gms | German Medical Science

54. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

07. bis 10.09.2009, Essen

An application of propensity score methods to estimate treatment effect of corticosteroids in patients with severe cutaneous adverse reactions

Meeting Abstract

Suche in Medline nach

  • Peggy Sekula - Universitätsklinikum Freiburg, Freiburg
  • Martin Schumacher - Universitätsklinikum Freiburg, Freiburg

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 54. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds). Essen, 07.-10.09.2009. Düsseldorf: German Medical Science GMS Publishing House; 2009. Doc09gmds103

doi: 10.3205/09gmds103, urn:nbn:de:0183-09gmds1034

Veröffentlicht: 2. September 2009

© 2009 Sekula et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen ( Er darf vervielf&aauml;ltigt, verbreitet und &oauml;ffentlich zug&aauml;nglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.



Introduction: Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are severe cutaneous adverse reactions marked by high mortality. Since randomized clinical trials are hardly feasible due to their rarity, experiences with different treatments are reported from single cases or small case series which are partly contradictory. One of the aims of the EuroSCAR-study comprising a large number of cases with SJS/TEN was thus to estimate the effect of treatments on mortality. Since the patients were not randomly assigned to treatment groups, these differ not only with respect to important confounding factors but also between countries. The use of standard methods was, therefore, not very satisfactory. Alternatively, propensity score (ps) methods are proposed for such situations and assessing their applicability is the aim of the following analysis.

Methods: For the analysis of treatment data observed in the EuroSCAR-study, ps methods are applied to deal with confounding. The presentation will focus on the comparison of two treatment modalities: supportive care only (SUPP) and corticosteroids (STER).

Results: 206 French and German patients were treated either with SUPP or STER. Imbalances between the treatment groups but also between the countries can be recognized. We did not succeed in developing a ps model for the full cohort that is satisfying concerning the balance between the treatment groups. In addition, the inclusion of a variable for the country of the patient leads to a separation of the patients by country. Thus, we decided to develop ps models for each country separately and estimate the treatment effect.

Conclusions: Although several problems occurred by analysing the data, the application of ps methods was successful and provides valuable information. We could confirm findings of the original analysis. In addition, the ps methods seem to be able to better cope with complex confounding.


Schneck J, Fagot J-P, Sekula P, et al. Effects of treatments on the mortality of Stevens-Johnson syndrome and toxic epidermal necrolysis: A retrospective study on patients included in the prospective EuroSCAR study. Journal of the American Academy of Dermatology. 2008; 58(1): 33-40.
Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983; 70(1): 41-55.
D’Agostino RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-reandomized control group. Statistics in Medicine. 1998; 17(19):2265-2281.