This page is part of the MII IG Kerndatensatz-Modul Onkologie (v2027.0.0-ballot.1: Release) based on FHIR (HL7® FHIR® Standard) R4. This is the current published version. For a full list of available versions, see the Directory of published versions

Guidance for Researchers

Guidance for Researchers

The data of this module originates from oncological tumour documentation along the oBDS reporting events (diagnosis, surgery, start/end of radio- and systemic therapy, follow-up, death) — the same documentation stream that feeds the German cancer registries. Two consequences of this origin matter for every analysis:

1. The data arrives with a reporting lag

Reporting events are documented and forwarded with delay — legal reporting deadlines, site-internal documentation backlogs and registry processing add up to months. For research this means:

• Recent time periods are systematically incomplete. A drop in case counts or therapy events in the most recent months is usually a reporting artefact, not an epidemiological signal.
• Define cohort cut-offs conservatively (data-freeze date well before the analysis date) and state the freeze date alongside every result.
• The absence of a recent event is weak evidence: a missing follow-up or therapy-end record may simply not have arrived yet.

2. A single site's extract can be incomplete

Oncological care is distributed: the resection may happen at a centre while follow-up and adjuvant therapy take place at other sites (practices, community hospitals, rehabilitation). A data extract drawn from one site therefore documents that site's share of the course of disease, not necessarily all of it:

• Absence of a record is not absence of treatment. A patient without a documented systemic therapy may have received it elsewhere.
• Signals inside the data can point to external treatment — for example a TNM classification carrying the y prefix (state during/after neoadjuvant therapy) without any documented systemic therapy episode.
• Be careful with denominators: "patients without adjuvant therapy" from a single-site extract mixes truly untreated patients with externally-treated ones. Prefer research questions robust to this, or use the population-based state cancer registries (Landeskrebsregister), whose cross-site consolidation is the completer source for full-course questions.
• In multi-site MII projects, record linkage across sites mitigates — but does not eliminate — this fragmentation.

Both properties are inherent to registry-fed clinical cancer documentation; they are not data-quality defects of an individual site. Analyses should name them in their limitations section.