Decision analyses for prophylactic replacement of the Björk-Shiley convexo-concave heart valve: an evaluation of assumptions and estimates.

Björk-Shiley convexo-concave (BScc) mechanical heart valves have a risk of outlet strut fracture. Decision-analytic models may support decision making regarding prophylactic replacement to avert the disastrous consequences of fracture. Assumptions and estimates from previous analyses were evaluated to develop an accurate new decision model, incorporating updated follow-up experience from 2,263 patients with BScc valves implanted between 1979 and 1985 in The Netherlands. The authors focus on estimation of fracture risks (50 events) and survival (883 deaths, excluding fractures). In previous analyses, fracture risk was estimated with widely varying combinations of patient, valve, and production characteristics. Two analyses assumed a constant fracture hazard during follow-up, while data from the present study suggest that risk decreases with increasing age during follow-up. An additive excess-risk model was applied in two analyses to estimate survival. The assumption of a constant additive excess risk was not supported by the Dutch data, which suggest that the life expectancies of patients currently alive with BScc valves may be shorter than previously believed. Based on the revised decision model, over 90% of the currently alive Dutch BScc valve patients do not benefit from replacement, since the fracture risk causes only a minor reduction of remaining life expectancy. The variation in essential assumptions of previous decision analyses directly affected the indication for prophylactic replacement. This study shows how detailed statistical analyses may guide modeling choices in a decision analysis.