Because the central question of PLCO was whether mortality due to prostate, lung, colorectal or ovarian cancer (P,L,C,O cancer) was reduced by screening, it was essential to confirm and collect data on all deaths among study participants during the trial. It was also necessary to ascertain whether a P,L,C,O cancer was the cause of death.
PLCO Screening Centers first learned about the occurrence of deaths from three main sources:
- Annual Study Update (ASU) forms were mailed to each active participant's address on an annual basis. If the participant had died, relatives or the postal service often notified the centers of the death. If no such notification was received, centers attempted follow-up contact with the participant and learned about more deaths that way. Some centers also consulted the Social Security Death Index or obituaries in conjunction with the study update forms.
- Newsletters and birthday cards mailed to participants sometimes prompted a notification of death from a participant's relative.
- A National Death Index search was conducted annually for participants who had not responded to repeated attempts to contact them.
Once the centers were aware that deaths had occurred, they obtained death certificates from state vital statistics bureaus and shipped them to centralized facilities for data compilation. All cause of death information listed on the death certificate was coded (in ICD-9), and a single underlying cause of death was determined by rules established by the National Center for Health Statistics.
For participants whose study data indicated possible death from a P,L,C,O cancer, the Death Review Committee (DRC) ascertained whether a P,L,C,O cancer was the cause of death. Deceased participants selected for DRC review included those with:
- Evidence of a P,L,C,O cancer
- Evidence of a condition associated with a P,L,C,O site
- Death of unknown cause
For most site-specific analyses (e.g., colorectal cancer mortality), death data can be obtained within the site-specific datasets (e.g., the colorectal dataset).