The Vagaries Of Public Support For Government Actions In Case Of A Pandemic

EXPERIENCE WITH SARS

Computer simulations and real-world experience with severe acute respiratory syndrome (SARS) show that social distancing can prevent or slow disease transmission^14,15 as long as at least 60 percent of the population complies with the policy.¹⁶ Implementing social distancing seems like common sense, and the policy was used successfully in Asia during the SARS epidemic. However, such policies may work in group-oriented Asian cultures but not in more individualistic countries such as the United States, Australia, and Western Europe.¹⁷

Canada used quarantine successfully in the Toronto area during the SARS outbreak, but SARS never threatened the general population there. About one-third of the approximately 23,000 people quarantined—some of them at their own request—were health care workers. Guarantees of paid sick leave while in quarantine were critical factors in Canadians’ compliance with the policy.¹⁸

RESPONSES TO H1N1 PANDEMIC

Abundant data are available about the public’s perceptions of risk, individual preventive behavior, and use of vaccine in the 2009–10 H1N1 pandemic.^19,20 However, relatively little research has been conducted on public support for government action in a pandemic, and most of that work predates the H1N1 outbreak. Some respondents did not know what the word pandemic meant, and none had experienced H1N1.

The limited pre-H1N1 data show that 70–80 percent of US respondents favored compulsory quarantine for flu patients in a pandemic.^21,22 Robert Blendon and coauthors²³ also reported that a large majority of working adults could stay home from work for up to ten days, and more than half could stay home for a month, without serious financial problems.

It is important to note that the questions that elicited those responses asked what people could do rather than what they would do. The questions thus did not address important intangible barriers to social distancing, including respondents’ perceptions of contagion risk and policy effectiveness, fear of job loss, and objections to policies based on civil liberty or mental health grounds.

ATTITUDES TOWARD SETTING VACCINATION PRIORITIES

Attitudes toward rationing or prioritizing of vaccines are another area in which little research has been conducted. Both Hye-Jin Paek and her coauthors²² and Catherine DesRoches and her colleagues²⁴ found that just over half the public would support government policies limiting the vaccine to high-risk groups.

An emerging virus would require the rapid development of a new vaccine, perhaps bypassing standard testing protocols. Both Sandra Crouse Quinn and her coauthors²⁵ and Paek and her coauthors²² found fewer than 40 percent would support vaccines that had not yet been fully approved. Other surveys conducted before^21,23,24 and during^19,20 the H1N1 outbreak did not address this question.

ROLE OF CURRENT STUDY

Our study examined levels of support for all of these actual and hypothetical policies in the early weeks of the outbreak, when concerns about the virus’s potential severity were high. Controlling for demographic and attitudinal factors to see the effect of each variable separately, we examined how risk perception, concern about H1N1, trust in the government’s handling of the crisis, and perceived fairness predicted support for government action.

Additionally, we examined attitudes about new or not-fully-approved vaccines and medicines—as noted, a topic that few previous surveys had covered. We also examined support for more-moderate measures such as encouraging sick people to stay home from work, and more-extreme measures such as quarantines and border closings.

Our study took place during the initial stages of the H1N1 epidemic, which is typically a pivotal time. In an epidemic’s initial stages, no one knows what trajectory the disease will take;²⁶ the public may be complacent even as public health officials are increasingly alarmed. Thus, our study took a snapshot of the time when uncertainty was highest, instead of gauging public response to a severe but hypothetical pandemic or documenting behavior and attitudes once people began to realize that the 2009 pandemic would be mild.

With a more virulent strain of influenza, the outcome could have been much different, and public compliance might have been a deciding factor in the outbreak’s severity. Therefore, our data are especially instructive in anticipating initial public responses to government action in future pandemics.

SURVEY INSTRUMENT AND MEASURES

Our survey instrument described H1N1 as “the current influenza outbreak” and used the term swine flu rather than H1N1. We developed measures of support for eight actual or potential government actions using a four-point scale ranging from “strongly oppose” to “strongly favor” (Exhibit 1). Based on the clustering of the distributions, we dichotomized responses to oppose (strongly oppose or oppose) or favor (strongly favor or favor).

Predictors of support for pandemic policies were standard demographic characteristics and self-reported attitudes regarding H1N1, including respondents’ perceptions of their susceptibility and the pandemic’s severity, and respondents’ concern about getting the disease. Additional independent variables included respondents’ confidence in the government’s handling of the pandemic, measured by indexing perceived levels of government openness, honesty, commitment, caring and concern, and competence in addressing H1N1; the extent to which respondents believed that the government’s response was in their personal best interest; and the degree to which respondents believed that the government would protect them from swine flu.^27,28 We calculated a mean score for the trust scale and categorized responses above that score as indicating high trust, and those below that score as indicating low trust.

We also measured respondents’ perception of how fairly all groups of Americans were treated with regard to H1N1, constructing a scale from four questions that asked about fairness, equal access to antiviral drugs and vaccines, and respondents’ degree of confidence that they would receive equal treatment in a quarantine. For the analysis, we divided the responses at the median value into categories of less and more perceived fairness. In addition, we asked respondents whether they “personally had ever experienced discrimination or been hassled when seeking health care because of your race, ethnicity or color,” and we categorized the responses as “yes” or “no.”

DATA ANALYSIS

We used the statistical software Stata, version 10.1, for complex survey analysis procedures incorporating weighting and stratification variables. For categorical measures in the bivariate analysis, we used adjusted Pearson chi-square tests. We examined the relationships between support for government actions and both demographic and attitudinal measures through binary logistic regression for dichotomous outcomes. In all analyses, a p value of <0.05 indicated a significant finding.

OVERALL SUPPORT FOR GOVERNMENT ACTION

Respondents’ views of some policies were nearly evenly divided (Exhibit 1). Slightly fewer than half supported encouraging people to stay home from work when sick; closing schools, stores, places of worship, and other gathering places; and setting priorities to determine who would get limited supplies of vaccines or drugs.

On the other hand, strong majorities supported quarantining people exposed to the flu to limit their contact with others; closing borders to visitors from countries with flu outbreaks; helping people care for sick family members at home so they wouldn’t have to go to the hospital; and distributing medicines or vaccines at designated public locations. Only about one in five respondents supported offering people new vaccines or drugs that had not yet been approved.

Our next step was to determine which demographic characteristics, such as age or race, and which attitudes, such as risk perception or trust, were most closely associated with support for these policies. Although public health messages often target specific demographic groups where particular risks or attitudes are more prevalent, ultimately, the goal is to change the attitudes and behaviors of individuals. Messages that focus on attitudes can reach individuals in multiple population groups simultaneously. We report attitudinal data for the entire sample in Exhibit 2, demographic predictors of support in Appendix Supplement 1, and attitudinal predictors of support in Appendix Supplement 2.²⁹

Generally, we found that although few people saw themselves as highly susceptible to H1N1, almost half felt that the pandemic could be severe and had concerns about getting the disease. A majority trusted the government’s ability to handle the pandemic and treat people fairly.

PREDICTORS OF SUPPORT FOR VARIOUS STRATEGIES

In an initial bivariate analysis (see Appendix Supplements 1 and 2),²⁹ we found that some demographic and attitudinal variables were associated with support levels for particular policies. For example, higher trust in the government’s handling of the pandemic appeared to be related to support for closing borders.

However, to make meaningful use of the data, it is important to go beyond bivariate analysis and evaluate each predictor individually, with all other variables being equal. We therefore used logistic regression to control for all other demographic and attitudinal variables,³⁰ singling out critical predictors of support that policy makers and public health officials should focus on (see Appendix Supplement 3).²⁹ We detail the results of this more precise analysis below.

OPPOSITION LEVELS

One finding that is particularly important for public health officials is that even the policies with the highest levels of support in our study were still opposed by roughly one-quarter of the respondents. Several basic, commonsense efforts to mitigate the spread of the disease—such as staying home from work while sick, or closing schools and other public gathering places—had surprisingly weak support, with fewer than half of the respondents in favor. It is even more surprising that although only 49.8 percent supported encouraging people to stay home from work, 71.5 percent supported closing borders. Social distancing is considered to be far more effective in a pandemic than closing borders, so this relatively low level of support for staying home from work presents a critical communication challenge for public health officials.

LACK OF SUPPORT FOR CLOSING PUBLIC PLACES

Only 44.2 percent of our respondents supported closing schools, day care facilities, places of worship, and stores in a pandemic. Given that many studies show schools as vectors of infection transmission, weak support for this policy is cause for concern. Policy makers should support solid research on the efficacy and timing of school closures in a pandemic, to acquire evidence that will support the policy of keeping children home.

However, it is also possible that answers to the survey question were affected by its mingling of religion, commerce, and education. This potential flaw should be corrected in future studies.

Typically, the components of risk perception—in this case, perceived susceptibility to disease and perceived severity of the disease—affect whether someone will take steps to mitigate risk. Indeed, support for a quarantine and for closing schools, stores, and places of worship was associated with higher perceived risk. However, increased perceived risk had no apparent impact on support for staying home from work; only about half of the people in any group supported this measure.

WORKERS’ ECONOMIC MOTIVATIONS

Officials responsible for giving the public information about risk should realize that even messages that raise levels of perceived risk may not translate into support for the policy of staying home. This is not to say that risk perception does not play a role, but that the most salient risk for many individuals is to livelihood, rather than health.

Lower socioeconomic groups are usually assumed to be less able to work from home or take sick leave, but a wide variety of jobs cannot easily be performed away from the job site. Even corporate executives are subject to cultural norms encouraging people to report to work even when they are sick, or discouraging or limiting the use of leave time when children are sick or child care is unavailable.

A February 2010 study by the Institute for Women’s Policy Research, based on statistics from the Centers for Disease Control and Prevention (CDC), found that an estimated eight million of the twenty-six million US adults thought to have contracted H1N1 worked while sick with the virus. Although the reasons for this cannot be determined, the institute blamed a lack of paid sick leave—which may disproportionately affect women, members of minority groups, and younger workers in low-wage jobs.³¹

Canadian public health officials found that addressing these issues was essential to the success of the SARS quarantine. Without laws and support mechanisms in place to make it financially feasible for workers to stay home when they are sick, efforts to increase public support for such policies may be futile.

COMMUNICATION STRATEGIES

In designing messages to promote compliance with policies in a pandemic, public health officials typically look at predictors of support and tailor messages to several audiences. However, it is worth noting that our survey found wide variability in levels of support and little consistency in the variables that predicted support. Factor analysis failed to reveal relationships between what might appear to be similar policies, so policy makers should not simply assume that there are logical continuums of support from less restrictive to more restrictive, or least effective to most effective, policies. Our results argue instead for a granular examination of each policy and each group.

For example, although women were far more likely than men to support quarantines and border closings, two methods of potentially separating the sick from the well, they were not significantly more likely to support other such policies, such as staying home from work when sick, closing schools and gathering places, or caring for the sick at home. It is possible that respondents perceived quarantines and border closings as things that other people might be subject to, whereas the three remaining policies seemed more likely to be applied to the respondents themselves.

Whatever the explanation, it is clear that no inferences should be made about support for one policy based on support for what public health officials may believe to be a similar policy.

AGE DIFFERENCES

Compared to the referent group of those ages 18–34, those over age 65 were far less likely to support social distancing and yet far more likely to support caring for the sick at home. There are various possible explanations for these differences, and it is not clear which of them is more likely. Perhaps older people have a more traditional work ethic when it comes to job absences, or perhaps they are more isolated and therefore place more importance on maintaining social connections at church and other gathering places. They may also have more experience or confidence in caring for the sick, and be more wary of hospitals as vectors for germs, compared to younger people.

People ages 35–64 also tended to support caring for the sick at home, but they did not agree with the older respondents on other questions. However, they were more likely than younger adults to support border closings. Again, no simple pattern emerges.

TRUST IN GOVERNMENT

Of the attitudinal characteristics, perceived severity and perceived trust in the government’s handling of H1N1 were the two most influential predictors of support for government action. Yet although those who perceived the severity of the crisis to be high were significantly more likely than others were to support quarantines, closing borders, and closing schools and public gathering places, perceived severity appeared to have no impact on other mitigation policies. Trust in the government was linked to support for staying home from work, quarantines, prioritizing limited supplies of medicine, and distributing medicines publicly, but not to support of not-fully-approved medicines or closing borders.

Based on these data, public health officials can make no assumptions about how a given group will respond to a policy. Instead, officials should set strategies on a group-by-group and policy-by-policy basis.

CHANGING ATTITUDES IN THE FACE OF A REAL THREAT

Our results contrast with the more consistent levels of support across demographic and attitudinal groups seen in surveys conducted prior to H1N1,^22,23 when the idea of a pandemic was hypothetical. That change suggests that the reality of H1N1 prompted quite different responses to potential policies and more varied responses across demographic groups. It may also represent confusion, as people grappled with unfamiliar policies and issues.