Figure 5. Temporal trends in 30-day perioperative death incidence in clinical registries.

Meta-regression using random effects model with bubble size proportional to study precision. There was similarly a statistically significant decrease in the 30-day incidence of mortality over the period 1991–2013, odds ratio 0.95 (95% credible interval 0.92–0.97, p < 0.001), translating into a 5% year on year mean risk reduction. The limits of the 95% credible interval and prediction interval have been drawn respectively in red and in blue. These data do not include preprocedural arteriography-related events.