search

eribul / thamortpred

Prediction of 90 day mortality after elective THA
https://erikbulow.shinyapps.io/thamortpred/
0 stars 0 forks source link

cemented THA #54

Closed eribul closed 4 years ago

eribul commented 4 years ago

Addressed in discussion: extrapolation to countries or practices where cemented THA is less common, especially in younger patients. In the United States, for example, cemented THA is typically reserved for elderly and osteoporotic patients and this would impact the generalizability of your findings.

eribul commented 4 years ago

VI skulle ev kunna använda den föreslagna modellen och tillämpa den på hela kohorten, utan filtreriung på fixation och sedan kolla AUC etc.

eribul commented 4 years ago

Bättre tillämpa den på samtliga fall utom cementerade. Får då dessutom ett extern dataset för valideringen.

eribul commented 4 years ago

Även med all annan datar sammanslagen har vi endast 28 dödsfall. GLM konvergerardock, vilket intre är fallet om vi stratifierar på hybrider/omvända hybrider.

eribul commented 4 years ago

Review: We agree that this is unfortunate. We did try to use the model on the Swedish patients with uncemented/hybrid/reverse hybrid THA as well. We only had 28 deaths among those patients, yielding a wide and therefore non-informative 95 % confidence interval for the area under the curve (AUC; 0.54-0.76). We also tried to stratify this analysis for the uncemented/hybrids/reverse hybrids as well, which was not possible due to non-convergence of the numerical procedure used for model fitting in the regression analysis. We have now further emphasized difficulties of transportability on line 276-277.