Therese Andersson

To understand the disease progression and survival after a cancer diagnosis, it is important to fully capture the complex disease pathways that can occur. More advanced statistical methods are needed to be able to do this in the best way. MPN is a chronic blood tumor disease, with 400 new cases in Sweden annually. Despite efforts to optimize treatment and take preventive measures against complications, patients with MPN have poorer survival than the general population and a high risk of complications such as thromboembolism and bleeding. Within this project, I will develop statistical models for survival data with several time scales and several possible outcomes. A major focus will be to enable reporting of the results of complex analyzes in a way that is easy to understand and to develop user-friendly software so that others can use the methods. The utility of the methods will be demonstrated for myeloproliferative neoplasias (MPNs), with unanswered, clinically important research questions. I will use a population-based cohort of MPN patients with long follow-up time and with matched population controls. 1. Measure how the risk of thromboembolism and bleeding among MPN patients changes over time since diagnosis and changes due to the fact that patients age. Determine whether the risk of thromboembolism increases among patients who have had bleeding and whether the risk of bleeding increases among patients who have had thromboembolism. Describe the entire disease chain from MPN to complications of thromboembolism and bleeding, and finally death, and the proportion of patients who suffered from thromboembolism and bleeding, or both, and died over time. 4. Calculate the life expectancy of MPN patients and lost life years.

Sweden has numerous linkable registers which are a goldmine for research in the medical and social sciences. However, as increasingly more complex research questions are being addressed, there are challenges to correctly and efficiently analyse such data, and further methods development is needed. We aim to develop and apply methods for register-based research that are relevant for a wide range of settings. We also aim to transfer new methodology to subject-matter scientists and train the next generation of biostatistical scientists. The long-term goal is to leverage the rich information within the registers for the benefit of patients, policy-makers and society. Specifically, we will develop, evaluate and apply methods for:1. Missing data in registers2. Causal mediation analysis3. Policy-relevant complex disease models and lifetime outcomes in health economics 4. Federated analyses and synthetic data to share information.Our proposal focuses on the development of statistical methods, but each aim is motivated by real problems in medicine and social science, e.g. improving cancer screening, understanding causal mechanisms for social inequalities in cancer survival, time trends in disease incidence trends and situations when individual-level data cannot cross borders. Our team of biostatistical scientists will work in close collaboration with an extended network of Swedish and Nordic register holders, epidemiologists, clinical and social scientists, and relevant stakeholders.

The prognosis following a diagnosis of cancer, or other diseases, is often measured using survival proportions. Recently, however, there has been an increased interest in presenting other measures of patient survival, that are easier to interpret and can give a better understanding of the impact a diagnosis has on an individual’s life span. I have previously developed such methods, namely methods for estimating the life expectancy for cancer patients and the loss in life expectancy (LLE) compared to cancer-free individuals. To further enhance the usefulness of the measures, further extensions are needed. The aim of this project is to further develop and apply the methods, specifically toCombine causal mediation analysis and LLE to quantify the impact of potential drivers of socioeconomic differences in survivalIncorporate uncertainty in general population mortality in the estimation of LLEInvestigate assumptions in survival extrapolation for estimation of LLEThe developed methods will be applied to cancer registry data, to answer research questions of clinical and public health importance. For example, to investigate socioeconomic inequalities in survival of patients with malignant melanoma and colorectal cancer, to estimate the LLE for hematological cancers with an indolent disease course, as well as for patients diagnosed with malignant melanoma at a young age. The methods have potential to be used also in other disease areas.

The prognosis for cancer patients is often measured using survival rates. Recently, however, there has been increased interest in presenting other measures of cancer patient survival, which are easier to interpret and which can provide a better understanding of the impact a cancer diagnosis has on an individual's lifespan. I have previously developed such methods, namely methods for estimating the average life expectancy of cancer patients and the loss in life expectancy (LLE) compared to cancer-free individuals. In order to further increase the usefulness of the measurements, they need to be further developed. The purpose of this project is to further develop and apply the methods. The new methods will be applied to cancer registry data, to answer research questions of clinical interest and which are important from a public health perspective. For example, to investigate socio-economic inequalities in survival in patients with malignant melanoma and colorectal cancer, to estimate the LLE for hematological cancers with an indolent course of disease, as well as for patients diagnosed with malignant melanoma at a young age. More specifically, I will: - Combining causal analysis and LLE to quantify the impact of potential causes behind socioeconomic disparities in cancer patient survival - Include uncertainty in general population mortality in the estimate of the LLE, so that confidence intervals for the LLE are correctly estimated - Examine assumptions made when estimating LLE, to see in which situations assumptions do not apply and design alternative assumptions

To understand the disease progression and survival after a cancer diagnosis, it is important to fully capture the complex disease pathways that can occur. More advanced statistical methods are needed to be able to do this in the best way. MPN is a chronic blood tumor disease, with 400 new cases in Sweden annually. Despite efforts to optimize treatment and take preventive measures against complications, patients with MPN have poorer survival than the general population and a high risk of complications such as thromboembolism and bleeding. Within this project, I will develop statistical models for survival data with several time scales and several possible outcomes. A major focus will be to enable reporting of the results of complex analyzes in a way that is easy to understand and to develop user-friendly software so that others can use the methods. The utility of the methods will be demonstrated for myeloproliferative neoplasias (MPNs), with unanswered, clinically important research questions. I will use a population-based cohort of MPN patients with long follow-up time and with matched population controls. 1. Measure how the risk of thromboembolism and bleeding among MPN patients changes over time since diagnosis and changes due to the fact that patients age. Determine whether the risk of thromboembolism increases among patients who have had bleeding and whether the risk of bleeding increases among patients who have had thromboembolism. Describe the entire disease chain from MPN to complications of thromboembolism and bleeding, and finally death, and the proportion of patients who suffered from thromboembolism and bleeding, or both, and died over time. 4. Calculate the life expectancy of MPN patients and lost life years.