Addi van Bergen

Social Exclusion Index-f or Health Surveys (SEI-HS) 87 4 The index and scale scores were trichotomised using 85 th and 95 th percentile scores, resulting in three categories of SE: “moderate to strong” exclusion (score > P95), “some” exclusion (P85 < score < P95) and “little or no” exclusion (score < P85). There are a number of reasons for selecting P85 and P95 as cut-off points. Firstly, using these cut-off points enhances the applicability of the instrument in public health policy. Municipalities prefer to target comprehensive (and costly) interventions at well- defined small population groups with the highest risk, while more general preventive policies may focus on wider population groups. 5% and 10%, respectively, are considered here as useful guidelines. Secondly, the categorisation fits the right-skewed distribution of the index scores, indicating that the largest part of the population is not excluded (Figure 2). Lastly, the choice of the two cut-off points does justice to the relative and continuous character of SE. It allows for the possibility of social groups being differentially included rather than suggesting an artificial dichotomy between included and excluded groups and avoids the stigma of labelling particular groups [7]. Despite this substantiation, the choice of P85 and P95 as cut-off points remains arbitrary. A certain degree of arbitrariness is inevitable in a continuous phenomenon such as SE, where there is no set point at which a person is or is not excluded. Using objective methods such as ROC curves for determining cut-off points would only disguise the inherent arbitrariness. Although the SEI-HS was designed specifically for inclusion in the Netherlands PHM, it is highly suitable for application in public health surveys in countries with similar physical, economic and social conditions where it complements the current validated SE measures. Because of its potential for calculating composite scores and the absence of health as a constituent part of the index, the SEI-HS. allows researchers to study the relationship between SE and health, knowledge indispensable for designing effective policies to diminish socioeconomic health inequalities. This is a promising development as SE provides a broader and thereby potentially more effective range of policy options than concepts like poverty and loneliness [3, 59, 60]. The SEI-HS can be used in identifying risk groups for targeting specific interventions and monitoring their impact over time [6, 7, 60], and in raising the profile and visibility of excluded groups and alerting professionals to the diverse causes and consequences of SE [13]. Finally, our approach to the development of a short embedded index with canonical correlation analyses, may serve as an example to the further development of key public health measures. CONCLUSIONS We have described the development of an instrument to measure the multidimensional concept SE and its validation in a major national public health survey. All four dimensions of SE could be measured and overall, the SEI-HS showed satisfactory to good psychometric properties. The SEI-HS enables researchers to take a next step in the advancement of much needed knowledge on SE and health. The study also provides valuable insights in how to develop embedded measures for public health surveillance.

RkJQdWJsaXNoZXIy ODAyMDc0