Autumn Conference of the Section Sociology of Medicine and Health of the German Sociological Association 03/04 October 2013 European University Institute, Florence, Italy


Call for Papers: Bias in Health Data

Public health research is based mainly on two different types of data: register data and survey data. Cases recorded in register data (such as hospital or insurance data) are the result of a selection process with several biases. For example, hospital data undergo several selection stages beginning with the definition of a mental or physical state such as being ill or dys- functional, the decision to use medical care, the subsequent diagnosis by a physician and the decision for or against a particular treatment and finally the choice for stationary versus ambulant treatment. Independently of medical or health issues, social mechanisms also bias the filtering process at each stage. This fact is of minor concern when using hospital data to describe the population of a hospital, however, the problems begin when data are used to estimate the prevalence of diseases depending, for instance, on gender, social class, or age: Gender is correlated with sensitivity to symptoms and the means of accessing medical care, while social class influences the conditions of health insurance and therefore also the cost of a particular treatment; older people are more likely to choose inpatient treatment since they may not be able to cope with the situation at home, which is also relevant for hospital data.

Survey data in health research is often based on answers from respondents who are medical laypeople. Respondents may systematically over- or underestimate the occurrence of a par- ticular disease, i.e. the statistical error (false positive, false negative) is systematically biased by the respondents ́ degree of health knowledge and sensitivity to symptoms, which in turn depends on the educational background, gender and so on. Educational background is highly correlated with the degree of health knowledge, which is needed to identify a physical or mental state as “disease”. The more complex the symptoms (as with mental diseases, for ex- ample), the more health knowledge is needed in order to classify the symptoms as being connected with a disease or condition.

The conference should shed more light on these social mechanisms creating systematic bias- es in health data. Presentations could focus on the following theoretical, empirical or meth- odological issues:

  • Which data type (register versus survey data) is affected by which biases?
  • Which systematic biases in register data occur due to systematic differences in...
    • access and usage of medical care?
    • physicians diagnoses?
    • the decision for and against medical treatment?
    • the decision for ambulant versus stationary treatment?
  • How is survey data biased by determinants such as health knowledge?
  • Regarding known biases of survey methodology: What is special about health data?

We have a number of preferences within the topic of "bias in health data": First, we are especially interested in presentations focusing on biases in measures of health. Biases in other data (e.g. prevalence of risk factors or socioeconomic status) are also relevant and presentations dealing with such related topics will also be considered. Second, we are mostly interested in biases that pertain to data used in health research and less in general biases already well-known from general survey methodology. Finally, we focus more on biases based on social mechanisms (e.g. due to different perceptions of health in different social groups) than e.g. on the technical problem of moving from one version of the International Classification of Diseases to the next.

We will also provide an extra time slot for presentations that deal with other topics of sociology of medicine and health.

Please submit your abstract (about one page) via email to rasmus.hoffmann@eui.eu and cgross@soziologie.uni-kiel.de by 31 May 2013. Letters of acceptance/rejection will be sent by 21 June 2013. A small conference fee will be charged to cover two meals, coffee, water and snacks. More detailed information about the location, conference schedule and accommodation options will be provided by the end of June. The best presentations will be selected and presenters will be invited to submit a full paper after the conference to be published in a special issue volume of an international health journal edited by Christiane Gross and Rasmus Hoffmann.

Organization:
Rasmus Hoffmann (European University Institute/Erasmus Medical Center) Christiane Gross (University of Kiel) 
See Older News...