REPORT TO THE PRESIDENT ON U.S. PREPARATIONS FOR 2009-H1N1 INFLUENZA
III. Anticipating the Return of 2009-H1N1: Envisioning Scenarios
While the course of the 2009-H1N1 pandemic cannot be accurately predicted, it is important to have a clear picture of our current knowledge and to envisage a range of specific scenarios against which to make plans and assess our capabilities.
Our current knowledge is that the virus is readily transmissible, especially to younger age groups, and causes severe clinical manifestations in a small but significant proportion of cases, with most of the severe cases in people under age 65. The proportion of influenza cases that ends in death appears similar to that for seasonal influenza (perhaps 1 per 1,000 patients seeking medical attention), but the absolute number of deaths is expected to be at least as high, if not substantially higher, than for seasonal flu because a higher proportion of the population is likely to become infected (perhaps 40 to 60 percent for pandemic flu versus perhaps 5 to 20 percent for seasonal flu). Moreover, the distribution of deaths is likely to cause a greater loss of expected years of life because the virus predominantly affects younger people. Some specific individuals appear to be at much higher risk, including patients with neurologi cal disorders, pregnant women, and patients with asthma. Certain ethnic groups also may be at higher risk, such as Native Americans. Notwithstanding these observations, there remains great uncertainty about the likely course of the pandemic.
The Working Group believes that planning activities would be aided by development of a small number of specific, shared scenarios describing the possible evolution of the pandemic.
We believe it would be valuable for DHHS to develop a limited number of specific scenarios for dis semination to Federal, state, local and private decision-makers, to be used for assessing capabilities and planning responses.
For planning purposes, we describe a plausible scenario in which the pandemic causes between 30,000 and 90,000 deaths and requires at its peak 50 to 100 percent of ICU beds in affected regions of the country, placing extreme stress on a system in which 80 percent of ICU beds are already otherwise occupied. Analysis of this scenario and alternative scenarios should facilitate decision making about the use of mitigation methods in response to new information about the epidemic.
Because the course of the 2009-H1N1 pandemic cannot be accurately predicted, it is important to have a clear picture of our current knowledge and to envision a range of specific scenarios against which to test our planning and capabilities. While changes in the virus remain possible, the current picture of 2009-H1N1 is as follows:
- The virus is transmitted readily between people at a rate comparable to that estimated for previ ous pandemic strains. In most places where surveillance is available, there is clear evidence of ongoing transmission even through the summer.
- Confirmed cases are concentrated in younger age groups, up to age 24 . According to CDC, infection risk in the 0 to 24 age group is 4 to 5 times greater than for those in the 25 to 49 age group, and 20 times greater than those over age 65.
- Almost all severe cases are in people younger than age 65 . To date, 83 percent of deaths and 71 per cent of hospitalizations from 2009-H1N1 in the United States have been in people between the ages of 5 and 64. This is in stark contrast to seasonal influenza, in which two-thirds of hospitaliza tions and almost 90 percent of deaths occur in persons 65 or older. This means that the years of anticipated life lost per death are much greater than is usual as a result of seasonal influenza.
- The case-fatality ratio (i.e., proportion of infected individuals who die as a result of the infection) appears to be similar to seasonal influenza—possibly on the order of 0.1 to 0.3 percent of medically attended cases (i.e., those infections requiring hospitalization or primary care), and perhaps 0.05 to 0.2 percent of all symptomatic cases, whether or not medical care is sought . However, these numbers are highly uncertain, in particular because the number of medically attended cases is not well measured and the number of mild cases that do not come to medical attention is essentially unknown.
- Despite a similar case-fatality ratio as for seasonal influenza, the number of deaths from 2009-H1N1 is likely to be substantially higher and the deaths and severe illness in the population will likely be concentrated among much younger people than is the case for seasonal influenza . Because most of the population lacks significant immunity to a new pandemic strain, the proportion of people infected in a pandemic is usually substantially higher than for seasonal flu (50 to 70 percent for pandemic flu versus perhaps 5 to 20 percent for seasonal flu). Second, as noted above, the consequences of infection in this epidemic are already known to be far more severe for children and young adults, and seemingly milder for people over age 65 (with deaths mainly among children and young adults, compared to seasonal influenza).
- Individuals with certain underlying medical conditions—including those with neurological disorders and pregnant women—appear to be at substantially elevated risk of severe outcomes . According to CDC, as many as one-third of fatal cases and one-fifth of hospitalizations have been in persons with neurological (e.g., neurocognitive, neuromuscular, seizure) disorders. Pregnant women accounted for 8 percent of deaths and 6 percent of hospitalizations, although they make up about 1 percent of the population. Asthma, diabetes, immunodeficiencies, chronic obstructive pulmonary disease (COPD), and other chronic conditions appear to be associated with severe outcomes as well.
- Certain populations appear to be at elevated risk of severe outcomes, including Native American groups . American Indians and Alaska natives historically are at high risk for severe respiratory infections; while it is unclear what toll they have suffered from 2009-H1N1, a cluster of severe 2009-H1N1 disease among First Nation people in remote Manitoba, Canada, suggests that these groups may be at high risk. Cases of 2009-H1N1 virus infection in these clusters have had rapidly progressive, diffuse, lower airway disease (compared to seasonal influenza, which more commonly involves the upper airway), resulting in development of acute respiratory distress syndrome (ARDS) and prolonged ICU admission.
The Need for Concrete Scenarios for Response Planning
The Working Group is concerned that uncertainty about the course of the 2009-H1N1 pandemic may be hampering planning. While uncertainty is inherent in pandemics, planning activities may be aided by development of a limited number of specific, shared scenarios that describe the possible evolution of the pandemic. Dissemination of a limited number of plausible scenarios would provide a framework against which decision-makers at the Federal, state, and local levels could test current capabilities and also structure specific plans and decision points. In the absence of such frameworks, decision-makers may fail to adequately assess capabilities relative to potential needs. They also may fail to foresee key decision points and be forced into hasty decisions in the “heat of battle.” A scenario-based approach already has been embraced by the United Kingdom , which has defined and made public its planning assumptions for a “reasonable worst case” scenario.
MAIN RECOMMENDATION (CHAPTER 3)
We recommend that DHHS rapidly develop a limited number of specific scenarios and disseminate them to Federal, state, local and private decision-makers for planning purposes. Components of these scenarios ideally would include:
ii. peak burden on primary care providers, emergency rooms, hospital admissions, and ICUs;
iii. number of doses and timing of vaccine availability;
iv. dosing requirements and efficacy of vaccine; and
v. efficacy and supply of antiviral drugs and medical materiel.
These scenarios would allow Federal, state, local and private entities to assess their capacity and develop plans for deployment and targeting of medical and non-medical interventions under the various scenarios.
In addition, it would be valuable for DHHS to define trigger points related to changes in circumstances (e.g., change in severity) to facilitate timely action, as well as the data and data streams that will be required to activate these trigger points.
To illustrate this approach, we describe in Box 3A a scenario that we consider to be a reasonable model for planning, followed by sample decision points that might be appropriate. We also suggest indicators and triggers to redirect decision making should an unanticipated event emerge within the scenario. The assumed characteristics of the model scenario are described in Table 3-1.
BOX 3A : A MODEL SCENARIO: A POSSIBILITY, NOT A PREDICTION
One plausible scenario is that there will be resurgence in transmission of 2009-H1N1 this fall that is comparable to that seen in spring-summer 2009 but with higher rates of transmission due to the resumption of school and the cooler, drier weather. Following a relatively steady or declining burden of cases in August, the number of new cases will begin to rise exponentially in the first week of September, growing 10-fold about every 10 to12 days. Hypothetically, the peak incidence of infection nationally will occur around October 15, with minor variations across the country such that peak incidence almost everywhere will occur during the month of October. At this peak, perhaps 1 to 2 percent of the population will become infected each day.
Predicting demand on the health care system during this peak is fraught with uncertainties, but the following numbers from one possible scenario are illustrative. During the peak, 1 or 2 out of every 2,000 Americans might be hospitalized. Cases requiring mechanical ventilation or intensive care could reach 10 to 25 per 100,000 population, requiring 50 to 100 percent or more of the total ICU capacity available in the United States and placing great stress on a system that normally operates at 80 percent of capacity. Because adult ICUs are not prepared to care for pediatric patients, there could be a particular shortage of facilities for sick children. In particular locations, the stress on the health care system could grow even more acute, as large outbreaks occur in prisons, schools, and isolated communities with limited health care access, such as Native American reservations. As awareness of the pandemic spreads, pressure on emergency departments could mount, with long lines and a need for triage of mild cases and non-influenza cases.
Alongside these health-related burdens, substantial absenteeism from work and school could occur, as sick children stay home, schools with large outbreaks close, and parents are forced to stay home either because of their own ill ness or to take care of sick children. Key members of the social infrastructure, such as police officers and firefighters, are increasingly home ill. Exposure of healthcare workers to sick patients is continual and antiviral supplies prove inadequate for ongoing prophylaxis of these workers. Retail pharmacies run out of antiviral supplies in late September or earlier, and states face the demand to replenish these supplies from state stockpiles and state Strategic National Stockpile allocations; however, many states lack the ability to move antiviral drugs into the retail supply chain and focus on delivery to hospitals. Hospitals face competing pressures to dispense antiviral drugs for prophylaxis of their workers, to provide them to patients appearing in the emergency room, or to save them for the sickest admitted patients. Debates intensify about the value of antiviral use for long-term prophylaxis or early treatment for mild infection in high-risk groups such as pregnant women and immunocompromised patients, treatment of severely ill patients, and prophylaxis of essential healthcare workers.
In this model scenario, around October 15, as the epidemic peaks, a major supply of 2009-H1N1 influenza vaccine becomes available. Immunization starts within days, with considerable geographical variation in the rate at which administration occurs. Immunization of priority groups is completed by early or mid-November, resulting in immunity in vaccinated adults by mid-late November, as the epidemic wanes in most populations. Children require two doses and do not acquire immunity until December, when new infections will have become rare.
By the end of 2009, 60 to 120 million Americans would have experienced symptomatic infection with 2009-H1N1; nearly 1 to 2 million would have been hospitalized, with about 150,000-300,000 cared for in ICUs; and somewhere between 30,000 and 90,000 people would have died, the majority of them under 50 years of age.
We emphasize that this is a plausible scenario, not a prediction. By way of comparison, it is less severe by a factor of three (in terms of expected deaths per capita) than the “reasonable worst case” planning assumptions, publicized by the UK government, for the H1N1 resurgence in that country.
Beyond this scenario, alternative scenarios are needed to take into account the possibility that major assumptions are incorrect. In particular, four variations are of notable importance:
- A milder scenario in which the number of deaths and severe cases is much lower than outlined here, perhaps because many mild cases or infections without symptoms were missed in the spring, leading to an overestimate of the severity.
- A modified scenario in which a large fraction (e.g., one-third) of 2009-H1N1 cases are resistant to oseltamivir (Tamiflu) by the peak of the epidemic, reducing the effectiveness of an important method for mitigating the epidemic.
- A more severe scenario, in which changes in the virus result in elevated rates of hospitaliza tion, intensive care demand, and death. In this case, the focus of severe disease may shift more toward the general population, making focused attention on groups that showed high-risk in the spring less of a priority.
- A delayed scenario in which transmission does not increase dramatically in the early autumn, so that vaccine availability precedes the peak of the epidemic, reducing the number of subsequent cases by conferring protection through immunization.
We emphasize again that the baseline scenario and the alternatives above are given as examples for planning purposes; they are not predictions of what will happen. DHHS should exercise its own expert judgment in defining the most relevant scenarios, with the caveat that scenarios other than the most likely also should be considered. In addition, planning should include at least one scenario in which the peak of the epidemic precedes the availability of significant vaccine supplies.
To illustrate the value of scenario-based analysis, it is useful to consider issues in vaccine allocation, since the timing of availability of significant quantities of vaccine is still uncertain.
- In the model scenario, we assume that vaccine administration will commence around the peak of the pandemic, with substantial population-level immunity occurring only 2 to 8 weeks after the peak. In this case, vaccination will have limited value in reducing transmission. There may be a strong rationale for vaccinating certain high-risk groups as rapidly as possible, by accelerating the availability of at least some vaccine.
- If transmission is substantially delayed compared to the model scenario, vaccination of children may be of high value epidemiologically: it may be possible to immunize many before exposure, protecting them and decreasing spread.
- Conversely, if an increase in severity is detected with the expected rate of transmission, broader administration of vaccine before complete clinical trial data are available may be appropriate, and the use of adjuvant (as discussed in Chapter 5) might offer an improved risk-benefit profile.
In addition to analyzing specific scenarios in advance, it will be important to define indicators that could trigger the need to make changes in plans, and to incorporate these indicators into scenarios. We believe that it would be valuable for DHHS to define specific triggers in advance to the extent possible, since this will allow orderly decision making when unexpected events arise. (It is worth noting that beyond these triggers, public pressure in response to events, such as a cluster of child deaths, may force certain communities to change their strategies.) Examples of potential triggers are shown in Box 3B.
BOX 3B: EXAMPLES OF INDICATORS THAT MIGHT SERVE AS TRIGGERS FOR ACTION
Indicators of unacceptable burdens on health care might trigger guidance to intensify community mitigation to spread out the peak burden.
Indicators of substantially increased severity that might justify changes in plans for antiviral use, vaccine formulation (adjuvant use), or community mitigation.
In addition, concentrations of unusually severe cases could occur in a population subgroup defined by geography/ethnicity (e.g., a remote Native American population), by underlying medical condition (e.g., pregnancy or a novel risk factor), or by place of residence (e.g., nursing home, prison). In such cases it will be urgent to provide adequate treatment for affected persons, consider accelerating vaccine delivery to similar groups, and ascertain reasons for this increased severity, distinguishing viral changes from infectious cofactors, host factors, or other reasons.
|Join the GlobalSecurity.org mailing list|