Use of Influenza Risk Assessment Tool for Prepandemic Preparedness

Stephen A. Burke; Susan C. Trock


Emerging Infectious Diseases. 2018;24(3):471-477. 

In This Article


During 2011–2017, SMEs evaluated 14 animal-origin influenza viruses using the IRAT. The emergence and impact scores are plotted for each virus (Figure 2). Of the viruses scored thus far by IRAT, avian influenza A(H7N9) A/Hong Kong/125/2017 ranked highest for potential risk. Other viruses, such as A/Indiana/08/2011, an influenza A(H3N2) variant (H3N2v), had a similar score for risk for emergence similar to that of A/Hong Kong/125/2017 but a much lower estimated risk for potential impact.

Figure 2.

Comparison of average emergence and impact scores for 14 animal-origin influenza viruses using the Influenza Risk Assessment Tool. Circle represents each virus: A, H7N9 A/Hong Kong/125/2017; B, H7N9 A/Shanghai/02/2013; C, H3N2 variant A/Indiana/08/2011; D, H9N2 G1 lineage A/Bangladesh/0994/2011; E, H5N1 clade 1 A/Vietnam/1203/2004; F, H5N6 A/Yunnan/14564/2015-like; G, H7N7 A/Netherlands/2019/2003; H, H10N8 A/Jiangxi-Donghu/346/2013; I, H5N8 A/gyrfalcon/Washington/41088/2014; J, H5N2 A/Northern pintail/Washington/40964/2014; K, H3N2 A/canine/Illinois/12191/2015; L, H5N1 A/American green-winged teal/Washington/1957050/2014; M, H7N8 A/turkey/Indiana/1573-2/2016; N, H1N1 A/duck/New York/1996. Additional information about virus scores and individual viruses is available at

Influenza A(H7N9) and the IRAT

On March 31, 2013, the China Health and Family Planning Commission notified the World Health Organization (WHO) of 3 cases of human infection with influenza A(H7N9).[14] Three viruses were isolated and analyzed at the China WHO Collaborating Center and the complete viral genome sequences deposited in a publicly accessible influenza database. After these reports, CDC used the IRAT to assist the US Department of Health and Human Services' Biomedical Advanced Research and Development Authority with the overall prepandemic risk assessment of these viruses.

Although laboratories had begun the animal transmission challenge work, study results were not available. Hence, the IRAT risk element of transmissibility in animal models lacked data. This element is ranked as the second most important and thus carries a high weight in the computation of a final score for the IRAT emergence question. Therefore, it was necessary to assign a score for this element without data. Point scores for the other 9 elements were gathered and used to populate the IRAT to generate a partial risk score.

Although information about the outcome of laboratory animal transmission studies was scarce, previous observations showed significant correlation between other IRAT risk elements (receptor-binding properties, genomic variation, and human infections) and this element. A moderate score for this element extrapolated from other elements would greatly improve the ability to compare this new virus with other viruses evaluated previously with the IRAT. Based on the evidence of increased α2,6 receptor binding, the presence of L226 in the HA receptor binding pocket and the ability to infect humans, this element was assigned a score of 7 in the moderate risk category (range 4–7). Uncertainty was captured by assigning the risk element transmissibility in animal models a score of 1, 7, or 10 (Table 1). Using these 3 possible scores for this risk element, the summary risk score for the emergence question would be 5.2, 6.4, or 7.0, respectively. Assigning these same scores to this risk element to calculate the impact score, the summary risk scores would be 7.1, 7.4, or 7.5, respectively (Table 2). The much greater range in emergence (1.7) than impact (0.3) score is understandable when the relative weight assigned to this risk element is considered in the 2 different scenarios.

Only minimal data were available for 2 other elements (global distribution in animals and infections in animals) in April 2013. For the purposes of the risk scoring, we gave the global distribution in animals element a score of 1 because this virus had been identified in only a few live-bird markets in China. Because of the lack of information, confidence scores were low for this element. SMEs gave infection in animals a higher risk (mean 6, moderate risk) and confidence scores probably because of other H7N9 viruses associated with avian species. Because these elements carry less weight in risk scoring for both questions, they did not heavily affect the final score. In general, the SMEs agreed about risk scores for these elements.

The uncertainty and the data gaps, particularly for the transmissibility in the animal models element, were presented to decision makers and discussed. Particular attention was given to explain the range of scoring generated about the emergence question. However, the SMEs agreed that the impact score was less influenced by the missing information and the risk score did not significantly affect the final summary score.

In May 2013, 1 month after the initial assessment, information became available to inform the risk element transmissibility in animal models. More information was available for all other elements as well. The viruses were rescored in May 2013. The resulting average summary risk score for the 2 similar influenza A(H7N9) viruses (A/Anhui/1/2013 and A/Shanghai/2/2013) was 6.4 for the emergence risk and 7.2 for impact on public health if this virus gains the ability to transmit from person to person. SMEs reported a higher level of confidence in their risk scores at this time, although most element risk scores did not change appreciably. Since May 2013, these viruses have been scored annually in 2014, 2015, 2016, and again in early 2017, with little to no change in scoring, but with higher levels of confidence in individual scores each year.

Adaptation of the IRAT to Assess Influenza A(H5N1) Viruses

In 2014, the IRAT was used to compare several influenza A(H5N1) viruses that circulated during 2011–2014. Use of the IRAT is predicated on the assumption that each risk element will independently assess some aspect inherent in or associated with the various viruses included in the assessment. Based on available information, 5 of the risk elements would have had virtually the same score for all the H5N1 viruses. Although these 5 risk elements are useful for discriminating among other viruses, when comparing H5N1 viruses, sufficient information is lacking to enable the distinctions among the viruses necessary for the IRAT. These 5 elements (disease severity, population immunity, antiviral treatment susceptibility, receptor-binding properties, and transmissibility in animal models) were therefore removed from the IRAT scoring.

To use the IRAT to compare these viruses with each other, we tailored specific questions for this effort. Two questions were generated that related specifically to prepandemic mitigation of the impact these viruses could have on public health. The risk assessment focused on 2 questions about the risk element of antigenic relationship and availability of vaccines:

  1. Considering the vaccine antigens that are in the US Strategic National Stockpile or are currently being generated for this purpose, what H5N1 viruses pose the greatest potential risk to public health?

  2. Considering what CVVs are available or in development, what H5N1 viruses pose the greatest potential risk to public health?

To answer these 2 questions, SMEs scored the element of antigenic relatedness twice, in relationship to 1) a currently available WHO CVV or 2) antigens already prepared and stockpiled. Stockpiled antigen would be more quickly available for use than antigen in early development as a CVV.

Ten H5N1 clades were considered to be circulating during 2011–2014. SMEs scored these 10 using the IRAT based on information available for 5 elements that could be used to distinguish between these related viruses. SMEs were asked to provide a risk score for 1) human infections, 2) antigenic relationship of the viruses, 3) global distribution in animals, 4) infections in animals, and 5) genomic variation. The elements are listed in order of importance (i.e., the most heavily weighted element is human infections, and the next most important is the antigenic relatedness).

Initial SME scores for the 5 elements were averaged and presented to the same SMEs, and consensus on the final scores was reached through discussion. These risk scores were then multiplied by the appropriate weighting factor to generate summary risk scores. Because this risk assessment comprises only 5 elements, weights were apportioned on the basis of 5 elements rather than on the standard 10 elements. The IRAT definitions for the elements remained the same.

When scoring the 10 H5N1 viruses for antigenic relatedness, the SMEs based their first risk score on knowledge of currently available WHO CVVs and applied the IRAT definition of antigenic relatedness. The same SMEs then provided a second risk score for antigenic relatedness to US stockpiled antigens when considering the same 10 clades. In some instances, the average risk scores for this element differed. In each case, the average risk score for this element was multiplied by 0.2567, providing 2 possible summary risk scores for each virus (Table 3, Table 4).

When SMEs considered risk scores associated with antigenic relatedness to CVVs, clade 1.1.2 was the only virus clade that scored >5.0. When considering summary risk scores when the antigenic relatedness element was based on already stockpiled antigens, SMEs scored 4 virus clades >5.0: clade 1.1.2, clade, clade, and clade (Table 5).

Ultimately, the US Department of Health and Human Services' Biomedical Advanced Research and Development Authority decided to base the antigen to add to the US stockpile on influenza A(H5N1) clade This clade did not score the highest through the IRAT, but additional information, such as production deadlines, availability of the CVVs, and contractual obligations, also was considered before a final decision was reached, reinforcing that the IRAT is just 1 input for decision makers.