Addi van Bergen

Construction of a multidimensional measure for social exclusion 49 3 Netherlands. We had access to the data collected in the surveys of 2008 with health questionnaires from the Community Health Services of the four largest cities in the Netherlands. Our ultimate goal is to develop a nationally validated and standardised measure to monitor social exclusion in routine public health surveys. METHODS Ethics statement Ethical approval was not required as this study relied on secondary anonymised data collected in the context of performing statutory tasks (Public Health Act of the Netherlands), in strict accordance with the national standard. At no point in time did the datasets contain direct identifiers. Codes to track response were removed from paper questionnaires directly upon receipt and processed separately, as were online access codes. The risk of re-identification of individuals from indirect identifiers such as age (in years) and sex, was very low. The datasets are freely available for non-commercial research purposes. Data source and participants We conducted secondary analysis on data of four public health surveys that were collected in 2008 by the local Community Health Services in the cities of Amsterdam, Rotterdam, The Hague and Utrecht, using a uniform research methodology. The content of the questionnaires was only dissimilar for items that were selected according to local policy priorities. In each city an a select sample was drawn from the non-institutionalised population aged 16 years and older, stratified by district, neighbourhood, age and ethnicity. A total of 42,686 persons received a questionnaire by mail. These questionnaires could be filled out in writing or via the Internet. Non-responders received a reminder after two weeks. In addition, difficult to reach groups such as non-Western immigrants and residents of deprived neighbourhoods were contacted after four weeks by telephone or home visit and invited to participate by mail or personal interview in the language preferred by the respondent. For Turkish respondents, the main non-Dutch speaking minority in the Netherlands, a translated questionnaire was available. The overall response rate was 50% (20,877 respondents) and ranged between 47% in Rotterdam and 54% in Utrecht. Despite the intensive follow-up, the response was lower among difficult to reach groups. Through oversampling these groups were still well represented in each of the four studies. [11]. In line with the age standard for public health surveys in the Netherlands, we limited our analyses to respondents aged 19 years and older (19,658 respondents).