Health threats such as epidemics (influenza, RS virus, COVID-19), heat waves and disasters such as earthquakes and major accidents can affect the mortality of a population. In order to respond to serious health threats, it is important that increases in the mortality are detected as early as possible. This is especially important for serious events where the mortality and disease burden are not known in advance from other data. Mortality is an objective and precise measure, and mortality surveillance has been used worldwide for decades and is therefore an important tool for monitoring the health status of a population. EuroMOMO monitors the mortality for the entire population as well as for the following age groups: 0-14 years, 15-44 years, 45-64 years, 65-74 years, 75-84 years and 85+ years. The monitoring of the mortality in the entire population (all ages) is used to quickly observe fluctuations in mortality across the population regardless of age, i.e., it provides an overall measure of the level of mortality in society. The monitoring of mortality in the age groups is used to assess whether there are age groups that are at particularly high risk and what impact a known threat or event may have on the mortality of a particular age group.
To respond to serious health threats, it is important to detect and quantify increases in mortality as quickly as possible. Therefore, it is important that the mortality surveillance is timely. There may be a slight delay in the registration of deaths, which is why this delay is corrected for.
All deaths are included in this surveillance, and the causes of death are not considered. This is necessary, among other things, to obtain a mortality surveillance that is as timely as possible, since registration of causes of death is often significantly more delayed than the registration of the actual death. Causes of death are also to some extent open to doctors’ interpretations, whereas changes in overall mortality are an objective measure of the impact of a serious health threat or event.
The mortality is estimated and uploaded on the website every Thursday, with data available up to last week.
European mortality surveillance can provide valuable insights in the overall health of populations in Europe and early detection of outbreaks and epidemics. Furthermore, it can contribute as a tool for public health interventions as well as for research and epidemiological studies.
Information on the date of death, when the death was registered, and the age of the deceased is obtained nationally each week.
The registered number of deaths is the number of deaths recorded in the national register at a given time, i.e., when the data is collected for the weekly mortality analyses.
The expected mortality is estimated as the number of deaths that can be expected when there are no serious health threats or epidemics. The mortality is typically higher in the winter due to respiratory pathogens and severe cold, and lower during the summer. Periods, when increases in the mortality can be expected, include times with higher risk of epidemics of certain infectious diseases such as influenza, RS virus, COVID-19 or extreme temperatures. Since several of these diseases are common, the expected mortality should be understood as the best level of mortality we can expect to achieve. This means, for example, that in wintertime, the mortality is rarely at the expected level since there are often circulating infections that can affect the mortality.
The purpose of the monitoring itself is to detect increases in mortality without unnecessary delays. Therefore, the expected number of deaths must be based on periods with favourable conditions. The expected number of deaths is estimated, using a statistical model, based on the number of deaths registered in the previous five years during periods of the year when elevated mortality is rarely observed. Typically, this involves week 16 through 25 in the spring and week 37 through 44 in the fall. In all cases, expected mortality will include seasonal variations. However, at the moment, when data from the three pandemic years (2020-2022) are excluded from the estimations, the oldest data currently used is up to 8 years old. See note on “The European mortality is at an expected level after 3 years of the COVID-19 pandemic and data from spring 2023 and onwards will from now on be included in the estimations of the expected mortality” in the bulletin.
It should be noted that the expected mortality is estimated separately for the total population and for each age group. This means that the summed expected deaths in the age groups may not necessarily be identical to the expected number of deaths calculated for the entire population (all ages).
There is a certain delay in the registration of deaths. Therefore, the registered deaths are adjusted (projected) for this delay based on the previous registration pattern.
The mortality is calculated both overall for the entire population (all ages) and separately for each age group (0-14 years, 15-44 years, 45-64 years, 65-74 years, 75-84 years, and 85+ years). The expected mortality is estimated for both the entire population and each age group separately and therefore cannot be directly summed. Monitoring mortality in the entire population (all ages) is used to capture fluctuations in mortality across the population independent of age, while the age-specific monitoring provides a picture of mortality in each age group.
There is greater statistical uncertainty associated with estimating the expected mortality in individual age groups because the number of deaths per group is smaller compared to the total population. This means that larger fluctuations in mortality are required in the age groups before the difference becomes large enough to exceed the threshold for elevated mortality. Therefore, it can happen that the model detects excess mortality in the entire population (all-ages), but not in any specific age group. Conversely, excess mortality in one age group may be “masked” in the mortality of the total population.
When the mortality has been statistically unusually high compared to optimal conditions (interpreted as two or more standard deviations higher than the expected level) for three consecutive weeks, the mortality is increased. The three-week rule is used to ensure that the elevated mortality is real and not just a random variation in the data. If the mortality has been four or more standard deviations higher than the expected level for three consecutive weeks, it is referred to as substantially increased. The magnitude of the elevated mortality compared to the expected mortality is called excess mortality.
If the mortality is increased for three consecutive weeks, excess mortality can be calculated by subtracting the expected number of deaths from the corrected registered number of deaths. If the corrected registered number of deaths is lower than the expected number of deaths, it is called negative excess mortality.
Periods of increased mortality are usually observed in the winter season, as the estimations of the expected mortality are based on periods where there have been no circulating respiratory infections or, for example, extreme cold. The expected mortality is therefore the expected mortality under optimal conditions. This is done to be able to catch increases in mortality quickly, but it will cause the effect that there will be observed increased mortality in many winter periods.
The mortality surveillance system is established to detect changes quickly and in a timely manner in the populations’ health status. To be timely, the surveillance system uses all deaths regardless of cause. Since the surveillance system is based on a statistical model, it cannot reveal other characteristics of those who have died. Therefore, the surveillance system cannot say anything about the cause of a possible increase in mortality, only that there is an increase.
All estimates of expected mortality are re-estimated every week based on updated data, i.e., historical data is not fixed and may change over time. Therefore, the expected number of deaths for a particular week may vary from week to week. Furthermore, the expected number of deaths may also differ due to different countries being included in the weekly analyses.
Sometimes it can occur that a country is unable to provide data for a particular week, e.g., due to technical problems etc. Only countries that contribute with data the week in question are included in the pooled analyses and have their national data displayed on the website. If for some reason a country is unable to provide data for a particular week, the country will not appear on the website, but will be displayed when they provide data again. What is the shaded area referred to as “normal range” in the graphs showing the pooled weekly number of deaths and the graphs showing the weekly z-score at the national level? The shaded area in the mortality graphs show the range of the expected level of mortality, i.e., the area in which mortality is expected to lie.
The data on national levels are shown as Z-scores and can be downloaded. For requests for additional national data please contact the country of interest directly (contact list).
Each country makes its own national analyses of its mortality and sends the results to the EuroMOMO hub. To obtain the total (pooled) number of observed and expected deaths at European level, the hub summarizes the forwarded numbers from all the participating countries as well as the associated uncertainties, by week and age group.
Generally, a Z-score is a statistical measure that describes how far away from the mean a given data point is. In this context, if the observed number of deaths is the same as the expected number of deaths, the z-score is 0. The greater difference the observed number of deaths is from the expected number (either above or below), the higher the z-score will be (with either positive or negative sign respectively).
When the expected number of deaths (baseline) is estimated, excess mortality is calculated as the difference between the observed number of deaths and the baseline.