The emergence of new infectious diseases has forced public health planners to refocus on the need for improved preparedness strategies. The 2009 H1N1 influenza pandemic led many policy makers to consider how they would respond to a potentially devastating disease outbreak.Reference Fauci1Reference Holmberg, Layton, Ghneim and Wagener2Reference Thomas, Dasgupta and Marinot345 The Centers for Disease Control and Prevention (CDC) reported that between April and November 14, 2009, approximately 9820 deaths occurred, 213,000 people were hospitalized, and 47 million people were infected with H1N1.6 Despite significant incidence, this new infection appears to have morbidity similar to seasonal influenza.7 There remains the possibility, however, that a viral mutation could occur in which influenza became highly contagious and its potential lethality could result in a panic with devastating worldwide health and economic impact.Reference Fauci145Reference Monto8Reference Juckett9 Researchers speculate that a variant of influenza, such as the H5N1 avian influenza virus, has the potential to become such a virus.Reference Iwami, Takeuchi and Liu1011 Because many believe it is only a matter of time until another deadly influenza pandemic occurs, most preparedness plans focus on methods of minimizing worldwide impact by containing the virus before widespread transmission can occur.Reference Holmberg, Layton, Ghneim and Wagener2Reference Juckett9Reference Oshitani12
US media reports about pandemic plans have prominently featured pharmacological responses to an influenza outbreak. These include the stockpiling of medications such as oseltamivir and efforts to develop a protective vaccine.Reference Holmberg, Layton, Ghneim and Wagener2Reference Thomas, Dasgupta and Marinot3Reference Enserink13 Responses to the 2009 H1N1 outbreak relied mostly on encouraging vaccination, especially for vulnerable populations, and promoting personal protective measures such as hand washing and using alcohol-based hand sanitizers. Less often reported are nonpharmacological countermeasures including social distancing, isolation, and quarantine.Reference Juckett914Reference Gostin15Reference Toner16 Considered a central component of the successful effort to end the severe acute respiratory syndrome (SARS) outbreak in 2003, many planners now regard these measures, especially isolation and quarantine, as essential tools to slow or prevent spread in the initial stages of an influenza pandemic.Reference Svoboda, Henry and Shulman17Reference Bell18
Quarantine has been used extensively in public health, but it has not been used in the United States for nearly 70 years.Reference Gernhart19 Legally, most explicit quarantine powers are granted to states, originating at a local level during the colonial era.Reference Barbera, Macintyre and Gostin20 Today, states may quarantine under a variety of laws addressing the health and safety of citizens.Reference Daubert21 The federal government also has more implicit quarantine authority under the Commerce Clause in the US Constitution (Article I, Section 8, Clause 3),Reference Swendiman and Elsea22 most notably in cases in which there is a risk of transmission of infectious disease across state linesReference Swendiman and Elsea22 or if a state requests the intervention of the federal government.23 Because many state quarantine laws are more than 100 years old and in response to the September 11, 2001 terror attacks and the SARS outbreak, public health experts developed the Model State Emergency Health Powers Act,Reference Brown24 which provides guidance to states on revamping emergency response plans, including the use of quarantine orders.Reference Swendiman and Elsea22 Some argue it expands the federal government's authority under an Executive Order25 to enact quarantine orders, including broad surveillance rights, mandatory vaccination and treatment, destruction and seizure of personal property, and isolation and quarantine.Reference Erickson, Gostin, Street and Mills26 Despite this expanded federal authority, a majority of states have enacted legislation based on these guidelines.Reference Brown24
Studies of past uses of quarantine, including during the 1918–1919 “Spanish flu” pandemic and the 2003 SARS outbreak, have identified several components of successful programs, as well as some shortcomings.Reference Markel, Lipman and Navarro27Reference Schabas28Reference Hawryluck, Gold and Robinson29Reference Cava, Fay and Beanlands30Reference DiGiovanni, Conley, Chiu and Zaborski31Reference Blendon, DesRoches and Cetron32Reference Gostin, Gravely, Shakman, Markel and Cetron33Reference Blendon, Benson, DesRoches, Hermann and Weldon34 Current CDC guidelines for pandemic influenza mitigation35 call for 3 primary strategies to combat the spread of influenza: vaccination, treatment of infected as well as exposed people, and implementation of infection control and social distancing measures, primarily quarantine in the home or a health care setting.35 This plan was used during the most recent H1N1 outbreak but primarily focused on closing schools as a way to produce social distancing; since then, the CDC has discontinued encouraging school closures and has instead stressed the importance of vaccination, personal protective measures, and voluntary quarantine at home.35 It is not clear how compliant the public has been with voluntary home quarantine. Even in situations in which people voluntarily comply with quarantine orders (eg, when the infectious agent is extremely lethal), the widespread use of quarantine would present many challenges, including the location, monitoring, care, and possible compensation of those placed under quarantine. Public support is vital for any program involving quarantine and isolation.Reference Blendon, Koonin, Benson and Cetron36Reference Janssen, Tardif, Landry and Warner37Reference Aledort, Lurie, Wasserman and Bozzette38Reference Richards, Burris and McNelis39 Addressing these challenges to enlist the support of the public is essential to optimize compliance in an environment of limited resources.
The purpose of this study was to investigate factors that may influence an individual's decision to comply with a quarantine order by using a hypothetical avian influenza pandemic as an example. This study examines not only the demographic characteristics of the individuals but also their attitudes about avian influenza, overall understanding of its transmission, and perceptions of its severity and the perceived risk of contracting the disease. Our results provide health planners with information that is useful for improving efforts to enact a quarantine order when the next pandemic occurs.
METHODS
A total of 1204 adult Pennsylvania residents participated in a random digit dial telephone sample in September 2006. The interviews were conducted by Abt SRBI Inc using a computer-assisted telephone interviewing system, under the direction of the Institute for Public Affairs, Temple University. Sample telephone numbers were called up to 8 times to complete an interview, and 1 attempt was made to convert each refusal. Within each household, a respondent was chosen according to a random-selection method. The response rate for the study was .28.
The interview data were weighted to accurately reflect population distributions across the state. These weights included a telephone weight to correct for unequal probabilities of selection due to a household having more than 1 telephone, a region weight to correct for population differences in the state, and a poststratification weight to represent the demographic characteristics of the population. These weighting strategies ensured that our sample's demographics matched the US Census estimates of the adult population. The overall margin of error attributable to sampling is ±2.8 percentage points. The sampling error for subgroups is slightly larger.
Survey
The survey instrument was developed based on existing surveys dealing with avian influenzaReference Blendon, Koonin, Benson and Cetron36Reference Janssen, Tardif, Landry and Warner37 and a literature review of quarantine.Reference Blendon, Koonin, Benson and Cetron36Reference Janssen, Tardif, Landry and Warner37Reference Aledort, Lurie, Wasserman and Bozzette38Reference Richards, Burris and McNelis39 All of the avian influenza–specific questions used a 0 to 10 scale in which respondents were asked to rate how much they disagreed or agreed with statements or how likely they were or were not willing to carry out specific behaviors. The survey was structured to elicit overall perceptions of avian influenza and concern levels without explaining the differences between avian influenza that is transmitted from bird to human and avian influenza that may mutate and be transmitted from person to person, leading to a highly infectious pandemic.
A statement was then read to the respondents describing the difference between avian influenza as it exists today and how it may change if it mutated (Table 1). The interviewer also explained what quarantine is and how the government may carry it out (Table 1). This was done to ensure that all of the respondents were answering questions about vaccination and quarantine using the same definitions of avian influenza and quarantine. After reading these statements, respondents were asked about willingness to comply with quarantine at varying levels of coercion:
TABLE 1 Survey Questions on Attitudes About Avian Influenza and Likelihood of Compliance with Quarantine
Level 1: respondent would voluntarily choose to stay at home for 2 weeks
Level 2: respondent would stay at home for 2 weeks if the US government requested
Level 3: respondent would travel to a quarantine facility if the US government requested
Level 4: respondent would travel to a facility if the US government ordered
Survey questions in the order in which they were asked are outlined in Table 1.
Analysis Strategy
Descriptive statistics were used to summarize the sample in terms of sociodemographic variables as well as responses to survey items. Means and standard deviations were reported for continuous measures of survey response items and frequencies were reported for categorical demographic measures. Two sample t tests and 1-way analysis of variance (ANOVA) models were used to compare continuous response items according to categorical independent variables including working status, sex, race, age, education, income, and religiosity. Linear regression models were then estimated to determine predictors of quarantine using sociodemographic and attitudinal variables. Post hoc tests were then used to compare pairwise differences in adjusted mean values when the independent variable had more than 2 levels.
RESULTS
Demographic Characteristics of the Sample
Table 2 presents the demographic characteristics of the survey sample. The sample comprised proportionately fewer men than women (45.8% vs 54.2%) and was primarily white (88.2%). It was also evenly split among the age-distribution groups, with roughly one third of the sample in each group. The majority of the sample (58.8%) was working, almost 45% had a household income between $30,000 and $74,000, and 56.5% had graduated from high school. Although the majority (62.2%) did not consider themselves born-again Christians, 41.6% indicated they attend religious services once per week or more often. In comparison with the 2000 Pennsylvania Census data, the survey sample is similar in demographics and differences were controlled for in the weighting scheme.
TABLE 2 Demographic Characteristics of the Sample
Likelihood of Compliance With Quarantine
Respondents were asked how willing they would be to comply with various levels of quarantine on a scale of 0 to 10, with zero meaning very unlikely and 10 meaning very likely. Using ANOVA, we examined compliance with each type of quarantine (dependent variables) using sociodemographic characteristics.
As shown in Table 3, ANOVA analysis revealed statistically significant differences in means by demographic group for all types of quarantine. Sex is significantly related to compliance with all 4 types of quarantine. Women consistently reported being more willing than men to adhere to each level of quarantine, including choosing to stay at home (F = 36.4, P < .0001), staying home if requested by the government (F = 43.45, P < .0001), traveling to a facility if requested by the US government (F = 6.77, P = .0094), and traveling to a facility if ordered by the US government (F = 7.37, P = .0067). A smaller difference was observed by race; nonwhites were more likely than whites to indicate that they would be willing to travel to a facility if asked by the US government (F = 4.71, P = .0302).
TABLE 3 Quarantine Variables by Independent Demographic Variables, ANOVA Analysis
There were also significant relations by age, education level, income, and religiosity:
• Respondents older than 65 years were more willing than younger respondents to choose to stay at home (F = 5.56, P =.004) or to stay at home if requested by the US government (F = 3.36, P = .0332).
• Respondents who had graduated from high school were more likely than those who did not to say they would choose to stay at home (F = 3.61, P = .0273), while both high school and college graduates were more likely than those who were not to stay at home if asked by the US government (F = 5.48, P = .0273). Those without a high school diploma, however, indicated that they would be more likely to travel to a facility if ordered by the US government (F = 3.16, P = .0426) compared with those with more education.
• Respondents making less than $30,000 or between $30,000 and $74,000 (F = 3.34, P = .0357) were more likely to say they would stay at home if the US government asked or travel to a government facility if the US government requested (F = 4.72, P = .0091) compared with those who made more than $75,000.
• Respondents who were not employed were more likely to say they would choose to stay at home (F = 23.98, P < .0001), stay at home if asked by the US government (F = 12.04, P = .0005), and travel to a facility if the US government asked (F = 4.77, P = .0291) compared with those respondents who were employed.
Significant differences were seen using the variable of frequency of religious service attendance. Respondents who indicated that they never attended religious services were much less willing to comply with quarantine orders than those who attended services. This group was significantly different from the other 2 groups in willingness to stay at home if asked (F = 5.02, P = .0067), travel to a facility if asked (F = 7.84, P = .0004), and travel to a facility if ordered (F = 16.33, P < .0001).
Multivariate Models Predicting Likelihood of Compliance With Levels of Quarantine
Regression analyses determined which survey items significantly predicted the likelihood of compliance with the 4 levels of quarantine. Regression analyses included all of the demographic variables used in ANOVA analyses as well as attitudinal and overall knowledge questions concerning avian influenza, as shown in Table 1.
As indicated in Table 4, a number of variables contributed significantly to predicting compliance with all of the quarantine variables. Specific predictors for choosing to stay at home included how closely respondents were following the news; in this case, not following the news was a significant predictor of respondents stating that they would choose to stay at home. In addition, how likely someone thought he or she or someone close to him or her may contract avian influenza predicted reported willingness to stay at home as did being female and not presently working. The overall model for this quarantine variable had an R 2 value of .13 (F = 4.55, P < .0001).
TABLE 4 Regression Analyses Predicting Reported Compliance With 4 Levels of Quarantine
For the quarantine variable stay at home if the US government asked, significant predictor variables included how closely someone was following the news (again, those who had not been following the news were more likely to say they would stay at home), how likely he or she believed he or she or someone close would contract the virus (after being told the difference between avian influenza as it exists today and what could occur if it becomes infectious), sex (being female), and age (being between the ages of 18 and 44). The overall model for this quarantine variable had an R 2 value of .12 (F = 4.13, P < .0001).
There were a number of predictors of respondents' willingness to travel to a facility, either by US government request or order. If the US government requested, only 1 predictor variable was significant, whereas there were 7 significant predictors of complying with the more coercive level of quarantine. The 1 significant predictor for requesting that individuals travel to a facility was the belief that “my doctor would be able to get enough vaccine” (R 2 = .11, F = 3.53, P < .0001). Significant predictors for the US government order level of quarantine included the belief that the respondent or someone close to him or her would contract avian influenza, the belief that someone could die from avian influenza, the belief that the respondent's doctor would be able to obtain enough vaccine if it were available, increased attendance at religious services, having less than a high school education, and not presently working. The overall regression model for this quarantine variable had an R 2 value of .10 (F = 3.24, P < .0001).
DISCUSSION
A significant finding is that sociodemographic characteristics only partially predict respondents' self-reported likelihood of complying with quarantine orders. Including both sociodemographic characteristics and attitudinal variables better predicts self-reported likelihood of compliance with each of the levels of quarantine in a hypothetical avian influenza outbreak.
Although differences can be seen across sociodemographic groups in response to quarantine questions, the ANOVA revealed significant differences between or among groups in willingness to comply, with the likelihood of willingness to comply decreasing as the level of quarantine increased. For example, although most respondents said they would choose to stay at home for 2 weeks, far fewer were willing to travel to a quarantine facility if requested or ordered to by the government. Significant differences in willingness were evident by sex, working status, education level, age, and income. Women overall were much more likely than men to indicate they would be willing to comply with quarantine orders. Although some women may indicate more willingness to stay at home because they are not working, employment status does not seem to explain why women are more willing to travel to a government facility. Perhaps because women exhibit more positive health protective behavior, they perceive quarantine in whatever form as a way to avoid contagion. Employment status shows a similar pattern. Again, it may be that those who are not working are more likely to stay at home because they do not see themselves as having to worry about losing their job if they stay at home or shelter in a government-provided facility for 2 weeks.
Age, education level, and income groups were also significantly different. Respondents who were older than 60 years were more likely than those who were 60 or younger to say they would stay at home by choice or if asked. Because many of those older than 60 may also not be working, it follows that they would be more willing to stay at home. Younger people, however, were less willing to stay at home, perhaps for fear of losing employment. An alternative explanation is that older people are, on average, more concerned about health issues and are also more familiar with the ravages of epidemics such as poliomyelitis and tuberculosis, which are unfamiliar to younger respondents. Interestingly, this is in opposition to those willing to be vaccinated for influenza. Many studies have reported a disparity in the number of those older than 65 years, as well as ethnic disparities, in obtaining seasonal influenza vaccination.40Reference Egede and Zheng41
Respondents who have earned a high school diploma were more likely to be willing to choose to stay at home than those with a college degree, but college graduates were more willing to stay at home if the government asked than those without a high school diploma. Finally, respondents without a high school diploma were more likely to say they would travel to a government facility if ordered to, compared to both those with a high school diploma or college degree. Because people with lower levels of education also tend to have lower incomes and less social capital, they may fear reprisal from the government by not complying more than people who have higher incomes and more social capital.
Similar significant differences were seen in income levels. Those with moderate levels of income were more willing to stay at home if asked to, whereas those making <$30,000 were more willing to travel to a facility if asked to by the government. It may be that those with lower incomes are either not working at all or do not believe they have a choice if the government tells them they must comply. Conversely, those with the highest incomes are the least likely to comply with any of the quarantine orders. Again, this may be a function of social capital or empowerment, wherein these respondents believe that they have the capacity to make a decision about what to do in an avian influenza epidemic and do not need the input of the government. They may also fear reprisal less.
Finally, large differences were seen in willingness to comply between those who never attended religious services compared to those who did. Respondents who never attended services were significantly less likely to stay at home if the government asked compared to those who attended services once per week or more. They were also less likely than either those who regularly attended religious services or periodically attended services to travel to a facility if asked or ordered. In fact, the more often they attended religious services, the more likely respondents were to indicate their likelihood of complying with each level of quarantine. It may be that respondents who regularly attend religious services are more trusting of institutions and thus more likely to accept government recommendations or orders. They may also have a greater concern for the well-being of communities than those who are less integrated into community institutions. Respondents who do not attend religious services may be less likely to comply with authority, including government recommendations and orders, and believe they prefer to make their own decisions about what to do in an avian influenza outbreak.
A number of knowledge/attitude variables also positively predicted the likelihood of complying with levels of quarantine. Closely following the news about avian influenza was negatively associated with likelihood of complying (ie, respondents who had been following the news about avian influenza were less likely to say they would comply with quarantine orders than those who had not followed the news closely). In contrast, respondents who believed that they were susceptible to the virus were more likely to indicate a likelihood of complying with recommendations or orders. This seems incongruent; however, respondents who had not followed the news may have obtained misinformation about avian influenza and believed themselves more susceptible than those who were informed about the virus and its effects. Thus, people who were the least informed may also have believed they were more likely to contract the virus and die of it than do those who followed the news and were less fearful about contagion. Finally, those who believed that their doctor would be able to obtain enough vaccine to distribute to them if an epidemic occurred were also more likely to say they would comply with quarantine orders. Perhaps those who had confidence that they would obtain access to vaccines have a generally positive attitude about health care and public health efforts, and perceived quarantine as an acceptable strategy to protect their health.
Because perception that doctors can acquire vaccine was significant in predicting the likelihood of complying with 3 of the 4 levels of quarantine, ensuring access to vaccines may be a particularly important element in successfully implementing quarantine. This may be part of the public's frustration with the distribution of the H1N1 vaccine, which was mostly provided by public health clinics or “vaccination events” rather than by private doctors.Reference Ehart and Geracimos42 Future efforts may be more successful if vaccinations were provided by a broader number of providers.
The regression analyses elucidate which variables best predict who would or would not comply with quarantine orders. Each of the levels of quarantine scenarios have slightly different significant predictors, suggesting that messages may have to differ depending on the level of quarantine action that public health officials believe needs to be taken. Although the overall fit of the model and the R 2 statistics are relatively low (ranging from .10 to .14), the models do identify specific variables that may help public health planners to develop media or risk communication messages to maximize public quarantine compliance.
Limitations
A major limitation of this study is that respondents had low levels of familiarity with avian influenza and thus may have had difficulty accurately envisioning their likelihood of complying with quarantine recommendations or orders. In addition, because R 2 statistics were low, other variables not captured in this survey may be important in determining how people may react in the event of an avian influenza outbreak in the United States. Although a 28% response rate is not untypical for a random digital dial telephone survey, the methodology may have inherent biases, including who answers the telephone and/or is willing to participate in a telephone survey. This may have affected the validity and reliability of the survey and its results. It should also be noted that the language used in the statement about quarantine and read to respondents did not clarify which level of government would be issuing a quarantine order (state or federal), although questions about willingness to comply with quarantine orders or recommendations used the words “United States government” (Table 1). This omission may have changed the answers of some respondents who distrust the federal government more than their state government and could have affected results. However, overall willingness to comply with quarantine orders was generally high and this effect may not have significantly affected results. Finally, as with any study of intentions, behavior may be different in an actual, rather than hypothetical, situation.
CONCLUSIONS
These results provide information that is potentially useful for preparing for an outbreak that may be contained with isolation and quarantine. Unlike the present H1N1 outbreak, a mutated H5N1 outbreak, which could have higher morbidity and mortality, would require further community mitigation besides vaccination efforts and personal behaviors including isolation and quarantine. In this study, respondents viewed varying levels of quarantine differently, suggesting that public health communications need to be crafted to take into account the likelihood of decreasing levels of cooperation with increasing levels of coerciveness. In general, the public would initially be cooperative with a generalized request to voluntarily quarantine themselves or quarantine themselves if asked to do so in the event of an avian influenza outbreak; and this could, if reinforced, be sustained through the initial stages of an outbreak. Cooperation levels would likely decrease, however, the longer the outbreak lasted. Thus, it would be important for public health planners to develop a clear plan that is communicated to the public and sets realistic guidelines for how the public is to comply with the request. Similar to the www.flu.gov Web site and the subsequent public information campaign on the H1N1 outbreak that encouraged vaccination and frequent hand washing, communication about what isolation and quarantine are and what is expected of the public must be clear, with an articulated expected result. This is certainly true when public health officials are competing with an around-the-clock news cycle that produces many contrary messages. The safety of the H1N1 vaccine was 1 message public officials have had to discuss and defend, resulting in public confusion about who should and should not be vaccinated.Reference Ehart and Geracimos42 In the case of mandatory quarantine, messages would be even more difficult to keep clear.
In this study, for example, if asked or ordered to seek shelter in a facility, respondents were much less willing to cooperate, although means were in the mid-range of the 0 to 10 scale. We can conclude from these results that, although not pleased with the idea of going to a facility, most respondents would not respond overly negatively toward the possibility, especially if the morbidity and mortality of the outbreak was high. This attitude may be enhanced further by targeted messages to those groups who indicated they were the least willing to comply. It should be noted that these means actually increase when the order is more coercive. This is encouraging and again suggests that levels of cooperation by the public during a pandemic would, at least initially, be more positive than expected. It may also indicate that the more specific public health planning is before an outbreak occurs, the better the response from the public.
Finally, public health preparedness planners may consider the possibility of increasing the public awareness of quarantine and the willingness of significant proportions of the population to comply with orders. Such awareness campaigns may foster a spirit of cooperation and expectation of being prepared to take positive action to contain an outbreak successfully.
Authors' Disclosures: The author reports no conflicts of interest.
