Publicatie
Publication date
Results of a cross-validation exercise of the Global Influenza B Study (GIBS) and WHO FluNet databases
Caini, S., Paget, J. Results of a cross-validation exercise of the Global Influenza B Study (GIBS) and WHO FluNet databases Utrecht: Nivel, 2018. 30 p.
Download the PDF
Background: The WHO FluNet database is a publically available database which is being used by researchers around the world to study the epidemiology of influenza. The Global Influenza B Study (GIBS) database was established in 2013 and considering it covers 30 countries and was developed independently from the FluNet database, we aimed to compare the two database in an effort to validate the WHO FluNet database.
Methods: The GIBS database was initially assembled in 2013-14, and contains epidemiological and virological influenza surveillance data from thirty countries around the world. In June 2018, we downloaded the influenza surveillance data available in the WHO-FluNet database for the same countries and years. We performed the following comparative analysis of public health-relevant characteristics of influenza epidemics: 1) the proportion of influenza B over all influenza cases in each country and season; 2) the proportion of B/Victoria and B/Yamagata lineages over all influenza B cases in each season and country; 3) the “typical” timing and amplitude of influenza epidemics (overall and separately for influenza A and B epidemics) in each country using the EPIPOI software
Results: Our analysis was based on 21 countries where there was comparable data from FluNet and GIBS for the post-pandemic period (i.e. from 2010 onwards). For 9 countries, we found the same or very similar data in both datasets. For 12 countries, the FluNet data was very different and, although we did not have enough information to clarify the exact reasons for the differences (e.g. what type of data was reported in the FluNet database (primary care, secondary care, or both)?), the data contained in the two databases were mostly not independent from each other.
Conclusions: After inspecting the two databases in detail, we concluded that it was not possible to validate the FluNet database using the GIBS dataset. The main reasons for this are that they presented different datasets for the same country but the independence of the two databases was 3 not likely and other factors (e.g. we only had very specific regional data for a country). In order to better interpret and validate the FluNet database in the future, we recommend that WHO carries out routine surveys of the national data sources in each country (e.g. define which populations are included in the database, define the national representativeness of the data) and makes this information available in the public domain.
Methods: The GIBS database was initially assembled in 2013-14, and contains epidemiological and virological influenza surveillance data from thirty countries around the world. In June 2018, we downloaded the influenza surveillance data available in the WHO-FluNet database for the same countries and years. We performed the following comparative analysis of public health-relevant characteristics of influenza epidemics: 1) the proportion of influenza B over all influenza cases in each country and season; 2) the proportion of B/Victoria and B/Yamagata lineages over all influenza B cases in each season and country; 3) the “typical” timing and amplitude of influenza epidemics (overall and separately for influenza A and B epidemics) in each country using the EPIPOI software
Results: Our analysis was based on 21 countries where there was comparable data from FluNet and GIBS for the post-pandemic period (i.e. from 2010 onwards). For 9 countries, we found the same or very similar data in both datasets. For 12 countries, the FluNet data was very different and, although we did not have enough information to clarify the exact reasons for the differences (e.g. what type of data was reported in the FluNet database (primary care, secondary care, or both)?), the data contained in the two databases were mostly not independent from each other.
Conclusions: After inspecting the two databases in detail, we concluded that it was not possible to validate the FluNet database using the GIBS dataset. The main reasons for this are that they presented different datasets for the same country but the independence of the two databases was 3 not likely and other factors (e.g. we only had very specific regional data for a country). In order to better interpret and validate the FluNet database in the future, we recommend that WHO carries out routine surveys of the national data sources in each country (e.g. define which populations are included in the database, define the national representativeness of the data) and makes this information available in the public domain.
Gegevensverzameling