COVID-19: information access, trust and adherence to health advice among migrants in Norway

As a part of the project Inncovid. Norge [14], an online survey was developed based on the Norwegian Citizen Panel (NCP) survey performed in March of 2020 in relation to the pandemic. The Inncovid. Norge survey included 45 questions about the participant’s and his/her family’s general health, risk assessment, sources of information about the outbreak, how they were affected by the pandemic, knowledge about and adherence to government recommendations, and the degree of trust they had in several Norwegian institutions to handle the pandemic. In addition, the survey contained eight socio-demographic questions. The initial questionnaire was developed in Norwegian and then translated into five different languages: Polish, Arabic, Somali, Tamil and Spanish. Two bilingual members of the project team performed translations for each language. We used web-based software (Nettskjema) to create and run the survey. This portal collects data directly into a secure server (TSD). All participants were registered with a serial number and no personally identifiable information was recorded. Data was collected between May 25th and July 1st of 2020. All participants received information about the study and provided consent before responding. The survey’s completion time was around 15 min.

We recruited migrants living in Norway who could complete the surveys in Polish, Arabic (mainly targeting Syrians), Somali, Tamil (mostly targeting people from Sri Lanka) and Spanish (primarily targeting migrants from Chile and Spain). These migrant groups were chosen as they cover different continents and the major religions in Norway, have different length of stay in Norway, reasons for migration and integration profiles. The largest group of migrants in Norway is comprised of more than 100,000 migrant workers from Poland. Migrants from Syria have had the strongest growth in Norway in recent years and, together with persons with Somali and Sri Lankan background, represent some of the biggest groups that arrived in Norway as asylum seekers/refugees. Spanish-speaking migrants constitute an important but heterogeneous migrant group in Norway, including Latin Americans and Spanish people [8]. As numbers of migrants are usually underrepresented in population surveys, we decided to obtain at least a similar proportion of migrants as the proportion of the majority population represented in the NCP. Thus, our goal was to include at least 500 of the approximately 200,000 migrants from the mentioned groups living in Norway (0.25%) to achieve a similar proportion of the population as the NCP survey, which was responded to by 12,051 individuals among 5.4 million persons (0.22%).

Project researchers, whose mother tongue is one of these five languages, reached the target groups through key members of the migrant communities, informal networks using a snowball sampling method. We encouraged migrants to further forward invitations -using a survey link- through Messenger, Viber, WhatsApp and other social media. Information about the project was also forecasted in local radio channels and posted on relevant Facebook groups used by migrants living in Norway and on the websites of the Norwegian Organization for Asylum Seekers (NOAS), Moja Norwegia (portal for Polish migrants in Norway) and Church City Mission.

This study reports the answers to 31 variables, eight of these were demographic. For reporting purposes and facilitating comparisons between groups, Likert scales from 16 variables were merged:

The analyses were conducted for all migrants together and in five separate language groups. We used R 3.4.4. for the statistical analysis. Unweighted descriptive analysis was used for demographics, COVID-19 cases and perception of health. We weighted responses by gender and age and assigned an adjustment weight to each survey respondent for questions about following authorities’ recommendations, source of information and trust [15]. The weights were calculated based on a known age-gender distribution in the Norwegian population obtained from the National Statistical Institute of Norway [16]. The “survey” package in R was used for the unweighted and weighted descriptive analysis [17].

The Inncovid. Norge project received ethical approval from the Regional Ethical Committee (REK number 132585). The participants signed a digital consent written in their native tongue before starting to fill the questionnaires.

Self-reported COVID-19 cases were similarly distributed among all groups. In total, 96.5% of the respondents presumed or confirmed not having COVID-19. There was otherwise variation in the perception of health and risk of infection. Health was perceived as “excellent or very good” twice as often in the Somalian group compared to the Tamil one. In addition, Tamils perceived that their infection risk in 2020 was “very or somewhat high” more often than the other groups. (Table 1).

The vast majority of respondents (82%) felt they had received sufficient information about the coronavirus from the health authorities (Table 1). Figure 1 shows the level of importance of seven different information sources. In general, press conferences from the government (Fig. 1A), Norwegian news media (Fig. 1B) and health authorities’ websites (Fig. 1F) were relevant sources of information for all groups. Information from social media (Fig. 1C and D) and conversations with friends and acquaintances (Fig. 1E) were less important, although there were differences among groups. The Somalian group considered information obtained via social media and conversations with friends more important than other groups. In contrast, the Poles considered these sources of information less relevant than the other groups and were more worried about their inaccuracy (Table 1).

Figure 2 shows the answers to five questions related to the extent of trust of the respondents. Overall, most participants reported trusting the health system (in general terms) (Fig. 2A and B). However, they reported less trust in how the health authorities have handled the pandemic (Fig. 2C). The majority agreed that the government treats all groups fairly (Fig. 2D). To a lesser extent, participants agreed when asked if the government listens to the opinions of all citizens (Fig. 2E). In addition, there were variations among the groups for these questions, with Poles indicating generally lower levels of trust and Tamil and Arabic speaking respondents reporting high trust in the health system. However, Arabic speaking respondents were less confident on receiving good medical treatment if becoming seriously ill.

Figure 3 displays the answers to four survey questions related to following authorities’ advice. Overall, all groups reported a high level of adherence to the authorities’ advice (Fig. 3A) but perceived that Norwegians’ followed recommendations to a lesser extent (Fig. 3B). Although the majority agreed that by following the recommendations, they avoid becoming sick, Spanish and Polish speaking were more skeptical (62 and 48% for Spanish and Polish groups respectively “strongly agreed” that by following recommendations they avoid getting sick) as shown in Fig. 3C. The Polish group also reported skepticism about avoiding making others sick by following the advice (63%) (Fig. 3D).

To our knowledge, this is the first study among migrants in Norway to study their access to information about COVID-19, trust in health authorities and adherence to recommendations. A strength in our study was that Inncovid. Norge is formed by researchers and health workers from the five migrant groups recruited. Knowing the communities and key persons in these environments facilitated the recruitment and achieving the targeted number of participants. However, the study has some limitations. First, the Somali speaking group is under-represented in our sample with only 33 respondents. This made it difficult to perform meaningful statistical comparisons among groups. Second, the results were obtained via a web-based questionnaire and although we used different channels and networks to reach the respondents within the five-selected migrant groups, selection bias may be present. Generally, respondents to our questionnaire were not representative of their populations, especially regarding their high employment status. They also differed in terms of gender and age, for which reason we weighted the results for the main outcomes. However, the different groups present characteristics as expected regarding length of stay and other variables. Thus, our results should be interpreted with caution for the generalizability of the findings to the migrant population in Norway. Third, given the survey nature of our study, we cannot disentangle the reasons behind our results, and further research is necessary to understand the differences in the outcomes we have identified among migrant groups. Last, in an ideal world, our results would be part of the NCP study, including representative samples of migrants, allowing the sound methodological comparison of results among groups and with the majority population, but this was not feasible at the time.