The validity of death certificates: routine validation of death certification and its effects on mortality statistics.

The 3478 death certificates (7.1% of all annual death certificates) of this study comprise those national death certificates in 1995 submitted for validation to the panel representing both medical and nosological expertise. As such, it is highly selected and represents, from the nosological point of view, the most inconsistently filled-in portion of Finnish death certificates. The routine validation procedure is essentially based on exploitation of the extra medical information, i.e. the case history, on the Finnish death certificate form. Altogether, 2813 (80.9%) out of 3478 certificates could be adjusted at the primary panel session; the rest required further clarification. The re-assignment of cause of death by the panel and the impact of panel adjustments on the national mortality statistics is assessed here by comparing the initial death certification and the finally registered underlying cause of death grouped into ICD-9 major categories with special reference to the subcategories of neoplasm, cardiovascular disease (HVD) and unnatural death. A statistically significant decline (p<0.0001) in deaths, both in the category of symptoms, signs and ill-defined conditions and in the pulmonary circulation disease subcategory of HVD with 37.6 and 35.1%, respectively, was observed. The decrease of 11.1% in the benign or NUD neoplasm subcategory and the increase of 8.6 and 7.0% in the categories of endocrine disease, and musculo-skeletal and connective tissue disease, respectively, are essential observations as to the quality of the cause of death register. The effect on the HVD major category was practically nil. At the HVD-subcategorial level, a decrease of 14.0% for diseases of the veins and lymphatics and other circulatory diseases and an increase of 3.5% for hypertensive diseases (HYP) were the two next most obvious alterations to the diseases of the pulmonary circulation, but were without statistical significance. For ischaemic heart disease and other subcategories, the effects were minor. The unnatural deaths as a whole increased in the final statistics with only 0.9%. In the study data, categorial changes ranged from the decrease of 75.2% for symptoms, signs and ill-defined conditions to the increase of 77.3% for endocrine diseases. In conclusion, the Finnish death certificate form, death certification practices and cause of death validation procedure seem to serve the coding of causes of death for mortality statistics appropriately. The results of the study form a relevant reference background to evaluation of epidemiological studies on mortality.