values for white. Thus, black reference values misclassify black people as having normal lung function despite having worse outcomes. When defining normality using LLN_white for all, people with normal FEV₁ had similar breathlessness and mortality in both white and black people