analysis-of-indications-for-selectively-missing-results-in-comparative-registry-based-studies