A Study of the Validity of the Diagnosis of Stroke in Mortality Data

Clinical records of residents of New Haven, Conn., who died from cerebrovascular disease between 1962–1964 were abstracted for data concerning the final strokes. Of 690 possible records, 620 were abstracted. The method by which the records were reviewed by computer is described in detail along with the criteria for diagnosis of the different types of strokes. The diagnoses for each decedent found in the autopsy report, clinical record, death certificate and by computer were compared with each other and graded as to whether they were identical, similar or dissimilar. The diagnoses from the autopsies and records (133 cases) agreed in 79% of cases; from autopsies and certificates in 65% of cases; and of diagnoses from all the records abstracted and which were not for decedents who had definitely died in the absence of cerebrovascular disease (607 cases), 74% agreed with the certificates. The reliability of the diagnoses was graded into five categories ranging from autopsy confirmation to no evidence available. Evidence from autopsy diagnoses and good clinical records indicated that cerebral hemorrhage was over-diagnosed on death certificates at the expense of thrombo-embolism. The hemorrhage-thrombosis ratio from certificate data was 2.7:1 and from autopsy and clinical data, .78:1.