Artikel
The ONKO-KISS study on bone marrow and peripheral bloodstem-cell transplantation and the occurrence of hospital-acquiredinfections during neutropenia: essential statistical issues.
Suche in Medline nach
Autoren
-
Jan Beyersmann
- Institute of Medical Biometry and Medical Computer Science, University Hospital Freiburg, Germany
-
M. Schumacher
- Institute of Medical Biometry and Medical Computer Science, University Hospital Freiburg, Germany
-
M. Dettenkofer
- Institute of Environmental Medicine and Hospital Epidemiology, University Hospital Freiburg, Germany
| Veröffentlicht: | 8. September 2005 |
|---|
Gliederung
Text
Bone marrow and peripheral blood stem-cell transplantationshave become a successful therapy for severe hematologicdiseases. However, with transplantation a patient enters a high riskphase, called neutropenia. Occurrence of infectious complications,mostly bloodstream infections or pneumonia, is more likely duringneutropenia, increasing morbidity and mortality. ONKO-KISS is aprospective multi-centre cohort study to provide reference data and toassess risk factors for the occurrence of bloodstream infections andpneumonia during neutropenia [Ref. 1]. It is part of the surveillance program of the German National Reference Centrefor Surveillance of Nosocomial Infections (KISS - German HospitalInfection Surveillance System; http://www.nrz-hygiene.de/).
A meaningful analysis of the data collected in this study has to judgethe (possibly different) effects a risk factor has on the competingevents `infection', `death' and `end of neutropenia'. Also, the timethese events occur is important. In statistics, this situation isknown as `competing risks' [Ref. 2]. An adequatehazard-based analysis using the well-known Cox model, say, isfeasible, but interpretationally difficult due to the competingrisks. We illustrate these difficulties and offer an alternativeanalysis using the Fine-Gray model for cumulative incidences [Ref. 3]. The cumulative incidence function for `infection', say,displays the proportion of patients infected as time progresses. Anexcellent interpretation of our results from the Fine-Gray model canbe achieved displaying these proportions for different constellationsof risk factors.
References
- 1.
- M Dettenkofer, S Wenzler-Röttele, R Babikir, H Bertz, W Ebner, EMeyer, H Rüden, P Gastmeier, FD Daschner: Surveillance of nosocomialsepsis and pneumonia in patients with a bone marrow or peripheralblood stem cell transplant. A multicenter project. ClinicalInfectious Diseases 2005 (40): 926-931
- 2.
- J Klein, J D Rizzo, M-J Zhang, N Keiding. Statistical methods forthe analysis and presentation of the results of bone marrowtransplants. Part~I: Unadjusted analysis. Bone Marrow Transplantation2001 (28): 1001-1011
- 3.
- J Fine, R Gray. A proportional hazards model for thesubdistribution of a competing risk. Journal of the AmericanStatistical Association 1999 (94): 496-509
