Sök artiklar i SveMed+

Tidskrift

Sammanfattning

Predictions of individual residual survival times are often requested. However, both clinical experience and focused studies show that it is hard to be even close to target. Might the widespread use of detailed statistical analysis of mortality from specific diseases be helpful in making such predictions more precise? We draw on general statistical considerations, illustrated by a concrete example, to show that for common mortality patterns even a completely correct statistical model without estimation uncertainty will not yield useful individual predictions. We briefly discuss the practical consequences and suggest a version of what one can tell the patient.

MeSH