Public acceptability of public health policy to improve population health: A population‐based survey

Abstract Background For public health policies to be effective, it is critical that they are acceptable to the public as acceptance levels impact success rate. Objective To explore public acceptance of public health statements and examine differences in acceptability across socio‐demographics, health behaviours (physical activity, diet, binge drinking and smoking), health status and well‐being. Method A cross‐sectional survey was conducted with a nationally representative sample (N = 1001) using a random stratified sampling method. Face‐to‐face interviews were conducted at homes of residents in Wales aged 16+ years. Individuals reported whether they agreed, had no opinion, or disagreed with 12 public health statements. Results More than half of the sample were supportive of 10 out of 12 statements. The three statements with the greatest support (>80% agreement) reflected the importance of: a safe and loving childhood to becoming a healthy adult, schools teaching about health, and healthier foods costing less. Individuals who engaged in unhealthy behaviours were less likely to agree with some of the statements (eg 39.8% of binge drinkers agreed alcohol adverts should be banned compared to 57.6% of those who never binge drink; P < .001). Conclusions Findings show an appetite for public health policies among the majority of the public. The relationship between supporting policies and engaging in healthy behaviours suggests a feedback loop that is potentially capable of shifting both public opinion and the opportunities for policy intervention. If a nation becomes healthier, this could illicit greater support for stronger policies which could encourage more people to move in a healthier direction.


| INTRODUC TI ON
Public health literature has extensively examined the environmental and social determinants of health, and has identified the substantial contributions that inequality and lifestyle choices (eg diet, physical activity and smoking habits) make to population health and well-being. [1][2][3] Equally, an expanding evidence base shows that health inequalities and pressures on health services can be reduced through the implementation of effective public health policies, including supporting early childhood development, poverty alleviation, progressive taxation, restricted advertising, sustained public health messaging and improved access to health care. 4 Investing in such solutions to improve health and well-being is an increasing strategic priority globally 5,6 as countries strive to reach the 2030 Agenda for Sustainable Development and its Sustainable Development Goals (SDGs). 5 Effective public health policies are required to help achieve these goals, and policies must invest across the life-course and empower the public to take charge of their health. 4 For public health policies to be effective and sustainable, it is critical that they are acceptable to the public. Despite this, health policy discussions can often occur in the absence of knowledge on public opinions on such policies or can be dominated by the views of patients rather than the public. The degree to which the public accepts a policy impacts its chances of success, 7 and public involvement in the policy development process can increase such acceptance. 8 Consequently, the importance of public involvement and an awareness of public opinion in policy making is increasingly recognized. 9 A growing body of research linking public opinions with a range of governmental and policy agendas shows some congruence, 10-12 and this can be stronger when public opinion is considered in the early stages of the decision-making process. 13 However, public opinions, expectations and demands are not always aligned with the priorities and resource allocation set by policy-makers. 14,15 Further, views can differ across population groups based on their experiences. 16 Thus, it is important to understand which public health policies are most likely to gain public support and which sections of the public may require more advocacy for public health policy and interventions.
Since the 1930s, politicians, researchers and the media have been using a variety of methods (eg face-to-face, telephone or online surveys; citizen juries or discussion groups) 17 to measure public opinion across a wide range of topics. 7,[18][19][20][21] Internationally, however, while research has explored the public acceptability of different mechanisms of delivering policy messages, 22,23 relatively few nationally representative public opinion surveys have attempted to identify public acceptance of different health priorities and policies. 9,14,24,26 Furthermore, while some studies have examined differences in public acceptance of public health policies by socio-demographic levels, less have explored differences according to current health status or health-related behaviours. In 2017, a Canadian survey which focused on health inequalities 21 found greater public support for public health interventions which specifically targeted children and older people than those targeting the entire population. Moreover, people who believed that low-income populations were to blame for health inequalities were less supportive of welfare interventions, be that targeted (ie those for children or older people) or population level. More recently (2019), an online survey with adults resident in England found levels of acceptability for nudge-based (eg altering portion sizes) and tax-based (eg increasing the price of a product) policies were dependent on the policy and the target behaviour; support for a policy was found to decrease as levels of engagement in the behaviour targeted by the policy increased. 27 For example, those who drank more units of alcohol in a week were less accepting of a policy targeting alcohol. Moreover, one study which conducted nationally representative surveys in six European countries found the UK and Italy (compared to Denmark, France, Germany and Hungary) to be the most disposed towards nudge-based public health policies. 22 Overall, however, there is a dearth of nationally representative surveys on the public's views and appetite for public health measures in the academic literature.
In Wales, public engagement in decision-making is an integral part of policy formation with novel legislation championing the public as a key independent stakeholder in the development of government policies and activities. 28 Thus, using co-design and the principles of participatory design, 29 we conducted a nationally representative survey of the public's views on current public health issues in Wales to engage individuals who would be affected by policy changes in the decision-making process. The purpose of the study was to co-create the long-term strategy for the national public health agency. Here, we report the public's acceptability of 12 public health statements, examining differences across participant characteristics including socio-demographics, health behaviours, health status and well-being.

| Ethical approval
Ethical approval was obtained from Bangor University Healthcare and Medical Sciences Academic Ethics Committee. Approval to conduct the study was also received from Public Health Wales Research and Development Office. Interviewers followed the Market Research Society Code of Conduct and adhered to the Declaration of Helsinki.
All participants provided informed verbal consent to take part.

| Data collection
Letters were issued to randomly selected households (N = 3041) to provide residents with information on the study and the opportunity to withdraw by phone or email. A total of 182 (6.0%) households opted out at this stage. Households that had not opted out were visited by trained researchers between 9AM and 8PM, across all days of the week. Each household was visited a maximum of five times.
Upon contact with residents, researchers presented a letter of authority and provided a second opportunity to withdraw from the study. In order to identify participants, the inclusion criteria applied were (a) resident in Wales; (b) aged 16+ years; and (c) cognitively able to participate (defined as at the door the participant was able to provide informed consent, that is mental capacity, competence and comprehension to understand the research, appreciate what participation would entail and ability to give adequate informed consent to research).
Only one individual from each household was able to participate and in houses with multiple occupants, the occupant with the next birthday was selected to participate.
Recruitment in each LSOA continued until the target sample was met. This resulted in contact being made with 1673 households.
Of those, 358 households declined to participate and four were ineligible, resulting in 1003 completed interviews and an overall completion rate of 60% (1003 agreeing from 1673). Analyses were undertaken with 1001 individuals for whom all socio-demographic information (ie gender, age, deprivation) was available.

| Questionnaire
The questionnaire was designed drawing on previous public opinion surveys 14,31 to capture the public's views on (a) what they identify as the largest contributors to poor health and well-being; (b) what public health issues require public service action; (c) where they source health information; and (d) their acceptability of different public health statements. Analyses here focus on the latter outcome.
The questionnaire was piloted for understanding and timing, and adjusted according to preliminary feedback.
Participants were asked to self-report whether they agreed or disagreed with 12 public health statements using a 5-point Likert scale (see Table 1). The 12 public health statements were selected based on public health priorities in Wales, 33,34

| Statistical analysis
Statistical analyses were conducted using SPSS v24. The proportion who agreed/strongly agreed, had no opinion or disagreed/strong disagreed was calculated for each statement. Consistent with the focus of the paper, 21 where analyses examined agreement, proportions were then dichotomized into those who agreed (agreed/strongly agreed) and those who did not agree (disagreed/strong disagreed/ neither agreed nor disagree). Chi-square was used for initial bivariate analyses to assess the relationship between agreement and par-

| RE SULTS
Ten of the 12 public health statements were supported (agreed/ strongly agreed) by the majority (>50%) of participants (Table 1).
Binary relationships between each statement and participant characteristics are shown online in Tables S2-S4. Significant independent associations found among participant characteristics in the multivariate analyses of key statements are described below. cost less than those living in deprivation quintile 1 (the most affluent area; Table 2). Nearly half of the participants (48.7%) agreed that 'advertising of alcohol should be banned to reduce alcohol problems', while a quarter (25.3%) disagreed (Table 1). Agreement was independently associated with being female, 70+ years old, never binge drinking, not smoking and being physically inactive (Tables 2-3). People who never binge drink were 1.7 times more likely to agree with the statement than those who regularly binge drink (Table 3).

| D ISCUSS I ON
The current study aimed to capture a Welsh national perspective on public health policy and interventions across 12 public health areas, and to identify differences across socio-demographic groups, public health statements (see Table 1).
A key finding of this study is that those who engage in unhealthy behaviours were least supportive of public health measures (eg 39.8% of binge drinkers agreed alcohol adverts should be banned; see Tables 3-4 health focused (eg nutrition) and that the public identify children as a deserving population group. 21 In our study, the statement receiving the greatest support was that 'a safe and loving childhood is essential to becoming a health adult' (87.6% agreed; see Table 1). This may reflect a recent prioritization in Wales of work to increase awareness of the harmful impacts of adverse childhood experiences 28 States which identified that a majority of parent's wished they had more parenting information. 36 This difference could be due to differences in culture and the provision of support. Further work is required as to how is best to share information with parents and how to improve the public acceptability of the messages delivered.
Differences in acceptability of the 12 individual statements across participant socio-demographics, health behaviours, general health and well-being were also found. Males and females were in agreement on the action of seven policies (see Table 2). However, females were significantly more likely to be in favour of banning adverts on junk food and alcohol, the introduction of 20mph speed limits and healthier foods costing less, while males were found to be more supportive of schools teaching children about how to live a healthier life. Female support for policy relating to alcohol and obesity identified here is consistent with previous studies. 7,27 Researchers have suggested that females are more supportive than males as they are more health conscious. 27,37 However, further research is required to identify what is driving the difference found between genders. Similar differences were found between age groups (see Table 2).
Older cohorts were more supportive than younger cohorts of banning alcohol and junk food adverts (as previously found 7 ), the introduction of 20mph speed limits and thinking that people should look after themselves. Younger cohorts were also the most supportive of policy to reduce climate change-consistent with previous research 38 and highlighting younger cohorts as champions for such change. This invested support is particularly helpful for achieving the SDG of reducing climate change. 5 Little association was found between agreement of policy measures and deprivation. 27 The only significant difference identified was that residents of the most deprived areas were more supportive of employers doing more to look after their workers' health than residents of the least deprived areas.
This could be due to individuals living in more deprived areas being in more physically demanding employment, having less flexibility or a greater reliance on employers for support as they may have less support external to work.
Previous literature has shown that people's acceptability of a policy decreases if they engage in the behaviour targeted by the policy. 7, 27 We explored how acceptability differed by levels of physical activity, binge drinking, smoking, and fruit and vegetable consumption. People who engaged in an unhealthy lifestyle (across all four behaviours) were typically less likely to support public health statements (see Table 3). Of the statements explored which related the behaviours measured, those who consumed the least fruit and vegetables were significantly less likely to agree that junk food adverts should be banned than those who consumed the most; and regular binge drinkers were significantly less likely to agree that alcohol adverts should be banned than none binge  Table 3) is declining (although not as quickly among individuals from lower socio-economic status backgrounds). Our results suggest that such a reduction may be accompanied by increased support for public health policies to tackle the same behaviour. Longitudinal data should be used to examine this hypothesis.
The final participant characteristic we explored was health and well-being. Optimistic individuals were significantly more in favour for seven of the statements than more pessimistic individuals; all seven statements were prevention-focused (eg more health campaigns, cost of food, loving childhood; see Table 4). Given that optimistic individuals have been found previously to engage in less unhealthy behaviour and report higher quality of life, it is unsurprising that they would support prevention health measures. 43,44 Critically, those who reported that they felt safe and secure in their community were significantly more likely to agree with statements on healthy foods costing less, climate change, schools teaching more on health, and the importance of a safe and loving childhood.
However, those who reported they felt isolated in their community were significantly more supportive of banning alcohol adverts and parents being given professional advice on child rearing. the language required and extent to which messaging is needed when targeting specific population groups. 46 Our results suggest that potentially, if a nation becomes healthier, a greater proportion of people may also be in favour of living healthier and happier lives.
Such positive feedback could create a tipping point as a majority in favour of stronger public health policies encourage even more people to move in a healthier direction. Equally, however, in countries where health-harming behaviours are prevailing or increasing, public health bodies may have more difficulty mustering public support for health improving polices. In their absence, a reduction in public health may occur further increasing resistance to public health policies.
This study is not without limitations. For instance, all data were self-reported and may be affected by recall bias and/or social desirability bias, where individuals do not disclose accurate reflections of their lifestyle or opinions. Research suggests that opinions on health priorities can alter when people are provided with the opportunity to discuss the issues before making a decision; 47 this opportunity was not provided and while the initial letter outlined the purpose of the study, it did not divulge the public health topics to be explored.
There were relatively few existing surveys to draw questions from, yet where possible questions were derived from validated measures, or adapted from national surveys and the questionnaire was piloted to ensure it suited the target audience. For analyses, participant characteristic questions were collapsed into categories meaning some relationships may have been masked. In addition, while the health behaviour, health status and well-being measures were selected a-priori based on hypothesized associations with attitudes on the public health statements, stepwise regression has a number of recognized limitations, especially where variables are highly correlated. 48 Appendix Table S5 identifies that there is relatively low correlation between the key variables used in our models (all r < .279). However, we cannot rule out that some variables excluded from our model on the basis of non-significance with the variables of interest may be associated with outcome variables in populations outside of our sample. 49 Importantly, further research is required to validate the relationships identified in these models.
While public health interventions have been implemented for several decades, there is an absence in the literature on their acceptability to the public. Understanding the public's views on public health policies is critical to their development, establishment and uptake, but potentially even more important to their sustainability as governments come and go. Here we identified a positive relationship between supporting public health policies and engaging in health improving behaviours. This raises the possibility of a feedback loop where increases in health behaviours are accompanied by more support for public health policies which themselves facilities further health behaviour gains. More work is required on understanding the interactions between health behaviour and policy support. However, our findings suggest that where population health is allowed to deteriorate, public support for the necessary public health policies to reverse such trends may be even harder to find.

ACK N OWLED G EM ENTS
We would like to express our sincere gratitude towards all the individuals in Wales who gave of their time freely to participate in this