Mayak workers study cohort. An inter-institutional comparison of causes of death in the cause-of-death register of Ozyorsk in the Russian Federation.

Background: The cause-of-death register at the Southern Urals Biophysics Institute (SUBI), Ozyorsk, Russia, was established to document the number and causes of deaths in the Mayak workers cohort, which includes all persons (N = 22,377) employed at Mayak nuclear facility between 1948 and 1982. Most workers were occupationally exposed to high doses of ionizing radiation and have been shown to have increased risks of various chronic diseases including cancer.

Objectives: To investigate the quality of cause of death coding in the SUBI register.

Methods: A random sample of 246 deaths (~1% of the total) was coded independently at the SUBI and the Danish Cancer Society using the International Classification of Diseases 9 (ICD-9). Proportions of matching codes were calculated.

Results: Overall, 233 deaths (95%) were identically classified using the ICD-9 main category matching. Excluding mismatches that were considered to be incorrectly coded during validation, the validity of the register increased to 98%. Using the specific ICD-9 first three-digit matching, 182 deaths were identically coded (74%) and the respective validity of the register was 85%. There were also some non-resolvable discrepancies demonstrating limitations of assigning one code for each death or using language-adapted ICD-9 version.

Conclusions: This validation study was an important quality check of a register used for mortality follow-up in a highly influential epidemiological study on radiation-related health effects. The results of the inter-institutional comparison were generally favourable; however, since the comparison revealed individual mismatches and some systematically differing coding practices, it is essential to repeat it on a regular basis in order to maintain a high quality.