Global epidemiological and genetic burden of rare neurological diseases in children

Summary

We assessed the global burden of rare childhood neurological diseases from 1990 to 2021 and projected trends to 2050 using Global Burden of Disease data, alongside an analysis of the genetic burden from ClinVar. Four neurological conditions were classified as rare childhood neurological diseases: motor neuron disease, multiple sclerosis, neuroblastoma and other peripheral nervous cell tumors, and other neurological disorders. From 1990 to 2021, the global age-standardized prevalence and disability-adjusted life-year rates of rare childhood neurological diseases increased, with higher burdens among boys and in low-middle sociodemographic index countries. Our customized forecasting model projected that 110.2 thousand children would be affected by 2050, representing an 11.83% increase from 2021. Annual and cumulative increases in newly reported disease-causing genes and pathogenic/likely pathogenic variants associated with monogenic neurological diseases were observed from 1990 to 2024. The increasing and disparate burden of rare childhood neurological diseases highlights the need for targeted prevention.

Graphical abstract

Highlights

• •Global burden of rare childhood neurological diseases increased from 1990 to 2021
• •Our forecasting model projects 110.2 thousand children affected by 2050
• •Higher burden observed in boys and low-middle-SDI countries
• •Disease-causing genes and pathogenic variants increased from 1990 to 2024

Health sciences; Medicine; Medical specialty; Neurology; Public health

Introduction

Rare diseases represent a heterogeneous group of diverse disorders, each affecting a small proportion of the global population—fewer than 1 in 2000 individuals in any World Health Organization region—but collectively exerting a substantial global impact, with more than 7,000 distinct types currently recognized.^1,2 In recent years, rare diseases have garnered increasing attention from governments, researchers, and international organizations due to their cumulative health and socioeconomic burden. With advances in modern genetic diagnostics,³ recent estimates suggest that the total number of distinct rare diseases may exceed 10,000, collectively affecting approximately 3.5%–5.9% of the global population, equivalent to about 263–446 million individuals worldwide.⁴ Nearly half of all rare diseases are neurological in naturel.⁵ These rare neurological diseases encompass a wide spectrum of disorders, including congenital and neurodevelopmental diseases, neurogenetic and neuroimmune diseases, cerebrovascular and neurodegenerative disorders, neuromuscular diseases, peripheral nervous system disorders, communicable neurological diseases, and cancers of the nervous system.⁶ Importantly, around 90% of rare childhood diseases involve major neurological manifestations.⁵ Approximately 80% of all rare diseases are genetic in origin, nearly 70% of which are present during childhood, and about 30% of affected children die before the age of five.² Consequently, rare neurological diseases in children represent a substantial proportion of the global rare disease burden.

Rare childhood neurological diseases are typically congenital, genetic, and life-threatening, characterized by high morbidity and mortality and a severe lack of effective or approved therapies.² Their diagnosis and management remain particularly challenging due to limited access to specialized screening, advanced genetic testing, and expert clinical care, especially in resource-limited regions.⁷ The Global Burden of Diseases, Injuries, and Risk Factors (GBD) Study provides a systematic framework for evaluating disease impact across populations and over time. The GBD 2021,^6,8 incorporates additional peripheral neuropathies, neurodevelopmental conditions, and other rare neurological disorders that were not fully captured in previous GBD versions. This expansion enables a more comprehensive assessment of the global burden and temporal trends of both common and rare neurological diseases, including incidence, prevalence, mortality, and disability-adjusted life-years (DALYs). Notably, GBD 2021 reported that the 37 neurological conditions analyzed collectively constitute the leading cause of overall disease burden worldwide.⁶ Despite these advances, the specific global, regional, and national burden of rare neurological diseases in children remains poorly defined. Previous GBD analyses have primarily focused on common neurological diseases^9,10 or aggregated data without age-specific stratification,^6,11,12,13 obscuring the unique epidemiological patterns of childhood rare neurological diseases.

To address these gaps, we utilized GBD 2021 estimates to quantify the global, regional, and national burden of rare neurological diseases in children aged 0–14 years from 1990 to 2021. We calculated age-standardized prevalence, mortality, and DALYs rates, stratified by age, sex, and sociodemographic index (SDI) quintiles. Temporal trends were assessed using the average annual percentage change (AAPC).¹⁴ Future prevalence projections for 2022–2050 were generated using a modified ensemble forecasting approach based on that of Su et al.,¹⁰ combining Poisson regression with an autoregressive integrated moving average (ARIMA) model (Figure 1). This study provides the first global assessment of the epidemiological burden and projected trends of rare neurological diseases in children. Additionally, we present the evaluation of the genetic burden of rare monogenic neurological diseases using ClinVar. The findings offer essential epidemiological and genetic evidence to inform clinical decision-making, guide research priorities, and support health policy development for this highly vulnerable population in the coming decades.

Figure 1.

Framework of the customized forecasting model

Step 1: for each location l, age group a, sex s, and year y, Poisson regressions were fitted using either SDI alone or SDI together with its quadratic term. Y_l,a,s,y represented the number of rare neurological disease cases, β₀ was the location-age-sex-specific intercept, β_a was the fixed coefficient for each age group, β₁ and β₂ were the fixed coefficients associated with SDI_l,y and its quadratic term, respectively. Ridge regularization was applied to β₁ and β₂, and log(population_l,a,s,y) served as the offset term. The inclusion of the quadratic term was determined based on 5-fold cross-validated RMSE. Step 2: the residual series {e_l,a,s,y} was modeled using an ARIMA process to obtain the forecasted residuals ${\hat{e}}_{l,a,s,y}$. The adjusted predictions, denoted as ${\hat{Y}}_{l,a,s,y}^{\left(2\right)}$, were then generated by adding the forecasted residuals to the initial Poisson regression predictions ${\hat{Y}}_{l,a,s,y}^{\left(1\right)}$. Step 3: the final prediction ${\hat{Y}}_{RMSE}$ was derived through RMSE-weighted model averaging, with weights w_m calculated as the inverse of each model’s RMSE. ARIMA, autoregressive integrated moving average; RMSE, root mean squared error; SDI, sociodemographic index.

Results

Global trends of rare neurological diseases in children

Globally, the number of prevalent cases of combined rare neurological diseases in children aged 0–14 years increased by 21.98%, from 80,784.65 cases in 1990 to 98,540.90 cases in 2021 (Table 1). The corresponding age-standardized prevalence rate (ASPR) increased by 7.17%, from 4.60 (95% uncertainty interval [UI], 3.93–5.26) per 100,000 population in 1990 to 4.93 (95% UI, 4.09–5.77) per 100,000 population in 2021, with an AAPC of 0.21% (95% confidence interval [CI], 0.17%–0.24%) (Figure 2A). Over the same period, the global DALYs increased by 30.46%, from 1.01 million to 1.32 million. The corresponding age-standardized DALY rate (ASDR) increased by 14.92%, from 57.23 (95% UI, 50.61–63.85) per 100,000 population in 1990 to 65.77 (95% UI, 54.85–76.69) per 100,000 population in 2021, with an AAPC of 0.45% (95% CI, 0.39%–0.50%) (Figure 2B; Table S1). In contrast, the global age-standardized mortality rate (ASMR) decreased by 6.98%, from 0.43 (95% UI, 0.40–0.47) per 100,000 population in 1990 to 0.40 (95% UI, 0.36–0.44) per 100,000 population in 2021, with an AAPC of −0.28% (95% CI, −0.34% to −0.24%) (Figure 2C; Table S2).

Table 1.

Age-standardized prevalence rate and AAPC of combined rare neurological diseases in children globally and by SDI fifths, 1990–2021

	Prevalence (95% UI)
	Absolute number in 1990	Age-standardized rate in 1990 (per 100,000)	Absolute number in 2021	Age-standardized rate in 2021 (per 100,000)	AAPC (95% CI)
Global	80784.65 (65499.13–96070.18)	4.60 (3.93–5.26)	98540.90 (77771.35–119310.45)	4.93 (4.09–5.77)	0.21 (0.17–0.24)
Sex:
Girls	37296.51 (28911.89–45681.13)	4.37 (3.63–5.12)	43364.30 (33590.55–53138.05)	4.48 (3.69–5.26)	0.05 (−0.01 to 0.10)
Boys	43488.14 (35695.53–51280.75)	4.81 (4.15–5.48)	55176.60 (41899.07–68454.13)	5.36 (4.29–6.43)	0.33 (0.30–0.36)
Age groupa:
<5 years	38436.59 (28209.43–48663.75)	6.20 (4.55–7.85)	45288.04 (30665.09–59910.99)	6.88 (4.66–9.10)	0.31 (0.25–0.36)
5–9 years	22635.78 (18380.14–26891.43)	3.88 (3.15–4.61)	28370.28 (22909.04–33831.52)	4.13 (3.33–4.92)	0.19 (0.16–0.22)
10–14 years	19712.28 (15310.34–24114.23)	3.68 (2.86–4.50)	24882.58 (19476.99–30288.18)	3.73 (2.92–4.54)	0.05 (0.03–0.06)
SDI:
High SDI	19268.72 (16875.22–21662.21)	10.40 (9.48–11.32)	16027.54 (14109.34–17945.75)	9.33 (8.53–10.12)	−0.37 (−0.44 to −0.32)
High-middle SDI	15434.48 (12859.89–18009.06)	5.64 (4.94–6.33)	13962.74 (11592.50–16332.97)	6.11 (5.35–6.87)	0.21 (0.13–0.26)
Middle SDI	22443.44 (17863.88–27023.00)	3.87 (3.27–4.47)	25982.93 (20603.01–31362.86)	4.65 (3.93–5.38)	0.58 (0.53–0.62)
Low-middle SDI	16500.29 (12237.18–20763.41)	3.43 (2.75–4.12)	25972.83 (18835.83–33109.83)	4.51 (3.51–5.51)	0.87 (0.83–0.91)
Low SDI	7059.58 (4579.01–9540.16)	2.95 (2.12–3.78)	16517.07 (10060.64–22973.49)	3.51 (2.34–4.69)	0.57 (0.52–0.61)
Combined group:
Motor neuron disease	26903.76 (20414.36–34670.00)	1.57 (1.29–1.84)	29854.32 (22409.92–39148.87)	1.47 (1.21–1.74)	−0.18 (−0.20 to −0.17)
Multiple sclerosis	4195.76 (2218.07–6438.85)	0.25 (0.14–0.36)	5275.62 (2920.61–7879.78)	0.26 (0.15–0.36)	0.05 (0.05–0.05)
NPNCT	39661.46 (29113.41–53743.48)	2.20 (1.64–2.76)	51762.40 (34704.53–70435.21)	2.62 (1.86–3.38)	0.54 (0.46–0.59)
Other neurological disorders	10023.68 (5554.65–15894.44)	0.58 (0.37–0.78)	11648.55 (6431.54–18432.75)	0.58 (0.37–0.78)	0.003 (−0.001 to 0.006)

AAPC, average annual percentage change; CI, confidence interval; NPNCT, neuroblastoma and other peripheral nervous cell tumor; SDI, sociodemographic index; UI, uncertainty interval.

^anot age-standardized.

Figure 2.

Global trends of combined and individual rare neurological disease in children at global level, 1990–2021

(A–C) AAPCs of ASPR (A), ASDR (B), and ASMR (C) for combined rare neurological diseases and each individual disease.

(D) Prevalence number and ASPR of combined rare neurological diseases by sex.

(E–G) AAPCs of ASPR (E), ASDR (F), and ASMR (G) by sex for the four individual rare neurological diseases.

Error bars in (A–C) and (E–G) represent 95% CIs of the AAPCs. In (D), error bars represent 95% UIs of prevalence numbers, and shaded areas represent 95% UIs of ASPRs. AAPC, average annual percentage change; ASPR, age-standardized prevalence rate; ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; DALYs, disability-adjusted life-years; CI, confidence interval; UI, uncertainty interval; MND, motor neuron disease; MS, multiple sclerosis; NPNCT, neuroblastoma and other peripheral nervous cell tumor; ONDs, other neurological disorders.

Among the four individual rare neurological diseases in 2021, neuroblastoma and other peripheral nervous cell tumor (NPNCT) had the highest ASPR at 2.62 (95% UI, 1.86–3.38) per 100,000 population, accompanied by the greatest increase in AAPC (0.54%) from 1990 to 2021. Multiple sclerosis showed a modest but statistically significant rise in ASPR, while other neurological disorders demonstrated a non-significant upward trend. Conversely, the ASPR for motor neuron disease declined by 6.37%. Regarding ASDR, other neurological disorders and NPNCT were the primary contributors to the global disease burden of rare neurological diseases in 2021, both showing sustained increases since 1990, with AAPCs of 0.73% and 0.26%, respectively. In contrast, ASDRs for motor neuron disease and multiple sclerosis decreased, with AAPCs of −2.48% and −0.30%, respectively. For ASMR, only NPNCT showed an upward trend (AAPC, 0.26%), while the other three rare neurological diseases exhibited a declining trend from 1990 to 2021.

Global trends by sex

Between 1990 and 2021, boys consistently exhibited higher prevalence, DALYs, and mortality for combined rare neurological diseases than girls worldwide (Figure 2D). Over this period, the global ASPR and ASDR increased in both sexes, with larger increases in boys, whereas the global ASMR declined in both sexes, with a greater reduction among girls (Tables 1, S1, and S2).

Motor neuron disease exhibited significant declines across all three age-standardized rates (ASRs) in both sexes, with steeper reductions in girls (Figures 2E–2G). Multiple sclerosis (AAPC, 0.08% in girls and 0.02% in boys) and NPNCT (AAPC, 0.27% in girls and 0.74% in boys) both showed statistically significant increases in ASPR (p < 0.05); however, the minimal annual change observed for multiple sclerosis may have limited epidemiological significance. For ASDR, other neurological disorders rose in both sexes (AAPC, 0.80% in girls and 0.67% in boys), while NPNCT increased only in boys. ASMR for other neurological disorders increased slightly in girls (AAPC, 0.11%), and NPNCT displayed an upward ASMR trend in boys (AAPC, 0.46%).

Global trends by age group

In 2021, children younger than 5 years exhibited the highest prevalence, DALYs rate, and mortality rate for combined rare neurological diseases, at 6.88, 69.31, and 0.59 per 100,000 population, respectively (Tables 1, S1, and S2). From 1990 to 2021, the prevalence rate increased across all age groups, with more substantial rises observed in children under 5 years of age. In contrast, the rates of DALYs and mortality declined among children under 5 years in both sexes but increased among those aged 5–14 years.

Among the four rare neurological diseases, motor neuron disease showed consistent declines in prevalence, DALYs, and mortality rates across all age groups and both sexes (Figures S1 and S2). Prevalence rates of both multiple sclerosis and NPNCT increased across all age groups, especially among those aged 10–14 years, with AAPCs of 0.04% and 1.25%, respectively. However, the minimal annual change observed for multiple sclerosis suggests limited epidemiological relevance despite statistical significance. For DALYs and mortality rates, NPNCT increased across all age groups and both sexes, except for a slight decline in girls under 5 years of age. Other neurological disorders showed rising DALYs and mortality rates among children aged 5–14 years in both sexes, while multiple sclerosis increased only among girls aged 10–14 years.

Global trends by SDI

In 2021, high-SDI countries showed the highest ASPR, ASDR, and ASMR for combined rare neurological diseases in children, at 9.33, 80.20, and 0.62 per 100,000 population, respectively (Tables 1, S1, and S2). However, all three ASRs showed persistent declines from 1990 to 2021. In contrast, low-middle-SDI countries experienced rising ASPR, ASDR, and ASMR over the same period, with sharper increases observed in boys (Figure S3). From 1990 to 2021, SDI values were strongly correlated with ASPR (r = 0.91, p < 0.001) for combined rare neurological diseases in children.

Although the four rare neurological diseases showed heterogeneous patterns across SDI levels, all three ASRs generally increased in middle-, low-middle-, and low-SDI countries (Figures 3A and S4). The ASPR of multiple sclerosis rose across all SDI levels and both sexes, with the sharpest increase observed in high-middle SDI countries. NPNCT showed rising ASPR across most SDI levels and sexes, except in high-SDI regions. For ASDR and ASMR, multiple sclerosis, NPNCT, and other neurological disorders all increased across both sexes in low-, low-middle-, and middle-SDI regions, with the most rapid growth observed in the low- and low-middle-SDI strata.

Figure 3.

Global trends in rare childhood neurological diseases by SDI quintile, region, and country, and decomposition analysis of prevalent cases from 1990 to 2021

(A) Global trends in ASPR, ASDR, and ASMR of individual rare neurological disease in children by SDI fifths and sex, 1990–2021.

(B) Regional AAPCs in ASPR, ASDR, and ASMR for individual rare neurological diseases in children across 21 global regions, 1990–2021.

(C) AAPCs of ASPR for combined rare neurological diseases in 204 countries and territories, 1990–2021.

(D) Decomposition of change in number of prevalent cases of rare neurological diseases between 1990 and 2021 globally and by SDI fifths.

AAPC, average annual percentage change; ASPR, age-standardized prevalence rate; ASDR, age-standardized DALY rate; ASMR, age-standardized mortality rate; DALYs, disability-adjusted life-years; NPNCT, neuroblastoma and other peripheral nervous cell tumor; SDI, sociodemographic index.

Regional trends of rare neurological diseases in children

From 1990 to 2021, 11 of 21 global regions exhibited increases in both prevalence and DALYs for combined rare neurological diseases in children, while mortality rose in 9 regions during the same period (Tables S3–S5). The most pronounced increase in ASPR was observed in South Asia (AAPC, 1.10%), whereas the largest rises in ASDR and ASMR occurred in Western sub-Saharan Africa.

In 2021, NPNCT was the predominant contributor to the prevalence of rare neurological diseases in children across most of the 21 regions, whereas other neurological disorders accounted for the majority of disability and mortality burdens globally (Figure S5). NPNCT showed particularly sharp ASPR increases in South Asia, Western sub-Saharan Africa, and the Caribbean (Figure 3B). Other neurological disorders demonstrated substantial increases in ASDR, primarily in Western sub-Saharan Africa, Oceania, and Southeast Asia.

National trends of rare neurological diseases in children

Between 1990 and 2021, rising ASPR was documented in 116 of 204 countries and territories, rising ASDR in 110, and rising ASMR in 77 for combined rare neurological diseases in children (Figures 3C and S6; Table S6). The steepest ASPR increase occurred in Guyana (AAPC, 4.23%). Tokelau experienced the largest rise in both ASDR and ASMR. In 2021, India reported the greatest number of prevalent cases (13,905.86), followed by China (10,146.92) and Pakistan (7,904.42).

Among the four specific diseases, multiple sclerosis showed the broadest global ASPR increase, with upward trends in 170 of 204 countries and territories, followed by other neurological disorders (164 countries) and NPNCT (120 countries). Increasing ASDR trends were detected in 94, 99, and 109 countries for NPNCT, multiple sclerosis, and other neurological disorders, respectively. Regarding ASMR, NPNCT displayed the widest increase, rising in 94 countries and territories.

Decomposition of changes in rare neurological disease prevalence

Decomposition analysis from 1990 to 2021 revealed that population growth and change in prevalence accounted for 73.33% and 35.28% of the global increase in combined rare neurological diseases cases among children aged 0–14 years, respectively (Figure 3D). In low- and low-middle-SDI countries, the rise in prevalent cases was primarily driven by population growth and higher prevalence. Among the 13 GBD regions with increasing prevalence, population growth was the dominant factor in nine regions (Table S7). Among the ten countries with the highest numbers of childhood cases in 2021, change in prevalence was the leading driver in five countries (Figure S7).

Across the four individual rare neurological diseases, population growth was the principal driver of global case increases for motor neuron disease (140.08%), multiple sclerosis (63.56%), and other neurological disorders (96.94%). In contrast, the rise in NPNCT was jointly driven by population growth (54.75%) and change in prevalence (65.19%). Across SDI strata, population growth primarily explained increases in low- and low-middle-SDI countries, while change in prevalence was dominant in middle-SDI region (Figure 3D). Similarly, among the ten countries with the highest numbers of cases for each disease, population growth was the main contributor in four, five, and four countries for motor neuron disease, multiple sclerosis, and other neurological disorders, respectively. In comparison, change in prevalence was the predominant driver in six countries for NPNCT.

Future trends of rare neurological diseases from 2022 to 2050

Globally, the number of prevalent cases of combined rare neurological diseases in children was projected to reach 110.2 thousand (95% UI, 80.6–141.9) by 2050, representing an 11.83% increase from 2021 (Figure 4A; Table S8). The prevalence rate was expected to rise to 5.51 (95% UI, 4.45–6.58) per 100,000 population, with higher rates in boys than in girls (Figure 4B). The ASPR of combined rare neurological diseases in children was forecast to reach 5.62 (95% UI, 4.52–6.71) per 100,000 population in 2050, corresponding to an AAPC of 0.40% from 2022 (Figure 4C). Consistent with historical trends from 1990 to 2021, the ASPR was projected to rise more rapidly in boys than in girls. Across age groups, all were expected to experience increases, with the most pronounced rise in children under 5 years of age (AAPC, 0.59%) (Figure 4D).

Figure 4.

Projected numbers, rates, and age-standardized rates of prevalence of rare neurological diseases in children, 2022–2050

(A–C) Projected prevalence of combined rare neurological diseases in children by sex: (A) numbers of prevalent cases, (B) prevalence rates, and (C) age-standardized prevalence rates.

(D) Projected age-specific prevalence rates of combined rare neurological diseases in children.

(E–H) Projected age-standardized prevalence rates by sex for (E) NPNCT, (F) multiple sclerosis, (G) other neurological disorders, and (H) motor neuron disease.

(I–L) Projected numbers of prevalent cases by sex for (I) NPNCT, (J) multiple sclerosis, (K) other neurological disorders, and (L) motor neuron disease.

Shaded regions represent 95% uncertainty bands for historical estimates and projected trajectories. NPNCT, neuroblastoma and other peripheral nervous cell tumor; UI, uncertainty interval.

Among the eight global regions projected to exhibit increasing trends between 2022 and 2050, the most rapid rise in the ASPR of combined rare neurological diseases in children was forecast for Western sub-Saharan Africa, followed by Southern sub-Saharan Africa and South Asia (Figure S8). At the national level, 132 countries were expected to show increasing ASPR trends over this period (Figure S9). Model validation indicated strong predictive performance, with a root mean squared error (RMSE) of 922.55, a mean bias of 76.83, a relative RMSE of 0.92%, an mean absolute error (MAE) of 625.42, and a mean absolute percentage error (MAPE) of 0.62% (Table S9; Figure S10). A comparison with the original framework proposed by Su et al. further showed improved predictive performance of the customized forecasting model (Table S9).

We further projected the future prevalence for four individual rare neurological diseases in children at global and national levels (Figures 4E–4L; Table S8). The ASPRs in 2050 were forecast at 3.45 (95% UI, 2.47–4.46) for NPNCT, 0.26 (95% UI, 0.13–0.41) for multiple sclerosis, and 0.60 (95% UI, 0.37–0.83) for other neurological disorders, with corresponding AAPCs of 0.86%, 0.09%, and 0.13% from 2022. Relative to the 1990–2021 trends, projected ASPRs for NPNCT, multiple sclerosis, and other neurological disorders were expected to increase gradually from 2022 to 2050 across most countries and territories, particularly for NPNCT and other neurological disorders.

Genetic burden of rare neurological diseases in ClinVar

Between 1990 and 2024, ClinVar recorded 65,331 pathogenic or likely pathogenic variants in disease-causing genes associated with rare monogenic neurological diseases (Figures 5A and S11). Variant submissions increased markedly over time, with 1,633 variants reported in 1990–2010, 18,851 in 2010–2020, and 44,847 in 2021–2024, corresponding to 183, 255, and 55 newly reported disease-causing genes, respectively. Both the annual and cumulative numbers of newly reported disease-causing genes and pathogenic or likely pathogenic variants increased throughout the study period. This increasing trend in reported genetic evidence was observed alongside the increasing global epidemiological burden of childhood rare neurological diseases since 1990. Notably, the largest numbers of pathogenic or likely pathogenic variants and newly reported disease-causing genes were observed within the category of other neurological disorders, which is largely composed of inherited neuromuscular diseases (Figures 5B–5L).

Figure 5.

Global genetic evidence of rare monogenic neurological diseases, 1990–2024

(A–D) Cumulative numbers of pathogenic or likely pathogenic variants and disease-causing genes reported for (A) combined rare monogenic neurological diseases, (B) motor neuron diseases, (C) NPNCT, and (D) other neurological disorders.

(E–L) Genetic evidence for eight subtypes of other neurological disorders: (E) autonomic nervous system disorders, (F) congenital myasthenic syndromes, (G) movement and spinocerebellar disorders, (H) muscular dystrophies, (I) primary myopathies, (J) neurodegenerative diseases, (K) peripheral nerve disorders, and (L) myopathies due to systemic diseases.

NPNCT, neuroblastoma and other peripheral nervous cell tumor.

Discussion

Globally, both the prevalence and DALYs associated with rare neurological diseases in children increased markedly between 1990 and 2021, while mortality modestly declined. The rise in prevalence was evident across all pediatric age groups, with the most pronounced increase observed in children under 5 years of age. These findings are consistent with broader global trends in rare diseases.² Since 2010, the number of recognized rare diseases, particularly those of genetic origin, has steadily grown, with approximately 250–280 newly described disorders reported annually.¹⁵ This surge largely reflects major advances in diagnostic technologies, including newborn screening, next-generation sequencing, and single-molecule sequencing, coupled with expanded access to specialized clinical evaluation.^{3,12,16,17,18,19} As most pediatric rare neurological diseases are monogenic or Mendelian and often cause severe, progressive disabilities, these diagnostic improvements, together with prolonged patient survival, have naturally led to higher recorded prevalence and DALYs over the past three decades. Although true changes in disease occurrence cannot be determined using GBD data alone, these trends provide valuable insights into the population-level burden of rare neurological diseases. Similarly, we observed sustained increases in both the annual and cumulative numbers of newly reported disease-causing genes and pathogenic or likely pathogenic variants associated with rare monogenic neurological diseases. These increases primarily reflect advances in sequencing technologies, expanded genetic testing, and continued growth of the ClinVar database. Although these findings do not represent the true underlying genetic burden, they may partly reflect the accumulation of genetic evidence underlying the burden of rare monogenic neurological diseases. Concurrently, the number of targeted therapeutics for rare diseases has increased substantially.²⁰ These breakthroughs likely contribute to the observed and expected reductions in mortality from rare childhood neurological diseases.

Over the past three decades, we identified distinct burden trajectories across the four individual rare neurological conditions studied: increasing burdens for multiple sclerosis, NPNCT, and other neurological disorders, but a decreasing burden for motor neuron disease. The decline in pediatric motor neuron disease contrasts with a recent GBD report,¹¹ which indicated an increasing global burden. Motor neuron disease primarily affects adults aged 50–70 years, and thus, the rising global burden reported previously reflects aging populations rather than pediatric trends. In children, motor neuron diseases comprise a heterogeneous group of disorders, with spinal muscular atrophy being a major inherited form.^21,22 Advances in prenatal diagnosis, newborn screening, and disease-modifying therapies have changed the clinical management of affected children in some settings.²¹ These advances may have contributed to the declining burden observed in our analysis. However, the observed decline should be interpreted cautiously, given the considerable uncertainty associated with GBD estimates for rare pediatric motor neuron diseases. Further studies based on disease registries and population-based surveillance data are needed to better characterize the long-term trend. The rising trend for NPNCT align with previous GBD 2021 analyses,^23,24,25 which also reported an increasing trend in NPNCT burden. Most other neurological disorders documented in the GBD 2021 database correspond to neuromuscular diseases, such as autonomic nervous system disorders, muscular dystrophies, and myopathies.⁶ Over half of all neuromuscular diseases are monogenic, and advances in molecular genetics have led to the identification of over 600 monogenic neuromuscular disorders.²⁶ The continuing expansion of genetic diagnostic technologies worldwide is expected to further broaden this spectrum and increase detection rates. Many of these disorders manifest early in life,²⁷ such as Duchenne muscular dystrophy and severe forms of spinal muscular atrophy. Consequently, neuromuscular diseases are poised to remain major contributors to the rising global burden of childhood rare neurological diseases.

We found a consistently higher burden of rare neurological diseases in boys and in low-SDI countries. This finding aligns with GBD 2021 results, which reported higher overall burdens for 37 neurological conditions in males compared with females across the lifespan.⁶ Similarly, a regional GBD analysis of 13 common neurological diseases in Asia from 1990 to 2019 also found that males experienced a greater burden than females.⁹ Although those studies covered all age groups,⁹ our results confirm that this sex disparity emerges early in childhood. However, the observed sex differences should be interpreted with caution, as they may partly reflect disparities in healthcare access and diagnostic ascertainment between boys and girls, particularly in low-SDI countries, in addition to disease-specific biological differences. While high-SDI countries exhibited the highest overall burden of rare childhood neurological diseases in 2021, their burden has declined since 1990, potentially reflecting improvements in early diagnosis, multidisciplinary management, and supportive care. Conversely, middle-, low-middle-, and low-SDI countries experienced increasing burdens over the same period, with the rate of increase inversely correlated with SDI. This trend may partly reflect improvements in diagnostic capacity, changes in disease incidence, demographic changes, and healthcare inequalities, particularly in low- and low-middle-SDI countries. Neurological health loss remains disproportionately concentrated in low- and middle-income countries, partly due to the higher prevalence of perinatal complications and infectious diseases among neonates and children under five.^6,28 Furthermore, improvements in early-life survival have paradoxically led to greater long-term neurological disability.^12,29 The persistent inequality in healthcare infrastructure, exacerbated by uneven economic and regional development, further compounds this issue. For example, Asia accounts for roughly 60% of the global population but only 20% of the world’s neurologists.⁹ Such shortages of specialized personnel and services severely restrict access to timely diagnosis, treatment, and rehabilitation, contributing to poorer outcomes and sustained neurological burden in these regions.²⁸ Notably, estimates for several low-SDI countries were accompanied by wide UIs. This pattern may largely arise from limited primary data availability in these regions, driven by restricted access to specialized clinical services, limited diagnostic capacity, and incomplete disease surveillance.^28,30 Where primary data are limited, GBD estimates rely more heavily on modeling and extrapolation, which increases uncertainty and results in wider UIs.^31,32 These broad UIs may limit the precision of burden estimates, highlighting the need for strengthened disease surveillance and clinical reporting in low-SDI countries to improve epidemiological evidence.

A strong positive correlation was observed between SDI values and the ASPR of rare neurological diseases in children, suggesting that SDI may significantly influence the spatiotemporal prevalence patterns. Within the GBD non-fatal disease modeling framework, SDI serves as a key predictor in Poisson regression models, which have been successfully applied to forecast diseases such as Parkinson’s disease, dementia, rheumatoid arthritis, diabetes, and lower back pain.^{10,33,34,35,36} Accordingly, SDI was incorporated as a key covariate in our forecasting framework. Decomposition analysis further identified population growth as the major driver of increasing case numbers, prompting the inclusion of population size as an offset term in the Poisson regression. To account for residual temporal dependencies not captured by SDI and demographic covariates, we applied an ARIMA model to regression residuals and integrated outputs from four models using RMSE-weighted averaging. This ensemble approach enhances predictive accuracy and robustness for long-term projections of rare neurological diseases. Although our projections were based on historical trends and SDI, future advances in newborn screening, genomic medicine, and gene therapies may influence long-term projections of rare neurological diseases. Expanded newborn screening and broader access to genomic testing may improve early diagnosis and case ascertainment, potentially increasing observed incidence and prevalence over time. Advances in disease-modifying gene therapies may reduce mortality and disability while prolonging survival, with potentially divergent effects on prevalence, mortality, and DALYs. Furthermore, the availability and accessibility of these advances may vary across regions and SDI levels, potentially influencing future disparities in disease detection and outcomes. As these developments were not explicitly incorporated into the forecasting framework, actual future epidemiological patterns may differ from the projected estimates.

Limitations of the study

This study, however, has several limitations. First, as the estimates were derived partly from extrapolations based on countries with available epidemiological data, caution should be exercised when interpreting these findings in real-world contexts. Second, the aggregation of heterogeneous conditions may obscure disease-specific epidemiological patterns. “Other neurological disorders” is a residual GBD category comprising multiple neurological and neuromuscular conditions that are not separately modeled, thereby limiting interpretation of disease-specific patterns within this category. Similarly, we combined the four individual disorders to characterize the overall burden of rare childhood neurological diseases, which may obscure differences in prevalence, mortality, DALYs, and temporal trends across diseases. Third, the aggregation of disease-specific estimates and uncertainty estimation assumed independence among diseases, without accounting for potential comorbidities or inter-disease dependence. However, because most rare neurological diseases have distinct genetic or disease-specific etiologies, the impact of this assumption is expected to be limited. As more comprehensive patient-level data become available, future studies could account for comorbidity patterns to refine estimates of the overall disease burden. Fourth, the reported genetic evidence derived from ClinVar data may be affected by ascertainment bias resulting from unequal testing availability, database submission and reporting practices, and the overrepresentation of populations from high-income countries. Variant classifications in ClinVar are also dynamic and may change as new evidence becomes available; therefore, our results reflect the classifications available at the time of data retrieval and should not be interpreted as definitive or immutable assessments of variant pathogenicity. Finally, SDI was selected as the principal predictor because it is consistently available across countries and over time. However, other factors associated with disease prevalence were not incorporated because comparable global data are unavailable. Future projections may be improved as additional predictors become available.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Zhiying Xie (xiezhiyingxzy@bjmu.edu.cn).

Materials availability

This study did not generate new unique reagents, biological materials, or experimental materials.

Data and code availability

The data that support the findings of this study are available from the Global Burden of Disease Study 2021 (GBD 2021) at http://ghdx.healthdata.org/gbd-2021, the GBD world population standard 2017–2100 at https://ghdx.healthdata.org/record/ihme-data/global-population-forecasts-2017-2100, and the January 2025 ClinVar Full Release XML file at https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/. The analysis code used in this study is provided in supplemental information. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We sincerely acknowledge the Institute for Health Metrics and Evaluation and the National Center for Biotechnology Information for providing valuable data access to the GBD and ClinVar databases, respectively, for noncommercial research purposes. This work was supported by Natural Science Foundation of Yangzhou Municipality (grant no. YZ2025131), National Natural Science Foundation of China (grant nos. 82201463, 82201553, and 82571596), and Peking University Clinical Scientist Training Program (BMU2026PYJH029).

Author contributions

Conceptualization, Zhihao Xie., C.S., and Zhiying Xie; methodology, Zhihao Xie and W.L.; data curation, C.S. and W.L.; formal analysis, Zhihao Xie, C.S., W.L., and X.Z.; validation, L.M., X.L., and Y.L.; visualization, C.L. and Z.Y.; interpretation of results, Zhihao Xie, C.S., and Zhiying Xie; supervision, Zhihao Xie, C.S., and Zhiying Xie; writing – original draft, Zhihao Xie, C.S., W.L., and Zhiying Xie; writing – review and editing, all authors. All authors contributed to this manuscript and approved the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Global Burden of Disease 2021 Study results	Institute for Health Metrics and Evaluation	http://ghdx.healthdata.org/gbd-2021
GBD world population standard	Institute for Health Metrics and Evaluation	https://ghdx.healthdata.org/record/ihme-data/global-population-forecasts-2017-2100
ClinVar	National Center for Biotechnology Information (NCBI, National Library of Medicine, USA)	https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/
Software and algorithms
R (version 4.5.1)	R Foundation for Statistical Computing	https://www.r-project.org/
glmnet R package	CRAN	https://cran.r-project.org/package=glmnet
forecast R package	CRAN	https://cran.r-project.org/package=forecast
Joinpoint Regression Program v 5.4.0	Surveillance Research Program, National Cancer Institute (NCI, USA)	https://surveillance.cancer.gov/joinpoint/

Experimental model and study participant details

This study was a secondary analysis of publicly available data from the GBD 2021 study and ClinVar. No human participants, animal experiments, cell lines, or clinical trials were involved or conducted.

Method details

Data sources and case definitions

The GBD 2021 study^8,37 provides a comprehensive and systematic assessment of health loss associated with 371 diseases and injuries across 204 countries and territories, encompassing key indicators of disease burden, including prevalence, incidence, mortality, and disability. Within the GBD 2021 framework, we evaluated 37 conditions affecting the nervous system⁶ and identified rare neurological disease categories in children according to three predefined criteria: (i) relevance to children aged 0–14 years; (ii) a prevalence threshold of no more than 1 in 2,000 persons (50 per 100,000); and (iii) representation as a distinct disease entity or category within the GBD 2021 cause hierarchy that permits population-level estimation. Based on these criteria, motor neuron disease, multiple sclerosis, neuroblastoma and other peripheral nervous cell tumors (NPNCT), and other neurological disorders were retained as rare neurological disease categories in children aged 0–14 years (Table S10). In the GBD 2021 cause hierarchy, other neurological disorders is a heterogeneous residual category rather than a single disease entity, including degenerative diseases, disorders of the autonomic nervous system, some movement disorders, spinocerebellar diseases, nerve root and plexus disorders, peripheral nerve disorders, neuromuscular disorders, and some muscle diseases such as muscular dystrophies, myopathies, and inflammatory myopathies. To characterize both the overall burden of rare childhood neurological diseases and the burden of individual disorders, estimates were generated for the four disorders combined and for each individual disorder. The included disorders have largely distinct etiologies and are treated as separate disease entities in the GBD framework, thereby reducing the likelihood of double counting in the combined estimates. From the GBD 2021 database,⁶ we extracted age-, sex-, and location-specific data on prevalence, DALYs, and mortality for these four rare neurological diseases, along with their 95% uncertainty intervals (UIs). GBD 2021 employed a cause-of-death ensemble model to estimate cause-specific mortality and the Bayesian meta-regression framework DisMod-MR 2.1 to generate non-fatal estimates.^6,38 Where primary epidemiological data are limited, GBD uses its established modeling framework to generate estimates; therefore, no additional imputation was performed in this study. To summarize age distribution patterns, children were grouped into three categories: <5 years, 5–9 years, and 10–14 years.

Age standardized rates of disease burden indicators

Age-standardized rates (ASRs) and corresponding 95% UIs for each disease burden indicator were estimated using the direct standardization method, applying the 2019 world standard population from GBD 2021 as the reference. Specifically, the age-specific population weights were derived as the proportion of each age group within the total global population of the 2019 standard. These weights were applied to age-specific rates for each rare neurological disease to compute disease-specific ASRs, which were then aggregated across all included diseases using the following formula:

$$\text{AS}{\mathrm{R}}_{\text{overall}}=\sum _{j=1}^m\frac{{\sum }_{i=1}^n{w}_i·r_{ij}}{{\sum }_{i=1}^n{w}_i}$$

where r_ij represents the disease rate for age group i and disorder j, w_i is the standard population weight for age group i, n is the total number of age groups, and m is the total number of rare neurological disorders included in the analysis. The ASRs for each burden indicator were obtained by summing disease-specific estimates under the assumption that diseases were mutually independent.

In estimating the uncertainty of the overall ASR, we used the delta method to propagate variance. First, we approximated the age-specific variances for each disease from the GBD reported UIs. The variance of each disease-specific ASR was then calculated as the weighted sum of these age-specific variances, scaled by the square of the total age-standardization weight. Finally, assuming independence across diseases, the variance of the overall ASR was obtained by summing the disease-specific ASR variances:

$$\text{Var}\left(\text{AS}{\mathrm{R}}_{\text{overall}}\right)=\sum _{j=1}^m\left(\frac{{\sum }_{i=1}^n{w}_i^2·\text{Var}\left(r_{ij}\right)}{{\left({\sum }_{i=1}^n{w}_i\right)}^2}\right),\text{Var}\left(r_{ij}\right)\approx {\left(\frac{{\text{Upper}}_{ij}-{\text{Lower}}_{ij}}{2\times 1.96}\right)}^2$$

where r_ij is disease rate for age group i and disorder j, w_i is the standard population weight, n is the total number of age groups, m is the total number of rare neurological disorders, and Upper_ij and Lower_ij are the upper and lower bounds of the 95% UIs for r_ij reported by the GBD 2021.

Decomposition analysis

We conducted a Das Gupta decomposition analysis³⁹ to quantify the relative contributions of population growth, changes in population age structure, and changes in age-specific prevalence to the change in number of rare neurological disease cases in children from 1990 to 2021. This method provides a systematic evaluation of the demographic and epidemiological determinants underlying the observed changes in cases over time.

Customized forecasting model

To project future prevalence of rare neurological diseases in children from 2022 to 2050, We developed a modified forecasting model based on the framework of Su et al.,¹⁰ which originally incorporated only SDI and population age structure as predictors. We implemented three principal methodological changes. To begin, population size was added as an offset in the Poisson framework, ensuring that differences in population magnitude were properly normalized in prevalence estimates. We then included a quadratic component for SDI to represent potential non-linear effects. Finally, ridge regularization was applied to mitigate overfitting and strengthen the model's predictive stability. The forecasting framework integrated Poisson regression with ARIMA modeling within a root mean squared error (RMSE)-weighted model averaging system (Figure 1). Four models were constructed to project the prevalence of childhood rare neurological diseases, each employing Poisson regression with SDI as the principal covariate.

The customized forecasting process comprised three key steps. In Step 1, two Poisson regression models were developed with SDI as the primary predictor. Model 1 accounted for variations in prevalence trends across countries and territories, while Model 2 captured the distinct trends across different age groups. Two SDI formulations were evaluated: a linear model without regularization and a quadratic model with ridge regularization. The optimal formulation was selected using five-fold cross-validated RMSE to ensure model parsimony and predictive accuracy. In Step 2, an ARIMA model was applied to the residuals from the Poisson regressions obtained in Step 1. The ARIMA (p, d, q) parameters were automatically selected based on the Bayesian Information Criterion (BIC). The predicted residuals were subsequently added to the Poisson regression forecasts to account for temporal dependencies not explained by SDI or demographic covariates. In Step 3, the outputs from the four models were integrated using the RMSE-weighted model averaging framework. RMSE values were computed for each model, and corresponding weights were derived inversely proportional to RMSE magnitude. Final forecasts were obtained by averaging model predictions according to these weights. Model performance was evaluated using an out-of-sample temporal validation strategy, in which data from 2011 to 2021 were withheld for validation. The forecasting models were fitted using data from 1990 to 2010, and forecasts for the validation period were compared with the corresponding observed values. Predictive performance was assessed using RMSE, mean bias, relative RMSE, mean absolute error (MAE), and mean absolute percentage error (MAPE). In addition, sensitivity analyses were performed by comparing the original forecasting framework proposed by Su et al. with our customized forecasting model using these performance metrics.

Uncertainty in future projections was quantified through a bootstrap approach with 1,000 iterations. In each iteration, input data were uniformly resampled between the lower and upper bounds of their original UIs, and forecasts were recalculated using the RMSE-weighted framework. The 2.5th and 97.5th percentiles of the resulting bootstrap distribution were used as the lower and upper bounds of the 95% UIs, respectively.

Querying and processing genetic data in ClinVar

We compiled a curated list of 520 disease-causing genes associated with rare monogenic neurological diseases (Table S11) based on published literature and established databases, including the Human Gene Mutation Database and the Online Mendelian Inheritance in Man. The conditions analyzed encompassed motor neuron disease, NPNCT, and other neurological disorders, including neurodegenerative diseases, autonomic nervous system disorders, movement and spinocerebellar disorders, peripheral nerve disorders, congenital myasthenic syndromes, myopathies associated with systemic diseases, primary myopathies, and muscular dystrophies. Genetic data were obtained from the January 2025 ClinVar Full Release XML file. Variants were included if they had a consensus classification of likely pathogenic or pathogenic from multiple submitters. Duplicate submissions for the same variant were consolidated before analysis to avoid double counting. For variants with multiple interpretations, variants were excluded if any submission classified them as Uncertain Significance, Likely benign, or Benign, regardless of whether Pathogenic or Likely pathogenic interpretations were also present. Only variants without any Uncertain Significance, Likely benign, or Benign interpretations were retained for analysis. No variant reclassification was performed. We quantified the annual number of newly assessed pathogenic or likely pathogenic variants and, in parallel, calculated the yearly number of newly reported disease-causing genes.

Quantification and statistical analysis

Disease burden was quantified using prevalence, mortality, and DALYs, with corresponding age-standardized rates and 95% UIs. The association between global ASPR and SDI was assessed using Pearson’s correlation coefficient (r), with the corresponding p value used to evaluate the statistical significance of the linear association. Long-term temporal trends were quantified using the average annual percentage change (AAPC).^14,40 A statistically significant upward or downward trend was defined as a 95% confidence interval for the AAPC that lay entirely above or below zero, respectively. All AAPC analyses were conducted using the Joinpoint Regression Program version 5.4.0 (National Cancer Institute, USA). All other statistical analyses were performed using R version 4.5.1.