Reducing stigma towards those with mental illness remains a key priority of many governments [1, 28, 29]. The Swedish Government’s commitment to reduce stigma demonstrates the need for appropriate measurements available in the local language to monitor how attitudes are changing. In this study, we wanted to determine if the American designed Attribution Questionnaire could be culturally translated for Swedish physicians. The Attribution Questionnaire could be a helpful tool for measuring and monitoring stigma in the Swedish setting. Overall, results showed that the questionnaire is suitable for measuring physicians’ attitudes in the Swedish context. However, modifications were necessary to make the questionnaire externally valid in the new setting.
Results from our SEM model are acceptable [27] but the correlations between the latent factors are weaker than previous studies [4, 14, 15]. Similar to Akyurek, Efe [17] we found a negative and weak relationship between the responsibility and pity latent factors compared to other translated versions that found a stronger relationship [14, 15]. However, our sample, comprised only of physicians, differed from the other studies that included mostly students [4, 18] one including medical students [19], or the general population [14, 15]. Physicians, in general, would have more contact with those with mental illness compared to the other studies’ sample populations. As such, the theoretical model may be more applicable to general populations than physicians. Furthermore, since we used a truncated version of the Attribution Questionnaire is difficult to directly compare results with previous version of the questionnaire.
The importance of the expert committee
The expert committee played a significant role in modifying the questionnaire, much more than the process of back-translation. Previous questionnaire translation emphasises the role of back-translation as best practice for translating questionnaires [30, 31]. However, Epstein, Osborne [24] argue that an expert committee can play more pivotal role when translating across cultures. Our findings support this assertion, indeed, the questionnaire benefited greatly from the use of an expert group. It was the expert committee not the back-translation that identified the semantic issues and found experiential equivalence. Overall, our study supports the inclusion of an expert committee for culturally translating a questionnaire.
Cultural adaptation and the validity of the questionnaire
Overall, we found that cultural adaptation makes the questionnaire more relevant to the local context. Our results support that questionnaires should be translated for language and culture [21,22,23, 32, 33]. External validation—whether the results can be generalised or useful outside of the research context—is essential. Cultural translation increases the external validity of the questionnaire and ultimately the usefulness of results. Internal consistency—how well a survey measures what you want to measure—is also important and can be impacted by cultural translation. Our results showed that Cronbach’s alpha for some factors, in particular, ‘Responsibility’ were much higher compared to the Italian and Spanish translated versions of the Attribution Questionnaire (Responsibility α: AQ-S = 0.819, AQ-I = 0.615 and AQ-E = 0.390) [14, 15].
Limitations
This study is part of a broader project on physicians’ attitudes towards disability pension applicants in Sweden. As such, the vignettes used with the survey were changed from the original American vignette of Harry with schizophrenia to Johan/Johanna with depression, alcohol dependence or low back pain. Such changes could explain why the internal consistency varies between the different translated versions. It is difficult to determine if the results from the SEM for the Attribution Questionnaire are not as strong due to the modifications made by the research team or the cultural differences between the contexts or both [15]. Due to feasbility contstraints, our sample included only two thirds of GPs registered in Sweden. A third party—the Hälso och sjukvårdens adressregister—provided the addresses and randomly selected the two thirds before releasing the data for our use. As such, this could impact on the quality of our sample. Survey response rate was low, which could bias the results. Nevertheless, the sample size was bigger than that of many other studies, and concerned two groups of physicians that more frequently deal with disability pension applicants in their work. We also could have included persons with lived experience of mental illness in the expert committee to get their views on the cultural translation. Finally, we were unable run the SEM analysis and the impact on the model of different type of providers e.g. GPs versus psychiatrists.
Strengths
To the best of our knowledge, only one other study has culturally translated the Attribution Questionnaire [17]. Additionally, compared to the other Attribution Questionnaire translations [14, 15], our sample size is much larger. As such, our results could be more reliable in terms of the SEM and internal consistency analysis. Our results also demonstrate that the Attribution Questionnaire can be modified to suit specific contexts or sub-groups and still remain a meaningful measure of stigma.
Implications for policy
Given the Swedish Government’s initiative to reduce stigma, we recommend that the AQ-S could be an efficient tool to measure and monitor stigma in the Swedish setting, especially among physicians but also the wider population. In particular, Swedish policy-makers could use the AQ-S to determine whether newly implemented anti-stigma strategies are working. However, the AQ-S should be validated among other sub-groups.
Implications for practice
Disability pension is not a universal program. Government adjudicators must decide who ‘deserves’ to receive disability pension and who does not. However, these decisions must be based on unbiased information and not discriminatory attitudes. The AQ-S is practical tool to help gauge whether disability pension applicants, regardless of their diagnosis, receive a fair assessment. While this study focuses on this in relation to physicians, the AQ-S could also measure stigma among adjudicators themselves.
Implications for future research
Further examination of intersectionality—the idea that various forms of social stratification can intersect to create cumulative discrimination or disadvantage [34] in the context of disability pension is vital. While this study explored the intersection of gender and disability, we intend to expand this exploration to include ethnicity and other factors e.g. socioeconomic status in future studies. We need a more nuanced understanding of what factors might lead to positive or negative attitudes among physicians and other policy-makers. The vignettes created in this study could be amended to explore such ‘intersectional stigma’ [35]. A better understanding of intersectional stigma and disability pension would also provide clearer pathways for policy intervention to address any inequalities that may exist.