LAUSR

Background Citizens living in disadvantaged neighborhoods experience poorer health than the majority, and this inequality is a public health problem even in a welfare state such as Sweden. Numerous initiatives aimed at improving health and quality of life in these populations are being implemented and evaluated. Given that these populations are predominantly multicultural and multilingual, an instrument such as the WHOQOL-BREF, which is cross-culturally validated and available in multiple languages, may be appropriate. However, this cannot be ascertained since the psychometric properties of WHOQOL-BREF have never been assessed in the Swedish context. Thus, the current study aimed at assessing the psychometric properties of the WHOQOL-BREF questionnaire in citizens from a disadvantaged neighborhood in Southern Sweden. Methods The respondents in this study were 103 citizens who participated in the health promotional activities of a Health promotional program and also responded to the 26-item, WHOQOL-BREF questionnaire as a part of an evaluation to assess the impact of the activities on the health-related quality of life of citizens. A Rasch model using WINSTEP 4.5.1 was used to assess the psychometric properties in this study. Results Five of the 26 items, including pain and discomfort, dependence on medical substances, physical environment, social support, and negative feelings did not display acceptable goodness-of-fit to the Rasch model. On removing these items, the 21-item WHOQOL-BREF scale had an improved internal scale validity and person-separation reliability than the original 26-item version for this group of citizens from the neighborhood. When assessing the individual domains, three of the five items that were misfits on analyzing the full model also showed misfits in relation to two respective domains. When these items were removed, the internal scale validity of the domains also improved. Conclusion WHOQOL-BREF seemed to be psychometrically inadequate when used in the original form due to internal scale validity problems, while the modified 21-item scale seemed better at measuring the health-related quality of life of citizens living in socially disadvantaged neighborhoods in Sweden. Omission of items shall be done but with caution. Alternatively, future studies may also consider rephrasing the items with misfits and further testing the instrument with larger samples exploring the associations between subsamples and specific item misfit responses.