Cancer Registry

Contains structured data on patients diagnosed or treated for malignant tumors (and some benign tumors) at the University of Michigan Hospital: demographic information, tumor characterization, treatment information, and outcomes (yearly follow-up). It is a curated data-set: data is abstracted from the medical record (MiChart) by Certified Tumor Registrars according to rules set out in the American College of Surgeons' (ACOS) FORDS manual (see link) on one hand and Michigan Cancer Surveillance Program handbook on the other. Additional data might come from the Radiation Oncology system, ARIA (manually accessed by the registrars). By federal mandate, data needs to be reported to the Michigan Cancer Surveilance Program, the state registry. Data also needs to be reported to the National Cancer Data Base for the Cancer Cancer to maintain ACOS certification. Database goes back to 1994 [need to check exact date]. Data is not abstracted in real-time, there is about a 6 months lag between patient's first date of contact with the hospital and abstraction.

Data and Resources

American College of Surgeons' FORDS manual
Explore
- Preview
- Download

Additional Info

Field	Value
ACCESS - Access Contact	Gretchen Miela (miela@umich.edu)
ACCESS - Access mechanism	Currently through direct request to Gretchen Miela. IRB approval needed for all research-related inquiries that need more than aggregated information or cohort numbers (i.e. inquiries asking for specific patients MRN's, etc.)
ACCESS - Access protocol	SQL
ACCESS - Terms of Use
COMPOSITION - Data Standards utilized	need to check ICD 10 (OO?3?) for histology information; NACCR standards for the coding / storing of data.
COMPOSITION - Data element definitions	FORDS, NAACCR
COMPOSITION - Data profile
COMPOSITION - Site Breadth
COMPOSITION - Sources	(1) Demographic Information comes from MiChart and follow-up letters to the patient (2) Tumor and Treatment Information comes from MiChart documents - clinic notes, labs, outside documents; some information on treatment details is extracted from ARIA (RadOnc system). (3) Follow-up information (vital status, updated treatment, recurrence info) can come from several sources: either from the patient or from the managing physician in the form of responses to letters mailed by the registry; from the Karmanos SEER registry for patients in the Tri-County area; from the Death Index.
COMPOSITION - Subject or Concept Area Coverage
STAKEHOLDERS - Attribution Citation	TBD
STAKEHOLDERS - Data Collectors	Certified Tumor Registrars abstract data from the medical records, code it, and manually enter it in the registry database. A prerequisite for the data collection is identifying cancer cases (casefinding). This is primarily done by reviewing pathology reports. Alternative sources for casefinding include tumor boards, radiology reports, etc. Casefinding is supported by Pathology Informatics and Cancer Center Informatics. Pathology Informatics supplies extracts of pathology reports identified as positive for cancer on the basis of a query that looks for key terms indicative of cancer. Cancer Center Informatics supports a custom-built application, CaFE, that presents the pathology reports to the registrars in a format suitable for quick review. CaFE is also currently being re-written to accept clinic notes from Clarity and filter them based on a list of key terms indicative of cancer.
STAKEHOLDERS - Data Manager(s)	Gretchen Miela = registry manager
STAKEHOLDERS - Data Steward(s)	TBD - Gretchen Miela = registry manager; Alfred Chang = faculty lead (ACOS?)
STAKEHOLDERS - Principal Investigator
STAKEHOLDERS - Publisher/Data Provider	Gretchen Miela = registry manager; Cancer Center Informatics (Frank Manion director, Cristina Moisa BA on the CC Registry projects)
SUMMARY - Data Set Unique Identifier
SUMMARY - Higher-level data models (logical and/or conceptual)	start with the FORDS manual as this captures all business rules and describes the data entry fields; would also be good to refer to the NAACCR data dictionary that forms the basis of the structures
SUMMARY - Personal Health Information Indicator	Yes
SUMMARY - Physical data model, reverse-engineered	does not exist at this time
SUMMARY - Study Description
SUMMARY - UMHS Data Asset Type	Disease Registry
TIMELINE - Initial Creation Date	? Ongoing for a long time? Reference date for the database in the 90's (meaning that all patients since the reference dates are being followed for outcomes)
TIMELINE - Last Modified Date	Ongoing
TIMELINE - Retention schedule
TIMELINE - Study End Date
TIMELINE - Study Start Date
TIMELINE - Update Schedule	Daily