The state of global disability metrics in 2026 — what gets counted, what doesn’t, and why the gap keeps closing slowly

Two decades after the Washington Group on Disability Statistics published its Short Set on Functioning, the world finally has a defensible headline number: 1.3 billion people, roughly 16% of the global population, living with significant disability, per the WHO’s 2024 monitoring update to the World Report on Disability. That figure rests on a measurement infrastructure that took twenty years to build — and still misses four populations almost entirely. Fifty-five-plus national statistical offices now run the Washington Group Short Set; more than 90 LMICs publish UNICEF/WG Child Functioning Module data; the 2025 UN DESA Compendium maps around 120 countries producing at least some disability-disaggregated SDG indicator reporting. The headline is firmer than it has ever been. What still slips through it is what the rest of this dossier is about.

01 · How disability is being counted in 2026

Three things changed between the 2011 World Report on Disability and the WHO’s 2024 monitoring update, and they all push the global prevalence figure in the same direction. The world is older — the over-60 population has grown by close to 240 million in 13 years, and significant functional limitation rises steeply after 60. Chronic-disease prevalence, particularly type-2 diabetes and its sequelae, has expanded faster than population growth across South Asia and sub-Saharan Africa. And measurement itself has improved: the Washington Group instruments, the UNICEF/WG Module on Child Functioning, and the Model Disability Survey have matured into reference tools that produce higher and more reliable counts than the legacy “do you have a disability?” yes/no question they replaced.

02 · The headline number, and what it actually measures

The WHO’s 16% prevalence figure (roughly 1.3 billion people) is therefore best read as the floor of a defensible estimate, not as a discovery that the world has somehow become more disabled in 13 years. Of that 1.3 billion, the 2024 update splits out around 190 million adults with very significant difficulties in functioning — the population for whom rehabilitative services, assistive technology, and personal-assistance entitlements are most plainly indicated and least adequately delivered. That sub-figure has barely moved between the 2011 and 2024 reports, which is itself a finding.

The number you do not see in the WHO headline is impairment-by-condition. The 2011 report leaned on the Global Burden of Disease (GBD) framework — years lived with disability by specific cause — and the 2024 update keeps GBD as an appendix rather than the headline. A functional-limitation count and a condition-by-condition count are not the same population: conflating them double-counts people who appear in several condition rows and undercounts people whose limitations have no neat ICD-11 home. The 2024 update treats functional limitation as the primary frame.

The range — 2.2% to 17.8% — is what the field’s long debate over comparability is actually about. It is not, mostly, that disability is genuinely seven times more common in the UK than in India. The instrument, the threshold at which someone is counted (some difficulty? a lot?), the inclusion or exclusion of people in institutions, and respondents’ willingness to self-identify all move the number more than the underlying epidemiology does. Where the WG-SS is deployed consistently — across most of Latin America in particular — cross-country numbers now sit in a far tighter band than a decade ago.

03 · The Washington Group instruments and where they actually run

The Washington Group Short Set on Functioning (WG-SS) is six questions: difficulties seeing, hearing, walking or climbing steps, remembering or concentrating, self-care, and communicating, each on a four-point severity scale (no difficulty, some, a lot, cannot do at all). It was designed in the mid-2000s to be short enough to bolt onto a national census, neutral enough to translate across cultures, and severity-graded enough to support cross-country comparison if used consistently. The Washington Group Secretariat’s 2024 inventory counts more than 55 national statistical offices that have deployed WG-SS verbatim or near-verbatim in their most recent census or flagship household survey.

”More than 55” sounds modest, and it is. There are around 200 jurisdictions that run a decennial or quinquennial census. The WG-SS users include most of Latin America, an expanding number of African NSOs, several large Asian countries, and a slowly growing share of OECD high-income members. Non-users split between countries that still rely on a legacy “are you a person with a disability?” single question (common in parts of South and Southeast Asia and several Gulf states), countries running a national instrument designed before WG-SS existed (the UK’s ONS uses the GSS harmonised disability question alongside WG-style questions), and countries whose most recent census predates the Washington Group’s last revision.

04 · The child-functioning instrument

For children, the canonical instrument is the Washington Group / UNICEF Module on Child Functioning (CFM), finalised in 2016 and now the reference module for the 2–4 and 5–17 age bands. UNICEF’s MICS (Multiple Indicator Cluster Survey) programme has included the CFM by default since MICS6; the 2024 global release carries CFM-based prevalence estimates from more than 90 low- and middle-income countries, the broadest comparable child-disability dataset that has ever existed. UNICEF’s 2024 Seen, Counted, Included update reports that around 1 in 10 children aged 5–17 in LMICs has a disability under the CFM definition — substantially higher than legacy estimates, and consistent with the WHO’s directional adjustment. The CFM has no analogue for children under 2: the developmental-milestone variation in the 0–24 month band is too wide for a short module to be useful, and that gap is the first of four populations the 2026 measurement infrastructure does not adequately count.

05 · The populations the headline numbers still miss

Four groups remain systematically under-counted even in the best-instrumented national surveys.

People in institutions

Most household-survey sampling frames exclude residential institutions, long-stay psychiatric facilities, prisons, and refugee camps — populations in which disability prevalence, particularly psychosocial and intellectual disability, is dramatically elevated. The European Commission’s 2023 institutional-population mapping found around 1.4 million Europeans in residential disability institutions alone, none of whom appear in standard household-survey estimates. UN DESA’s 2025 Compendium calls the exclusion a “first-order undercount” in its methodology annex.

Persons with intellectual or psychosocial disabilities

The six WG-SS questions are functional-domain questions and good at picking up what they cover. Intellectual disability shows up partially through “remembering or concentrating”; psychosocial disability shows up partially through “communicating” and partially nowhere. The Washington Group’s Extended Set and the WG-SS Enhanced module add questions on anxiety, depression, upper-body function, and fatigue, but Enhanced is in use in fewer than two dozen countries as of 2024. The WHO 2024 update flags this as the single largest measurement gap in the current instrument family.

Children under 2

The CFM begins at age 2 and no internationally standardised instrument exists for younger children. UNICEF and the WHO’s Early Childhood Development team have piloted a hybrid clinical-screening approach in seven countries since 2023, but the work is methodological rather than ready for national rollout.

People in conflict-affected and displacement settings

UNHCR’s 2024 mid-year statistical release counts more than 117 million forcibly displaced people worldwide. On the limited evidence available, disability prevalence in displaced populations runs substantially higher than in settled populations of the same origin — a function of injury, deprivation, and the differential mobility of disabled people in crisis. Almost no NSO’s sampling frame covers displaced populations adequately. The Humanitarian Inclusion Standards working group has been pressing UNHCR and IOM since 2022 to embed WG-SS in registration intake, with patchy results.

The instruments exist. The question, as in so much of disability policy, is whether the data will be collected.

06 · The double-counting debate, in four countries

The single most contested operational question in disability statistics in 2026 is also the most technical-sounding: when you ask the six WG-SS questions, and a respondent reports difficulty in more than one domain, do you count them once at the highest severity, once at any severity above the threshold, or do you sum the severity scores? National statistical offices treat this differently — and the difference moves the headline number by several percentage points.

Bangladesh’s 2022 census reported 2.8% using a “lot of difficulty / cannot do at all” threshold and a highest-domain rule — the conservative end of the WG-SS spectrum. South Africa’s 2022 census used the same instrument with a more inclusive threshold (some difficulty in two or more domains, or a lot in one) and reported 6.0%. Brazil’s IBGE Census applied a moderately inclusive threshold and reported 8.9%. Mexico’s INEGI captures any reported “some difficulty” in any domain alongside the more restrictive tier, and reports 16.5% — close to the WHO global figure, far higher than its South Asian neighbours using ostensibly the same instrument.

None of these offices is wrong. Each has defended its choice in publicly available methodology and each has reasons rooted in previous-cycle data continuity. The Washington Group’s 2024 guidance, for the first time, recommends reporting at least two thresholds in parallel — a “narrow” count at the highest severity and a “broad” count including some-difficulty responses — precisely so the cross-country comparison becomes possible.

07 · SDG-disaggregated reporting and the 2025 Compendium

The UN’s Sustainable Development Goal Indicator 17.18.1 tracks the share of indicators in the SDG framework that are disaggregated by, among other things, disability. The UN DESA Disability Statistics Compendium first published in 2018, with major updates in 2022 and most recently 2025, is the field’s single best inventory of which countries are producing what.

The 2025 Compendium’s headline finding is that around 120 countries now produce at least some disability-disaggregated SDG indicator reporting — up from 76 in the 2018 baseline. Depth is variable: most report disability-disaggregated education-attainment data (SDG 4.5.1) and labour-force-participation data (SDG 8.5), far fewer report disability-disaggregated maternal-health, justice-system, or political-participation data, and almost none report disability-disaggregated climate-resilience indicators. The 2025 Sendai Framework mid-term review called this gap out: disaster-risk-reduction reporting under SDG 11.5 and 13.1 remains very thin on disability disaggregation, despite Sendai’s own 2015 commitment.

08 · What changed in 2025–26 and what 2026 still misses

Three concrete developments are reshaping the 2026 measurement landscape. First, the WHO 2024 monitoring update reset the headline to 1.3 billion and 16% and made functional limitation the primary frame; reporting in any other frame now requires extra justification. Second, the UN DESA 2025 Compendium published the first country-by-country matrix of disability-disaggregated SDG indicator reporting — making gap-mapping straightforward for donors and civil society in a way it was not before. Third, the Washington Group Secretariat’s 2024 inventory has begun publishing not just whether a country uses WG-SS, but how — which threshold it applies, whether it deploys the Extended Set or Enhanced module, and whether its micro-data is publicly available for re-analysis. The 2025 update added nine new countries, including the first publicly funded micro-data release from a major South Asian NSO.

Three structural blind spots are unlikely to close on the current trajectory.

• Comparable institutional-population data. The European Commission’s 2023 mapping is the closest any region has come; no comparable global dataset exists. Without it, the 16% headline misses tens of millions of people whose prevalence is the highest of any sub-population the field counts.
• Comparable psychosocial-disability prevalence. The Extended Set and WG-SS Enhanced have the questions; deployment runs at a fraction of the WG-SS user base. Until that catches up, the global figure systematically undercounts the population whose access barriers most plainly require non-physical accommodations.
• Disability-disaggregated climate and disaster data. The Sendai Framework’s 2015 commitment to disability-disaggregated DRR reporting is, in 2026, almost entirely unmet outside a handful of pilot countries. As climate-driven disaster frequency continues to rise, this is becoming the single largest gap in the SDG-disaggregation matrix.

The through line

Two decades after the Washington Group published the Short Set, global disability prevalence is finally being counted in a way that produces defensible, broadly comparable numbers — 16%, around 1.3 billion people, with around 190 million adults experiencing very significant difficulties. The 2024 WHO update and the 2025 UN DESA Compendium have aligned around that frame. What remains is the long tail of populations the headline misses: people in institutions, children under 2, displaced populations, and people with intellectual or psychosocial disabilities whom the six functional-domain questions only partly capture. Closing those gaps is a national-statistics-office investment decision, not a research problem. The instruments exist. The question, as in so much of disability policy, is whether the data will be collected.

Read more from Disability World on the CRPD, on national regulations, and on the wider 2026 reporting record.