Global, regional, and national burden of meningitis, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016

1Meningitis is an inflammatory condition involving the membranes (meninges) covering the brain and spinal cord. It can have infectious causes, such as bacteria, mycobacteria, viruses, fungi, or parasites, or be associated with autoimmunity, cancer, or reactions to medication. Risk factors that predispose individuals to meningitis and epidemics include malnutrition, household overcrowding, HIV infection, absence of immunisation, indoor air pollution, and sickle cell disease. Most deaths due to meningitis have infectious causes, but the clinical severity of disease varies with the causative organism. Bacterial meningitis can rapidly become fatal and lead to severe disability in those who survive. Survivors might have complications, such as cognitive impairment, behavioural problems, hearing loss, motor weakness, paralysis, incoordination, or seizure disorder; although few data are available from low-resource settings, one study found as many as a quarter of survivors of bacterial meningitis had neuropsychological sequelae 3–60 months after hospital discharge., Both bacterial and viral meningitides can place a substantial burden upon families, communities, and societies.,, Although bacterial meningitis has often been associated with persistent intellectual impairment, some viral pathogens, such as parechovirus, have also been associated with impaired developmental attainment.

2Evidence before this studyMeningitis is a disease with a high case-fatality rate that is known to occur in epidemics, requires timely and appropriate diagnosis and treatment to avoid death, and can lead to lifelong disability among survivors. In addition to previous Global Burden of Diseases, Injuries and Risk Factors (GBD) studies, some studies have investigated the causes and outcomes of meningitis in large cities and countries, but not comprehensive reporting on meningitis burden by cause, age, geography, or over time across continents. We searched PubMed without language restrictions for five components of epidemiology information related to meningitis: (1) incidence of all meningitis (from database inception to December, 2013), (2) the proportion of meningitis deaths (from database inception to November, 2016) and cases (from database inception to December, 2013) due to each of its four causes (as classified in GBD), (3) case-fatality ratio and long-term mortality experience of survivors with chronic complications of meningitis (from database inception to December, 2013), (4) the proportion of survivors with long-term complications, along with the nature and distribution of those complications (from database inception to December, 2013), and (5) the distribution of sequela severity by type of long-term outcome (from database inception to December, 2013). The following search string was used in PubMed: (((“Meningitis”[MeSH] OR “Meningitis, pneumococcal”[MeSH] OR “Meningitis, Haemophilus”[MeSH] OR “Meningitis, Meningococcal”[MeSH] OR “Meningitis, viral”[MeSH] OR “Meningitis”[Title/Abstract]) AND ((“etiology”[Title/Abstract] OR “causes” Title/Abstract] OR “cause pattern”[Title/Abstract] OR “aetiology”[Title/Abstract] OR “cause”[Title/Abstract]) AND (“fatality”[Title/Abstract] OR “mortality”[Title/Abstract] OR “death”[Title/Abstract]) AND 1985/01/01[PDAT]:3000/12/31[PDAT]) AND “humans”[MeSH]).Added value of this studyThis study reports on incidence, mortality, and disability of overall meningitis and its subtypes for each of 195 countries, 23 age groups, and both sexes from 1990 to 2016. It leverages the strength of the overall GBD 2016 study to generate estimates of disease burden that are internally consistent within each cause and across all causes, contextualises them within the sociodemographic development spectrum, and reveals the populations with the greatest need for improved meningitis prevention and treatment. Global burden from meningitis has reduced, but substantial disparities across geographic regions and age groups persist. Cause-specific estimates allowed for examination of trends in specific infections and their relationship to vaccine programmes. By quantifying long-term disability, this study also highlights the importance of health systems and societies being prepared to support survivors of meningitis.Implications of all the available evidenceThis study estimates that the largest concentration of meningitis mortality remains in the meningitis belt, which includes 26 countries across sub-Saharan Africa—from Senegal in the west to Ethiopia in the east—but many other countries also continue to have a high burden of meningitis. Vaccine coverage should be increased and existing vaccination schedules optimised to maximise population protection. Incorporating meningitis vaccines into routine vaccination schedules and ensuring these vaccines are available and administered in more countries should also be considered. Development of new vaccines and reinvestment in existing vaccines will be important to address the potential of serotype replacement in pneumococcal meningitis, to deal with epidemics, and to combat forms of meningitis that are not currently preventable. In particular, use of pneumococcal conjugate or polysaccharide vaccine in more adults and use of available or broader valency meningococcal vaccines could result in marked reductions in the burden of meningitis. Enhanced availability of appropriate antibiotics and health-care services, and low-cost diagnostics at the point of care or point of surveillance, are paramount for detecting meningitis, improving survival, and reducing disability—especially in low-resource settings where the burden of meningitis is greatest. Increased ability to detect antibiotic resistance will also become more relevant as antibiotic use and resistance continue to rise. More robust vaccine coverage and cause-specific surveillance data are crucial to improve the geographical and temporal resolution of meningitis estimates going forward.

3Evidence before this study

4Meningitis is a disease with a high case-fatality rate that is known to occur in epidemics, requires timely and appropriate diagnosis and treatment to avoid death, and can lead to lifelong disability among survivors. In addition to previous Global Burden of Diseases, Injuries and Risk Factors (GBD) studies, some studies have investigated the causes and outcomes of meningitis in large cities and countries, but not comprehensive reporting on meningitis burden by cause, age, geography, or over time across continents. We searched PubMed without language restrictions for five components of epidemiology information related to meningitis: (1) incidence of all meningitis (from database inception to December, 2013), (2) the proportion of meningitis deaths (from database inception to November, 2016) and cases (from database inception to December, 2013) due to each of its four causes (as classified in GBD), (3) case-fatality ratio and long-term mortality experience of survivors with chronic complications of meningitis (from database inception to December, 2013), (4) the proportion of survivors with long-term complications, along with the nature and distribution of those complications (from database inception to December, 2013), and (5) the distribution of sequela severity by type of long-term outcome (from database inception to December, 2013). The following search string was used in PubMed: (((“Meningitis”[MeSH] OR “Meningitis, pneumococcal”[MeSH] OR “Meningitis, Haemophilus”[MeSH] OR “Meningitis, Meningococcal”[MeSH] OR “Meningitis, viral”[MeSH] OR “Meningitis”[Title/Abstract]) AND ((“etiology”[Title/Abstract] OR “causes” Title/Abstract] OR “cause pattern”[Title/Abstract] OR “aetiology”[Title/Abstract] OR “cause”[Title/Abstract]) AND (“fatality”[Title/Abstract] OR “mortality”[Title/Abstract] OR “death”[Title/Abstract]) AND 1985/01/01[PDAT]:3000/12/31[PDAT]) AND “humans”[MeSH]).

5Added value of this study

6This study reports on incidence, mortality, and disability of overall meningitis and its subtypes for each of 195 countries, 23 age groups, and both sexes from 1990 to 2016. It leverages the strength of the overall GBD 2016 study to generate estimates of disease burden that are internally consistent within each cause and across all causes, contextualises them within the sociodemographic development spectrum, and reveals the populations with the greatest need for improved meningitis prevention and treatment. Global burden from meningitis has reduced, but substantial disparities across geographic regions and age groups persist. Cause-specific estimates allowed for examination of trends in specific infections and their relationship to vaccine programmes. By quantifying long-term disability, this study also highlights the importance of health systems and societies being prepared to support survivors of meningitis.

7Implications of all the available evidence

8This study estimates that the largest concentration of meningitis mortality remains in the meningitis belt, which includes 26 countries across sub-Saharan Africa—from Senegal in the west to Ethiopia in the east—but many other countries also continue to have a high burden of meningitis. Vaccine coverage should be increased and existing vaccination schedules optimised to maximise population protection. Incorporating meningitis vaccines into routine vaccination schedules and ensuring these vaccines are available and administered in more countries should also be considered. Development of new vaccines and reinvestment in existing vaccines will be important to address the potential of serotype replacement in pneumococcal meningitis, to deal with epidemics, and to combat forms of meningitis that are not currently preventable. In particular, use of pneumococcal conjugate or polysaccharide vaccine in more adults and use of available or broader valency meningococcal vaccines could result in marked reductions in the burden of meningitis. Enhanced availability of appropriate antibiotics and health-care services, and low-cost diagnostics at the point of care or point of surveillance, are paramount for detecting meningitis, improving survival, and reducing disability—especially in low-resource settings where the burden of meningitis is greatest. Increased ability to detect antibiotic resistance will also become more relevant as antibiotic use and resistance continue to rise. More robust vaccine coverage and cause-specific surveillance data are crucial to improve the geographical and temporal resolution of meningitis estimates going forward.

9Immunisation programmes targeting the major bacterial pathogens—Haemophilus influenzae type b, Neisseria meningitidis, and Streptococcus pneumoniae—have successfully reduced cases of meningitis in Africa, the Americas, Asia, Australasia, and Europe.,, The H influenzae type b vaccine is now part of immunisation programmes across 19 countries and the pneumococcal vaccine is part of immunisation programmes across 134 countries. The MenAfriVac mass vaccination campaign delivered over 270 million doses of menin-gococcal A protein-polysaccharide conjugate (MenA) vaccine to children and young adults in 21 of the 26 countries of the meningitis belt between 2010 and 2018, which resulted in a marked decline in cases of disease; the meningitis belt was first defined as a “region south of the Sahara between latitudes 4 and 160N” and was later expanded to include the Sahelian parts of Benin, Cameroon, Ethiopia, The Gambia, Ghana, Mali, and Senegal (appendix). However, maintaining control of MenA epidemics in the long term depends on these countries implementing the vaccine routinely—a strategy taken up by only seven countries within the meningitis belt to date.,

10Contemporary information on country-specific incidence, mortality, and long-term disease burden of meningitis by cause is needed by governments and funders to identify areas of improvement and stagnation, strengthen screening and treatment programmes, and adequately plan to support the ongoing needs of survivors. Such data can help determine country-specific immunisation needs, and countries eligible for funding can use it to prioritise needs and assist with applications for support from Gavi, the Vaccine Alliance, UNICEF, and the Bill & Melinda Gates Foundation, among others.

11The Global Burden of Diseases, Injuries, and Risk Factors (GBD) 2016 study, generated comprehensive estimates of meningitis burden by cause at the national, regional, and global level from 1990 to 2016. This report details the approach, methods, and results of the GBD 2016 analysis of the global meningitis burden to highlight levels and trends by age and geography, over time, and with respect to sociodemographic development status. The aim is to increase awareness and understanding of these estimates for stakeholders who play various roles in the identification, management, and prevention of meningitis, from researchers and clinicians to policy makers, advocates, and planners of neurological services. This report also aims to highlight some limitations and gaps in the data, with a goal of catalysing collaboration to improve future global estimates for meningitis through improved data collection, timeliness, and implementation of evidence-based policies to reduce the burden of meningitis globally.

12The GBD 2016 estimation strategy for both mortality (deaths and years of life lost [YLLs]) and non-fatal health outcomes (prevalence, incidence, and years of life lived with disability [YLDs]), began with estimation of the epidemiology of total acute meningitis and then apportioning of that total to four sub-causes. Three specific pathogens were considered separately given their relevance to public health policy, availability of effective vaccines, and comparatively robust data sources: S pneumonia (pneumococcal), N meningitides (meningococcal), and H influenzae type b. The remaining pathogens (eg, bacteria, fungi, and viruses) were grouped together as other meningitis; future analyses will consider specific pathogens for disaggregation from this category. Complete descriptions of GBD 2016 methods and inputs are available in corresponding summary publications., A detailed description of methods, including an analytic flowchart, data source search and extraction criteria, and model development details with respect to meningitis, is available in the appendix.

13The overall goal of GBD estimation methods is to combine all available data into an internally consistent analytic framework for describing death and disability due to each cause of disease or injury. This methodological approach was applied to each of the 195 countries and territories (11 of which were analysed at the first subnational administrative level [eg, states in the USA and provinces in China]), 23 age groups, both sexes, and for each year from 1990 to 2016 (inclusive). In addition to deriving specific estimates for each age group, location, sex, and year, GBD 2016 updated the method for describing epidemiological transition that is based on the Socio-demographic Index (SDI), a composite metric based on total fertility rate, education, and per-capita income, that is designed to be a summary measure of societal development and is scaled between 0 and 100. All GBD locations were grouped into SDI quintiles.

14We estimated mortality using vital registration and verbal autopsy data that were extracted from all sources available at the time of the study and we processed the data using a set of standardised algorithms that were used across GBD 2016 to account for incompleteness, misclassification, and stochastic variability (appendix). International Classification of Diseases, ninth Revision (ICD-9) and tenth Revision (ICD-10) codes used as the case definition for meningitis were 036–036·9/A39–A39·9 (menin-gococcal meningitis), 320·0/G00·0 (H influenzae type b meningitis), 320·1/G00·1 (pneumococcal meningitis), and 047–049/320·2–322·9 (other meningitis). We then modelled all available data using cause of death ensemble modelling (CODEm) with ten candidate covariates: (1) the proportion of population living in meningitis belt, (2) proportion of children younger than 5 years who are underweight (≤2 weight-for-age Z scores), (3) proportion of households with improved water, (4) the Healthcare Access and Quality Index, (5) health system access, a composite of vaccine coverage and pregnancy services, (6) diphtheria-tetanus-pertussis vaccine coverage, (7) lag distributed income (in international dollars per capita), (8) MenAfriVac vaccine coverage from 2010 to 2012, (9) improved sanitation (proportion with facilities that hygienically separate human excretia from contact with humans), and (10) maternal education (years per capita). We chose the CODEm models with the highest out-of-sample predictive validity and then combined them with estimates from all other specific causes of death to ensure the sum matched total all-cause mortality for each age group, sex, location, and year. We then assigned meningitis cause mortality in a mutually exclusive and collectively exhaustive fashion using a separate set of models of the proportion of meningitis deaths due to each cause. These models were informed by a combination of vital registration and scientific literature data identified via systematic review. YLLs were calculated by multiplying age-specific deaths and remaining global life expectancy at age of death, which was estimated by GBD 2016 to be 86·6 years at birth.

15Non-fatal health outcomes were informed by a series of complementary data sources and models. First, we generated internally consistent estimates of overall meningitis incidence, prevalence, remission, and mortality using DisMod-MR 2.1, a Bayesian meta-regression tool developed for GBD. This model was informed by published literature identified via systematic review and supplemented with inpatient hospital and claims data that were corrected for readmission rates and ICD code position (primary vs non-primary) at discharge. ICD-coded hospital data served as the reference category. We then apportioned overall prevalence and incidence to underlying causes using a second set of DisMod-MR 2.1 proportion models informed by scientific literature and surveillance data reporting on specific infectious causes. We derived viral meningitis estimates by multiplying total meningitis and the ratio of viral-to-total meningitis from inpatient hospital sources. We calculated the incidence of each long-term cause-impairment combination by age group, sex, location, and year by multiplying the cause-specific case-fatality ratio by cause-specific incidence estimates, then applying post-discharge proportions of health consequences by cause, both of which were derived from a meta-analysis. We paired these data with corresponding long-term mortality data to derive prevalence estimates of long-term meningitis complications for each population. We then paired and multiplied final prevalence estimates with corresponding disability weights from the GBD Disability Weights Survey to calculate YLDs, which we then processed through a microsimulation framework to account for comorbidity between diseases. Long-term complications of meningitis included intellectual disa-bility, motor impairment, vision problems, hearing loss, epilepsy, and behavioural problems. The sum of YLLs and YLDs is disability-adjusted life-years (DALYs), a composite measure of overall burden.

16We calculated attributable burden of meningitis to the risk factors low birthweight and short gestation—the only two risk factors of the 84 risk factors that met GBD 2016 causal criteria for inclusion—but only for the early and late neonatal periods as these were the only age groups for which evidence was sufficient. We quantified the population attributable fractions of low birthweight and short gestation by pairing the estimated exposure to each component with the corresponding relative risk of mortality. Details on low birthweight and short gestation estimation are in the GBD 2016 Comparative Risk Assessment paper.

17We captured and propagated uncertainty through all calculations by sampling 1000 values (called draws) for each prevalence, death, YLL, YLD, or DALY estimate and summing draws across age, cause, and location for all intermediate calculations. 95% uncertainty intervals (UIs) were defined by the ordinal 25th and 975th draw values. We calculated change statistics on the mean values and at the draw level to derive UIs; significant increases and decreases were defined as 975 or more of the 1000 draws having a positive or negative value, respectively.

18The funder of the study had no role in study design, data collection, data analysis, data interpretation, or the writing of the report. All authors had full access to the data in the study and final responsibility for the decision to submit for publication.

19Summary results for meningitis burden are presented here. Complete results for GBD 2016 by age, sex, geography, year, and cause, including further detail not provided in this manuscript, are available in a set of interactive online visualisations called GBD Compare and are available for download from the GBD Results Tool at the Global Health Data Exchange.

203268 unique data sources were used for cause-specific mortality and 1348 unique sources for the estimation of non-fatal health outcomes from meningitis. More geographical and causal detail on dataset coverage for each component of estimation is shown in the appendix. All input sources and citations for each component are available from the GBD Input Data Sources Tool. The information on all the DisMod-MR 2.1 models, including input data and model performance, are explorable online, with corresponding information for CODEm also available.

21Incident cases of meningitis globally increased from 2·50 million (2·19–2·91) in 1990 to 2·82 million (2·46–3·31) in 2016. Age-standardised incidence was highest in the meningitis belt—ie, the peri-Sahelian countries of sub-Saharan Africa stretching from Senegal through South Sudan and Ethiopia (figure 1). Overall incidence in 2016 varied over 350 times, from a high of 207·4 (95% UI 183·9–233·9) per 100 000 population in South Sudan to a low of 0·5 (0·4–0·7) per 100 000 population in Australia. Much of central and western sub-Saharan Africa continues to have a high burden of meningitis, with few locations showing signs of sustained improvement and several countries in particular—Gabon, Equatorial Guinea, Nigeria, South Africa, and Zambia—having much higher meningitis death rates than would be expected on the basis of their SDI.

22The global number of meningitis deaths was estimated to have decreased by 21·0% from 403 012 (95% UI 319 426—458 514) in 1990 to 318 400 (265 218–408 705) in 2016 (table). Most (269 422 [84·6%] of 318 400) of these deaths occurred in countries in the low SDI (134 983 [95% UI 108 808–183 517]) and low-middle SDI (134 439 [109 866–171 733]) quintiles in 2016. The highest mortality rates were estimated for countries in the meningitis belt (table). However, of the top ten countries with the greatest absolute number of meningitis deaths—India, Nigeria, Ethiopia, Pakistan, Democratic Republic of the Congo, Uganda, Tanzania, Niger, Afghanistan, and China—four are located outside the meningitis belt (India, Pakistan, Afghanistan, China). Of the countries with at least 1000 meningitis deaths, in 2016 the countries with the largest declines were, in descending order, Brazil, China, Sudan, USA, Nepal, Haiti, Egypt, Indonesia, Myanmar, Niger, and India, all of which saw age-standardised declines in meningitis mortality of at least 50% between 1990 and 2016. Of these 11 countries, only Sudan and Niger are in the meningitis belt. Only Zimbabwe was estimated to have a significant increase in age-standardised meningitis mortality rate from 1990 to 2016 (table).

23Of the causes of incident meningitis, Haemophilus influenzae type b decreased the most (–49·1%), from 780 070 cases (95% UI 613 585–978 219) globally in 1990 to 397 297 cases (291 076–533 662) in 2016. Meningococcus was the leading cause of meningitis mortality in 1990 (192 833 deaths [95% UI 153 358–221 503] globally) but by 2016, the other causes category was the leading cause for both deaths (136 423 [112 682–178 022]) and incident cases (1·25 million [1·06–1·49]). Globally in 2016, 1·48 million (1·04–1·96) YLDs were due to meningitis compared with 21·87 million (18·20—28·28) DALYs, indicating that the contribution of mortality to meningitis burden is far greater than the contribution of disabling outcomes. Pneumococcus caused the largest number of YLDs (95% UI 634 458 [444 787–839 749]) in 2016, owing to its more severe long-term effects on survivors.

24Trends in deaths, incidence, and YLDs for causes of meningitis showed distinct patterns with respect to proportion of burden by age (figure 2), rates by age (figure 3), and changes in age-standardised rates from 1990 to 2016 (figure 4). In younger ages, meningococcal meningitis and other meningitis were the predominant cause categories in incident cases and deaths. Death rate and incidence increased, as did YLD rates, in the oldest age groups, with other meningitis and meningococcal meningitis causing most of the burden in people 80 years and older. Global mortality decreased the most for meningococcal meningitis between 1990 and 2016 to the extent that, by 2016, the proportion of meningitis deaths in the other cause category (eg, due to Listeria, Staphylococcus, Gram-negative bacteria, fungal, and viral infections) was slightly higher in cause fraction to meningococcus. The time trend of incidence and YLDs was less stark. With decreasing incidence and YLD rates of H influenzae type b and pneumococcal meningitis, we estimated increasing rates of other meningitis, a trend that started around 2005. Although 54·0% (1·53 million [1·23–1·95]) of incident cases and 46·0% (146 322 [114 583–216 388]) of all deaths occurred in children younger than 5 years and risk of neurological disability was highest under 1 year of age, the accumulated disability burden was estimated to be highest in adolescents and young adults, with sequelae such as intellectual disability, epilepsy, vision impairment, and motor impairment.

25Increasing SDI was consistently associated with lower age-standardised DALY rates due to meningitis (figure 5). Sub-Saharan Africa, without exception, had higher than expected age-standardised rates of meningitis DALYs, along with the Caribbean, north Africa, and the Middle East. Central Asia, southeast Asia, and eastern Europe all had rates in the 1990s that were higher than expected but improved compared with expectation by 2016. East Asia, high-income Asia Pacific, Andean Latin America, and central Latin America were the only regions with age-standardised meningitis DALY rates that were consistently lower than expected on the basis of SDI.

265·8% (4·5–7·0) of total meningitis DALYs were attributable to low birthweight and short gestation in 2016, with the highest rates of attributable DALYs occurring in the meningitis belt countries, Burundi, Malawi, Somalia, Zambia, and Afghanistan (appendix). All attributable burden was in neonates younger than 28 days, as described in the methods section.

27Global mortality from meningitis has declined, but substantial disparities in geographical distribution and age groups persist. In 2016, nearly 0·6% of all-ages deaths and close to 3% of the total in children younger than 5 years were due to meningitis (figure 3A). The estimates indicate that, while most incident cases are now due to infections in the other meningitis category, meningococcal infections continue to be a major cause of meningitis mortality, and most long-term disability is due to the after-effects of pneumococcal infection. The peak age of incident meningitis is during the neonatal period, deaths are highest in children younger than 5 years, and meningitis death rates are highest in many of the same locations where overall younger than 5 years mortality is highest. Rates of disability among survivors are highest in the youngest children, with a peak of prevalent disability therefore accumulating in adolescents and young adults.

28On the one hand, the overall figures provided by GBD 2016 are encouraging. Age-standardised death rates due to meningitis are estimated to have dropped globally by 21·0% from 1990 to 2016. On the other hand, the corresponding figures for other vaccine-responsive and preventable pathogen-based diseases have declined (eg, measles by 93·0%, tetanus by 90·7%, and syndrome-based diseases such as diarrhoea due to rotavirus by 57·9%), suggesting that progress in meningitis could be substantially faster. The need for accelerated action on meningitis across the globe was recognised in a high-level meeting in May, 2017, of more than 50 global experts, led by the Meningitis Research Foundation in collaboration with WHO, resulting in a commitment to formulate a global roadmap to defeat meningitis by 2030—a plan that required broad global commitments to strengthen systems for prevention, treatment, and surveillance.

29Vaccination is a successful intervention for preventing meningitis. H influenzae type b meningitis has historically been a common cause of childhood hearing loss and neurological disability, affecting 15%–30% of infected children, but widespread vaccination has virtually eliminated H influenzae type b in many high-income countries and has significantly reduced its incidence in many low-income countries. Widespread pneumococcal conjugate vaccination has also resulted in significant declines in invasive pneumococcal disease beyond the age groups targeted for vaccination, although post-licensure studies have detected an absence of efficacy against serotype 3. Pneumococcal conjugate vaccine generates herd protection by preventing nasopharyngeal carriage of bacteria targeted by the vaccine in people who are immunised, although corresponding increases in nasal carriage and infection by pneumococcal serotypes that are not covered has also been widely documented, suggesting that continued progress will require new and improved, higher-valency vaccines or vaccines that provide universal coverage against all pneumococcal strains.

30Epidemics remain a persistent threat in the meningitis belt countries of sub-Saharan Africa. Six of the ten countries with the highest number of meningitis deaths are in the meningitis belt, where a high incidence of meningococcal meningitis in particular has been linked to climatic factors—low humidity, dry winds, and high levels of dust in the air. Mass vaccination for MenA with MenAfriVac across the belt led to steep declines in incidence of meningococcal meningitis, but epidemics due to MenC, MenW, MenX, and pneumococcus still occur. Nearly 18 000 cases of MenC were reported in Niger and Nigeria in 2017, the largest recorded epidemic. As affordable MenC polysaccharide conjugate vaccines used to control epidemics are being phased out of production and the potential remains high for epidemics from other serogroups, development of an affordable multivalent meningococcal vaccine is needed for mass vaccination campaigns and introduction into routine vaccination schedules.

31Meningitis burden continues to be highest in resource-poor settings, which is not surprising given the confluence of factors predisposing to infections, epidemics, and poor outcomes. The proportion of survivors experiencing neurological disability is also estimated to be higher in people living in low-resource settings, owing to poorer access to acute syndromic care and fewer resources available for support of survivors with neurological sequelae. Even for people who can access services, out-of-pocket costs are the largest source of health expenditures in many low-income and middle-income countries; meningitis can lead to catastrophic health expenditures for poor households, which can lead to delays in seeking treatment or failure to seek treatment at all.,,,,

32In many areas of the world, especially rural areas, the capacity to do traditional diagnostic assays such as culture or PCR on CSF or blood samples is limited or non-existent. Limitations in capacity can be especially problematic in places where other infections, such as cerebral malaria, are common and can lead to delayed or missed diagnosis of bacterial or viral meningitis with resulting misclassification of deaths and under-representation of meningitis burden., Additionally, diagnosis of meningitis in children, especially neonates, is difficult; the symptoms of meningitis in neonates are often subtle and mirror those seen in neonates with sepsis or other infections, resulting in potential under-reporting. Current assays also have important limitations: sensitivity to detect pathogens varies by pathogen; no assay is able to detect or characterise N meningitidis serogroup X; latex agglutination tests can be expensive and at times unreliable; and the widely used Pastorex agglutination test, although able to detect antigens from H influenzae type b, S pneumoniae, and N meningitidis serogroups A, C, Y, W, B, and group B Streptococcus, requires cold-chain storage. Country-specific regulations that allow only already overburdened medically-trained providers to do lumbar punctures should be reconsidered because such policies can delay needed care. Efforts should be made to enhance training for other providers to search comprehensively for the source of severe acute infections and do lumbar punctures as needed. To facilitate such a change in practice, development and deployment of an affordable, sensitive, and specific rapid diagnostic test for the main bacterial meningitis pathogens—one that does not require specialised laboratory equipment or refrigeration—is urgently needed.

33Acceleration of declines in meningitis burden will require improved disease surveillance and more attention to the long-term implications and impairments for survivors. Enhanced national and regional surveillance would mean that countries can monitor meningitis locally and in turn improve the precision of global estimates. Although high-quality surveillance data are largely unavailable for most countries in sub-Saharan Africa, MenAfriNet is a counterpoint that can serve as a model for other countries. Comprising Mali, Niger, Burkina Faso, and Togo—four meningitis belt countries—MenAfriNet is a network that began doing high-quality case-based surveillance after MenA vaccine introduction. As MenAfriNet looks to expand, data sharing is crucial both to improve estimates and to show other countries how effective and accurate surveillance can be done. Improved awareness of the disease by populations and governments is also key, as is improved support for survivors and their families.

34Aside from GBD, no other global, cause-specific meningitis estimates are available across all age groups. WHO estimates cause-specific mortality due to meningitis using GBD cause-fractions for countries that do not have high-quality vital registration data and released cause-specific estimates for 2000 and 2016. WHO estimated only 141 deaths (95% CI 119–150) globally for meningitis in neonates younger than 28 days, 45 064 deaths (35 854–54 523) in infants aged 1–11 months, and 37 573 deaths (29 690–43 857) in children aged 1–4 years compared with GBD 2016, which estimated 8911 deaths (6074–12 348) in infants aged 0–6 days, 10 157 deaths (7812–15 141) in infants aged 7–27 days, 70 841 deaths (55 018–104 486) in infants aged 1–11 months, and 56 412 deaths (42 274–88 597) in children aged 1–4 years (see GBD Compare). The total number of all ages meningitis deaths estimated by WHO in 2016 (278 812 [226 461–323 217]), was also substantially lower than GBD 2016 (318 400 [95% UI 265 218–408 705]), although cause fractions were in closer agreement (0·49% for WHO in 2016; 0·58% [95% UI 0·48–0·75] for GBD 2016). Other publications reporting on meningitis epidemiology for particular pathogens or locations revealed some differences from the present analysis. First, in neonates, GBD 2016 estimated more meningococcus and less other meningitis than two studies that found most meningitis cases were due to group B Streptococcus and Escherichia coli, along with Listeria (in the high-income setting) or S pneumonia and Klebsiella (in the low-income and middle-income setting)., Second, combining mortality and non-fatal results to derive case-fatality ratios, GBD 2016 also estimated meningococcal meningitis to have a higher case-fatality rate than pneumococcal meningitis, a finding that contrasts with several studies that suggest lower case-fatality rate in meningococcal meningitis,, except occasionally in the setting of epidemics. These inconsistencies suggest that although case-fatality rate might be higher for meningococcal meningitis overall, case-fatality rate might vary by pathogen across specific age groups and geographic settings. It is important to note the comparatively high uncertainty of cause-specific meningitis estimates and to acknowledge the need to ensure all available evidence on mortality is used to inform non-fatal estimates and vice versa.

35This study has several limitations. First, it shares the same limitations as the overall GBD 2016 study, which are detailed elsewhere.,,,, A major limitation is the relative paucity of data on distribution and severity of long-term sequelae in survivors, which for meningitis is especially notable given the high rates of neurological disability that can follow neonatal meningitis. Learning-related disabilities have not been included in GBD, potentially leading to burden underestimation in conditions such as meningitis. GBD criteria for risk-outcome pair criteria, which assess causal links from the perspective of risks rather than outcomes, are not ideal for assessing comparatively rare outcomes like meningitis. Second, cause-specific meningitis estimates have a high degree of uncertainty owing to a relative paucity of pathogen-specific data. Although surveillance data can be extremely variable in quality, comprehensiveness, and completeness, if methods could be developed to identify and incorporate high-quality surveillance data into GBD modelling, such data could further improve the temporal and geographical precision of overall and pathogen-specific estimates. Third, because GBD 2016 provides estimates for only three of the major bacterial causes of meningitis, important neonatal and infant pathogens, such as group B Streptococcus, Listeria, Staphylococcus, and E coli, are all grouped together; we aim to generate separate estimates for additional pathogens in future analyses. With release of a group B Streptococcus vaccine on the horizon, it will become crucial to obtain country-specific burden estimates to guide countries in planning and implementing vaccine strategies. Meningitis due to Mycobacterium tuberculosis and HIV-associated opportunistic infections were not included because GBD classifies them with the underlying diseases. Fourth, although SDI is a useful tool for exploring epidemiologic transition, it is incomplete because it does not capture important aspects of societal function, such as political stability, gender equity, urbanicity, technology penetrance, or infrastructure, which are important social determinants of health.

36Meningitis burden remains unacceptably high in 2016, despite evidence that population-level vaccination campaigns can effectively prevent many causes of meningitis. Detailed country-specific meningitis burden estimates are crucial advocacy tools, empowering patients, caregivers, and civil society organisations to ask more from governments and funders to help target improved vaccination rates, to foster development of—and improve access to—rapid diagnostic tests, and to ensure that appropriate antibiotics for the most common causes of meningitis are included in the list of essential medicines. High rates of meningitis disability shine a spotlight on the need to strengthen health systems to provide additional educational support, aids and therapy for vision and hearing problems, and adaptation of the built environment for those with mobility issues. The inclusion of reliable surveillance data as inputs into GBD modelling will improve future meningitis estimates and will help prevent more people from dying from or having to live with the sequelae of meningitis.

37Correspondence to: Dr Nicholas J Kassebaum, Global Health, Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, 98121, USA nickjk@uw.eduorProf Joseph R Zunt, Departments of Neurology, Global Health, Medicine (Infectious Diseases) and Epidemiology, University of Washington, Seattle, WA 98104, USA jzunt@uw.edu

38Correspondence to: Dr Nicholas J Kassebaum, Global Health, Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, 98121, USA nickjk@uw.edu

39or

40Prof Joseph R Zunt, Departments of Neurology, Global Health, Medicine (Infectious Diseases) and Epidemiology, University of Washington, Seattle, WA 98104, USA jzunt@uw.edu