Arthritis & Rheumatism, Volume 63,
November 2011 Abstract Supplement

Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Chicago, Illinois November 4-9, 2011.


Consensus Statements Concerning the Use of Administrative Data in Rheumatology Research and Surveillance.

Lacaille1,  Diane, Lix2,  Lisa, Bernatsky3,  Sasha, O'Donnell4,  Siobhan, Bombardier5,  Claire, Administrative Data Rheumatology Research & Surveillance,  

Arthritis Research Ctr; University of British Columbia, Vancouver, BC
University of Saskatchewan, Saskatoon, SK
McGill UHC/RVH, Montreal, QC
Public Health Agency of Canada, Ottawa, ON
Institute for Work & Health, Toronto, ON

Background/Purpose:

Administrative health databases (for remuneration & management) exist in each province & territory in Canada. They are a great resource for rheumatic disease surveillance & research, but standards are needed to ensure comparability of findings. Our purpose was to develop consensus statements in this regard.

Methods:

We convened 52 decision makers, clinicians, & researchers to a 2-day workshop. Eight months in advance, participants were organized into working groups covering 3 themes: case definitions, methods, & co-morbidity/outcomes. Each conducted systematic & scoping reviews to define themes. At the workshop, Delphi-type consensus building techniques were used to develop statements for endorsement.

Results:

13 consensus statements were endorsed. Regarding case definitions for rheumatic diseases, these should be justified based on purpose, validity assessment, & feasibility; validation studies should adhere to published guidelines on conduct & reporting; and authors should acknowledge limitations of administrative data for case ascertainment

Regarding methods, authors must address confounding by indication; use appropriate methods to address other common confounders & biases; clearly define & justify exposure risk windows: and acknowledge limitations of administrative data.

For comorbidity/outcomes, our statements note that osteoporosis diagnostic codes should not be used alone due to low sensitivity. Hospital discharge data, & physician & procedure data when available, can be used to accurately identify hip fractures. Fractures not requiring hospitalization can be identified by combining physician billing diagnoses & procedure codes. For vertebral fractures, additional research is needed.

Regarding using administrative data (exclusive of cancer registries) to define cancer outcomes, authors should use an algorithm with good sensitivity & excellent specificity in a comparable population. Implications of imperfect case definition should be discussed.

Hospitalization diagnoses can be used to ascertain serious bacterial infections. Current data is not sufficient to recommend administrative data to identify opportunistic infections.

For cardiovascular disease, hospitalization data can be used to ascertain acute myocardial infarction, but there are significant limitations for congestive heart failure.

Regarding renal disease, administrative data can be used to identify kidney disease requiring dialysis. However, current data do not support using hospitalization data for acute or chronic kidney disease as a co-morbidity or outcome.

Conclusion:

Our recommendations are consistent with other recent guidelines including the ISPOR report and the EULAR Points paper to address specific needs of rheumatic disease biologics registries. Our consensus statements include other issues as well, and some Canada-specific details. Their usefulness and implications for surveillance and research extends beyond Canada's borders.

Dissemination is in progress; for background documents see our current website archive https://connect.mcgill.ca/r41824168/

To cite this abstract, please use the following information:
Lacaille, Diane, Lix, Lisa, Bernatsky, Sasha, O'Donnell, Siobhan, Bombardier, Claire, Administrative Data Rheumatology Research & Surveillance, ; Consensus Statements Concerning the Use of Administrative Data in Rheumatology Research and Surveillance. [abstract]. Arthritis Rheum 2011;63 Suppl 10 :1881
DOI:

Abstract Supplement

Meeting Menu

2011 ACR/ARHP