Artikel
Risk factors based on routine administrative hospital data for the prediction of in-house-mortality in brain tumour surgery – Stausberg Score
Risikofaktoren für die Prädiktion der In-House-Mortalität bei Patienten nach Hirntumor-Operation, basierend auf Krankenhaus-Routine-Daten – die Rolle des Stausberg Scores
Suche in Medline nach
Autoren
-
Andreas Jödicke
- Vivantes Klinikum Neukölln, Klinik für Neurochirurgie, Berlin, Deutschland
| Veröffentlicht: | 4. Juni 2021 |
|---|
Gliederung
Text
Objective: Indicators for treatment quality (quality indicator, QI; e.g. outcome) are prone to inaccuracies. For longitudinal intra- or cross-sectional inter-departmental comparison, input variables as risk factors, other than age and gender, seem to be multiple and their share of the value of a QI is difficult to quantify. Stausberg and Hagn developed a score for the QI In-House-Mortality based on 42 ICD-10 categories, which can be easily drawn from routine hospital administrative data. We analyzed its applicability for patients having undergone brain tumor surgery.
Methods: We analyzed a retrospective mono-center cohort of patients (n=411) being operated upon a brain tumor from 2015 to 2018 on In-House-Mortality as the primary outcome. Age, gender, PCCL, relative weight (RW), length of stay (LOS), nosocomial infection (NI), type of admission (emergency room (ER), non-ER), event of re-operation and complications and the Stausberg Score on demission were retrieved from routine hospital data banks. Univariate and multivariate binary regression analyses were performed. Statistical significance was set as the probability of Type 1 error below 5% (p<0.05).
Results: In-House-Mortality in surgically treated patients with brain tumors was 2,9%. Univariate regression identified LOS, PCCL, RW, Stausberg Score, NI and type of admission as significant risk factors. Multivariate analysis revealed only Stausberg Score as a significant risk factor (OR 1,150 (95%CI 1,022 – 1,295); p=0,021; Nagelkerke’s R2: 0,403) with an 15% increase in odds per higher score point with a meaningful predictive effect (AUROC 0,85 (95%CI 0,74 – 0,93)).
Conclusion: Events with low prevalence are difficult to model and to predict. Stausberg Score seems to be a promising candidate for the prediction of In-House-Mortality in surgically treated brain tumor patients.
