gms | German Medical Science

MAINZ//2011: 56. GMDS-Jahrestagung und 6. DGEpi-Jahrestagung

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V.
Deutsche Gesellschaft für Epidemiologie e. V.

26. - 29.09.2011 in Mainz

Does additional confounder information alter the results of a database study on the risk of bleeding associated with phenprocoumon

Meeting Abstract

Suche in Medline nach

  • Sigrid Behr - BIPS Universität Bremen, Bremen
  • Walter Schill - BIPS Universität Bremen, Bremen
  • Iris Pigeot - BIPS Universität Bremen, Bremen

Mainz//2011. 56. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds), 6. Jahrestagung der Deutschen Gesellschaft für Epidemiologie (DGEpi). Mainz, 26.-29.09.2011. Düsseldorf: German Medical Science GMS Publishing House; 2011. Doc11gmds234

doi: 10.3205/11gmds234, urn:nbn:de:0183-11gmds2343

Veröffentlicht: 20. September 2011

© 2011 Behr et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.de). Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.


Gliederung

Text

Background: Claims databases are a substantial source for pharmacoepidemiological studies although they often lack information on important confounders.

Objectives: Estimating the bleeding risk in patients treated with phenprocoumon by combining claims data with additional information on BMI and smoking behavior using two-phase methodology.

Methods: We conducted a nested case-control study using German health insurance claims data of the years 2004-2007 (phase 1). Cases were patients hospitalized for bleeding. Twenty controls without bleedings were selected for each case. Additional confounder information was obtained from interviews in a sub-set of 505 insurants (phase 2). The odds ratio (OR) of bleeding for phenprocoumon treatment was calculated by multivariate logistic regression using data from the complete case-control dataset, including available confounder information. In addition, a two-phase analysis was conducted estimating the OR of bleeding under consideration of phase 2 data on BMI and smoking behavior.

Results: The phase 1 case-control study included 1,261 cases and 25,220 controls sampled from a study cohort of 190,746 insurants. In the phase 1 analysis, we observed an adjusted bleeding OR of 4.38 (95% confidence interval (CI) 3.04-6.30) for males aged 55 taking phenprocoumon. The bleeding risk associated with phenprocoumon use decreased significantly (p<0.05) with increasing age and there was a trend for a higher risk in females. The two-phase analysis revealed current smoking and a high BMI (≥30 kg/m²) as independent risk factors for bleeding. While BMI did not alter the effect of phenprocoumon on bleeding risk, a significant interaction was observed for smoking with a higher risk for non-smokers. Ignoring the smoking interaction in the two-phase model, the OR for phenprocoumon use was comparable to the phase 1 estimate (OR 4.49, 95% CI 2.56-7.88).

Discussion: Phase 2 information on smoking and BMI added valuable information to the study of bleeding risk associated with phenprocoumon use. The observed interaction between smoking and phenprocoumon use might be explained by the short-term effect of smoking on blood clotting. However, more research is needed on the magnitude of this interaction.