The original claim on X, with Eurostat chart showing "Healthy life years at birth" for selected EU countries.
Why the Eurostat HLY indicator is a survey artefact, not a health crisis
Claim source: @1goodtern on X, citing Eurostat table hlth_hlye
Analysis by Claude Opus 4.6 (Anthropic), prompted by @dobssi on X · March 2026
"Absolutely staggering drop in Healthy Life Expectancy in Sweden according to Eurostat. From 73.3 years of healthy life expectancy down to 66.2. Sweden. Can no one understand how much trouble we're in?"
The original claim on X, with Eurostat chart showing "Healthy life years at birth" for selected EU countries.
The post accompanies a chart from Eurostat's hlth_hlye table showing Sweden's "Healthy Life Years at birth" falling from a peak around 73 years (2016–2020) to roughly 66 years by 2022–2023. The implication is that Sweden has experienced a catastrophic deterioration in population health. This analysis examines what the metric actually measures and whether the claim is consistent with objective Swedish health data.
Healthy Life Years (HLY), also known as disability-free life expectancy or Sullivan's Index, is a composite indicator that combines two inputs: standard life tables (mortality data) and a single self-reported question from the EU Statistics on Income and Living Conditions (EU-SILC) survey called the Global Activity Limitation Indicator, or GALI.
The GALI question asks survey respondents: "For at least the past six months, to what extent have you been limited because of a health problem in activities people usually do? Would you say you have been: severely limited? limited but not severely? not limited at all?" That single subjective question — answered by a sample of private households — is then fed into the Sullivan method to calculate a life expectancy weighted by the proportion reporting no limitations.
This is not a clinical measurement. It is not derived from hospital records, disease registries, cause-of-death data, or any objective health metric. It reflects how a sample of people answered one question in a household income survey. The indicator was designed as a rough policy monitoring tool for cross-country comparisons — not as a precise measure of national health status, and certainly not as something from which year-on-year catastrophes should be inferred.
Eurostat's own methodology page acknowledges this directly [7]: the indicator is derived from self-reported data and reflects respondents' subjective perceptions as well as their social and cultural background. Furthermore, EU-SILC does not cover people living in health and social care institutions — those most likely to face activity limitations — meaning the underlying data systematically under-estimates the share of people with limitations.
Sweden has historically recorded among the highest HLY values in the EU. In 2016, Sweden topped the entire table at 73.3 years for women and 73.0 for men [10]. In 2020, it was still the highest: 72.7 for women, 72.8 for men [10]. Eurostat's own press release noted that extreme values can partly be explained by the way activity limitation is measured in each country.
Academic research has been more direct. A peer-reviewed study found that Sweden and Norway show considerably higher HLY values [4] than other European countries, and concluded this likely reflects differences in the national implementation of the GALI question rather than genuine health differences. Methodological factors such as question formulation, data collection method, and the number of response categories can all produce comparatively high or low prevalence of reported limitations.
In other words, Sweden's HLY was probably overstated for years before it "dropped." A correction toward the EU mean (around 63 years) might simply mean the measurement became more realistic, not that health deteriorated.
EU-SILC is implemented nationally, and Sweden has among the worst response rates in Europe. Eurostat's own methodology documentation records that the overall individual non-response rate in EU-SILC varied from 5.7% in Romania to 49.5% in Sweden [9]. When nearly half of selected individuals do not respond, the results become highly sensitive to who responds and how the survey is administered.
An academic study in the European Journal of Public Health examined attrition bias in EU-SILC and found that the 95% limits of agreement for HLY estimates reach ±2 years [1]. This means that even without any real health change, a single country's HLY can fluctuate by up to 2 years purely from sampling variation and panel attrition. The authors explicitly concluded that attrition can lead to significant error in single-country HLY estimates and wrong conclusions in comparative studies of population health. Sweden's apparent 7-year drop is dramatic, but in a survey with this level of non-response, even large swings cannot be taken at face value without corroboration from objective data.
The entire EU-SILC legal and methodological framework was revised under the Integrated European Social Statistics regulation (2019/1700) with effect from 2021 [6]. This introduced multiple changes to data collection, variable definitions, and survey implementation across member states. The timing of Sweden's HLY "drop" — which begins accelerating around 2021–2022 — coincides directly with this regulatory transition.
The impact of survey methodology changes on HLY is not theoretical. A 2025 European Parliament study on promoting healthy ageing [11] explicitly documented how changes to GALI question phrasing in Germany and other countries led to inconsistencies in data, generating significant fluctuations in reported HLY and undermining its reliability as a structural health measure. The study went so far as to recommend either simplifying the GALI to a yes/no format or replacing it entirely with chronic health problem data from SILC, precisely because question wording changes across countries and years have produced wild swings in national estimates.
The pandemic forced many countries to shift EU-SILC data collection from face-to-face computer-assisted personal interviewing (CAPI) to telephone interviewing (CATI). This methodological change is known to affect how people respond to subjective health questions. Sweden, as a "register country" where most EU-SILC data comes from administrative databases but the health questions (including GALI) still require personal interviews for subjective information, would be particularly sensitive to such mode effects. Any shift in who responds — or how they interpret the question over the phone versus in person — directly alters the disability prevalence estimate that feeds into the HLY calculation.
If Sweden had genuinely lost 7 years of healthy life expectancy between 2016 and 2023, this would represent an extraordinary health catastrophe — visible across every objective data source. It would mean a population suddenly experiencing disability at rates comparable to Latvia or Slovakia, countries with fundamentally different health profiles. The companion reports in this suite examine exactly these objective data sources.
The companion mortality analysis examined natural-cause mortality across all age groups using Socialstyrelsen (National Board of Health and Welfare) data from 2000 to 2024, with a quadratic 2000–2019 baseline and 95% prediction intervals.
No excess mortality from natural causes in Swedish children (ages 0–19) at any point during or after the pandemic. A single brief breach occurred in working-age adults (20–64) and the elderly (65+) during the acute pandemic year of 2020, after which rates returned to or below baseline. By 2024, multiple age bands were recording the lowest natural-cause mortality rates ever measured in the dataset. Sweden's life expectancy at birth barely dipped in 2020 and recovered to pre-pandemic levels by 2021 — one of the few EU countries where this was the case.
A 7-year drop in healthy life expectancy would require either a dramatic rise in mortality or a dramatic rise in disability. The mortality data show neither. Swedish mortality continued its long-term improving trend with only a brief pandemic interruption.
The diagnosis overview examined age-standardised diagnosis rates across all 14 ICD-10 chapters using Socialstyrelsen patient registry data from 2008 to 2024. Of the 14 chapters analysed with linear 2012–2019 baselines, only two showed sustained above-baseline breaches in the 0–19 age group: D50–D89 (Blood & immune disorders) and E00–E90 (Endocrine, nutritional & metabolic). The remaining 12 chapters — covering everything from cancers to cardiovascular disease to respiratory conditions to musculoskeletal disorders — showed no sustained post-pandemic elevation in children or indeed across most age groups.
The monthly deep dive into those two flagged chapters identified the specific driver: D80–D89 (immune disorders), with a clear age gradient strongest in 0–4 year olds, peaking at +82% ASMR excess in 2022 and still elevated at +31% in 2025. Catch-up diagnosis was excluded (no 2020 dip, no plateau over 5 years). E10–E14 (diabetes) showed a divergent pattern — rising in 0–4 year olds but declining in 10–19 year olds. E65–E68 (obesity) showed broad increases but with a pre-existing upward trend complicating pandemic-specific attribution.
These are real and important findings — but they describe specific signals in specific diagnosis categories in specific age groups. They are the opposite of the broad-based population health collapse that a 7-year HLY drop would imply across all ages and all conditions.
| Indicator | Type | Post-2020 signal |
|---|---|---|
| Life expectancy at birth | Objective (mortality) | Recovered by 2021; among highest in EU |
| Natural-cause mortality, ages 0–19 | Objective (mortality) | No breach of 95% PI; continued improvement |
| Natural-cause mortality, all ages | Objective (mortality) | Brief 2020 breach only; 2024 at record lows |
| 12 of 14 ICD-10 diagnosis chapters | Objective (registry) | No sustained above-baseline elevation |
| D80–D89 Immune disorders, ages 0–19 | Objective (registry) | Sustained breach since Q4 2020; real signal |
| Eurostat "Healthy Life Years" | Subjective (survey) | ~7-year drop 2016→2023 |
The HLY indicator is the sole outlier suggesting catastrophic decline. Every objective data source tells a consistent story of a country with excellent health outcomes, a brief pandemic disruption, and continued long-term improvement — with one specific, well-characterised morbidity signal in paediatric immune disorders.
The "staggering drop" in Swedish Healthy Life Years is almost certainly a survey measurement artefact, not evidence of population health deterioration.
The HLY indicator rests on a single subjective question in a household income survey with a ~50% non-response rate in Sweden. The drop coincides with a major EU-SILC framework revision in 2021 and pandemic-era changes to survey administration mode. Sweden was already a known outlier on the high side, with academic literature attributing its extreme values to national survey implementation differences rather than genuinely superior health. Eurostat's own metadata warns that extreme values are partly explained by measurement differences between countries.
Meanwhile, every objective Swedish health metric — mortality rates across all age groups, hospital registry data across 14 ICD-10 chapters, cause-of-death statistics — shows either continued improvement or stability, with one documented exception (D80–D89 immune disorders in young children, characterised in the companion deep dive).
Presenting the HLY drop as evidence of a population health crisis, without acknowledging any of these well-documented methodological limitations, contributes to statistical misinformation. The data that matter — the data derived from actual clinical encounters and death certificates rather than survey responses — do not support the claim.
This report synthesises findings from the three companion analyses (which use 2000–2024 mortality data and 2008–2025 patient registry data from the Swedish National Board of Health and Welfare) with Eurostat's published HLY methodology documentation, EU-SILC quality reports, and peer-reviewed literature on the GALI indicator. No new statistical modelling was performed for this report; the objective data references draw on the existing suite's baselines and prediction intervals (quadratic 2000–2019 for mortality; linear 2012–2019 for diagnoses; per-month linear 2012–2019 for the monthly deep dive; European Standard Population 2013 for age standardisation throughout).
[1] Muszyńska-Spielauer MM, Spielauer M. "The effect of sample attrition in the EU Statistics on Income and Living Conditions on the estimates of Eurostat's Healthy Life Years." European Journal of Public Health, 33(3):378–380, 2023. DOI: 10.1093/eurpub/ckad069
Finding: 95% limits of agreement for HLY reach ±2 years; attrition can lead to significant error in single-country estimates.
[2] Luy M, Di Giulio P, Minagawa Y. "The impact of interpersonal reporting heterogeneity on cross-country differences in Healthy Life Years in Europe." European Journal of Public Health, 33(6):1060–1064, 2023. DOI: 10.1093/eurpub/ckad176
Finding: Differential item functioning shifts country HLY rankings by up to 3 positions; Sweden's respondents rate identical disability vignettes as less limiting than the European average.
[3] Berger N et al. "Assessing the validity of the Global Activity Limitation Indicator in fourteen European countries." BMC Medical Research Methodology, 15:1, 2015. DOI: 10.1186/1471-2288-15-1
Finding: GALI is a valid general measure, but differs significantly between countries in how it reflects disability — likely due to implementation variations, perception differences, and question understanding.
[4] Spitzer S, Greulich A, Gagnon A. "The impact of population's educational composition on Healthy Life Years: An empirical illustration of 16 European countries." SSM – Population Health, 15:100857, 2021. DOI: 10.1016/j.ssmph.2021.100857
Finding: Sweden and Norway show considerably higher HLY than other European countries, likely reflecting GALI implementation differences rather than actual health differences.
[5] Bogaert P et al. "The use of the global activity limitation indicator and healthy life years by member states and the European Commission." Archives of Public Health, 76:30, 2018. DOI: 10.1186/s13690-018-0279-z
Finding: GALI question wording has changed over the years since introduction; survey organisers use their own adaptations, limiting comparability.
[6] Wirth H, Pforr K. "The European Union Statistics on Income and Living Conditions after 15 Years." European Sociological Review, 38(5):832–848, 2022. DOI: 10.1093/esr/jcac024
Finding: EU-SILC was revised with effect from 2021 under new framework regulation 2019/1700; Sweden uses stratified single-stage random sampling with register data for most variables but interviews for subjective health.
[7] Eurostat. "Healthy life years statistics." Statistics Explained, updated 2024. Link
Official methodology description; notes that the indicator is self-reported, reflects subjective perceptions, and excludes the institutionalised population.
[8] Eurostat. "Healthy life years by sex (from 2004 onwards) — metadata." hlth_hlye ESMS. Link
Documents GALI implementation differences between countries; notes the way the question was implemented may hamper cross-country comparisons.
[9] Eurostat. "Health variables in SILC — methodology." Statistics Explained. Link
Reports Sweden's individual non-response rate at 49.47% — the highest in the EU.
[10] Eurostat press releases: ddn-20190204-1 (2016 data, Sweden highest at 73.3); ddn-20220613-1 (2020 data, Sweden still highest at 72.8); ddn-20240925-1 (2022 data); ddn-20250808-1 (2023 data, Sweden at 67.2 for men).
Extreme values noted as partly explained by how activity limitation is measured in each country.
[11] European Parliament, EPRS. "Promoting healthy ageing in the EU." Study PE 765.798, 2025. PDF
Documents how GALI question phrasing changes in Germany and other countries generated significant HLY fluctuations; recommends simplifying GALI to yes/no or replacing with chronic health problem data.