Evaluations of statistical methods for outlier detection when benchmarking in clinical registries: a systematic review