Population-based cross-sectional online surveys of adults were conducted in 2017 in England (n = 3,499) (13), the Netherlands (n = 2,173) (14), Sweden (n = 3,000) (17) and in 2018 in Norway (n = 1,208) (15, 16) (Table 1). Data on socio-demographic characteristics, educational level, alcohol consumption, and attitudes towards and experiences with strategies for addressing alcohol in routine healthcare were collected. Individuals aged 18 years or older were eligible for inclusion in all countries other than England where the lower age limit was 16 years (13). Data collection methods specifying details for each country are described in previously published articles (13–17).

Age, gender, and educational level were also collected, with educational level (coded “lower” for high-school graduate or below, and “higher” for college/university education or above).

Alcohol consumption was measured using the Alcohol Use Disorders Identification Test (AUDIT) (23) in England and AUDIT-C in Norway, Sweden and the Netherlands. The definition of risky drinking is primarily based on the level of alcohol consumption that is associated with negative consequences, such as health problems, accidents, and social issues. However, the definition of risky drinking varies across countries and might depend on the cultural and social context (24). Risky drinking was analyzed using country-specific definitions of risky drinking (13–17).

(i) In England, the definition of risky drinking was based on the 10 AUDIT questions relating to: alcohol consumption (items 1–3); alcohol dependence (items 4–6) and alcohol-related harm (items 7–10). Based on their responses, two drinking categories were made: lower-risk drinkers (score of <8) or risky drinkers (score of ≥8) (13). One standard drink in England contains 8 g of pure alcohol.

The distribution of sample characteristics and attitudes towards discussion in healthcare was estimated for each country. Differences in proportions were compared between countries using the chi-squared test, and differences in continuous variables between countries were compared using analysis of variance or Kruskal-Wallis test.

A varimax-rotated principal components factor analysis was performed among the different attitude variables to derive simplified beliefs and attitude components. Two composite dimensions were derived from the factor analysis. The factors identified by the factor analysis were dichotomized at the upper tertile and used as dependent variables in logistic regression. A logistic regression was first performed using the data set with the four countries using country as an independent variable, and odds ratios (OR) were estimated with 95% confidence intervals. The logistic regression analysis was performed using two different models: model I provides crude odds ratios of the following study determinants: age, gender, education, risky drinking, and country; and model II provides odds ratios multivariate adjusted for age, gender, education, risky drinking and country. A test of interaction was performed between study determinants and country using the likelihood test. The interaction was highly statistically significant (p < 0.001) both for factors I and II as outcomes, and therefore the logistic regression was performed separately for each country. Results were considered statistically significant at p < 0.05 using two-tailed tests. Statistical analyses were performed with SPSS 28 and Stata 17.

Table 1 presents sociodemographic and drinking characteristics by country. Alcohol consumption (proportion of adults with episodic drinking and overall alcohol consumption) amongst survey respondents was comparable with official national data sources in each country (1). Patterns of educational level were similar to national data in England and Sweden, but a larger proportion of respondents in Norway (70% vs. 43%) and the Netherlands (63% vs. 37%) were higher educated compared to national data (26). Age and gender of respondents were similar to official data sources in most countries except the Netherlands, where respondents were comparatively older (42% over 60 years vs. 22% aged 65 or more) (27, 28).

Population characteristics from each country regarding age, sex, education, and alcohol consumption were published in previous publications (13–17).

The descriptive statistics for the different attitudes about alcohol prevention in routine healthcare are shown in Table 2. This was presented in previous publications for England, Norway, and Sweden, but not for the Netherlands.

Factor analysis suggested there were two different factors affecting how alcohol conversations in healthcare are perceived. I. Negative attitude towards discussing alcohol with healthcare providers (including 1. routine, 2. private, 3. symptoms, 4. patient) and II. Perceived honesty and confidence in healthcare (including 1. routine, 5. honesty) (Table 2).

Together, the two derived factors captured 59% of the total variance. The first factor “negative attitudes towards preventive alcohol conversations” captured 38% of the variance. The second factor “Perceived honesty and confidence in healthcare” captured 21% of the variance.

A high level of education reduced the likelihood of having negative attitudes towards discussing alcohol in healthcare in all countries (OR 0.65, CI 0.59–0.71) (Table 3). Netherlands had the highest OR of negative attitudes (OR 1.50, CI 1.33–1.70) compared to the other countries. Risky drinkers were more likely to report negative attitudes towards discussing alcohol in healthcare in Sweden (OR 1.56, CI 1.32–1.86), Norway (OR 1.61, CI 1.23–2.11), and the Netherlands (OR 1.31, CI 1.08–1.59), but there was no association in England (OR 0.92, CI 0.74–1.15) (Table 5). Women were less likely to report negative attitudes towards discussing alcohol in healthcare, except in the Netherlands where there was no significant association (0.86, CI 0.72-1.03; p = 0.10) (Table 5). There was no association between age and negative attitudes towards alcohol discussion in healthcare except in the Netherlands, where those older than 50 were 1.4 more likely to report negative attitudes towards healthcare compared to the younger age group [50–59 years (OR 1.41, CI 1.03–1.95); ≥60 years (OR 1.43, CI 1.08–1.89)] (Table 5).

There was no association between education and the dimension perceived honesty and confidence towards healthcare (OR 0.95, CI 0.87–1.04) (Table 4). Respondents from the Netherlands (OR 0.78, CI 0.69–0.88) and England (OR 0.65, CI 0.59–0.73) displayed lower level of perceived honesty and confidence in healthcare (Table 4). Risky drinkers were more likely to report low perceived honesty and confidence towards healthcare in all four countries compared to lower risk drinkers (OR 0.70, CI 0.63–0.78) (Table 4). Women reported lower levels of the dimension perceived honesty and confidence towards healthcare compared with men in England (0.68, CI 0.59–0.80; p < 0.001) and in Norway (0.69, CI 0.54–0.88; p = 0.003). In Sweden (0.87, CI 0.75–1.01; p = 0.069) and the Netherlands (0.91, CI 0.71–1.05; p = 0.137) there was no statistically significant association between gender and perceived honesty and confidence towards healthcare (Table 6). Being older than 60 years increased the OR for the dimension perceived honesty and confidence towards healthcare to 1.62 times (CI 1.21–2.18) in the Netherlands, while it decreased it to 0.79 times (CI 1.08–1.89) in England (Table 6).

We found similarities in attitudes towards alcohol conversation in healthcare across England, Norway, Sweden and the Netherlands. People with a lower educational attainment had a higher level of negative attitudes, and risky drinkers had less perceived honesty and confidence in healthcare in all countries. However, Sweden and Norway had the highest level of positive attitudes towards alcohol conversations as routine, and a higher perceived honesty and confidence in healthcare.

There were also differences between countries in attitudes towards alcohol conversation in healthcare. Respondents in the Netherlands were significantly more likely to have negative attitudes towards alcohol prevention compared to those from other countries (OR = 1.5 compared to Sweden), and Norway and England were like Sweden (Table 3). The level of perceived honesty in healthcare was also significantly lower in the Netherlands (OR = 0.78) and in England (OR = 0.65) compared to Sweden and Norway (Table 4). Based on findings from a previous study comparing differences in public trust in healthcare between Germany, the Netherlands, England and Wales, the varied attitudes towards alcohol conversations we observed between countries might result from a combination of cultural differences and differences in how healthcare is organized (38).

The study has several strengths. The data sets are large and similar population-based surveys were performed in four different countries. This is the first study to include four countries in a study of attitudes in addressing alcohol conversations in healthcare. This research is valuable to inform preventative alcohol policies in the future and to improve strategies to tackle the higher alcohol-related morbidity and mortality in lower socio-economic groups in future.

The study has several limitations that are important to consider. In the Netherlands and Norway, the study samples had more educated respondents compared to national population estimates, which might have reduced the association between attitudes alcohol prevention and education. Furthermore, the original studies used different definitions of risky drinking between countries. The study used different educational coding systems in each country, necessitating that the 7-level variable on education had to be reduced to two-levels when comparing the countries. In England, the higher level includes only university education. Despite this loss of information, the study was able to detect large difference between low and high educated respondents in attitudes towards alcohol prevention in healthcare.

We identified socio-demographic and behavioural determinants of attitudes towards discussing alcohol in healthcare in four European countries, with some key similarities and differences between countries. The results suggest that trust in healthcare is an important underlying mechanism of attitudes towards alcohol conversations in healthcare. We also found a higher likelihood of having negative attitudes towards discussing alcohol in healthcare among respondents with low education in all four countries, as well as lower perceived honesty and confidence in healthcare among risky drinkers.

Our findings suggest that the implementation of alcohol prevention in healthcare targeting all patient groups has resulted in inequalities in attitudes towards discussion about alcohol in healthcare. Respondents with low education and risky drinkers are more negative towards routine delivery of alcohol prevention activities in healthcare settings. This highlights the need for more public information on alcohol prevention together with better training on alcohol issues for healthcare professionals to ensure that discussions about alcohol preventative measures are adapted to all patient groups. Given the evidence of increasing social inequalities in health (20), a stronger commitment is needed from the authorities in reducing alcohol consumption in the population. The results of this study provide insights into the complex relationship between alcohol and culture, which means that policy interventions to address alcohol-related harms should be adapted to each country.

Future research to further explore patients’ experiences and views towards alcohol conversations in healthcare is warranted, as well as the development of more diverse methods to implement preventive measures in healthcare. Less educated and risky drinkers have more negative attitudes: this implies that this should be studied further. Sociodemographic factors should be considered when implementing alcohol prevention in healthcare.