Randomized controlled trials of switching to a new medication vs continuing a chronic medication may underestimate the risk of adverse events in those continuing because of asymmetric depletion of susceptibles