Correlation versus linear regressionBMJ 2013; 346 doi: https://doi.org/10.1136/bmj.f2686 (Published 26 April 2013) Cite this as: BMJ 2013;346:f2686
- Philip Sedgwick, reader in medical statistics and medical education
- 1Centre for Medical and Healthcare Education, St George’s, University of London, Tooting, London, UK
A recent statistical question described how researchers investigated the association between right ventricular size and pulmonary hypertension.1 A cross sectional study design was used. Participants were 190 patients referred to a pulmonary hypertension clinic.2
Measurements of right ventricular size included right ventricular end systolic area (RVESA) recorded echocardiographically. Pulmonary artery systolic pressure (PASP) was used to indicate the extent of pulmonary hypertension. A scatter plot of pulmonary artery systolic pressure against right ventricular end systolic area was presented (figure⇓). Linear regression analysis was used to examine the association between right ventricular size and degree of pulmonary hypertension, with the resulting fitted linear regression line given by PASP=2.7133RVESA+15.717. A significant correlation existed between right ventricular end systolic area and pulmonary artery systolic pressure (r=0.74; P<0.001).
Which of the following statements, if any, are true?
a) The regression line facilitated the prediction of pulmonary artery systolic pressure from right ventricular end systolic area
b) The regression line implied there was a causal association between pulmonary artery systolic pressure and right ventricular end systolic area
c) Correlation quantified the strength of the linear …