Abstract

STUDY OBJECTIVE: To evaluate the completeness of notification of deaths by the National Health Service Central Register (NHSCR) for England and Wales. DESIGN: Deaths for a birth cohort were ascertained through scanning the relevant volumes of NHSCR. Attempts were made to confirm these deaths and additional deaths were ascertained through searching local records. Logistic regression was used to investigate how the probability of a death being missed by NHSCR varied with the year of birth, age at death, sex, and social class. SETTING: Deaths up to the end of 1989 in the CA postal area among 264,046 children born between 1950 and 1989 to mothers living in Cumbria. RESULTS: NHSCR originally ascertained 4139 deaths; local searches confirmed 3338 (81%) of these and found an additional 342. Most deaths missed by the NHSCR were neonatal deaths in the 1950s and 1960s. In the 1950s, 31% of children who died in the neonatal period either were not entered on NHSCR or, if they were entered, there was no record of their death. For children born from 1970 onwards, ascertainment of deaths through NHSCR was over 99% complete. CONCLUSIONS: The NHSCR was started in 1948 for the administration of records of National Health Service patients. It seems that many babies who died soon after birth were not therefore recorded. In parallel with the increasing use of NHSCR for epidemiological purposes, there has been a substantial and continuing improvement in its clerical procedures since the mid 1960's.