Incomplete and inaccurate death certification - the impact on research

Background The objectives of this study were (1) to investigate the extent of erroneous and/or omitted information on death certificates of patients implanted with Bjork-Shiley Convexo-Concave (BSCC) heart valves; (2) to determine whether this information could be associated with a possible under-reporting of acute mechanical failure of this valve. Methods A review was carried out of death certificates and clinical notes for patients implanted in the United Kingdom with BSCC valves. This was a multicentre study (38 hospitals) based at the Cardiothoracic Department, NHLI, imperial College School of Medicine at Hammersmith Hospital, London. The subjects were 478 patients implanted with a BSCC valve between 1979 and 1986 who died in the following years: 1984, 1987, 1990, 1993 and 1996. The main outcome measures were: (1) percentage of death certificates that record the presence of a valve prosthesis; (2) percentage of death certificates that record the presence of a valve prosthesis for patients who had a post mortem; (3) percentage of death certificates that record inaccurate or incomplete information related to the surgery; (4) percentage of death certificates that do not record a post mortem where one is known to have been performed. Results Twenty-one per cent (101/478) of the total number of death certificates record the presence of the valve prosthesis. Thirty-five per cent (43/123) of the death certificates for patients who had a post mortem record the presence of a valve prosthesis. Six per cent (30/478) of death certificates report inaccurate information related to the valve surgery. Twenty-five per cent (118/478) of the total number of death certificates recorded a single cause of death. Twenty-three per cent (110/478) of ail death certificates reviewed recorded only the mode of dying. Eight per cent (10/123) of the total number of death certificates for patients who had a post mortem did not record a post mortem. Conclusions The relatively high number of death certificates that do not record the presence of a valve prosthesis and the observed under-reporting of post mortems may lead to inaccurate reporting of the number of BSCC valves that fail. Previous recommendations to improve accuracy in death certification appear to have gone unheeded, and changes in the way certificates are completed for patients with implanted cardiac devices should be considered.