Glossary

A

Agency

On this site, a source type: a report or release from a public health body such as the UK Health Security Agency, the European Centre for Disease Prevention and Control or the US Centers for Disease Control and Prevention. Agency reports are timely and authoritative on surveillance; they are not peer-reviewed.

Age–period–cohort

The three kinds of explanation for a change in a population: something to do with how old people are (age), something that happened in a calendar year to everyone (period), or something shared by people born in the same years (cohort). They cannot be fully separated by statistics alone. See the primer on the Lexis diagram.

Age-standardised rate

A rate — of deaths, cases, anything counted per head — recalculated as if the population had a fixed age structure. It removes the effect of one population being older than another, or than itself ten years earlier, so that a difference in the rate means a difference in risk. Most comparisons on this site are age-standardised where the source allows; where they are not, the card says so.

Attributable fraction

The share of deaths or cases that would not have happened without a given exposure, if the association is causal. "Attributable" is a calculation from relative risk and prevalence, not an observation, and it depends on the counterfactual chosen.

B

Baseline

In excess-mortality work, the expected number of deaths against which the observed number is compared. The choice of baseline — an average of past years, a trend, with or without the pandemic years — is the main reason excess estimates differ. See the primer on excess mortality.

Birth cohort

Everyone born in the same year or years. Cohorts carry shared exposures through life, which is why a change that tracks year of birth rather than calendar year is read as a cohort effect.

C

Case fatality rate

Deaths divided by diagnosed cases of a disease. Sensitive to how many cases are found: more testing lowers the case fatality rate without anyone's risk changing. Distinct from the infection fatality rate, which divides by all infections including those never diagnosed.

Cohort life expectancy

The average length of life for people born in a given year, allowing for the death rates they will actually experience as they age — which for a cohort born today means projected rates decades ahead. Higher than period life expectancy where mortality is improving, and a projection rather than a measurement.

Confidence interval

The range within which the true value of an estimate is thought to lie, given the data. A 95 per cent interval is the conventional width. Two estimates whose intervals overlap substantially are not clearly different, however different their point estimates look.

Confirms

A verdict on an item: its finding is consistent with the existing weight of evidence on the driver it is filed under.

Contested

A driver status on this site: credible evidence points in different directions and the question is open. See also emerging, building, established, fading.

Contradicts

A verdict on an item: its finding is inconsistent with the existing weight of evidence. Contradicting items are shown in ember on cards and on timelines because they are the ones most worth reading.

D

DALY

Disability-adjusted life year. One year of healthy life lost, whether to early death or to time lived with illness weighted by its severity. The unit used by the Global Burden of Disease project to compare conditions on one scale.

Data watch

One of the site's three streams: a new or changed source of data — a dataset, registry, linkage, API or release format — and what it makes possible. Distinct from a finding.

Direction

The recent movement of a driver's evidence: strengthening, steady, weakening or mixed. A judgement of the evidence, not of the consequence.

Disability-free life expectancy

Years a person can expect to live without a long-standing condition that limits their activities, under current conditions. One of several healthy-life measures; see the primer.

DOI

Digital object identifier: the permanent address of a published paper, in the form 10.xxxx/yyyy. Every item's source is given with its DOI where one exists.

Driver

One of the drivers of future risk the site follows: a standing question, such as whether early-onset cancer is rising or better found. Each has a page with a status, a direction, a state-of-the-evidence paragraph and every item filed under it.

E

Emerging

A driver status: a question only recently studied at scale, where the evidence is thin and could go either way.

Established

A driver status: the main finding is settled and new evidence refines it rather than overturns it.

EuroMOMO

The European mortality monitoring network, which publishes weekly excess-mortality bulletins for participating countries on a common method. One of the site's methods-watch sources.

Eurostat

The statistical office of the European Union, publishing population, mortality and health data on definitions harmonised across member states.

Excess mortality

Deaths in a period above the number expected without a given event. See the primer.

F

Fading

A driver status: interest or evidence has waned and the question is no longer moving.

Fertility rate, total

The average number of children a woman would have if she experienced the current year's age-specific birth rates throughout her childbearing years. A period measure, like period life expectancy, and affected by changes in the timing of births as well as their number; see tempo effect.

G

Global Burden of Disease

A project led by the Institute for Health Metrics and Evaluation that estimates deaths, illness and risk factors for every country and cause on a consistent basis. Its figures are modelled, reassign ill-defined causes, and differ from national statistics for that reason.

H

Hazard ratio

The ratio of the rate at which an event happens in one group to the rate in another, over the period of a study. A hazard ratio of 1.3 means a 30 per cent higher rate; it is a relative measure and says nothing about how common the event is.

Healthy life expectancy

Years a person can expect to live in good health, under current conditions, by one of several definitions. See the primer.

Human Mortality Database

The reference source for comparable long-run death rates and life tables for around fifty countries, maintained by demographers at Berkeley and the Max Planck Institute. Its Short-Term Mortality Fluctuations series provides weekly deaths for excess-mortality comparison.

I

ICD

The International Classification of Diseases, maintained by the World Health Organization: the code list and rules by which every death is assigned an underlying cause. Most countries use the tenth revision; the eleventh is being adopted slowly. See the primer on coding.

Incidence

New cases of a condition in a period, usually per head of population. Distinct from prevalence, the number of existing cases at a point in time.

Item

One of the site's three streams: a finding from the research literature, an agency or a statistical office, in two or three sentences with a line on why it is interesting.

L

Lexis diagram

The demographer's chart of age against calendar year, on which a life is a diagonal line. Explained on this site in its primer.

Life expectancy at 65

The average further years of life for someone aged 65 under current death rates. The measure most relevant to pensions and later life, and less affected by infant and child mortality than life expectancy at birth.

Life table

The schedule of survival and death by age from which life expectancy is calculated. Period life tables use one year's death rates; cohort life tables follow a birth year.

M

Mortality improvement

The rate at which death rates fall from year to year, usually quoted as a percentage per year at a given age. Its slowing after about 2011 in several countries is one of the drivers the site follows.

Multiple causes of death

All the conditions mentioned on a death certificate, not just the underlying cause. Published by a growing number of countries and used to see what people die with as well as of.

N

Net migration

Immigrants minus emigrants over a period. The least predictable component of population change, and the assumption that moves projections most.

New

A verdict on an item: a question not previously studied at this scale or in this population.

O

Old-age dependency ratio

People above a chosen age, usually 65, divided by people of working age. A structural measure of ageing that says nothing about whether the older people are dependent.

P

Peer-reviewed

A source type: a paper published in a journal after review by other researchers. The review checks reasoning and methods; it rarely re-analyses data. See the primer on preprints.

Period life expectancy

The average length of life a hypothetical person would have if they experienced one year's death rates at every age. A summary of that year's mortality, not a forecast of anyone's life. See the primer on the Lexis diagram.

Polygenic risk score

A single number summarising a person's genetic predisposition to a condition, built from many small-effect variants. Discriminates modestly, performs worse in people of non-European ancestry, and is one of the drivers the site follows.

Population projection

A calculation of future population size and structure from assumed fertility, mortality and migration. Not a forecast in the statistical sense: the assumptions are choices, and the site's projection pages show how past assumptions fared.

Preprint

A research paper posted publicly before peer review. A source type on this site, with its own primer and a page tracking what became of each.

Prevalence

The number or share of people with a condition at a point in time. Rises when incidence rises, and also when survival improves.

R

Refines

A verdict on an item: consistent with the existing evidence, but changing the size of an effect, the population it applies to, or the mechanism.

Registry

A systematic record of every case of something in a defined population: cancers, births, deaths, prescriptions. The Nordic countries' linked national registries are the reason so much of the strongest evidence comes from there.

Relative risk

The rate of an outcome in an exposed group divided by the rate in an unexposed group. A relative risk of 2 doubles a risk that may be tiny or large; the absolute risk is needed to know which.

Replacement fertility

The total fertility rate at which a population replaces itself without migration, about 2.1 children per woman in low-mortality countries. Every country the site follows is below it.

Revision

One of the site's three streams: an official series changing its own past — a re-estimated baseline, a census rebasing, a coding change, late registrations — and what the change moves.

S

Standardised mortality ratio

Observed deaths in a group divided by the deaths expected if it had the death rates of a reference population, expressed as a ratio or percentage. A ratio above one means higher mortality than the reference after allowing for age.

Statistics

A source type on this site: a release from a national statistical office or a registry, such as the Office for National Statistics or Statistics Denmark.

Status

Where a driver's evidence stands: emerging, building, established, contested or fading. Set by judgement of the evidence and revised as it changes; earlier readings stay on the page, dated.

Sullivan method

The standard calculation for healthy life expectancy: splitting each year of the life table into healthy and unhealthy time using the share of people at each age who report themselves healthy.

T

Tempo effect

The distortion of a period fertility rate when women delay births. Postponement lowers the rate in the years it happens even if the eventual number of children is unchanged; whether recent declines are tempo or a real fall in family size is one of the drivers the site follows.

Test-negative design

The standard study design for vaccine effectiveness against medically attended illness: comparing vaccination among people who test positive and people who test negative for the same symptoms. See the primer.

U

Underlying cause of death

The disease or injury that began the chain of events leading to death, as chosen from the death certificate by the rules of the ICD. The basis of cause-of-death statistics. See the primer on coding.

V

Vaccine effectiveness

The percentage reduction in an outcome — infection, illness, hospitalisation or death — among vaccinated people compared with unvaccinated, in the real world. See the primer.

Verdict

The one-word judgement on each item of what it does to the existing evidence: new, confirms, refines or contradicts.

Vintage

The version of a data series as published at a particular date. Official series revise their past; a figure quoted with its vintage can be reproduced, and the site's trackers keep vintages for that reason.

Y

Years of life lost

A measure of premature mortality: for each death, the years between the age at death and a reference life expectancy, summed. Gives more weight to deaths at younger ages than a simple count.