WHO Global Situational Alert System: a mixed methods multistage approach to identify country-level COVID-19 alerts

Abstract

Background: Globally, since 1 January 2020 and as of 24 January 2023, there have been over 664 million cases of COVID-19 and over 6.7 million deaths reported to WHO. WHO developed an evidence-based alert system, assessing public health risk on a weekly basis in 237 countries, territories and areas from May 2021 to June 2022. This aimed to facilitate the early identification of situations where healthcare capacity may become overstretched.

Methods: The process involved a three-stage mixed methods approach. In the first stage, future deaths were predicted from the time series of reported cases and deaths to produce an initial alert level. In the second stage, this alert level was adjusted by incorporating a range of contextual indicators and accounting for the quality of information available using a Bayes classifier. In the third stage, countries with an alert level of 'High' or above were added to an operational watchlist and assistance was deployed as needed.

Results: Since June 2021, the system has supported the release of more than US$27 million from WHO emergency funding, over 450 000 rapid antigen diagnostic testing kits and over 6000 oxygen concentrators. Retrospective evaluation indicated that the first two stages were needed to maximise sensitivity, where 44% (IQR 29%-67%) of weekly watchlist alerts would not have been identified using only reported cases and deaths. The alerts were timely and valid in most cases; however, this could only be assessed on a non-representative sample of countries with hospitalisation data available.

Conclusions: The system provided a standardised approach to monitor the pandemic at the country level by incorporating all available data on epidemiological analytics and contextual assessments. While this system was developed for COVID-19, a similar system could be used for future outbreaks and emergencies, with necessary adjustments to parameters and indicators.

Keywords: COVID-19; Epidemiology; Public Health.

Figure 1

Multistage process for the weekly Global Situational Alert System (GSAS). PHSM, public health and social measure.

Figure 2

Summary of algorithm components responsible for alerts and the proportion of weekly final alerts identified using contextual information, by region and variant of concern (VOC). (A) Number of alerts missed using only reported cases and deaths (B). The Delta period incorporates assessments from July to December 2021 and the Omicron period incorporates assessments from January to May 2022. (C, D) The comparison is only for the weekly alerts between 3 January 2022 and 18 April 2022, when the process incorporated the standardised context assessment. (C) Note that a ‘Critical’ alert level could not be raised, by definition, at stage 1.

Figure 3

(A) Reported deaths per million population across alert levels and WHO regions. Note that reported deaths are shown on a log scale. (B) Timeliness of identifying countries to include on the watchlist before the reported peak in cases, shown by WHO region. (C) Timeliness of identifying countries to include on the watchlist before the reported peak in deaths, shown by WHO region.

Figure 4

Trends in numbers of cases in hospital per million population versus the situational alert level for a subset of countries in the European Region, the Region of the Americas and the Western Pacific Region for which this information is available. External data source: Our World in Data. Note that the hospitalisation data (availability and quality) vary among WHO member states based on a variety of factors and the data above are not representative of all member states.

Figure 5

Summary of trends in final alert levels, cases and deaths, hospitalisation and intensive care unit (ICU) numbers and vaccination coverage for Romania during the ‘Delta wave’ between August and December 2021.