The risk of coronary death associated with Type A behaviour (Framingha
m definition) and social contact was examined using a retrospective ca
se-control design. Logistic regression analysis of 134 male cases and
339 controls showed that increased risk was associated with Type B beh
aviour, and independently with limited social contact. Discussion focu
ses on issues surrounding the use of retrospective, proxy data; and on
the speculation that Type A behaviour may increase risk of non-fatal
coronary events, but protect against coronary death.