Addi van Bergen

Chapter 4 86 have a strong theoretically basis, but also a strong empirical basis. The findings of this study supported our choice. The SCP items perfectly complemented the existing PHM items. Together, they covered the full width of the theoretical construct and produced an empirically sound and valid instrument. Another strong point is the study’s large and representative sample. Over half a million adults were invited to participate in this study and data from over 250.000 respondents were available for analysis. The widespread participation allowed us to extend the generalisability of the SEI-HS to the whole Dutch adult population and calculate national reference data, by sex, age group, urbanicity, ethnical background and educational level; thus providing a benchmark for Community Health Services and municipalities to compare their local data with [50] c 2 . The high number of Community Health Services that took part in this study not only advanced the quality of the research, it also indicates the pertinence of SE to the field of public health in the Netherlands. The fact that 19 out of 28 Dutch Community Health Services (covering over 70 percent of the Dutch population) made space available in their surveys for additional SE items is illustrative of the importance given to SE. Most Community Health Services have since published local figures and reports on SE, with local policy recommendations [51-56]. This provides a good demonstration of the value and potential of a SE measure for the public health sector. The response rate of this study was 45.7%, which is typical for population surveys in the Netherlands [57, 58]. The Dutch PHM employs a systematic strategy to minimise non-response error. The strategy includes measures to increase the general response rate such as pre-survey notification and media coverage in e.g. local newspapers and social media, a mixed mode approach combining web and paper questionnaires, multiple reminders and specific measures to increase representation of hard to reach groups e.g. home visits, translated questionnaires, assistance in completing the questionnaire and oversampling. Lastly, it includes robust weighting procedures to reduce non-response error. We believe that sample representativity is sufficiently guaranteed by the taken measures, particularly for our purpose, the estimation of the parameters of the SEI-HS measure. Although additional analyses (not shown) indicate that the level of SE in the study population has relatively limited effect on the parameters of the SEI-HS, we recommend to retest the SEI-HS in different samples with full inclusion of population groups that are particularly vulnerable to SE. As is common practice in population health surveillance, only persons living in private households were included into the Dutch PHM, thereby excluding groups such as homeless persons and detainees. In the Netherlands, 0.2% of the adult population was estimated in 2012 as being homeless and 1.6% lived in an institutional household, mostly elderly persons [CBS Statline]. Prevalence rates should therefore be interpreted with caution. 2 The syntax to calculate SEI-HS index and scale scores are available from the corresponding author.