Influenza Severity Assessment PISA project
IHR Review Committee WHO should develop and apply measures that can be used to assess severity of every influenza pandemic. http://apps.who.int/gb/ebwha/pdf_files/wha64/a64_10-en.pdf. By applying, evaluating and refining tools to measure severity every year, WHO and MS will be better prepared to assess severity in next pandemic An early assessment followed by ongoing re-assessment as the pandemic evolves and new information becomes available, bearing in mind that severity will likely vary by place and over time; Quantitative values to define descriptive terms (e.g. mild, moderate and severe) to facilitate consistency; Use of a basket of indicators (e.g. hospitalization rates, mortality data, identification of vulnerable populations and an assessment of the impact on health systems) derived from a pre-agreed minimum data set; The expression of confidence and uncertainty around any estimates. 2
Severity indicators Transmission Seriousness of disease Impact (on society and health care systems) 3
Sources of information Routine influenza surveillance Special investigations Modelling 4
Steps in the PISA approach Step 1 Choose the parameters that will be used to assess severity indicators Step 2 Set the thresholds for each parameter Step 3 On a weekly basis, apply the thresholds to assess severity Step 4 Report the severity assessment findings and adapt control measures if needed 5
Which ones do you trust most Availability of subset of samples confirmed for influenza Timeliness Representative, stable over time Historical data, number of years of data Age-stratified data Denominator information available Co-morbidity data available 6
Proposed parameters TRANSMISSION: Weekly ILI or MAARI (medically attended acute respiratory illness) cases as a proportion of total visits, or incidence rates. Combination of weekly ILI or MAARI *weekly percentage positivity rates for influenza. 7
Proposed Parameters SERIOUSNESS of DISEASE : Cumulative death: hospitalization ratio (ideally for confirmed influenza) Cumulative ICU: hospitalization ratio (ideally for confirmed influenza) SARI/ARI or ILI ratio % % % 8
Proposed Parameters IMPACT: on the healthcare system weekly or monthly number or proportion of SARI cases with percentage flu-positive among SARI cases weekly excess Pneumonia &Influenza (P&I) or all-cause mortality (stratified by age). weekly number of confirmed influenza cases admitted to ICU weekly number of confirmed influenza cases admitted to hospital. 9
Proposed Parameters Other possible parameters reflecting more the impact on the society are: School closures Hospital beds occupied Work absenteeism School absenteeism 10
Absolute values are not comparable between countries When put into context with historical data, it is possible to assign them a category and compare parameters within and between countries 11
12 TITLE from VIEW and SLIDE MASTER 19 June 2017
Transmission Seriousness of disease Impact No activity or below seasonal threshold Low moderate high Extra-ordinary Level of confidence Low Medium High No activity or below seasonal threshold Low Moderate high Extra-ordinary Level of confidence Low Medium high No activity or below seasonal threshold Low moderate high Extra-ordinary Level of confidence Low Medium high 13 13
TRANSMISSION 14
Northern hemisphere 2016/2017 Transmission Generally low to moderate Comments: Higher transmissibility in elderly Seriousness Half of countries reported moderate to high; mainly in European countries Comments: High especially in elderly Impact Moderate to high in all countries reporting Comments: Elderly possibly at risk for high impact Spain: high impact but low seriousness 15
Findings Parameters and threshold setting for transmission % samples positive for influenza not well related to the other transmission parameters. Most countries had parameters including virological data or composite parameters. Not always enough data to disaggregate between age groups Might be influenced by health seeking behaviour during pandemic therefore local interpretation is important. Tropical countries: more work to be done on threshold setting. 16
Findings Seriousness of disease parameter Best to use cumulative ratio s and not weekly values. Report a judgment at the time the parameter is stabilizing; normally from peak of influenza activity onwards. To set cut-offs: check historical values. Most countries defined cut-offs derived from data from previous years and literature review. When data allows define thresholds by age-groups (<15 y of age, 15-64, 65+, or categories fitting these 3 groups) and all agegroups combined. Admission criteria might change during pandemic. 17
Findings: impact Impact parameters such as hospital admissions, ICU admissions (confirmed influenza), excess mortality worked well. Threshold setting for SARI, ICU admissions, hospitalizations worked ok Lack of standardized thresholds for the excess all-cause mortality 18
Complementing PISA special studies and modelling The special study group was revived Inventory will be completed Prioritization for the different stages of a pandemic Need to pool capacities and access to human resources 19
Acknowledgment The WHO technical working group for PISA Technical experts involved in TIPRA pilot testing and development Global influenza Programme colleagues Kaat Vandemaele Aspen Hammond Gina Samaan Bikram Maharjan Wenqing Zhang Regional office colleagues 20
Thanks Gracias 21
Pilot testing Countries participating: Australia, Bangladesh, Canada, Chile, China, Germany, France, Egypt, India, Ireland, Iran, Japan, Madagascar, New Zealand, Norway, Portugal, Spain, Singapore, South Africa, Thailand, UK (England and Scotland), US. Additional countries providing data to test the MEM method: Belgium, Georgia, Romania, Russian Federation, Ukraine, Hong Kong. Tomas Vegas and colleagues from the Foundation Institute of Health Sciences Studies of Castilla y León (IECSCyL) in Spain : explore MEM method for threshold setting for different parameters to allow categorisation of indicators Steve Rilley: Data needs for modelling 22
TRANSMISSION 23
Modelling data needs 24