Consequently, our findings regarding diseases of the various health systems may have favored the severe end of the spectrum (e.g., hypertension associated with stroke or coronary heart disease), although the testing of this hypothesis would have required access to comparable non-hospital morbidity data.