This was an invited lecture at the STRATOS session of the annual conference of the International Society for Clinical Biostatistics. The lecture gave an overview of the most commonly met issues in survival analysis. Many of these issues can be seen from two points of view. On a basic level, the misunderstanding of the issues and usage of overly simplistic approaches can lead to considerable biases and misleading results. Two important examples of this are the immortal bias and the fact that a competing event preventing the primary event from happening is falsely recognised as right-censoring. On the second level, a better understanding of the available methodology can lead to a more complex analysis that allows for optimal usage of the information in the data. Here we shall address the estimation of transition intensities and transition probabilities. As a further example, we shall turn to issues in the relative survival field, which can be seen as a special case of competing risks where the cause of event is unknown.
B.04 Guest lecture
COBISS.SI-ID: 32835545The lecture was given at an important international conference, which united three areas of population based survival analysis, and was as such of extreme relevance to our work. Our talk focused on estimation of net survival, where only recently, a consistent estimator has been introduced. Its use in practice has revealed an excessively large variance when estimating net survival of older age groups. We first simplified the problem by considering a non-censored case to show that the problem of large variance is intrinsic to the definition of net survival and not a property of a specific estimator. We then continued from the definition of net survival and generalize it to the censored case by the use of pseudo-observations. The estimator developed in this way has all the desired properties, we also provided a formula for its variance. Since pseudo-observations are available in several statistical packages, this new estimator is easy to implement. It has several interesting theoretical properties, its main advantage in practice is the fact that it does not require numerical integration. This also implies it can be directly used with life-table data, i.e. data grouped in intervals of time. We illustrated the properties of our proposal with simulations and a real data example of colon cancer patients.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 32836313Editorial work in the Slovene Medical Journal - Maja Pohar Perme
C.04 Editorial board of an international magazine
COBISS.SI-ID: 32893696