Global, regional, and national burden of disorders affecting the nervous system, 1990–2021: a systematic analysis for the Global Burden of Disease Study 2021

This is not an intervention study. It is a systematic epidemiological analysis that quantified the global health burden of 37 distinct disorders affecting the nervous system. The researchers tested nothing in the experimental sense. Instead, they estimated five key metrics for each condition across 204 countries from 1990 to 2021:

**Mortality** (deaths per year)

**Prevalence** (total number of people living with the condition)

**Years Lived with Disability (YLDs)** — a measure of non-fatal health loss

**Years of Life Lost (YLLs)** — a measure of premature death

**Disability-Adjusted Life-Years (DALYs)** — the sum of YLLs and YLDs, representing total health loss

The conditions included classic neurological diseases (stroke, Alzheimer's, epilepsy, migraine, meningitis, Parkinson's, multiple sclerosis, brain cancer) plus neurodevelopmental disorders (autism spectrum disorder, attention-deficit/hyperactivity disorder, intellectual disability) and neurological complications from other diseases (diabetic neuropathy, cognitive impairment from COVID-19, neurological damage from preterm birth, neonatal jaundice, malaria, syphilis, Zika virus, and congenital conditions).

The key innovation was a "sequela-level analysis" — meaning they only counted cases where actual nervous system damage occurred, rather than counting every case of the parent disease. For example, they did not count all COVID-19 cases, only those with documented cognitive impairment or other neurological sequelae.

The entire global population from 1990 to 2021, stratified by:

**204 countries and territories** (all UN member states plus territories with >50,000 population)

**Age groups**: 0–6 days (neonatal), 7–27 days, 28–364 days, 1–4 years, then 5-year age bands up to 95+ years

**Sex**: male and female separately

**Time**: annual estimates from 1990 through 2021

The study did not recruit individual participants. It synthesised data from thousands of sources: vital registration systems, verbal autopsy studies, hospital discharge records, health surveys, disease registries, epidemiological studies, and published literature. The total number of data sources used is not explicitly stated in the abstract, but the Global Burden of Disease (GBD) study typically draws on 100,000+ data sources across all conditions.

The researchers used a standardised epidemiological modelling framework called the GBD Comparative Risk Assessment. Key instruments and metrics included:

**DALYs** (0 to infinity, higher = worse): The primary summary metric. One DALY equals one year of healthy life lost. Calculated as YLLs + YLDs.

**YLLs**: Calculated by multiplying deaths at each age by the remaining standard life expectancy at that age (based on a reference life table with life expectancy at birth of 86.6 years).

**YLDs**: Calculated as prevalence × disability weight. Disability weights are values between 0 (perfect health) and 1 (death), derived from population surveys where people rank health states. For example, "severe Alzheimer's disease" has a disability weight of 0.449, while "mild migraine" has a weight of 0.030.

**Prevalence**: The total number of cases alive in a given year, estimated using DisMod-MR 2.1, a Bayesian meta-regression tool that synthesises prevalence, incidence, remission, and mortality data.

**Age-standardised rates**: Rates adjusted to a world standard population to allow comparison across time and between countries with different age structures.

**95% Uncertainty Intervals (UIs)**: Every estimate includes a range reflecting uncertainty from all sources — sampling error, measurement error, and modelling assumptions.

For the sequela-level analysis, the team identified specific health outcomes (sequelae) that represent nervous system damage. For diabetic neuropathy, they only counted cases with confirmed peripheral nerve damage. For COVID-19, they only counted cases with post-acute cognitive impairment or Guillain-Barré syndrome. They then recalculated YLDs for just those neurological sequelae, excluding the non-neurological burden of the parent disease.

**Study design:** This is a systematic analysis of epidemiological data — a descriptive epidemiological study, not an experiment. It is part of the Global Burden of Disease Study 2021, which uses standardised methods to estimate disease burden across all conditions.

**Data sources:** The team used a "multiple cause of death" approach for mortality data, drawing on vital registration systems (where available), verbal autopsy studies (for countries without complete death registration), and published literature. For non-fatal health loss, they used systematic reviews of prevalence studies, hospital discharge data, health surveys, and disease registries. Each data source was assessed for quality and bias.

**Modelling approach:**

For each condition, a Bayesian meta-regression model (DisMod-MR 2.1) estimated prevalence, incidence, remission, and mortality simultaneously, using all available data and borrowing strength across countries and regions.

For causes of death, the team used the GBD Cause of Death Ensemble model (CODEm), which tests multiple modelling strategies and selects the best-performing one.

For risk factors (e.g., high blood pressure for stroke, smoking for Alzheimer's), they used the GBD Comparative Risk Assessment framework, which estimates the burden attributable to each risk factor using population-attributable fractions.

**Comorbidity correction:** Because many people have more than one neurological condition (e.g., someone with both stroke and dementia), the team used a comorbidity correction to avoid double-counting. They simulated 1000 hypothetical individuals for each age-sex-country-year group, assigned them conditions based on prevalence, and then calculated the total DALYs by summing disability weights across conditions using a multiplicative model (since disability weights are not additive — having two conditions is worse than one but not simply twice as bad).

**What this design can prove:**

It can estimate the absolute and relative burden of nervous system disorders globally, regionally, and nationally.

It can track trends over time (1990–2021).

It can compare burden across age groups, sexes, and countries.

It can identify which conditions contribute most to health loss.

**What this design cannot prove:**

It cannot establish causation between risk factors and outcomes (though it can estimate attributable burden).

It cannot tell us about individual-level trajectories or treatment responses.

It cannot capture all aspects of disease burden — for example, it does not measure caregiver burden, economic costs, or quality of life beyond the disability weight framework.

It cannot detect conditions that are poorly diagnosed or underreported, especially in low-income countries with limited healthcare infrastructure.

It cannot tell us about the effectiveness of specific interventions — that requires randomised controlled trials.

**Major methodological weaknesses:**

Data quality varies enormously across countries. For many low-income countries, prevalence estimates are based on sparse data and statistical modelling rather than direct measurement.

Disability weights are derived from population surveys in a limited number of countries and may not reflect how different cultures value different health states.

The sequela-level analysis, while innovative, depends on assumptions about what proportion of cases have neurological damage — these proportions are often poorly known.

The study cannot account for undiagnosed cases, especially for conditions like migraine or dementia where many people never seek medical care.

The comorbidity correction is a statistical model, not a direct measurement, and may misrepresent how different conditions interact in real patients.

**Overall burden (2021):**

3.40 billion people (95% UI: 3.20–3.62 billion) had at least one nervous system disorder — 43.1% (40.5–45.9%) of the global population.

Total DALYs: 443 million (378–521 million) — the leading cause of disease burden worldwide, surpassing cardiovascular diseases and cancers.

Total deaths attributed to nervous system disorders: not explicitly stated in the abstract, but the age-standardised death rate decreased by 33.6% (27.6–38.8%) from 1990 to 2021.

**Trends over time (1990–2021):**

Global DALY counts increased by 18.2% (8.7–26.7%) — driven primarily by population growth and ageing.

Age-standardised DALY rates decreased by 27.0% (21.5–32.4%) — meaning the burden per person, adjusted for age, has fallen.

Age-standardised death rates decreased by 33.6% (27.6–38.8%).

Age-standardised prevalence was nearly stable: +1.5% (0.7–2.4%) — meaning the proportion of the population affected has barely changed.

**Top 10 conditions by age-standardised DALYs (2021):**

1. Stroke

2. Neonatal encephalopathy (brain damage in newborns)

3. Migraine

4. Alzheimer's disease and other dementias

5. Diabetic neuropathy

6. Meningitis

7. Epilepsy

8. Neurological complications due to preterm birth

9. Autism spectrum disorder

10. Nervous system cancer

**Key patterns by age and sex:**

Neonatal encephalopathy and preterm birth complications dominate in early life.

Migraine peaks in working-age adults (20–50 years) and is more common in women.

Alzheimer's and other dementias dominate in older age (60+ years).

Stroke burden is higher in men than women at younger ages, but the gap narrows with age.

Diabetic neuropathy is increasing rapidly, driven by the global diabetes epidemic.

**Regional variation:**

Stroke burden is highest in East Asia, Southeast Asia, and Eastern Europe.

Neonatal encephalopathy burden is highest in sub-Saharan Africa and South Asia.

Alzheimer's burden is highest in high-income countries (due to population ageing) but is rising fastest in low- and middle-income countries.

Migraine burden is relatively consistent across regions, though diagnosis rates vary.

The most striking finding is the sheer scale: **nervous system disorders now affect 43% of the world's population** — roughly 3.4 billion people. To put this in perspective, if you gathered everyone on Earth, nearly half would have at least one neurological condition at any given moment.

The **443 million DALYs** represent 443 million years of healthy life lost in a single year. That is equivalent to every person on Earth losing about 5.6 weeks of healthy life annually due to nervous system disorders alone.

The **18.2% increase in total DALYs** since 1990 means the absolute burden has grown substantially, even though the per-person rate has fallen by 27%. This is a classic demographic pattern: we are getting better at preventing deaths (age-standardised death rates down 33.6%), but population growth and ageing mean more people are living with neurological conditions for longer.

The **near-stable age-standardised prevalence** (+1.5%) is important: it means the proportion of people affected has barely changed in 30 years. We are not seeing a massive increase in neurological disease per person — we are seeing more people overall, and they are living longer with these conditions.

The **top 10 list** reveals a mix of preventable conditions (stroke, neonatal encephalopathy, meningitis), chronic conditions (migraine, Alzheimer's, epilepsy), and emerging threats (diabetic neuropathy, autism spectrum disorder). Stroke alone accounts for a huge share of the burden, driven by hypertension, smoking, and dietary risks.

**Acknowledged by the authors:**

Data gaps in low- and middle-income countries, where many neurological conditions are underdiagnosed and underreported.

The sequela-level analysis is novel and requires validation — the proportion of cases with neurological damage is uncertain for many conditions.

Disability weights may not capture the full impact of neurological conditions, especially cognitive and psychiatric symptoms.

The study does not include all conditions that affect the nervous system (e.g., chronic pain syndromes, some mental health disorders).

The comorbidity correction is a statistical model and may not reflect real-world interactions between conditions.

**Critical reader notes:**

The study is funded by the Bill & Melinda Gates Foundation, which has a strong interest in global health priorities. While this does not invalidate the findings, it is worth noting that the framing (nervous system disorders as the "leading cause of burden") supports arguments for increased funding in this area.

The GBD study uses a specific definition of "health loss" that may not align with how individuals experience their conditions. For example, mild migraine (disability weight 0.030) is treated as a much smaller burden than severe stroke (disability weight 0.552), but a person with daily migraines might disagree.

The study cannot capture the full societal impact — caregiver burden, lost productivity, and economic costs are not included in DALYs.

The COVID-19 neurological sequelae estimates are preliminary and based on limited data from the first two years of the pandemic.

The study does not distinguish between different severities of the same condition in a granular way — "Alzheimer's disease" includes everything from mild cognitive impairment to end-stage dementia.

For someone running their own n=1 experiment, this paper is not directly actionable in the way an intervention trial would be. However, it provides crucial context for any self-experiment related to brain health, cognition, or neurological function.

**What to test:**

Given that stroke is the #1 cause of neurological burden, and hypertension is the leading risk factor, any intervention that lowers blood pressure is worth testing. This includes dietary changes (reducing sodium, increasing potassium), exercise (aerobic and resistance), stress reduction (meditation, biofeedback), and supplements (magnesium, CoQ10, beetroot juice).

Given that diabetic neuropathy is #5 and rising, any intervention that improves insulin sensitivity or blood sugar control is relevant. This includes time-restricted eating, low-glycaemic-index diets, berberine, cinnamon, alpha-lipoic acid, and exercise.

Given that migraine is #3 and affects working-age adults, interventions for migraine prevention are high-impact. This includes magnesium glycinate (400–600 mg/day), riboflavin (400 mg/day), CoQ10 (150–300 mg/day), feverfew, butterbur, and trigger elimination diets.

Given that Alzheimer's is #4 and rising with population ageing, interventions for cognitive maintenance are worth testing. This includes aerobic exercise (150+ minutes/week), Mediterranean diet, cognitive training, sleep optimisation, and supplements like omega-3s (1–2 g/day EPA+DHA), curcumin, and vitamin D.

**Minimum meaningful duration:**

For blood pressure interventions: 4–8 weeks to see a stable change in resting blood pressure (measured at the same time of day, after 5 minutes of seated rest).

For blood sugar interventions: 2–4 weeks for fasting glucose and HbA1c (though HbA1c reflects the previous 2–3 months).

For migraine prevention: 8–12 weeks minimum — migraine frequency often takes 2–3 months to stabilise on a new intervention.

For cognitive interventions: 12–24 weeks — cognitive changes are slow and require consistent practice.

**What to measure:**

**Primary metric:** Your personal "neurological burden score" — a composite of symptoms relevant to you. For example, if you have migraines, track frequency (days per month), duration (hours), and severity (1–10 scale). If you are concerned about cognition, track processing speed (e.g., reaction time test), working memory (e.g., digit span test), and subjective cognitive complaints.

**Secondary metrics:** Blood pressure (systolic and diastolic, morning and evening), fasting blood glucose, HbA1c (if available), sleep quality (Pittsburgh Sleep Quality Index or