Pan American Health Organization · 2020
Guidance for national health authorities on improving COVID-19 mortality surveillance using all-cause mortality analysis, May 2020.
- Issuer
- Pan American Health Organization
- Edition
- 2020
- Audience
- National health authorities, epidemiological surveillance teams, public health emergency responders, civil registration officials, national statistics institutes
- Evidence
- Expert consensus, drawing on international surveillance methodology and WHO ICD-10 classification standards
- Reviewed
- 19 Oct 2026 · GMJ
What this guideline says
GMJ summary
This guidance document, issued by the Pan American Health Organization in May 2020, addresses the challenges of accurately measuring COVID-19 mortality in Latin America and the Caribbean. It proposes a multi-system surveillance approach, with particular emphasis on all-cause excess mortality surveillance as a comprehensive measure of the pandemic’s true impact, especially in settings with limited testing capacity.
The document outlines four complementary COVID-19 mortality surveillance sub-systems: COVID-19-specific mortality surveillance (confirmed and probable cases), SARI sentinel mortality surveillance (building on existing influenza surveillance infrastructure), COVID-19 mortality from national vital registration systems (using ICD-10 codes U07.1 and U07.2), and all-cause excess mortality surveillance. The latter compares weekly observed deaths against expected baselines derived from 3–5 years of historical data, allowing detection of excess mortality that may not be captured through testing alone.
The core methodology involves three steps: establishing a working group to assess existing mortality information systems; improving or developing systems for weekly collection of all deaths (from vital registration, civil registration, sentinel facilities, or community-based reporting); and conducting analysis to calculate excess deaths by comparing observed deaths to expected thresholds. The guidance emphasizes data quality review (coverage, completeness, timeliness, accuracy), appropriate disaggregation by age, sex, geography and ethnicity, and timely weekly reporting to inform public health response.
Key points
For clinicians — doses and thresholds as stated in the guideline
- Establish all-cause mortality surveillance using historical data from the previous 3–5 years to calculate expected weekly deaths as a baseline for comparison
- Calculate excess mortality as observed deaths minus expected deaths; use the 75th percentile of historical data or upper limit of 95% confidence interval as alert threshold
- Report deaths weekly (by epidemiological week) with minimal disaggregation by age group, sex, district of residence, place of death, and ethnicity where feasible
- Assess data quality indicators: coverage should reach ≥90% (all geographical units reporting), completeness ≥90% (all data points present), and establish average reporting delay in days or weeks
- Use ICD-10 emergency codes U07.1 (COVID-19, virus identified by laboratory) and U07.2 (COVID-19, virus not identified, clinical/epidemiological diagnosis) for cause-of-death coding
- For countries with incomplete vital registration, adjust mortality data by dividing historical deaths by UN World Population Prospects estimates to derive percent completeness as annual adjustment factor
- Implement sentinel facility-based surveillance at SARI sentinel sites, collecting daily line lists of all deaths with four core variables (sex, age, district of residence, ethnicity) submitted weekly
- Account for reporting delays when recent weeks show apparent declines; correct observed deaths based on average proportion registered over defined periods when administration is open
Summary indicators of interest for all-cause mortality surveillance
Table extracted from the guideline
| Indicator | Calculation |
|---|---|
| Excess mortality | weekly observed deaths – weekly expected deaths |
| Range for excess mortality (upper end) | weekly observed deaths – p25 weekly expected deaths |
| Range for excess mortality (lower end) | weekly observed deaths – p75 weekly expected deaths |
| Percent excess | (weekly observed deaths – weekly expected deaths) / weekly expected deaths |
| Deaths over the threshold | weekly observed deaths – upper threshold value |
| Percent over the threshold | (weekly observed deaths – upper threshold value) / upper threshold value |
Source: Pan American Health Organization, Enhancing COVID-19 Mortality Surveillance in Latin America and the Caribbean through All-Cause Mortality Surveillance. Reproduced in summary form.
From the guideline
Verbatim
“All-cause excess mortality surveillance can be a reliable and comprehensive measure of the overall impact of COVID-19 mortality.”
Editor’s perspective
Why this matters for Georgia and the region
Prof. Giorgi Pkhakadze, MD, MPH, PhD
Editor-in-Chief, Georgian Medical Journal · Chair, Public Health Institute of Georgia · WHO consultant, health and migration
As someone who has worked extensively on strengthening disease surveillance systems in Georgia and the region, I recognize that this PAHO guidance offers a practical framework directly applicable to our South Caucasus context. While Georgia has made significant investments in electronic health records and vital registration since 2013, the completeness and timeliness of mortality data remain heterogeneous, particularly in rural Adjara, Samtskhe-Javakheti, and mountainous regions where civil registration may lag by weeks. The recommendation I would most emphasize to Georgian epidemiologists is the systematic weekly compilation and analysis of all-cause deaths with minimal disaggregation—this is achievable with existing Public Service Development Agency data and NCDC surveillance capacity, and would provide early warning of outbreak spread to underserved areas where testing coverage is limited. The key implementation challenge for Georgia will be reconciling multiple data streams: hospital-based reporting through the Infectious Disease Surveillance System, civil registration data, and community deaths in areas where the deceased never reached a health facility. We must also ensure that excess mortality calculations account for baseline variations in winter respiratory mortality and the ongoing tuberculosis burden, which differs substantially between urban Tbilisi and rural regions.
ORCID 0000-0001-7609-4515 · Perspective written for the GMJ Guidelines Library, 19 Oct 2026
Applicability
Georgia and the South Caucasus
For Georgia and the South Caucasus, this guidance remains highly relevant for strengthening COVID-19 and future pandemic surveillance systems. Georgia has invested in vital registration and electronic health information systems, but completeness of mortality reporting varies between urban and rural areas, and among hard-to-reach populations including internally displaced persons. National programmes should assess the coverage and timeliness of civil registration data from the Public Service Development Agency, evaluate whether SARI sentinel surveillance sites (established for influenza monitoring) capture representative mortality data, and consider establishing community-based mortality reporting in mountainous regions and areas with ethnic minority populations where death registration may be delayed or incomplete. The methodology for calculating excess mortality using 3–5 years of historical data is directly applicable, though adjustments may be needed for 2008 conflict-related displacement and subsequent population movements. Integration with Georgia’s existing infectious disease surveillance infrastructure and the National Center for Disease Control would facilitate implementation.
Questions this guideline answers
Quick answers
Why use all-cause mortality surveillance instead of relying only on confirmed COVID-19 deaths?
All-cause mortality surveillance captures deaths that may be missed by testing-dependent systems, including deaths in communities with limited access to testing, deaths where COVID-19 was not suspected, and indirect deaths due to disrupted health services or delayed care-seeking during the pandemic. It provides a more comprehensive measure of the pandemic's true burden, especially in settings with diagnostic uncertainty or low testing coverage.
How do I calculate the expected number of deaths if I don't have 3–5 years of historical vital registration data?
You can use the median of whatever historical data are available, apply age- and sex-specific death rates from UN World Population Prospects 2019, WHO, or Global Burden of Disease estimates to your reference population, or use the first set of observations from a newly established all-cause surveillance system as the baseline for tracking trends forward. If previous years had events affecting mortality (natural disasters, outbreaks, famine), consider excluding those years.
What is the minimum data I need to collect for each death in facility-based or community-based surveillance?
The minimum essential variables are: sex, age at death, district or township of usual residence, and ethnicity (where feasible). Additional useful variables include place of death (facility vs. community), date of death, and epidemiological week. Deaths should be aggregated and reported weekly to enable timely trend analysis.
How do I set the alert threshold for excess deaths?
Use either the 75th percentile of historical data (when using medians for expected death point values) or the upper limit of the 95% confidence interval (when using regression models or averages for weekly expected values). When observed deaths exceed this threshold, it signals a high level of excess mortality requiring investigation and response. Assess the sensitivity and specificity of different threshold methods using your historical data.
Related on GMJ
Interconnected resources
Conditions A–Z
GMJ News coverage
- UK flu and COVID-19 surveillance data to May 2027 shows continued monitoring amid seasonal trendsOct 2026
- Community Health Workers Boost COVID-19 Testing Among Formerly Incarcerated PeopleAug 2026
- Cambridge researchers identify three distinct severe pneumonia subtypes, opening path to personalised treatmentSep 2026
- Machine Learning Model Predicts Child Pneumonia Hospitalisation Risk in Malawi Primary CareJun 2026
Quality & accreditation
Other guidelines on this topic
- COVID-19: Chloroquine and hydroxychloroquine research
- Regulatory considerations on authorization of the use of convalescent plasma (CP) to address the COVID-19 emergency
- Interpretation of laboratory results for COVID-19 diagnosis
- Dead Body Management in the Context of COVID-19
- Ethics guidance on issues raised by the novel coronavirus disease (COVID-19) pandemic
- Issuer
- Pan American Health Organization
- Type
- Procedures, manuals, guidelines
- Status
- Current
- Published
- 2020-01-01
- Languages
- English
- Scope
- COVID-19 mortality surveillance in Latin America and the Caribbean using all-cause mortality analysis
- Audience
- National health authorities, epidemiological surveillance teams, public health emergency responders, civil registration officials, national statistics institutes
- Licence
- CC BY-NC-SA 3.0 IGO
- Evidence
- Expert consensus, drawing on international surveillance methodology and WHO ICD-10 classification standards
- Record history
- Created 19 Oct 2026 · updated 19 Oct 2026 · next check Apr 2027
Pan American Health Organization. Enhancing COVID-19 Mortality Surveillance in Latin America and the Caribbean through All-Cause Mortality Surveillance. Washington, D.C.: PAHO; 2020. (PAHO/IMS/PHE/COVID-19/20-0035). Available from: https://iris.paho.org/handle/10665.2/52308GMJ Guidelines Library records are original summaries written by the GMJ editorial team from the issuer’s published document and reviewed by a physician. They do not replace the full guideline. The document is the copyright of its issuer and is linked, not hosted. Spotted an error or a newer edition? Tell the Guidelines Desk.
Was this article helpful?
