Data consistency in summary measures of population health

We explored these two research questions on the basis of empirical data for two common diseases in the Netherlands: breast cancer and major depression. We chose these disorders for two reasons: they are both important health problems and they allow us to study the research questions from different perspectives. Breast cancer is important in particular because of the mortality it causes, while depression causes mainly morbidity. Also, for breast cancer epidemiological data are easily available and regarded as relatively reliable, whereas for major depression the data still suffer from several problems. Finally, the disease staging for breast cancer and major depression are based on different concepts. For major depression, stages were differentiated according to severity classes (e.g. mild, severe), while for breast cancer phases in the disease pathway were used (e.g. diagnosis and therapy, metastasised). The research in this thesis consists of two parts. In part A we address the first research question: the validity and usefulness of disease models. This question can best be studied using data for a disease with well-described epidemiology. As cancer incidence and mortality are registered on a regular basis in the Netherlands and are regarded as relatively reliable, such data provide a good basis for studying this question. Chapter two therefore studies the validity and usefulness of IPM models using relatively reliable and complete data sets on breast cancer and three other common types of cancer. The results of these analyses showed us that time-trends in the epidemiological frequency data bias the outcome of these models. For breast cancer many additional epidemiological data (e.g. survival, prevalence, etc.) are available, allowing us to quantify, in chapter three, the impact of data problems and trends on the model for breast cancer. The last chapter of part A, chapter four, describes an application of a disease model to the less well monitored epidemiology of major depression to obtain internally consistent estimates for the epidemiological parameters of major depression. The second part of this thesis, part B, is concerned with tailoring health status valuations to the epidemiology and assessing their impact on the resulting summary measure. Since tailoring is a problem especially in diseases that are heterogeneous and/ or have unclear case-definitions, we thought it relevant to study this problem for major depression. In the Netherlands, the Netherlands Mental Health Survey and Incidence Study (NEMESIS) provided a good database for major depression with information on both prevalence and health status by severity class. These data enabled us to use the severity classes in the tailoring of the DWs to the epidemiology. In chapter five we compare disability between the severity classes. Chapter six uses this information to derive health status values per severity class that we subsequently used in a burden of major depression calculation. A comparison of the results with studies using nontailored values gives an impression of the importance of health status values on the overall burden of disease calculation. For breast cancer the health status valuations can be tailored using a modelling approach. This approach is used in chapter seven to calculate and compare the burden of breast cancer in six European countries and to study its sensitivity to variations in health status values. Chapter eight, the general discussion, integrates and discusses the results from these studies.

• View at Worldcat

Free Full Text ( Final Version , 5mb )