Should trials that are highly vulnerable to bias be excluded from systematic reviews?