Conversation
Five questions on cancer registries with Selam Bekele
A registry is only as good as the records it can join up. The lead author of AJCR's Addis Ababa linkage study on duplicates, missing deaths and why a person still checks the uncertain matches.
In this story: 3 people · 1 institution · 1 paper · 1 country
People
- Selam Bekele Addis Ababa University · 2 AJCR papers
- Dawit Tesfaye Addis Ababa University · 1 AJCR paper
-
Naledi Maseko
University of Cape Town · 2 AJCR papers
Institutions
- Addis Ababa University Ethiopia · 2 AJCR papers
Papers
Countries
- Ethiopia 2 AJCR papers
Selam Bekele of Addis Ababa University led a before-and-after study of digital record linkage in cancer registration, with Dawit Tesfaye and Naledi Maseko of the University of Cape Town. We asked about the work behind it.
Why start with completeness rather than new data?
Because the records already exist. A patient appears in a pathology report, a treatment log and sometimes a death register. Completeness is the work of recognising that these belong to one person, and of not counting that person twice.
What did linkage change?
In the study's 1,842 records from nine reporting facilities, the share meeting the registry's completeness criteria rose from 61% to 91%. Most of the gain came from matching treatment and mortality records that had previously sat in separate systems.
Why keep a person in the loop?
Automatic matching handles the clear cases. The uncertain ones, such as a misspelt name, a missing date of birth or two people with the same name at one hospital, go to a registrar. A wrong merge is worse than a missed one, because it quietly corrupts two histories.
What would you tell a registry starting out?
Agree on identifiers before you buy software. Most of the hard problems are about which fields each facility records, and how consistently.
And the limits?
The study is a synthetic example, so its numbers describe a modelled registry, not Ethiopia's. Real linkage work also has to settle consent and data protection first, and those questions are not technical ones.
Sample conversation for website testing. Selam Bekele and the study are fictional, and the answers were written for this demonstration. Addis Ababa University is real but was not involved.
