The threat of a human influenza pandemic has greatly increased over the past several years with the emergence of highly virulent avian influenza viruses, notably subtype H5N1 (9), and the more recent emergence of subtype H1N1. Federal agencies have modeled the high probability of a serious pandemic influenza outbreak and have begun to institute national and state plans to reduce transmission and mitigate the disease (10). The inadequate supply of some vaccines and antiviral medications and insufficient community mitigation planning have led to concern that the United States is inadequately prepared to deal with a pandemic (11).

Improving pandemic preparedness is critical, given the catastrophic consequences of influenza pandemics that have occurred in the past century, in 1918, 1957, and 1968; the severity of all past pandemics was substantial, ranging from 700,000 deaths (in 1968) to >50 million deaths (in 1918) (1,11). Evidence to determine the most effective nonpharmaceutical intervention strategies is limited (12). Some strategies being suggested include targeted, layered containment (13), which involves antiviral drug treatment for identified case-patients and prophylaxis for and quarantine of their household members, school closures, and social distancing in the community and workplace (2,8,13,14). Research has suggested that US adults seem to possess a broad willingness to comply with response strategies that include social distancing, although some segments of the population will likely be less able to comply with isolation recommendations (8), particularly those related to isolation from the workplace.

This study may provide public health authorities with realistic expectations for the success or failure of proposed mitigation measures, given that some population subgroups may have less ability to comply with recommendations because of real or perceived job insecurity and financial problems associated with missing work. Our findings suggest that some employment characteristics (inability to work from home, lack of paid sick leave) are associated with working adults’ ability to comply with recommendations and will be major workplace intervention points (areas to target) in the event of a serious outbreak. In addition, sociodemographic characteristics (particularly low-income status) put some workers at disproportionate risk of contracting and spreading pandemic influenza because of their perceived inability to miss work. These assessments may help identify the conditions under which some groups will be disproportionately likely to fail to comply and may help with workplace efforts to plan accordingly and communicate effectively in the event of a serious outbreak of pandemic influenza.

Job insecurity, whether real or perceived, is a real consideration for many working adults. US health authorities recommend that to prepare for a pandemic, businesses should establish policies for nonpunitive liberal leave and flexible worksite accommodations (2). However, we know of no legal precedent for mandatory job protection in the event of public health emergencies. Our study found that employees without paid sick leave, those with low income, and those who live in urban areas fear losing their jobs should they comply with recommendations to stay home in the event of a serious outbreak of pandemic influenza. Those respondents who said that they would not be paid if kept from work were almost 5× more likely as those who would receive pay to say that they would lose their job or business as a result of having to stay home from work. We were not surprised by this finding, given the long history of social epidemiology literature (e.g., the Whitehall studies) that has documented the effect of occupational status or grade, organizational injustice, job stress, and workplace power differentials on both job insecurity and disease outcomes (15–20). The effect of lack of paid sick leave provides insight into a measure of inequality in the work force, such that some groups of employees (e.g., those in minimum wage jobs or without paid sick leave), because of concerns about job security that stem from their workplace status, lack the power to choose to stay home from work in the event of an outbreak. Notably, across all income categories, low- and middle-income workers were significantly more likely than high-income workers to say that they would be likely to lose their job or business as a result of staying home for 7–10 days in the event of an outbreak. Those respondents living in urban areas also were 60% more likely than those living in rural areas to fear job insecurity. This fear could pose substantial problems for pandemic influenza mitigation because those in urban areas may be strongly encouraged to remain isolated to avoid virus spread in conditions of population density and crowding.

Financial problems also are likely to weigh heavily on the minds of US workers during a pandemic, and these problems may be part of the complexity of factors that comprise compliance considerations. Although US health authorities have recommended that businesses develop policies for employee compensation in the event of an influenza pandemic that causes workplace absences (2), we know of no precedent requiring that paid sick leave be granted (by employers or state or federal government) to employees who comply with isolation recommendations and miss work in the event of a public health emergency. Our study has elucidated some employment characteristics that are associated with the likelihood that workers think they would experience serious financial problems if they had to miss work; inability to work from home and lack of paid sick leave were associated with reports of experiencing serious financial problems if isolated from work over the 3 periods: 7–10 days, 1 month, and 3 months. Respondent income was another significant predictor of serious financial problems that may limit ability to comply with isolation recommendations. Even relatively short periods of isolation from the workplace (7–10 days) would be a problem for low-income workers, and if an outbreak were serious enough to warrant 1-month isolation recommendations, persons in low- and middle-income groups would have more difficulty complying than would upper-income groups, thus limiting the effectiveness of mitigation strategies. Moreover, at 3 months’ of isolation, persons from all income levels, especially low-income, would likely experience serious financial problems.

Strengths of our study include its practical significance; our findings may help preparedness planners find work-specific strategies that may increase the likelihood of compliance with isolation recommendations. These strategies may include working with employers to ensure work-from-home or sick leave capabilities for nonessential employees and planning to provide state or federal supplementary income support and job protection for workers who would not be paid if they missed work because of official pandemic mitigation recommendations. Other countries have implemented similar measures in emergencies; for example, during the 2003 pandemic of severe acute respiratory disease, the government of Singapore provided financial support to citizens who had to stay home to prevent the spread of the disease (21). In the United States, no such measures have been taken to prevent the spread of infectious diseases, but for other emergencies such as hurricanes and floods, federal income support has been provided to victims by way of disaster relief funds. In the event of a serious outbreak of pandemic influenza, when timely action will be needed to encourage and ensure isolation compliance, an existing mechanism for delivering financial support to affected persons is unemployment insurance. Currently, eligibility requirements for this benefit are limited to employees who involuntarily lose their job, but this requirement could be changed to use an existing system to disperse lump-sum payments to those financially affected by a pandemic, if the severity of an outbreak warranted isolation from the workplace for long periods.

Findings from our study should be considered in light of a few limitations, including the 36% response rate. Low response rates can bias samples, reflecting systematic differences between responders and the population from which they were drawn, thus limiting the external validity of estimates (extrapolation to the general population). However, the 1-month period of the survey (and thus limited time for callbacks) may mirror what might be necessary in the event of a pandemic, in which public surveys with a rapid turnaround time are necessary to gauge public knowledge and resource needs in an emergency situation. We point to research that suggests that the results of weighted data from surveys of shorter duration are similar to those based on surveys of longer duration and higher response rates and can be used without an unacceptable risk for bias (22,23). Furthermore, the HSPH Pandemic Influenza Survey, as it relates to our findings about job insecurity, did not assess perceptions of job loss versus reality of job loss, nor did it assess reasons why some respondents perceived that job loss would be a consequence for missing work for 7–10 days in the event of a serious outbreak. Future population surveys could attempt to disentangle these beliefs to inform policy and communication aimed at enabling compliance with workplace isolation strategies to quell the spread of a future pandemic.