Of the 134 experts in the 16 countries who were recruited into the Delphi process, 126 completed the last round. The eight experts who dropped out did so after the second round. Data on participating countries and the professional background of the experts is presented in Table .
The local reviewing process to remove repetition at each site prior to round two reduced these lists to between 11 and 40 factors for each country.
Final national consensus lists after the third round included between 10 and 16 factors with the ten highest scores, resulting in a total of 186 high scoring factors across all countries (Table ).
Number of factors of good practice listed per Delphi round and range of final scores
The 186 factors fall into nine thematic categories, which are summarized below in order of frequency, and are detailed further in Table . The first eight themes figured in the final national good practice lists for more than half of the participating countries. The themes represent general principles, some generic but with specific aspects for migrants. The nine general principles were:
Major themes in the 16 country-specific, final factor lists
• Accessibility: easy and equal access to health care (mentioned by all 16 countries)
This theme was mentioned on consensus lists of all countries as the number one priority. All countries mentioned the need for an easily accessible general health care system for all citizens (Table ). Although they had initially been asked to propose good practice principles for migrants with regular, legal incomes and speaking the local language, experts from several countries also specifically prioritised equal access for refugees and undocumented migrants.
• Empowerment of migrants (15 countries)
Migrants should be informed about their rights and the functioning of the health care system. This might involve a special consultation the first time people access the health care system. (Table )
• Culturally sensitive health care (14 countries)
Optimising culturally sensitive care several specific measures were proposed. (Table ) Although most countries' experts underlined the need for health care services to take into account the cultural or religious habits of migrants, others considered that migrants should be encouraged to understand the habits and culture of local health care systems.
• Quality of individual care (12 countries)
Services should consider the patient as an individual and not stereotype them with the characteristics of the cultural group they are perceived of as belonging to. (Table )
• Patient-health care provider communication (11 countries)
High quality interpreter services, either in person or by telephone, should be easily accessible. (Table ).
• Respect towards migrants (9 countries)
Practitioners should show respect, create trust, be interested and address patients without prejudice and with an open mind. (Table )
• Networking and interdisciplinarity (8 countries)
Meeting the health care needs of migrants requires networking within health care services and between health and social services. (Table )
• Targeted outreach activities (8 countries)
Outreach activities in health education, screening, prevention and promotion with difficult to reach migrant groups were mentioned in eight countries.
• Availability of data (6 countries)
Health care services should be provided with relevant knowledge on health and risk factors concerning the populations they are dealing with. Health registries should record and monitor migrant health to facilitate migrant health research. (Table )
If a similar analysis is conducted on only the three most important principles in each country, instead of ten, accessibility features in eleven countries, culturally sensitive care in ten, communication in nine and empowerment in eight. All the remaining priorities featured among the three most important ones, but in less than half of the countries.
Discordance within and between countries concerning factors experts considered to be important was not infrequent. Of the 120 round 2 scores containing discordant scores, 99 (83%) were voiced by 13 of the 126 experts who completed the Delphi process in 7 of the 16 countries. None of the four expert categories (academia, NGO, policy makers, practitioners, and no professional category) were over-represented in these discordant voices. Final factors containing discordant scores were frequent in: Greece (10/12), Austria (7/11), UK (7/12) and Portugal (7/13). However, the discordance in these final factors was all due to one single expert in each country, with the exception of the UK with 2 discordant experts. No other country had more than 3 final factors containing discordant scores.
The whole question of migrant-specific health care is seen as a false problem by discordant experts in Belgium ("the key difference is socioeconomic"), Finland ("everyone is culturally different, not just migrants") and France ("pinpointing cultural needs creates health care ghettos"). Within country discordance, as defined in the Methods section above, remained with respect to 15 good practice factors in seven countries (Table ). In Austria and Belgium, discordance even remained in the final list of ten most important factors. In Austria, there was disagreement on the need for employing staff speaking migrant languages, translated information material, and measures directed at preventing and diminishing discrimination. In Belgium, discordance persisted on the importance of providing specific information about health insurance and other financial support measures. In Lithuania, experts disagreed about the need for providing information about the health care system. In France and Poland, experts disagreed about the need for specific epidemiological information on areas with high migrant populations. French experts also disagreed about the need for having experts providing extensive training to health care professionals to be aware of ethnic and cultural issues, with opponents considering this to be a trap, arguing that the most important principles for health professionals are to take sufficient time and have access to an interpreter.
Consensus and discordance concerning factors of good practice in health care for migrants in Europe.
German experts disagreed on the need for improving contextual societal factors, such as the social acceptance and social support of migrants as ways of improving accessibility and quality of health care. The presence of a discordant voice generally resulted in that factor having a significantly lower final average score. This was the case, for example, in Germany for the factor concerning the need for health care providers to reflect on their own cultural backgrounds. In the Netherlands, experts had highly opposing views on the need for customized care for migrants, even on a small-scale, action-specific and temporary basis. In Portugal, promoting positive attitudes towards migrants and fighting against discrimination in health care professionals was excluded from the final national good practice list due to two discordant votes (both from academics).
Discordance also existed between the 16 participating countries. For example, coherent national, regional and local governance was seen as an advantage in Belgium and as a disadvantage in the UK, where political correctness in health care policy might have created resistance and discord.