The reliability of Swedish mortality statistics for rheumatoid arthritis

The reliability of Swedish mortality statistics for rheumatoid arthritis (RA) was investigated. All Swedish RA death certificates for the years 1971 and 1975 were collected. 1224 such death certificates were identified. Physicians' causal statements were compared with the corresponding National Central Bureau of Statistics' (NCBS) registration. The NCBS was found to have registered a twofold RA mortality increase for men, from PMR 0.044 to 0.112, and a threefold increase for women, from 0.180 to 0.577, whereas physicians had reported a slight decrease for men, from 0.057 to 0.043 and practically no change at all for women, from 0.213 to 0.216. The proportions of inadequately completed certificates were 34% in 1971 and 31% in 1975, and the proportions of altered certificates were 16% and 32% respectively. The diverging agreement between physicians and the NCBS was due mainly to an increased tendency on the part of the NCBS to favour RA in the registration, and not to physicians' formal errors on the certificate forms.