# Helsen Institute for Public Research — full content > The Helsen Institute for Public Research is an independent, nonpartisan research institute that publishes clear, verifiable syntheses of public evidence on health systems, demographic change, public trust, and research methods. All content below is licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Attribute to “Helsen Institute for Public Research” with a link to the cited page. Canonical site: https://helsen-institute.vercel.app --- # Excess Mortality: What It Measures and Why It Matters - Type: Explainer - URL: https://helsen-institute.vercel.app/research/excess-mortality-explained - Published: 2026-08-12 - Citation: Helsen Institute for Public Research (2026). “Excess Mortality: What It Measures and Why It Matters.” Helsen Institute Explainer, published August 12, 2026. https://helsen-institute.vercel.app/research/excess-mortality-explained **Summary.** Excess mortality is the difference between the number of deaths observed in a period and the number that would have been expected based on past trends. Because it does not depend on how individual deaths are coded, it is one of the most robust ways to measure the total impact of a crisis — but the estimate always depends on how the expected baseline is calculated, and it says nothing by itself about causes. **Key takeaways:** - Excess mortality compares observed deaths with an expected baseline estimated from historical data; it captures both direct and indirect effects of a crisis. - It is robust to differences in death certification and cause-of-death coding, which makes it well suited to international comparison. - The World Health Organization estimated approximately 14.9 million excess deaths associated with the COVID-19 pandemic in 2020–2021 — well above the count of confirmed COVID-19 deaths in the same period. - Every excess-mortality figure depends on modelling choices for the baseline; reputable estimates publish those choices and an uncertainty range. - Excess mortality identifies that more people died than expected, not why — attributing causes requires additional evidence. ## What excess mortality measures > **Definition:** Excess mortality is the difference between the number of deaths from all causes observed during a period and the number of deaths that would have been expected in that period based on historical patterns. The measure asks a deliberately simple question: compared with a normal year, how many more (or fewer) people died? Because it counts deaths from all causes, it does not depend on death certificates being filled in consistently, on testing capacity, or on how borderline cases are classified. That simplicity is its main strength. Excess mortality is usually reported either as an absolute number of deaths, as a rate per 100,000 people, or as a P-score — the percentage by which observed deaths exceed expected deaths. A P-score of 20% means one-fifth more people died in the period than the baseline predicted. ## How the expected baseline is estimated The “expected” number of deaths is a statistical estimate, not an observation, and different producers estimate it differently. The two most common approaches are: - Historical averages. The average number of deaths in the same weeks of several preceding years — transparent and easy to communicate, but blind to population growth, ageing, and long-run mortality trends. - Statistical models. Regression models that project the pre-crisis trend forward, typically accounting for seasonality, population size and age structure. Monitoring networks such as EuroMOMO and academic estimates generally use this approach. The choice matters. In a country whose mortality was falling before a crisis, a simple five-year average will understate expected improvement and therefore understate the excess. In an ageing population, the same average may point the other way. Serious estimates state their baseline method and publish uncertainty intervals; figures quoted without either should be treated with caution. ## Why researchers rely on it Cause-specific death counts depend on local certification practice: what one country codes as a COVID-19 death, another may code as pneumonia or an unspecified cause. Excess mortality sidesteps this problem entirely, which is why it became the standard yardstick for comparing the pandemic’s toll across countries. It also captures indirect effects that cause-specific counts miss: deaths from postponed treatment, overwhelmed health systems, or economic disruption, as well as reductions in deaths — for example from traffic accidents during lockdowns or mild influenza seasons. ## What excess mortality has shown Two episodes illustrate the measure’s value. During the COVID-19 pandemic, the World Health Organization estimated approximately 14.9 million excess deaths associated with the pandemic in 2020–2021 — substantially more than the roughly 5.4 million confirmed COVID-19 deaths reported for the same period, indicating significant undercounting in official cause-specific statistics. Earlier, the European heatwave of August 2003 produced an estimated 70,000 excess deaths across Europe according to the most widely cited assessment. Many of those deaths were never attributed to heat on death certificates; only the all-cause comparison revealed the event’s true scale, and it reshaped European heat-preparedness policy. ## How to read an excess-mortality figure 1. Check the baseline. Which years or model produced the “expected” number? Does it account for population change and pre-existing trends? 2. Prefer rates and P-scores for comparison. Absolute counts reflect population size; a per-100,000 rate or P-score is comparable across countries. 3. Look for the uncertainty range. Excess mortality is an estimate; reputable producers publish intervals, not single numbers. 4. Mind registration delays. Recent weeks are always incomplete in death-registration data; provisional figures are routinely revised upward. 5. Do not read causes into it. The measure shows that mortality departed from the baseline, not why — attribution requires separate analysis. ## Limitations > **Scope and limitations:** Excess mortality estimates are sensitive to baseline choice, unreliable for small populations where random variation is large, and delayed by death-registration lags. They may also reflect mortality displacement — deaths brought forward by weeks or months among the very frail — which can be followed by below-baseline periods. None of these caveats undermines the measure; they define how precisely it can be read. ## Sources and further reading - World Health Organization — global excess-mortality estimates associated with the COVID-19 pandemic, 2020–2021, with published methodology and uncertainty ranges. - EuroMOMO ([euromomo.eu](https://www.euromomo.eu/)) — the European mortality monitoring network, publishing weekly model-based excess-mortality bulletins for participating countries. - Human Mortality Database ([mortality.org](https://www.mortality.org/)) — the Short-Term Mortality Fluctuations series of weekly all-cause deaths used in much comparative research. - Our World in Data ([ourworldindata.org](https://ourworldindata.org/excess-mortality-covid)) — accessible cross-country excess-mortality charts with documented methods. - Robine et al. (2008), “Death toll exceeded 70,000 in Europe during the summer of 2003”, Comptes Rendus Biologies — the standard assessment of the 2003 heatwave. --- # Randomised Controlled Trials in Public Policy: A Primer - Type: Primer - URL: https://helsen-institute.vercel.app/research/rct-policy-primer - Published: 2026-08-05 - Citation: Helsen Institute for Public Research (2026). “Randomised Controlled Trials in Public Policy: A Primer.” Helsen Institute Primer, published August 5, 2026. https://helsen-institute.vercel.app/research/rct-policy-primer **Summary.** A randomised controlled trial assigns people or places to a programme by chance, creating groups that are comparable in expectation, so that later differences can be attributed to the programme itself rather than to who selected into it. Landmark trials — the RAND Health Insurance Experiment, Mexico’s PROGRESA, Oregon’s Medicaid lottery study, Moving to Opportunity — reshaped policy debates precisely because randomisation removed the usual doubts. RCTs answer ‘did this programme cause this effect, here’ with unmatched credibility; they are silent on values, weak on generalisation, and often impractical — which is why modern evaluation pairs them with quasi-experimental designs. **Key takeaways:** - Randomisation solves selection bias: because chance decides who receives a programme, treatment and control groups are comparable in expectation, and outcome differences can be attributed to the programme. - Landmark policy RCTs include the RAND Health Insurance Experiment (1970s–80s), Mexico’s PROGRESA conditional cash transfers (1997), the Oregon Health Insurance Experiment (2008), and Moving to Opportunity (1990s). - The 2019 Nobel Memorial Prize in Economics recognised Abhijit Banerjee, Esther Duflo and Michael Kremer for making randomised experiments central to development economics. - RCTs establish what happened in the trial’s setting; whether results transfer elsewhere (external validity) is a separate, empirical question. - When randomisation is infeasible, quasi-experimental designs — natural experiments, difference-in-differences, regression discontinuity — provide the next-strongest causal evidence. ## The problem randomisation solves Comparing people who joined a programme with people who did not almost never isolates the programme’s effect, because joiners differ from non-joiners in motivation, health, income and countless unmeasured ways. This is selection bias, and it contaminates naive before–after and participant–non-participant comparisons alike. > **Definition:** A randomised controlled trial (RCT) is a study in which eligible people, households, schools or areas are assigned to receive an intervention or to a control condition by a random procedure, so that the groups differ only by chance — and, after the intervention, by its effects. Randomisation makes the control group a credible counterfactual: an estimate of what would have happened to the treated group without the programme. No amount of statistical adjustment of observational data can guarantee this, because adjustment only handles the differences that were measured. ## Landmark trials in public policy - RAND Health Insurance Experiment (United States, 1971–1986). Families were randomly assigned to health-insurance plans with different levels of cost-sharing. Higher cost-sharing reduced use of care substantially, with little measured effect on average health for most participants — though with adverse effects among poorer participants with existing conditions. It remains the reference point for insurance-design debates. - PROGRESA / Oportunidades (Mexico, 1997– ). A conditional cash-transfer programme rolled out with randomised phase-in across villages. Evaluations found increased school enrolment and improved child health, and the programme’s published, independent evaluation became a template copied across dozens of countries. - Oregon Health Insurance Experiment (United States, 2008). Oregon allocated scarce Medicaid places by lottery, creating a natural RCT. Coverage increased health-care use, reduced depression and virtually eliminated catastrophic medical expenses; over the first two years it produced no statistically significant improvement in several measured physical outcomes — a nuanced result that both sides of the US coverage debate learned from. - Moving to Opportunity (United States, 1994–1998). Housing vouchers enabling moves to lower-poverty neighbourhoods were randomly assigned. Long-run follow-ups found substantial adult earnings gains for those who moved as young children, reshaping research on neighbourhood effects. The method’s spread beyond rich countries was recognised when the 2019 Nobel Memorial Prize in Economic Sciences went to Abhijit Banerjee, Esther Duflo and Michael Kremer “for their experimental approach to alleviating global poverty”. Networks such as J-PAL (founded 2003) have since run hundreds of policy trials worldwide. ## Reading an RCT result 1. Intention-to-treat. Good trials compare groups as randomised, whether or not everyone complied — preserving the benefit of randomisation and estimating the effect of offering the programme. 2. Uncertainty. Effects come with [confidence intervals](/glossary#confidence-interval); “no statistically significant effect” in a small trial is not evidence of no effect. 3. Pre-registration. Trials that state outcomes and analyses in advance are far less vulnerable to selective reporting. 4. Outcome choice. Check what was actually measured, over what horizon; many programmes plausibly affect outcomes trials were too short or too small to detect. ## What RCTs cannot do - Generalise automatically. An RCT estimates the effect in its setting, population and era. Transporting results requires argument or replication — the external-validity problem. - Capture system-wide effects. Small trials miss general-equilibrium consequences: a job-search programme that helps participants may partly displace non-participants. - Answer value questions. Trials estimate effects; they do not decide whether those effects justify costs or override competing aims. - Always be feasible or ethical. One cannot randomise constitutions, pandemics or central-bank policy; for many questions, quasi-experimental methods are the realistic ceiling. ## When randomisation is impossible Where trials are infeasible, researchers exploit natural experiments — policy discontinuities, lotteries, staggered roll-outs — using designs such as difference-in-differences, regression discontinuity and instrumental variables. The 2021 Nobel Memorial Prize (David Card, Joshua Angrist, Guido Imbens) recognised precisely this toolkit. The logic is the same as an RCT’s: find variation in exposure that is plausibly unrelated to the outcome, and defend that assumption openly. > **Scope and limitations:** This primer summarises trial logic and canonical examples for non-specialists; it simplifies technical matters (compliance adjustment, clustering, multiple testing) that matter in practice. Descriptions of individual trials compress large literatures — follow the sources for full results and debates. ## Sources and further reading - J-PAL ([povertyactionlab.org](https://www.povertyactionlab.org/)) — trial registries, evaluations and teaching resources on randomised evaluation. - RAND Corporation — the Health Insurance Experiment archive and summaries. - Finkelstein et al., the Oregon Health Insurance Experiment papers (Quarterly Journal of Economics, 2012; New England Journal of Medicine, 2013). - Cochrane ([cochrane.org](https://www.cochrane.org/)) — the systematic-review organisation whose evidence standards grew from clinical trials. --- # Measuring Public Trust in Institutions: Surveys, Findings, and Pitfalls - Type: Review - URL: https://helsen-institute.vercel.app/research/measuring-public-trust - Published: 2026-07-29 - Citation: Helsen Institute for Public Research (2026). “Measuring Public Trust in Institutions: Surveys, Findings, and Pitfalls.” Helsen Institute Review, published July 29, 2026. https://helsen-institute.vercel.app/research/measuring-public-trust **Summary.** Public trust in institutions is measured by several long-running survey programmes — including the OECD Trust Survey, Eurobarometer, the World Values Survey, the Gallup World Poll and national series such as the US General Social Survey — which differ in question wording, scales and coverage. Some findings are robust across sources: trust varies enormously between countries, local institutions tend to be trusted more than national ones, and trust in the US federal government has declined markedly since the 1960s. But many dramatic headline claims rest on comparing incompatible surveys or misreading single-year movements as trends. **Key takeaways:** - No single number measures “trust in institutions”; results depend on which institution is named, how the question is worded, and what response scale is used. - Long-running sources include the OECD Trust Survey, Eurobarometer, the World Values Survey, the Gallup World Poll, the Edelman Trust Barometer and the US General Social Survey. - Robust cross-survey findings: large and persistent differences between countries; higher trust in local than national government in many countries; markedly lower trust in the US federal government than in the early 1960s. - Common errors include comparing levels across surveys with different questions, treating single-year movements as trends, and reading “trust in government” as if it covered all public institutions equally. - A trust statistic should always be read with its question text, scale, population and year attached. ## Why measurement details matter “Trust in institutions is collapsing” may be the most repeated empirical claim in contemporary politics. Whether it is true depends entirely on which institutions, which countries, which years and — critically — which survey. Trust questions are unusually sensitive to wording: asking whether people trust the government “to do what is right”, whether they “have confidence” in it, or whether they trust it “to use their data responsibly” yields different numbers from the same respondents. This review documents the main measurement programmes and separates findings that replicate across them from artefacts of method. It is a companion piece to our explainer on [reading statistics critically](/insights/five-questions-before-trusting-a-statistic). ## The major surveys **Principal recurring surveys measuring institutional trust** | Survey | Run by | Running since | Coverage | Approach | | --- | --- | --- | --- | --- | | OECD Survey on Drivers of Trust in Public Institutions | OECD | 2021 | 30+ OECD countries | 0–10 trust scales for government, civil service, parliament, courts, media; plus questions on drivers such as responsiveness and integrity | | Eurobarometer | European Commission | 1970s | EU member states | “Tend to trust / tend not to trust” for national governments, parliaments and EU institutions | | World Values Survey | WVS Association | 1981 | Global (waves) | Four-point “confidence” scale across many institutions; enables long-run and cross-cultural comparison | | Gallup World Poll | Gallup | 2005 | 140+ countries | Binary confidence questions (national government, judiciary, military, media) | | Edelman Trust Barometer | Edelman | 2001 | 25+ countries | Online survey of trust in government, business, media and NGOs; widely cited, methodology differs from official statistics | | General Social Survey (US) | NORC, University of Chicago | 1972 | United States | Three-point “confidence in the people running” institutions battery; the longest consistent US series | The Pew Research Center’s long-running compilation of US polling deserves separate mention: it tracks trust in the federal government “to do what is right” back to surveys from 1958, providing one of the few six-decade trend lines in existence. ## Findings that hold up across sources - Between-country differences are large and persistent. Across surveys, trust in public institutions is consistently higher in, for example, the Nordic countries and Switzerland than in most other high-income democracies — differences of tens of percentage points that persist across decades and survey houses. - Local beats national in many countries. The OECD Trust Survey and national studies repeatedly find higher trust in local government and in front-line public services than in national governments and parliaments. - The long US decline is real. Trust in the US federal government fell from roughly three-quarters of the public in the early 1960s to around one-fifth to one-quarter in recent years — a decline visible across independent survey series, with the steepest fall in the 1960s–70s. - Institutions differ more than eras. Within most countries, the gap between the most-trusted institutions (often science, the military, or the police, depending on country) and the least-trusted (often political parties and legislatures) is larger than most changes over time. ## Methodological pitfalls 1. Cross-survey level comparisons. A 4-point confidence scale, a 0–10 scale and a binary question produce non-comparable percentages. Levels should only be compared within a survey; trends can cautiously be compared across them. 2. Trend claims from short windows. Trust measures move with elections, scandals and crises (including a widely observed temporary “rally” early in the COVID-19 pandemic). One-year movements are weather, not climate. 3. Bundling institutions. “Trust in government” often means the national executive; extending the claim to courts, civil servants or scientists is unsupported — these routinely score very differently. 4. Mode and sampling effects. Online opt-in panels, phone samples and face-to-face surveys reach different populations; changes in mode can masquerade as changes in trust. 5. Cultural response styles. Willingness to use scale extremes differs across countries, complicating fine-grained international rankings even within one survey. ## How to read a trust statistic A defensible reading of any trust figure requires four attachments: the exact question text, the response scale (and which responses were counted as “trust”), the population and mode (who was asked, and how), and the comparison being made (same survey over time, or across countries within one wave). Claims that arrive without them — most viral charts — cannot be evaluated and should not be repeated. > **Scope and limitations:** This review covers recurring, publicly documented survey programmes; it does not attempt a comprehensive league table of trust levels, precisely because cross-survey level comparisons are unreliable. Survey landscapes change; consult each programme’s current documentation for coverage and methods. ## Sources and further reading - OECD — Survey on Drivers of Trust in Public Institutions, main results reports (2022 and 2024), with full questionnaires and methodology. - World Values Survey ([worldvaluessurvey.org](https://www.worldvaluessurvey.org/)) — data and documentation for all waves since 1981. - Pew Research Center ([pewresearch.org](https://www.pewresearch.org/)) — the long-run “Public Trust in Government” series for the United States. - Eurobarometer — the European Commission’s recurring surveys of trust in national and EU institutions. - NORC General Social Survey — the US confidence-in-institutions series since 1972. --- # Health System Models Compared: Beveridge, Bismarck, and Beyond - Type: Explainer - URL: https://helsen-institute.vercel.app/research/health-system-models - Published: 2026-07-22 - Citation: Helsen Institute for Public Research (2026). “Health System Models Compared: Beveridge, Bismarck, and Beyond.” Helsen Institute Explainer, published July 22, 2026. https://helsen-institute.vercel.app/research/health-system-models **Summary.** Health systems are conventionally sorted into four financing models: the tax-funded Beveridge model, the social-insurance Bismarck model, single-payer national health insurance, and out-of-pocket payment. The typology describes how money is raised and pooled — not how good care is — and every real system mixes elements of several models. The United States is the clearest example, operating versions of all four for different population groups. The models remain useful shorthand, provided they are not asked to explain outcomes on their own. **Key takeaways:** - The standard typology distinguishes four financing models: Beveridge (tax-funded), Bismarck (social health insurance), national health insurance (single public payer, private providers), and out-of-pocket. - The Beveridge model, named for the 1942 report that shaped Britain’s NHS (founded 1948), funds care from general taxation; Nordic countries, Spain and Italy use variants. - The Bismarck model, dating to Germany’s 1883 sickness-fund law, funds care through mandatory insurance contributions; Germany, France, Japan and the Netherlands use variants. - The United States combines all four models for different groups — veterans’ care, employer insurance, Medicare, and the uninsured. - The typology classifies financing and pooling, not quality or efficiency; health outcomes depend heavily on factors outside the financing model. ## Why classify health systems at all Every health system must answer three questions: who pays, how is the money pooled, and who delivers care? The classic four-model typology — popularised in comparative health policy and by T. R. Reid’s The Healing of America — groups countries by their dominant answers. It is best understood as a map of financing architecture: useful for orientation, misleading if mistaken for a ranking. ## The four models ### Beveridge: tax-funded national health services Named after William Beveridge, whose 1942 report laid the groundwork for Britain’s National Health Service (founded 1948), the Beveridge model finances care from general taxation, with the state as the dominant payer and often a major provider. Coverage follows residence, not employment. The United Kingdom, the Nordic countries, Spain, Italy and New Zealand run recognisably Beveridge-type systems. ### Bismarck: social health insurance The Bismarck model descends from Germany’s Health Insurance Act of 1883 under Chancellor Otto von Bismarck — the world’s first national social health insurance scheme. Financing comes from mandatory contributions, historically shared between employers and employees, collected by non-profit insurers (“sickness funds”). Providers are largely private; insurance is compulsory and benefits are standardised. Germany, France, Belgium, Japan and (in reformed, competitive form) the Netherlands and Switzerland follow this family. ### National health insurance: single payer, private providers The national health insurance model combines elements of the other two: providers are mostly private, but a single public insurer pays the bills, financed by taxes or premium-like contributions. Canada is the archetype; Taiwan (since 1995) and South Korea run single national insurers with broad coverage. Single-payer purchasing concentrates negotiating power over prices, at the cost of debates over budgets and waiting times. ### Out-of-pocket: paying at the point of care Where no pooling mechanism covers most people, care is paid for out of pocket. This is the default in many low-income countries and the situation of uninsured people everywhere. High reliance on out-of-pocket payment is the pattern most strongly associated with catastrophic health spending and forgone care, which is why reducing it is central to the World Health Organization’s definition of [universal health coverage](/glossary#universal-health-coverage). **The four financing models at a glance** | Model | Main funding | Pooling | Typical providers | Examples | | --- | --- | --- | --- | --- | | Beveridge | General taxation | Single national pool | Largely public | UK, Nordics, Spain, Italy, New Zealand | | Bismarck | Mandatory insurance contributions | Multiple regulated funds | Largely private (non-profit and for-profit) | Germany, France, Japan, Netherlands | | National health insurance | Taxes / mandatory premiums | Single public insurer | Largely private | Canada, Taiwan, South Korea | | Out-of-pocket | Direct payment by patients | None | Mixed | Dominant in many low-income settings | ## The United States: all four at once The United States is the standard illustration that the models describe subsystems, not countries. Veterans receive care in a government-run system (Beveridge-like); most working-age adults hold employment-linked private insurance (Bismarck-like in structure, though voluntary for employers and typically for-profit); Medicare operates as national health insurance for those 65 and over; and the uninsured pay out of pocket. Much of American health policy debate is, in effect, a debate over the boundaries between these subsystems. ## What the labels do not explain - Outcomes. Life expectancy and mortality differences across high-income countries correlate weakly with financing model; behaviour, social conditions, and system performance within each model matter more. - Spending levels. Both tax-funded and insurance-based systems appear among high and moderate spenders; the outlier status of US spending is not explained by any one model label. - Delivery and quality. The typology says little about primary-care strength, digitalisation, workforce, or waiting times — dimensions on which countries within the same model differ widely. - Hybridisation. Most systems have converged toward mixes: tax subsidies inside Bismarck systems, patient charges inside Beveridge systems, private insurance layered over public cores nearly everywhere. ## Measuring coverage rather than labels For comparative purposes, coverage measures are more informative than model labels. The World Health Organization defines universal health coverage as all people having access to the health services they need without financial hardship, and — with the World Bank — tracks a service-coverage index and the incidence of catastrophic out-of-pocket spending. These indicators cut across the typology and expose gaps that model labels conceal. > **Scope and limitations:** This explainer summarises a conventional typology for orientation. Real systems are hybrids, reform constantly shifts details (contribution rates, coverage rules, private-sector roles), and country assignments describe dominant features, not the whole system. It does not rank systems or recommend a model. ## Sources and further reading - World Health Organization ([who.int](https://www.who.int/health-topics/universal-health-coverage)) — the definition and monitoring framework for universal health coverage. - OECD, Health at a Glance — recurring comparative statistics on financing, coverage and outcomes across member countries. - European Observatory on Health Systems and Policies — the Health Systems in Transition country profiles describing each system in detail. - T. R. Reid, The Healing of America (2009) — the accessible account that popularised the four-model framing. --- # Population Ageing in High-Income Countries: A Primer - Type: Primer - URL: https://helsen-institute.vercel.app/research/population-ageing-primer - Published: 2026-07-15 - Citation: Helsen Institute for Public Research (2026). “Population Ageing in High-Income Countries: A Primer.” Helsen Institute Primer, published July 15, 2026. https://helsen-institute.vercel.app/research/population-ageing-primer **Summary.** Populations age when people live longer and have fewer children. Nearly every high-income country now combines rising life expectancy with fertility below the replacement level of roughly 2.1 births per woman, shifting population structure toward older ages. The change is gradual, highly predictable over decades, and consequential for pensions, health systems and labour supply. Policy can shape how societies adapt to ageing far more than it can reverse the trend itself. **Key takeaways:** - Population ageing is driven by two forces: rising longevity and fertility below the replacement level of about 2.1 births per woman. - Standard measures include the share of the population aged 65 and over, the median age, and the old-age dependency ratio; conclusions can shift depending on which is used and how “working age” is defined. - Japan has the oldest population among major economies, with roughly three in ten residents aged 65 or over. - Ageing that took more than a century in early-industrialising countries such as France is now unfolding within a few decades in parts of East Asia and other later-developing regions. - Migration and family policy can moderate ageing but rarely reverse it; adaptation — pensions, participation, productivity, care systems — is where policy has most leverage. ## What population ageing is > **Definition:** Population ageing is a rise in the share of older people in a population, produced by longer lives, lower birth rates, or both. It is a change in population structure, not merely in size. Ageing is often discussed as if it were a surprise. It is the opposite: because everyone who will be 65 in the year 2060 has already been born, the broad outline of population ageing is among the most predictable facts in social science. Uncertainty attaches to the margins — future fertility, migration and longevity gains — not to the direction of travel. ## How ageing is measured **The standard indicators** - **Share aged 65+:** The percentage of the population aged 65 or over — the simplest headline measure of ageing. - **Median age:** The age that splits the population in half; it summarises the whole age distribution in one number. - **Old-age dependency ratio:** People aged 65+ per 100 people of working age. Definitions of “working age” vary (commonly 15–64 or 20–64), which changes the ratio substantially. - **Total fertility rate (TFR):** The average number of children a woman would have under current age-specific birth rates. Replacement level is about 2.1 in low-mortality countries. Each indicator answers a slightly different question, and headline claims often depend on the choice. A country can have a rising 65+ share while its dependency ratio is temporarily flattered by a large working-age cohort; “working age” cut-offs of 15, 20 or 65 are conventions, not facts about people’s actual working lives. Careful analysis states the definition being used — a habit we examine across statistics in [Why Definitions Decide Debates](/insights/definitions-decide-debates). ## Why populations age Two long-run forces drive ageing. The first is rising longevity: global life expectancy at birth has risen from below 50 years in 1950 to roughly 73 years today, according to United Nations estimates, with high-income countries typically above 80. The second is fertility decline: the global average has fallen from around five births per woman in the early 1960s to a little over two today, and nearly all high-income countries are below the replacement level of about 2.1. A third mechanism, population momentum, shapes the timing: age structures change slowly, so today’s births and deaths echo decisions and conditions decades old. Even if fertility rose to replacement level tomorrow, populations would continue ageing for decades because the large older cohorts are already in place. ## How fast it is happening The pace of ageing differs sharply across countries. In France, the share of the population aged 65 and over took more than a century to double from 7% to 14%. In parts of East Asia the same transition has occurred, or is occurring, within roughly a quarter of a century. Japan currently has the oldest population among major economies, with about three in ten residents aged 65 or over; South Korea, with the world’s lowest fertility rates in recent years, is projected to age exceptionally quickly. Speed matters as much as level: countries that aged slowly built pension and care systems over generations, while fast-ageing countries must adapt institutions within a single working lifetime — often at lower income levels than early-ageing countries enjoyed. ## What ageing means in practice - Pensions. Pay-as-you-go systems transfer income from workers to retirees; a rising retiree-to-worker ratio requires some combination of higher contributions, lower relative benefits, later retirement, or larger transfers from general budgets. - Health and long-term care. Per-person health spending rises steeply at older ages, and demand for long-term care grows with the population aged 80 and over — the fastest-growing age group in most high-income countries. - Labour markets. Slower or negative working-age population growth constrains labour supply, raising the economic weight of participation rates, skills, later retirement and productivity growth. - Public finances. Age-related spending (pensions, health, care) forms the largest predictable pressure on long-run budgets in most high-income countries’ fiscal projections. ## What policy can and cannot change Evidence from decades of policy experience supports a clear division. Policies aimed at reversing ageing have modest effects: family policies (parental leave, childcare, transfers) can support parents and may raise fertility somewhat, but no high-income country has durably returned to replacement level through policy alone. Migration lowers the average age and expands the workforce, but because migrants also age, plausible inflows moderate ageing rather than undo it — a conclusion reached, among others, by the United Nations’ well-known “replacement migration” calculations. Policies aimed at adapting to ageing have far more leverage: raising effective retirement ages alongside healthy-lifespan gains, increasing labour-force participation, investing in productivity, prefunding parts of pension promises, and building long-term-care systems before peak demand arrives. ## How population projections work Long-run population figures come from cohort-component projections: statisticians take today’s population by age and sex, then advance it year by year using assumed rates of fertility, mortality and migration. The United Nations’ World Population Prospects is the standard global source, publishing central estimates alongside variant scenarios. Projections are conditional statements — “if these rates hold, this structure follows” — and are most reliable over one to three decades, where most of the relevant people are already alive. > **Scope and limitations:** This primer describes broad patterns across high-income countries; individual countries differ in level, pace and institutional starting points. Long-run projections beyond mid-century carry substantial uncertainty in fertility and migration assumptions, and dependency ratios based on fixed age cut-offs overstate “dependency” where health and working lives are lengthening. ## Sources and further reading - United Nations, World Population Prospects ([population.un.org/wpp](https://population.un.org/wpp/)) — the standard global demographic estimates and projections, with full methodology. - OECD — comparative work on pensions (Pensions at a Glance) and old-age dependency across member countries. - Human Mortality Database ([mortality.org](https://www.mortality.org/)) — detailed, validated mortality and population series for research use. --- # What Is Evidence-Based Policymaking? - Type: Explainer - URL: https://helsen-institute.vercel.app/research/evidence-based-policymaking - Published: 2026-07-08 - Citation: Helsen Institute for Public Research (2026). “What Is Evidence-Based Policymaking?.” Helsen Institute Explainer, published July 8, 2026. https://helsen-institute.vercel.app/research/evidence-based-policymaking **Summary.** Evidence-based policymaking is the practice of grounding public decisions in the best available research evidence about what works, borrowing its core ideas from evidence-based medicine. In practice it means testing programmes before scaling them, grading the strength of evidence rather than counting studies, and building institutions — evaluation offices, ‘what works’ centres, statutory evidence requirements — that make learning routine. Evidence can discipline claims about consequences; it cannot settle questions of values, and it travels imperfectly across contexts. **Key takeaways:** - Evidence-based policymaking applies the logic of evidence-based medicine to public decisions: systematic use of the best available research on what works, for whom, at what cost. - Evidence varies in strength by design, not by conclusion — a well-run randomised trial supports causal claims that an observational correlation cannot. - Dedicated institutions now embed the practice: the UK’s What Works Network, the US Foundations for Evidence-Based Policymaking Act of 2018, and systematic-review bodies such as the Campbell Collaboration. - Evidence informs but does not decide: policy choices also involve values, priorities and trade-offs that no study can settle. - Findings are context-dependent; a programme that worked in one setting is a reason for a well-designed trial elsewhere, not a guarantee. ## The core idea > **Definition:** Evidence-based policymaking is the systematic use of the best available research evidence — about effectiveness, costs and side-effects — to inform the design, funding and reform of public policies and programmes. The phrase is borrowed from evidence-based medicine, articulated in the 1990s by David Sackett and colleagues as “the conscientious, explicit and judicious use of current best evidence” in clinical decisions. Applied to policy, the same discipline asks of any proposed programme: what is the best evidence that it achieves its aims, how strong is that evidence, and what would it take to find out? The practical shift it demands is subtle but real: from asking whether a study exists that supports a position, to asking how strong the overall body of evidence is — including the studies that point the other way. ## What counts as evidence Research designs differ in how well they can support causal claims. [Randomised controlled trials](/research/rct-policy-primer) and strong natural experiments can isolate a programme’s effect; observational studies can establish associations and generate hypotheses; qualitative work explains mechanisms and context. So-called evidence hierarchies rank designs by their resistance to bias — a useful shorthand, provided it is read as a statement about causal questions specifically, not about the worth of research in general. Modern practice grades bodies of evidence rather than single studies. Frameworks such as GRADE, developed for health guidelines and widely adapted since, rate the certainty of evidence (high to very low) by considering study design, consistency, precision and risk of bias together. [Systematic reviews](/glossary#systematic-review) and [meta-analyses](/glossary#meta-analysis) are the workhorses: they gather all studies meeting pre-stated criteria, appraise them, and synthesise the result. ## The institutions built around it - What Works centres (United Kingdom). Since 2013 the What Works Network has funded centres that commission and translate evidence for specific fields — education, policing, early intervention, local growth — into ratings and toolkits for practitioners. - The Foundations for Evidence-Based Policymaking Act (United States, 2018). Requires federal agencies to publish evidence-building plans, appoint evaluation officers, and improve researcher access to government data. - Systematic-review organisations. The Campbell Collaboration produces systematic reviews in social policy, education and crime, applying the model its sibling, Cochrane, established for health care. - Evaluation and audit bodies. National audit offices, budget offices and dedicated evaluation units increasingly assess not only whether money was spent lawfully but whether programmes achieved measurable results. ## What it looks like in practice In day-to-day government, evidence-based practice usually means: piloting programmes before national roll-out; building evaluation into programme design rather than bolting it on afterwards; pre-stating success measures; publishing results whether or not they flatter the programme; and sunsetting interventions that repeated evaluation finds ineffective. Mexico’s PROGRESA conditional cash-transfer programme, launched in 1997 with a randomised roll-out, remains the canonical example: its published evaluations shaped social policy across many countries. ## Honest limits - Evidence cannot settle values. Studies can estimate a policy’s effects; they cannot decide how much weight a society should give to competing goals. “What works” always implies “works for a stated objective”. - External validity is never free. Effects measured in one country, era or population need not transfer. The disciplined response is replication and local piloting, not assumption. - Timescales collide. Political decisions are often required in months; rigorous evaluation takes years. Interim decisions must be made under uncertainty — the evidence-based habit is to say so explicitly. - Evidence can be selectively deployed. Cherry-picking congenial studies is the failure mode the whole apparatus of systematic review exists to prevent. > **Scope and limitations:** This explainer describes the practice and its institutions; it does not evaluate any specific policy. Terminology varies — many practitioners prefer “evidence-informed” to acknowledge that evidence is one input among several legitimate ones. ## Sources and further reading - Sackett et al. (1996), “Evidence based medicine: what it is and what it isn’t”, BMJ — the founding statement of the evidence-based approach. - UK What Works Network — official documentation of the centres and their evidence standards. - Campbell Collaboration — systematic reviews of social and policy interventions. - Foundations for Evidence-Based Policymaking Act of 2018 — the US statutory framework for federal evidence-building. --- # What Makes Research Citable — by People and by Machines - Type: Insight - URL: https://helsen-institute.vercel.app/insights/what-makes-research-citable - Published: 2026-08-14 **Summary.** Research is cited when it can be found, understood, verified and referenced without friction. In practice that means stable URLs, explicit publication and update dates, self-contained summaries, clearly licensed reuse, and named sources for every claim. The same properties that help human readers also determine whether search engines and AI systems represent work accurately — so citability is now a publishing discipline, not an afterthought. Most research that goes uncited is not wrong — it is merely hard to use. A finding locked in a PDF without a summary, hosted at a URL that changes with every site redesign, published without a date or a licence, will lose to a weaker finding that is easy to quote. Citability is a property of publishing practice, and it can be engineered. ## Five properties of citable research 1. Stable, human-readable references. A claim needs an address that will still resolve in five years: persistent URLs, meaningful slugs, anchors for individual sections. Link rot is the quiet killer of citation. 2. Explicit dates. A dated claim can be placed in time and superseded honestly; an undated page can only be distrusted. Publication and last-updated dates should be visible on the page and in its metadata. 3. Self-contained summaries. A two-to-four sentence summary that survives being quoted out of context does the citing reader’s work for them — and is usually what gets excerpted. 4. Clear licensing. Reuse terms decide whether educators, journalists and databases can carry a finding forward. Open licences such as CC BY, which require attribution, actively convert reuse into citation. 5. Named sources for every claim. A statement whose origin is checkable gets repeated; a statement resting on ‘studies show’ gets discarded by careful readers and, increasingly, by careful machines. ## Machines now read over your shoulder A growing share of first contact with research happens through search snippets and AI assistants, which extract and recombine text. Pages that state facts in complete, attributable sentences — with consistent terminology, real headings and structured metadata — are extracted accurately; pages that rely on layout, tone or surrounding context to carry meaning are extracted wrongly or not at all. Writing for accurate extraction is now part of writing for the public record. None of this substitutes for being right. But between two correct findings, the one published with dates, licence, summary and stable address will be the one the world actually uses. Our own practices along these lines are documented in our [methodology](/methodology). --- # Five Questions to Ask Before Trusting a Statistic - Type: Insight - URL: https://helsen-institute.vercel.app/insights/five-questions-before-trusting-a-statistic - Published: 2026-08-07 **Summary.** Most misleading statistics fail one of five checks: an unclear source, an unstated definition, a missing denominator, vanished uncertainty, or a rigged comparison. Asking these five questions takes a minute and filters out the majority of numbers that should not be repeated. Few people have time to audit every number they encounter; fortunately, most bad statistics fail quickly under a handful of standard questions. The five below cover the failure modes we encounter most often in public debate. ## 1. Who produced it, and can I find it? A statistic without a findable source is a rumour with digits. Trace the number to its producer — a statistical office, a named survey, a published study — and check that the producer actually says what is claimed. Numbers frequently mutate as they pass through headlines and social media. ## 2. How is the thing defined? “Unemployment”, “poverty”, “migrant”, “excess death” — every measured concept has a technical definition that may differ sharply from the everyday word. Definitional choices routinely move headline numbers more than real-world change does; we examine this at length in [Why Definitions Decide Debates](/insights/definitions-decide-debates). ## 3. What is the denominator? Raw counts grow with population; “a record number of X” may mean nothing but a record number of people. Ask: per how many? Compared with the relevant base? A risk that doubles from one in a million to two in a million has doubled — and is still tiny. Percentages of small or shifting bases mislead in both directions. ## 4. Where did the uncertainty go? Estimates come with margins of error and [confidence intervals](/glossary#confidence-interval); headlines strip them. Small movements inside the margin of error — common in polls — are reported as swings. If a source publishes no uncertainty at all, treat precision as false. ## 5. Compared with what? Every “more”, “worse” or “record” embeds a comparison. Check the baseline: is it the relevant period (or a cherry-picked trough)? Are the units and definitions the same on both sides? Cross-country comparisons in particular routinely compare numbers produced by incompatible methods — a pitfall our review of [trust surveys](/research/measuring-public-trust) documents in detail. A number that passes all five checks may still be wrong — but it has earned provisional trust. A number that fails any of them has not earned repetition. --- # Why Definitions Decide Debates in Public Statistics - Type: Insight - URL: https://helsen-institute.vercel.app/insights/definitions-decide-debates - Published: 2026-07-31 **Summary.** Many statistical controversies are definitional disagreements in disguise. Whether unemployment counts discouraged workers, whether poverty is measured against a fixed basket or median income, and whether migrants are counted by birthplace or citizenship each change headline figures substantially — with every option defensible for some purposes. The remedy is not a single correct definition but explicit ones. When two commentators cite wildly different numbers for the same phenomenon, the explanation is usually not that one dataset is fake. It is that the two numbers measure differently defined things. Three familiar examples show the pattern. ## Unemployment: jobless is not the same as unemployed Under the International Labour Organization definition used for headline rates, a person is unemployed only if they are without work, actively seeking it, and available to start. People who have stopped searching, or who work a few hours while wanting more, fall outside the headline count — which is why statistical agencies publish broader measures of labour underutilisation alongside it. Neither number is “the truth”; they answer different questions. ## Poverty: absolute or relative? The World Bank’s international extreme-poverty line measures consumption against a fixed threshold adjusted for purchasing power — revised in 2025 to $3.00 a day at 2021 prices — and is designed to track absolute deprivation. The European Union’s headline indicator instead counts people below 60% of national median income: a relative measure that tracks inequality’s lower tail. Absolute poverty can fall while relative poverty rises, and vice versa; debates that ignore the distinction talk past each other by construction. ## Migration: foreign-born or foreign citizens? Migrant statistics may count the foreign-born (a fixed fact of biography) or foreign citizens (which changes with naturalisation), and flows may or may not include returning nationals, students, or short stays. Countries differ in which they emphasise, and the resulting figures can differ by large factors while all being individually correct. ## The practical rule Definitions are not deceptions — they are unavoidable choices, each fit for some purposes and unfit for others. The integrity test is whether the definition is stated, stable over time, and appropriate to the claim being made. A statistic quoted without its definition is unfinished; our checklist for such cases is in [Five Questions to Ask Before Trusting a Statistic](/insights/five-questions-before-trusting-a-statistic). --- # Correlation, Causation, and How Public Debate Confuses Them - Type: Insight - URL: https://helsen-institute.vercel.app/insights/correlation-causation-public-debate - Published: 2026-07-24 **Summary.** A correlation between two things is compatible with at least three explanations besides causation: confounding by a third factor, reverse causality, and selection effects. Causal claims require designs that rule these out — randomised trials where possible, credible natural experiments otherwise — plus transparent assumptions. ‘Correlation is not causation’ is the beginning of the analysis, not the end. Everyone has heard that correlation is not causation. The more useful knowledge is why a correlation arises when causation is absent — because each mechanism suggests its own test. ## Three ways a correlation lies - Confounding. A third factor drives both variables. Ice-cream sales and drowning deaths rise together because summer causes both. In observational health data, socioeconomic status confounds almost everything. - Reverse causality. The arrow points the other way. Do people become unhealthy because they stop exercising — or stop exercising because they became unhealthy? Cross-sectional snapshots cannot say. - Selection. The comparison groups formed themselves. People who choose a programme differ from those who do not, in ways usually unmeasured — the core problem randomisation exists to solve, as we explain in our [RCT primer](/research/rct-policy-primer). ## What supports a causal claim The strongest support is design: randomised assignment, or natural experiments in which exposure varied for reasons plausibly unrelated to the outcome. Where only observational data exist, researchers triangulate — checking dose-response patterns, temporal order, robustness to adjustment, and consistency across settings. The epidemiologist Austin Bradford Hill’s famous 1965 “viewpoints” (strength, consistency, temporality, plausibility, and others) remain a useful checklist, so long as they are treated as considerations for judgement rather than a scoring rubric. ## Reading claims in the wild Practical heuristics: be most sceptical when the claimed cause is something people select into (diets, apps, neighbourhoods); check whether the study design could distinguish direction; and notice whether effect sizes survive adjustment or shrink toward zero as controls improve — a classic signature of confounding. And symmetrically: correlation plus strong design is how nearly everything we reliably know about causes was learned. The slogan should raise the bar for evidence, not lower it to nihilism. --- # Open Data: The Promise and the Limits - Type: Insight - URL: https://helsen-institute.vercel.app/insights/open-data-promise-and-limits - Published: 2026-07-17 **Summary.** Open data — data anyone can access, use and share — enables verification, secondary research and public tools, and the FAIR principles (findable, accessible, interoperable, reusable) describe what makes it genuinely usable. But openness is not sufficiency: open datasets can be low-quality or unusable in practice, privacy limits what should be opened at record level, and publication without documentation or capacity produces transparency in name only. The strongest argument for open data is epistemic: claims based on closed data must be taken on trust, while claims based on open data can be checked, re-analysed and corrected. Open publication is how errors get found — and how findings earn durable credibility. ## What ‘good’ open data means The widely adopted FAIR principles (Wilkinson et al., 2016) hold that data should be findable (indexed, with persistent identifiers), accessible (retrievable by standard protocols), interoperable (standard formats and vocabularies) and reusable (documented, with clear licences). The principles matter because raw openness fails without them: an undocumented spreadsheet at an unstable URL is technically open and practically useless. ## Three limits to keep in view - Privacy. Record-level data about people can often be re-identified even after names are removed, especially when combined with other sources. Aggregation, access controls for sensitive microdata, and formal disclosure-control methods exist precisely because ‘open everything’ is not a responsible default for personal data. - Quality. Openness and accuracy are independent properties. Open publication makes errors findable; it does not prevent them. Documentation, versioning and known provenance are what allow users to judge fitness for purpose. - Capacity and ‘open-washing’. Publishing data is cheap; making it usable — cleaned, documented, maintained, answerable — is not. Portals full of stale, undocumented files deliver the appearance of transparency without its substance. The practical conclusion: open data is infrastructure, and like all infrastructure it works when maintained. For research organisations, the same logic applies to publications themselves — openly licensed, documented and stably addressed work is work that others can verify and build on, a theme we develop in [What Makes Research Citable](/insights/what-makes-research-citable). --- # Focus areas ## Public Health & Health Systems URL: https://helsen-institute.vercel.app/focus-areas/health-systems Public health and health systems research examines how countries organise, finance, and deliver health care, and how population health is measured and compared — including indicators such as life expectancy, excess mortality, and coverage of essential services. ## Population & Demographic Change URL: https://helsen-institute.vercel.app/focus-areas/population-change Demographic research studies how populations change in size and structure through births, deaths, and migration, and how indicators such as the old-age dependency ratio, median age, and total fertility rate describe those changes. ## Public Trust & Institutions URL: https://helsen-institute.vercel.app/focus-areas/public-trust Research on public trust examines how confidence in institutions such as government, media, science, and the courts is measured through surveys, how those measures differ, and what patterns they show across countries and over time. ## Evidence & Research Methods URL: https://helsen-institute.vercel.app/focus-areas/research-methods Evidence and research methods work explains how knowledge for public decisions is produced and graded — including randomised controlled trials, quasi-experimental designs, systematic reviews, and the frameworks used to judge how strong a body of evidence is. ## Data & Statistical Integrity URL: https://helsen-institute.vercel.app/focus-areas/data-integrity Data and statistical integrity research examines how official statistics are defined and produced, how definitional choices shape public debate, and what practices — provenance, open licensing, stable references — make data and research trustworthy and reusable. --- # Glossary URL: https://helsen-institute.vercel.app/glossary ## Cohort study A cohort study follows a defined group of people over time, comparing outcomes between those exposed and not exposed to a factor of interest. Because exposure is observed rather than assigned, cohort studies can establish temporal order but remain vulnerable to confounding. ## Confidence interval A confidence interval is a range of values, calculated from sample data, that expresses the statistical uncertainty around an estimate; a 95% interval is produced by a procedure that captures the true value in 95% of repeated samples. Wide intervals signal imprecise estimates. An interval that includes zero (or no difference) means the data are compatible with there being no effect — not proof that none exists. ## Cross-sectional study A cross-sectional study measures a population at a single point in time, providing a snapshot of conditions and associations but no information about temporal order. Snapshots cannot distinguish whether A preceded B or the reverse, which limits causal interpretation. ## Evidence hierarchy An evidence hierarchy ranks study designs by their resistance to bias when answering causal questions, typically placing systematic reviews and randomised trials above observational designs and expert opinion. Hierarchies apply to causal effectiveness questions specifically; for questions of mechanism, prevalence or lived experience, other designs are the appropriate tool. ## Excess mortality Excess mortality is the difference between the number of deaths from all causes observed in a period and the number expected based on historical patterns, capturing the total mortality impact of a crisis. See our full explainer: Excess Mortality: What It Measures and Why It Matters. ## Healthy life expectancy Healthy life expectancy estimates the average number of years a person can expect to live in good health, combining mortality data with information on illness and disability. The gap between life expectancy and healthy life expectancy indicates years lived with health limitations; it is central to debates about raising retirement ages. ## Life expectancy Life expectancy at birth is the average number of years a newborn would live if current age-specific death rates remained constant throughout its life — a summary of present mortality conditions, not a forecast for any individual. Because it is built from current death rates, period life expectancy can move sharply in crisis years and does not incorporate future medical progress. ## Meta-analysis A meta-analysis statistically combines the results of multiple comparable studies into a single pooled estimate, increasing precision beyond what any individual study provides. Pooling is only as good as the studies pooled: combining biased studies yields a precise biased answer. Meta-analyses are usually conducted within systematic reviews. ## Old-age dependency ratio The old-age dependency ratio is the number of people aged 65 and over per 100 people of working age; conventions for “working age” (commonly 15–64 or 20–64) vary and materially change the figure. The ratio counts ages, not actual work or dependency — a limitation that grows as working lives lengthen. See our population ageing primer. ## Peer review Peer review is the evaluation of research by independent experts in the same field before publication, used by journals to assess validity, rigour and clarity. Peer review is a quality filter, not a guarantee: reviewed papers can be wrong, and important work sometimes appears first as preprints. ## Preprint A preprint is a research manuscript shared publicly before formal peer review, typically on servers such as arXiv, medRxiv or SSRN. Preprints accelerate scientific communication but should be read as provisional; findings may change after review. ## Randomised controlled trial (RCT) A randomised controlled trial assigns participants to an intervention or control group by chance, so the groups are comparable in expectation and outcome differences can be attributed to the intervention. See our full primer: Randomised Controlled Trials in Public Policy. ## Replacement-level fertility Replacement-level fertility is the average number of births per woman at which a population exactly replaces itself between generations without migration — about 2.1 in low-mortality countries. The level exceeds 2.0 because slightly more boys than girls are born and some children do not survive to reproductive age; it is higher where child mortality is higher. ## Systematic review A systematic review identifies, appraises and synthesises all studies meeting pre-specified criteria on a defined question, using an explicit, reproducible search method. The pre-stated protocol is what distinguishes systematic reviews from narrative reviews, guarding against cherry-picking congenial studies. ## Total fertility rate (TFR) The total fertility rate is the average number of children a woman would bear if she experienced the current year’s age-specific birth rates throughout her reproductive life. TFR is a synthetic period measure: it can dip when births are postponed and later recover, without completed family sizes changing as much. ## Universal health coverage (UHC) Universal health coverage, as defined by the World Health Organization, means all people have access to the health services they need — promotion, prevention, treatment, rehabilitation and palliative care — without suffering financial hardship. UHC is measured along two dimensions: a service-coverage index and the incidence of catastrophic out-of-pocket health spending. --- # Frequently asked questions URL: https://helsen-institute.vercel.app/faq ## What is the Helsen Institute for Public Research? The Helsen Institute for Public Research is an independent, nonpartisan research institute founded in 2026. It publishes clear, verifiable syntheses of public evidence — explainers, primers, briefings and reviews — on public health and health systems, demographic change, public trust in institutions, research methods, and statistical integrity. The Institute explains evidence and methods; it does not campaign, endorse candidates or parties, or lobby. ## How is Helsen Institute research produced and reviewed? Every publication is drafted against named, checkable sources — official statistics, peer-reviewed research and primary documents — and passes internal editorial review for accuracy, sourcing and clarity before release. Publications state their scope and limitations explicitly, carry publication and last-updated dates, and are revised when the underlying evidence changes. Full details are on our methodology page. ## Can I reuse or republish Helsen Institute content? Yes. Unless otherwise noted, all Helsen Institute publications are licensed under Creative Commons Attribution 4.0 (CC BY 4.0). You may copy, redistribute, translate and adapt our work — including commercially — provided you credit the Helsen Institute for Public Research and link to the original page. ## How should I cite the Helsen Institute? Cite the Institute as author, with the publication title, year, and the page URL — for example: Helsen Institute for Public Research (2026), “Excess Mortality: What It Measures and Why It Matters”. Each publication page includes a ready-made citation in its “How to cite” section. Because every page carries a stable URL and explicit dates, citations remain checkable over time. ## Is the Helsen Institute politically affiliated? No. The Institute is nonpartisan and independent. It takes no positions on parties, candidates or campaigns, and its publications separate description of evidence from recommendation — in general, we explain what the evidence shows and how it was produced, and leave value judgements to readers. ## How is the Helsen Institute funded? The Institute is independently governed and does not accept funding conditioned on the direction or conclusions of its research. Funders have no editorial involvement, and funding disclosures are maintained on our About page. ## What happens when the Helsen Institute makes a mistake? Errors are corrected in the published text as soon as they are established, with the page’s last-updated date revised. Substantive corrections are noted on the page itself, and we welcome reports of suspected errors at any time via our contact page. ## May search engines and AI systems index and cite Helsen Institute content? Yes — the site is built for it. All content is openly licensed (CC BY 4.0), published with structured metadata, stable URLs and explicit dates, and our robots policy permits reputable search and AI crawlers. We ask that automated systems, like human readers, attribute content to the Helsen Institute for Public Research and link to the source page.