Global, regional, and national burden of spinal cord injury, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019

Summary Background Spinal cord injury (SCI) is a major cause of health loss due to premature mortality and long-term disability. We aimed to report on the global, regional, and national incidence, prevalence, and years of life lived with disability (YLDs) for SCI from 1990 to 2019, using data from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2019. Methods Using GBD 2019 data pooled in DisMod-MR 2.1, a Bayesian meta-regression tool, we systematically derived numbers and age-standardised rate changes with 95% uncertainty intervals (95% UIs) for the incidence, prevalence, and YLDs for SCI from 1990 to 2019 for the whole world, 21 GBD regions, and 204 countries and territories. We report trends based on age, sex, year, cause of injury, and level of injury. Findings Globally, 20·6 million (95% UI 18·9 to 23·6) individuals were living with SCI in 2019. The incidence of SCI was 0·9 million (0·7 to 1·2) cases with an estimated 6·2 million (4·5 to 8·2) YLDs. SCI rates increased substantially from 1990 to 2019 for global prevalence (81·5%, 74·2 to 87·1), incidence (52·7%, 30·3 to 69·8), and YLDs (65·4%, 56·3 to 76·0). However, global age-standardised rates per 100 000 population showed small changes in prevalence (5·8%, 2·6 to 9·5), incidence (–6·1%, –17·2 to 1·5), and YLDs (–1·5%, –5·5 to 3·2). Data for 2019 shows that the incidence of SCI increases sharply until age 15–19 years, where it remains reasonably constant until 85 years of age and older. By contrast, prevalence and YLDs showed similar patterns to each other, with one peak at around age 45–54 years. The incidence, prevalence, and YLDs of SCI have consistently been higher in men than in women globally, with a slight and steady increase for both men and women from 1990 to 2019. Between 1990 and 2019, SCI at neck level was more common than SCI below neck level in terms of incidence (492 thousand [354 to 675] vs 417 thousand [290 to 585]), prevalence (10·8 million [9·5 to 13·9] vs 9·7 million [9·2 to 10·4]), and YLDs (4·2 million [3·0 to 5·8] vs 1·9 million [1·3 to 2·5]). Falls (477 thousand [327 to 683] cases) and road injuries (230 thousand [122 to 389] cases) were the two leading causes of SCI globally in 2019. Interpretation Although age-standardised rates of incidence, prevalence, and YLDs for SCI changed only slightly, absolute counts increased substantially from 1990 to 2019. Geographical heterogeneity in demographic, spatial, and temporal patterns of SCI, at both the national and regional levels, should be considered by policy makers aiming to reduce the burden of SCI. Funding Bill & Melinda Gates Foundation.


Introduction
3][4] In the USA, for instance, SCI is estimated to cost more than US$9•7 billion annually. 5,6In many previous studies, epidemiological data for SCI have been relatively heterogeneous, reporting the annual incidence as from 1•2 to 5•8 cases per 100 000 population. 7The annual incidence of SCI in developing countries (based on the definition of developing countries by the International Monetary Fund) consists of an even wider range, from 0•2 to 13•0 cases per 100 000 population. 3,8This heterogeneity has been mainly rooted in diverse datagathering methods, case-defining approaches, and socioeconomic structures of different countries. 9,10ccurate and up-to-date estimates of the incidence and prevalence of disorders constitute the backbone of evidence-based health-care planning and resource allocation.Because no effective curative treatment for individuals with SCI has been identified, 11,12 prevention is crucial, and investigating epidemiological patterns would be the first step towards reaching this goal. 13To provide organised and targeted health-care support for people with SCI, it is also essential to have a comprehensive understanding of the prevalence and incidence of these injuries.Global data and comparisons between countries Trinidad and Tobago 0 (0 to 0)             Where super-regions contain only one region, data for the region are listed in the header row for the super-region.95% UI=95% uncertainty interval.YLDs=years of life lived with disability.

Table: Incidence, prevalence, and YLDs for spinal cord injuries for all ages in 2019 and change in age-standardised rate per 100 000 population from 1990 to 2019 for 204 countries and regions
The analytical cascade of DisMod-MR 2.1 software, a Bayesian meta-regression tool developed by the GBD team to run the estimations, is also shown in appendix 1 of the GBD 2017 capstone paper on incidence, prevalence, and YLDs. 19e used the SCI International Classification of Diseases (ICD) code list provided by the GBD modelling team for data processing.This list includes all conditions considered as SCI, both traumatic and non-traumatic, that were included in this study (appendix pp 2-6).For SCI, three categories are of interest in the GBD category hierarchy: spinal cord lesion at neck level, spinal cord lesion below neck level, and the parent category, spinal injuries.Our approach for estimating the burden of SCI was developed within the GBD 2019 methodology framework. 16We extracted detailed estimations for incidence, prevalence, and YLDs for SCI from the GBD Results Tool, 20 at neck level, below neck level, and the aggregate of these injury causes with disaggregation of age, sex, year, and cause of injury.Estimates were retrieved for the world, 21 GBD regions, and 204 countries and territories.A detailed description of region and country categorisation is available in the appendix (p 10).We present estimates by number (ie, counts) and agestandardised rates per 100 000 population using the GBD standard population structure.We then compared the agestandardised rates between 1990 and 2019 and investigated temporal and spatial patterns according to age, sex, year, and geographical location.Global data for the cause of injury in 2019 are provided, in addition to changes from 1990.Annualised rates of change from 1990 to 2019 represent the mean percentage change per year during this period.

Statistical analysis
Consistent with the GBD framework, we provide 95% uncertainty intervals (UIs) for all estimates, using the mean estimate across 1000 draws, with the 25th and 975th ranked values across all 1000 draws as the lower and upper bounds of the 95% UIs.The analysis contained in this Article is the result of an amalgamation of numerous surveys done with varying levels of quality and representativeness; we therefore report estimated results to a level of precision that conveys the true See Online for appendix uncertainty of the GBD process.We present the count data in thousands and rates to one decimal place.We also used round outwards (ie, lower limit down and upper limit up) to maintain 95% uncertainty.

Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or the writing of the report.

Regional and national trends
Regionally, high-income North America had the highest age-standardised incidence rate in 2019 with 22 (95% UI 17 to 29) cases per 100 000 population, followed by 14 (11 to 17) per 100 000 in Australasia and 14 (10 to 19) per 100 000 in tropical Latin America (appendix p 7).
Incidence, prevalence, YLDs, and the corresponding percentage changes in age-standardised rates for SCI varied across countries between 1990 and 2019 (table).Agestandardised incidence rates of SCI in 2019 varied widely between countries, with the highest rates observed in  (Figure 2 2A).The greatest national increases in the age-standardised incidence rates per 100 000 population for SCI from 1990 to 2019 were for Syria, Yemen, Afghanistan, and Libya.By contrast, the greatest national decreases in the age-standardised incidence rates per 100 000 population from 1990 to 2019 were for Ethiopia, Iran, Eritrea, and Kuwait (  2C).Burundi, Syria, Haiti, Central African Republic, Congo (Brazzaville), and Rwanda had the greatest increases in the age-standardised YLDs rate from 1990 to 2019, respectively.By contrast, Eritrea, El Salvador, and Lebanon had the greatest decreases.

Age-sex-specific patterns
Distribution of global incidence, prevalence, and YLDs for SCI from all causes in 2019 according to age groups is shown in the appendix (p 8).Compared with 1990, the incidence chart shows an important difference: one peak exists at age 20-24 years in 1990, whereas in 2019, incidence increases with age and remains higher than 50 000 cases from around age 20 years to around age 84 years, with two peaks, at age 30-34 years and age 50-54 years, after which it decreases (figure 3).The prevalence and YLDs charts show similar patterns and changes.In 1990, both prevalence and YLDs had one peak at the age of 35-39 years, whereas in 2019, the prevalence chart shows a higher peak at around 50-54 years and the YLD chart shows a higher peak at around 45-59 years (figure 3).Global incidence of SCI in children and people younger than 25 years seems to have been lower in 2019 than in 1990.

Injury patterns
In 2019, the global incidence of spinal cord lesions at neck level was 492 thousand (95% UI 354-675) and the incidence of spinal cord lesions below neck level was 417 thousand (290-585) new cases (ie, 54

Discussion
From 1990 to 2019, the global numbers of incidence, prevalence, and YLDs for SCI increased substantially.However, their age-standardised rates showed only slight changes.The increases in prevalence and YLDs might be attributable to several factors, including population growth, ageing, and improvements in global health-care access and quality. 21According to the GBD 2019 injuries hierarchy, SCIs are the type of injury that causes the highest mean long-term disability (ie, measured by YLDs).In this categorisation, spinal cord lesion below neck level is ranked first in terms of level of disability, and spinal cord lesion at neck level is ranked fourth. 16Because a major proportion of prevalence and YLD estimates come from chronic disabling conditions, SCI could therefore contribute to a major proportion of the burden of disease among all injuries.Although life expectancy of individuals with SCI, even with optimal medical management, is lower than that of the general population, several long-term studies have suggested that mortality for people with SCI has decreased since 1944. 22,23However, this effect might not be true for people from low-income and middle-income countries (LMICs) who cannot afford the long-term treatment and care. 21The increase of more than 80% in prevalence from 1990 to 2019 according to our study could be a reflection of increasing life expectancy and population ageing.However, the increase in absolute incidence, which is the main cause of the increasing YLDs and prevalence, warrants consideration.Whereas traumatic cases of SCI are more common in young people (eg, aged 15-29 years) and older people (eg, aged ≥65 years), 24 the incidence of non-traumatic SCI steadily increases with age.According to our results, the peaks of incidence, prevalence, and YLDs are shifting towards older ages over time, and an ageing population with increased lifespan can consequently lead to the increase in the absolute number of SCIs.
Although data for Socio-demographic Index (SDI) is not currently available for SCI in the GBD database, owing to the absence of statistical support for presenting such data, our findings support previous research indicating that incidence, prevalence, and YLDs of SCI are higher in high-income regions than in lower-income regions.High-income North America and Australasia, for example, had the highest age-standardised incidence, prevalence, and YLD rates in 2019.One reason for this finding might be the high quality of data coming from these regions, in addition to the number of reports and methods of reporting.In a study by Tropeano and colleagues, 25 among 2304 articles from 2008 to 2018, North America (843 [36•6%]) and Europe (833 [36•2%]) were the major publishing regions for studies of traumatic SCI.A systematic review in 2018 by Kumar and colleagues 26 showed the incidence of traumatic SCI to be higher in LMICs (13•6 cases per 100 000 people) than in high-income countries (8•7 cases per 100 000 people).This wide variation might reflect true regional differences in the incidence of SCI, but also the quality of studies included in the analysis.A deeper look into the data at a national level could help to improve the understanding of these differences.The four countries with the greatest increase in age-standardised incidence rates from 1990 to 2019 (ie, Syria, Yemen, Afghanistan, and Libya) are located in North Africa and the Middle East and were all involved in war and conflicts during this period.As depicted in the GBD 2016 paper, 15 North Africa and the Middle East was the only GBD region in which conflict and terrorism had a greater contribution than falls to age-standardised incidence of SCI.The wide UIs in these countries also reflect the heterogeneity of data sources for the estimates of SCI.We conceptualise that the burden of SCI, as assessed with counts of incidence, prevalence, and YLDs, is increasing worldwide, and it is essential to consider local epidemiological data in planning preventive strategies or allocation of resources.The difference in incidence rates of nations and regions can depend on the accuracy of the detecting bodies in these nations to record SCI.Low incidence in sub-Saharan Africa and high age-standardised incidence rates in Greenland might be due to scarcity of data.Notably, some of the countries with the most substantial changes in age-standardised prevalence of SCI are those for which data are scarce or absent in GBD 2019.Greater resources might need to be invested in the collection and registration of data to improve the accuracy of future estimates.
Globally, incidence, prevalence, and YLDs were higher in males than in females throughout the period from 1990 to 2019, and the size of the difference between males and females remained almost stagnant for all the three indices, with a slight increase for both sexes during this period.Although the male-to-female ratio for SCI has been reported to vary in previous studies according to their methods, case definitions, and locations, 8 males have always had higher rates of SCI than females.Men are usually more exposed to the causes of SCI, especially to the traumatic causes. 8Studies have shown that the maleto-female ratio of traumatic SCI decreases after age 54 years, suggesting that traumatic SCIs are more frequent among younger men than younger women, and they become progressively more relevant among women in later life, despite remaining more than twice as common among men. 27Moreover, GBD 2019 showed that males have divergent incidence patterns compared with women younger than 60 years but these patterns of change overlap in older age. 15In our study, however, 53•4% of the global incidence of SCI was in men in 2019, which showed a slight decrease in comparison with 1990 (57•0%).However, under-reporting of SCI among females due to non-inclusion of females in studies might be responsible for variation in injuries among males and females in 1990 as compared with 2019. 28There is an apparent reduction in global incidence of SCI between 1990 and 2019 among children and adolescents younger than 25 years.Whereas previously patients with SCI were thought to be relatively young, 8 a shift towards older ages is seen in comparison with 1990, in addition to an increase in incidence rate.In our study, falls were the leading contributor to incidence numbers of SCI globally, followed by road injuries.
However, this ranking might differ according to region, as a systematic review and meta-analysis by Golestani and colleagues emphasised that two leading causes of traumatic SCI in developing countries (based on the definition of developing countries by the International Monetary Fund) were motor vehicle crashes (43•2%) and falls (34•2%). 29A previous GBD subanalysis study also showed that transport injuries, one of the major causes of SCI, are more common in countries with a high SDI than in countries with a low SDI. 30HO has estimated an annual incidence of SCI at between 250 000 to 500 000 people worldwide. 31However, depending on the region, the estimated incidences and prevalence can vary widely. 32van den Berg and colleagues, for instance, reported annual incidences of SCI from 1•2 cases per 100 000 population in the Netherlands to 5•8 cases per 100 000 population in Portugal. 24The annual incidence of SCI in developing countries (based on the definition of developing countries by the International Monetary Fund) yields an even wider range, from 0•2 cases per 100 000 population in Saudi Arabia up to 13•0 cases per 100 000 population in Bulgaria. 8This heterogeneity has been mainly rooted in diverse data-gathering methods, case-defining approaches, and socioeconomic structures of different countries. 9The global incidence of SCI in our study was more than 900 thousand cases in 2019 with a rate of 11•5 (95% UI 8•9-14•6) cases per 100 000 population.We believe that one main reason for the heterogeneity could be that most studies in this area on SCI focus on patients with traumatic SCI, and surveys of local and national incidences have been conducted in this context, 32,33 mostly neglecting patients with non-traumatic SCI.Furthermore, data access and quality are heterogeneous across different locations, and data in LMICs are particularly sparse.However, a few highincome countries, such as Australia, Canada, Switzerland, and the USA, have developed SCI systems 34 that enable them to accurately monitor and periodically report epidemiological data.In some countries, many patients with SCI do not appear in the statistics because of the difficulty in assessing incidence, for example, owing to the absence of prehospital mortality data.
Efforts have been made to describe the true epidemiology of SCI in Iran, 35 but there is still a strong need to understand epidemiological characteristics to promote practical preventive strategies in many other areas with poor data.Although single-centre series are being published from LMICs, some high-income European countries continue to update their knowledge periodically.However, the epidemiology of SCI varies in different countries and the results of studies from high-income countries are not applicable to LMICs. 36,37n a systematic review for example, by use of an extensive search strategy, Jazayeri and colleagues identified 101 reports regarding traumatic SCI incidence up to 2013 from 41 countries, representing about 20% of countries worldwide. 34Moreover, most of the available literature evaluated the epidemiological aspects of SCI in high-income countries.In 2023, Jazayeri and colleagues updated their search strategy and found 58 new reports for 31 countries, 38 implying that data compilation efforts are increasing globally.However, there is still a huge gap to fill to draw a more accurate global map for SCI incidence and prevalence, especially for LMICs.Another study using GBD 2019 data for SCI by Ding and colleagues was published in 2022. 39In comparison with their paper, which emphasises global and regional results with limited discussion of age and sex, we tried to provide a comprehensive display of global, regional, and national results as well as trends by sex and age, and we prepared our report with input from a network of global researchers as part of the GBD Protocol.
General limitations of GBD, which have been discussed previously, 40 such as difficulties in accurately quantifying all sources of uncertainty, lags in data availability, heterogeneity in coding procedures, and other biases, fully apply to this report.GBD estimation of SCI is reconstructed by mathematical models using various sources of differing quality and is not a collection of measured data.This reconstruction might deviate from the actual data, particularly in some areas without adequate and reliable population-based data, such as sub-Saharan Africa or Greenland.These limitations might account for differences between these data and those collected by other governmental or disease-specific organisations.The quality of the data used in the predicting tool can also have an effect on the outcomes, which were mostly derived from the modelled data through the procedure in the DisMod-MR 2.1 tool.GBD 2019 estimates incidence, prevalence, mortality, years of life lost (YLLs), YLDs, and disability-adjusted life-years (DALYs) for 369 diseases and injuries.YLLs, which is the number of years lost due to premature death and represents a fatal burden, is not calculated by GBD 2019 for injuries categories and mortality data are not available; therefore DALYs, which are the sum of YLLs and YLDs, was not retrievable for SCI from the GBD 2019 database.The comparison of YLDs without consideration of socio economic factors might provide misleading data.However, presenting YLDs in our study could be considered an advantage over the GBD 2016 paper. 15Another methodological limitation is that data for the type of SCI (ie, traumatic vs non-traumatic), which can affect outcome substantially, are not currently retrievable from the GBD database.Traumatic SCI is abrupt and sudden, whereas non-traumatic SCI is usually progressive and gradual.This classification of data could be informative in many ways, including indicating age and sex differences.One other important issue regarding studies addressing SCI is differentiation between complete and incomplete lesions.There is a spectrum of clinical findings for SCI, which can be classified according to the American Spinal Injury Association score and Frankel grading.In other words, not all SCIs are permanent and associated with lost years of life and disability.People with minimal and mild injuries, and even some moderate injuries, can fully or near fully recover and return to healthy function and life.The nature of SCI affects the results and the burden of the disease significantly.Whereas incomplete SCIs are associated with better outcomes and less disability, complete lesions are permanent and carry substantial social, economic, and emotional burdens.Thus, these two entities should be separated when analysing and reporting the results of GBD.In future rounds of GBD, researchers should provide other epidemiological indices, such as YLL, DALYs, mortality, and life expectancy, for SCI so that the total (ie, fatal and non-fatal) burden of SCI can be provided, broken down by the cause of injury (ie, traumatic vs non-traumatic), and also type of the injury (ie, complete vs incomplete).The unavailability of data for SDI for SCI in the GBD database is another notable limitation of our study.Such data could have provided important information for economic and administrative planning purposes.
According to our results, epidemiology of SCI is increasingly affecting people at older ages, and the number of people affected is increasing globally.Although the age-standardised rates of incidence, prevalence, and YLDs showed only slight changes from 1990, the absolute counts of cases increased substantially.Our findings could aid health-care professionals and policy makers at the global or national levels in providing preventive interventions and advance planning for resource allocation to prevent and reduce the burden of SCI.Future studies should focus on considering the attributable risk factors and differences according to the severity and type of injuries, in addition to the potential effect of socioeconomic status on the burden of SCI.
Medicine Conference 2022 and from the International Society of Travel Medicine for the Travel Medicine Conference, Basel 2023; and leadership or fiduciary roles in board, society, committee, or advocacy groups, paid or unpaid with Kidsafe as Director, Farmsafe as Director, Auschem as Director, the Public Health Association of Australia Injury Prevention Special Interest Group as Convenor, and the International Society for Agricultural Safety and Health as a member of the Governance Committee, all outside the submitted work.E Trinka reports grants or contracts from the Austrian Science Fund (FWF), Oesterreichische Nationalbank, the EU, GSK, Biogen, Eisai, Novartis, Red Bull, Bayer, and UCB; consulting fees from Angelini, Clexio, Argenx, Arvelle, Epilog, Ever Pharma, UCB, Biogen, GSK, Bial, Eisai, Takeda, Newbridge, GW Pharma, Sunovion, Liva Nova, Marinus, Medtronic, Novartis, Sandoz, and Sanofi; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from UCB, Eisai, Biogen, Novartis, Bial, Sunovion, Ever Pharma, Liva Nova, Sanofi, Hikma, Newbridge, Arvelle, GW Pharma, and Sandoz; and other support from Neuroconsult as CEO, all outside the submitted work.All other authors declare no competing interests.

Data sharing
Data used in this article are available for download on the Global Health Data Exchange tool, 20 which is permitted to be used, shared, modified, or built on by non-commercial users via the Open Data Commons Attribution License.All GBD 2019 data are publicly available and can be downloaded via the Global Health Data Exchange tool 20 and from the GBD Compare Visualisation Tool. 41 (-23•2 to -11•2) (

Figure 2 :
Figure 2: Global age-standardised incidence, prevalence, and YLDs rates per 100 000 population for spinal cord injuries, 2019 (A) Age-standardised incidence rate.(B) Age-standardised prevalence rate.(C) Age-standardised YLDs rate.Countries for which no data (and therefore no modelled results) are available are shown in black.YLDs=years of life lived with disability.
Table continues on next page) Table continues on next page) Table continues on next page) Table continues on next page) Table continues on next page)