Analysis of mortality in French diabetic patients from death certificates: a comparative study.

This study was implemented in France to determine the causes of death in diabetic patients, whether diabetes was mentioned or not on the death certificate, and to assess the underestimation of the prevalence of diabetes at death. Two stratified random samples of death certificates were selected in the national mortality data base. The first included certificates mentioning diabetes as a cause of death (cases). The second, included certificates with no mention of diabetes (controls). For each certificate, a record form was sent to the certifying physician to ascertain diabetes in the first group and to trace unrecorded diabetes in the second group. In case of diabetes, the characteristics of the patient and his disease were collected (age at onset, treatment, complications ...). We obtained complete data for 325 cases and 959 controls. Among cases, 1% of the subjects were not confirmed as diabetic, while almost 10% of the controls were identified as having diabetes. The corresponding ratio of the corrected prevalence at death to that provided by the French statistics was estimated to 4.0 in men and 3.1 in women. Particular features are that 2% of the total diabetic decedents died from acute metabolic complications (diabetic or hyperosmolar coma, acidoketosis, or acute hypoglycemia), and that 33% of the unreported diabetic decedents under 45 died from trauma or poisoning. These results show that in France, the death rates published in the statistics for diabetes dramatically underestimate the impact of diabetes. A high risk of death is linked to this disease, particularly in people aged under 45, a problem that health deciders should address.