Clinical features and long-term outcomes of children with pneumococcal meningitis in China: a 10-year single-centre retrospective analysis (2025)

Background

Pneumococcal meningitis is a central nervous system infection responsible for high rates of mortality and long-term disability worldwide. We aimed to summarise the clinical characteristics, as well as determine the risk factors for mortality and long-term progression of paediatric pneumococcal meningitis (PM).

Methods

This retrospective study included children with microbiologically confirmed PM who were admitted to Beijing Children's Hospital between 2012 and 2021. The laboratory examination results and clinical characteristics of the enrolled patients were analysed. All were followed-up with over the phone, and their long-term prognoses were scored based on the Glasgow Outcome Scale-Extended scale. Factors associated with long-term unfavourable outcomes were identified using single-factor and multivariate analyses.

Results

Totally 301 children with PM were included. Only 22 (7.31%) were vaccinated against Streptococcus pneumoniae (SP) prior to disease onset. The median age at admission was 12.00 (7.00–44.08) months. The median hospitalisation length was 22.00 (14.50–33.00) days. A total of 190 patients (63.12%) experienced neurological complications during their clinical course. A total of 278 children were followed up for a median of 61.61 (41.84–85.42) months, 44 (15.83%) died, and 75 (32.05%) of the 234 surviving children experienced sequelae. In terms of their Glasgow Outcome Scale-Extended (GOS-E) Scale scores, there were 199 cases (71.58%) in the good prognosis group (1–2 points) and 79 cases (28.42%) in the poor prognosis group (3–8 points). Multivariate analysis showed that sex, recurrent seizures or convulsive status, muscle tone changes, dyspnoea, and lower peripheral white blood cells (WBCs) count in the acute phase were independent risk factors for poor prognosis (all P < 0.05).

Conclusions

PM is associated with high rates of fatality and morbidity in children. Approximately 1/3 of children who survive PM experiences neurological sequelae and require long-term rehabilitation training that poses significant burdens on the society and country.

Invasive pneumococcal disease (IPD), usually characterized by the isolation of SP from normally sterile bodily sites, is associated with high morbidity and mortality. PM is the most serious type of IPD with the highest mortality rate, and can have potentially devastating consequences [1]. In the Defeating Meningitis by 2030 plan, the World Health Organisation reported on the global meningitis disease burden, as evaluated via the "years of life lost due to disability" indicator, where PM ranked first [2].

PM is often accompanied by severe neurological complications during its clinical course, leading to 30–50% of surviving patients experiencing permanent neurological sequelae that poses significant burdens on their families and societies, including dyskinesia, hearing impairment, vision impairment, epilepsy, and others [3]. Identifying the risk factors that led to poor long-term prognosis in children with PM may aid in stratifying the management strategies used for their diseases and improving their prognoses through more individualised treatment approaches. A systematic review found that disturbances in consciousness, dyspnoea, and decreased white blood cell count were associated with poor prognosis [4]. Although several studies on the epidemiology and microbiology of PM in children have been conducted, relatively few reports have described the long-term outcomes and determining factors of the long-term prognoses of children with PM in China.

This study aimed to summarise the clinical characteristics, evaluate the long-term prognoses of children with PM in China. Additionally, we identified the main risk factors associated with poor long-term clinical prognosis in this patient demographic.

Research design and definitions

This was a single-centre retrospective observational study of microbiologically- confirmed PM in children aged 29days to 18years who had been admitted to the Beijing Children's Hospital between January 2012 and December 2021. PM was defined by the following four criteria: (1) Acute fever (axillary temperature > 38.0℃ or rectal temperature > 38.5℃); (2) One or more of the following clinical symptoms: headache, positive meningeal irritation sign, and consciousness change; (3) At least one of the following findings in the cerebrospinal fluid (CSF): WBC > 100 × 10⁶/L; WBC (10–100) × 10⁶/L with increased protein (> 1g/L) or decreased glucose level < 40mg/dL (2.22mmol/L); (4) Blood or CSF culture containing SP, or a positive SP antigen test of the CSF.

Data collection

The following data were retrospectively obtained from the medical records of each patient: (1) baseline characteristics (e.g., age, sex, time of admission and discharge, vaccination status); (2) clinical symptoms (e.g., fever, vomiting, convulsions); (3) acute period (within 1week of symptom onset) laboratory tests (e.g., peripheral blood C-reactive protein [CRP], blood WBC count, CSF WBC count); (4) pathogen information (e.g., pathogen identification method, drug susceptibility tests); (5) treatments (e.g., antibiotics and/or glucocorticoids); (6) complications (e.g., subdural effusion, hydrocephalus, ependymitis); (7) outcomes (all patients underwent neurological examinations and the outcome was scored according to the Glasgow Outcome Scale [GOS] [5] at discharge; Supplementary Table1).

Identification and serotyping of SP

Identification: (1) Streptococcus pneumoniae antigen detection (Cat. No. 600107, Beijing Bell Biotechnology Co., Ltd.) was used to identify SP in CSF or blood, (2) Isolation of SP in CSF or blood: the collected samples were inoculated on Columbia blood agar plates containing 5% sheep blood and cultured at 35°C in an incubator containing 5% CO2 for 16–20h. Then the morphology of the colonies was examined. The Optochin susceptibility test identified SP (inhibition zone diameter > 14mm). For the identified SP, the colony morphology described in the Manual of Clinical Microbiology, 10th edition (Spellberg and Brandt, 2011) [6].

Serotyping: Pure SP suspension obtained after overnight incubation was subjected to capsule swelling reaction test with Danish SP rabbit antiserum (SSI, Denmark) to identify the serotype of S. pneumoniae. A positive capsule swelling test is defined as a more obvious "swelling" of SP when a specific type of antibody binds to the capsule of SP and causes a change in its refractive index [7].

Antimicrobial susceptibility testing

Antimicrobial susceptibility was tested by establishing the minimum inhibitory concentration (MIC) value using an E-test (Liofilchem, Italy) for penicillin ceftriaxone, clindamycin, erythromycin, tetracycline, levofloxacin, trimethoprim/sulfamethoxazole, and vancomycin.

Follow-up

The children were followed-up with via telephone by a trained paediatrician. The follow-up time was at least 12months, and their prognoses were scored according to the Children's GOS-E Scale [8]. Good outcomes were defined as scores of 0–2, and scores of > 3 points were shown to correlate with poor prognosis. The sequelae mainly included hearing impairment, visual impairment, dysgnosia, dyskinesia, cognitive impairment, affective impairment, language impairment and epilepsy.

Statistical analysis

Statistical analysis was performed by using SPSS version 25.0. Continuous variables are described using medians with standard deviations for normally distributed data, and by medians with interquartile ranges (IQR) for non-normally distributed data. Student’s t-tests or Mann–Whitney U tests were used for quantitative data. Qualitative data are described as the number of cases (%), and comparisons between groups were performed using Chi-squared or Fisher’s exact tests. Statistical significance was set at P < 0.05 (two-tailed). Factors with a P value < 0.05 in univariate analysis were included in multivariate analysis, and forward selection was used to screen independent variables to explore the independent risk factors affecting the long-term prognoses of children with PM.

Patient characteristics

Between February 2012 and December 2021, a total of 301 children who were diagnosed with PM and admitted to our hospital were included in this study (Fig.1). Of these, 189 were male (62.79%) and 112 were female (37.21%). The median age was 12.00 (7.00–44.08) months and 142 (47.18%) were < 1year. Totally, 101 cases (33.55%) occurred duringthe winter months. Only 22 patients (7.31%) had been vaccinated against SP before disease onset. The median duration of hospitalisation was 22.00(14.50–33.00) days. A total of 92 patients (30.56%) had risk factors for developing PM, among which 37 (12.29%) had intracranial structural malformations, 36 (11.96%) had CSF leakage and 23 (7.64%) had traumatic brain injury. In addition, 10 children (4.31%) had primary immunodeficiency and five (1.66%) had secondary immunodeficiency (Table1). All of the children had fever (100%), followed by weak mental responses in 240 (79.73%), vomiting in 177 (58.80%), consciousness changes in 163 (54.15%), muscle tone changes in 82 cases (27.24%), and positive signs of meningeal irritation were observed in 139 cases (46.18%; Table1).

Complications

In total, 190 of the children (63.12%) developed neurological complications during the clinical course of their disease, the most common of which were subdural effusion (113 cases, 37.54%), hearing impairment (66 cases, 21.93%), hydrocephalus (55 cases, 18.27%), cerebral haemorrhage (20 cases, 6.64%), and cerebral hernia (16 cases, 5.32%; Table2). Of these, 28.32% (32/113) of the children with subdural effusion received subdural effusion puncture or subdural effusion drilling and drainage as treatment. A total of 44.44% (24/55) of the children with hydrocephalus underwent external ventricular drainage or ventriculoperitoneal drainage. Totally, 223 patients (74.09%) had concurrent infections in other parts of their bodies. Pneumonia was the most common infection (122 cases, 40.53%), followed by sepsis (112 cases, 37.21%) and mastoiditis (33 cases, 10.96%; Table2).

Laboratory test

The median peripheral blood WBC count was 13.58 (7.82–20.90) × 10⁹/L, and median CRP was 125.00 (43.50–170.00) mg/L. The median CSF WBC was 878.00 (215.50–2,726.50) × 10⁶/L. The median CSF protein and glucose levels were 2.02 (1.15–2.94) mg/L and 0.72 (0.18–2.00) mmol/L, respectively (Table1).

Serotype distribution and antibiotic susceptibility

All the selected cases had a clear pathogenic diagnosis. Among them, 187 (62.13%) were positive for CSF culture, 114 (37.87%) were positive for peripheral blood culture, and 217 (72.09%) tested positive for the SP antigen in CSF samples. Overall, 171 cases had drug susceptibility results, among which erythromycin (97.01%), clindamycin (96.61%), and tetracycline (88.52%) had the highest non-sensitivity rates; followed by cefepime (66.29%) and penicillin (64.67%). All the strains (100%) were sensitive to vancomycin, linezolid, and moxifloxacin (Table3).

Serotyping results for 52 strains of SP showed that 19F (36.54%) was the most common (Supplementary Table2). The coverage rate of the 13-valent vaccine was 92.31%. Twenty strains were subjected to multilocus sequence typing (MLST), of which three were found to be novel strains (Supplementary Table3).

Therapies and clinical outcomes

All patients received antimicrobial therapy, and the median time until the start of empirical therapy was 2.00 (1.00–3.00) days. The most used drugs were carbapenems (60.43%), third-generation cephalosporins (57.91%), and glycopeptides (52.88%). Empirical antibiotic therapy met the recommendations of the guidelines (using third-generation cephalosporins combined with vancomycin) [9]. Between 2012 and 2016, the compliance rate of empirical medication was 56.99%, and 64.94% between 2017 and 2021. The most used target drugs were third-generation cephalosporins (63.60%), glycopeptides (61.84%), linezolids (60.07%), and carbapenems (55.65%). Glucocorticoids were used as an adjunctive treatment in 208 children (69.10%), including 171 cases treated with dexamethasone (56.81%), 54 treated with methylprednisolone (17.94%), and five treated with prednisone (1.66%).

According to the GOS results, 217 patients (72.09%) had good outcomes (GOS = 5 points) at discharge, while 84 patients (27.91%) had poor outcomes at discharge (GOS = 1–4 points), including 29 dead patients (9.63%).

Characteristics of recurrent cases

PM recurred occurred in 10 of the cases (3.32%), including five cases where it recurred for the second time, four cases for the third time, one case for the fourth time, and one case for the sixth time. Compared to the non-recurrence group, the recurrence group had more female children (χ² = 6.323, P = 0.012). The median age of the recurrence group was higher than that of the non-recurrence group (82.50 [61.75–119.00] vs. 12.00 [7.00–41.00] months, P < 0.001). The children in the recurrence group had more intracranial structural malformations (χ² = 37.726, P < 0.001) and CSF leakage (χ² = 39.038, P < 0.001). However, this group also had fewer neurological complications (χ² = 6.737, P = 0.009), and only two out of the 10 cases experienced neurological complications—specifically, hydrocephalus and hearing impairment (Supplementary Table4).

Long-term follow-up

We attempted to conduct long-term follow-ups via telephone, following discharge from hospital, for all 301 enrolled children. Of these, the parents of 278 were successfully contacted and thus had follow-up data. The median follow-up time was 61.61 (41.84–85.42) months. Among these, 44 (15.83%) died, for an overall mortality rate of 15.83%. The mortality rate for the children < 1year old was 21.21% (28/132), and that of the children ≥ 1year old was 11.64% (17/146). Amongst the 234 surviving children, long-term sequelae occurred in 75 (32.05%), with the most common being hearing impairment (41 cases), dyskinesia (28 cases), language impairment (23 cases) and dysgnosia (21 cases; Table2).

According to the children's GOS-E scale scores, 199 (71.58%) were classified into the good prognosis group (1–2 points) and 79 (28.42%) were in the poor prognosis group (3–8 points). The proportion of patients with poor prognoses between 2012 and 2016 was 28.04%, whereas between 2017 and 2021 was 28.65%.

Prognosis and risk factors analysis

Univariate analysis showed that female sex, age < 1year, changes in consciousness, recurrent seizures or convulsive status, muscle tone changes, dyspnoea, pneumonia, otitis media, hydrocephalus, cerebral haemorrhage, cerebral veno-sinus thrombosis, hearing impairment, cerebral hernia and antibiotic treatment (> 3days) were associated with poor prognosis (all P < 0.05; Table4).

The concentration of CSF protein in the acute phase in the poor prognosis group was higher than that in the good prognosis group (2,770.00 [1,361.50–4,279.00] mg/L vs. 1,705.50 [1,057.53–2,568.25] mg/L, P < 0.001). The CSF glucose level in the poor prognosis group was lower than that in the good prognosis group, (0.38 [0.07–1.36] mmol/L vs. 1.00 [0.20–2.15] mmol/L, P < 0.05). The peripheral blood WBC count in the poor prognosis group was lower than that in the good prognosis group in the acute phase (7.94 [4.11–15.49] × 10⁹/L vs. 15.98 [10.64–21.56] × 10⁹/L, P < 0.001). And the CSF WBC count in the poor prognosis group was lower than that in the good prognosis group (427.00 [97.50–1,167.00] × 10⁶/L vs. 1,074.00 [396.50–3,137.00] × 10⁶/L, P < 0.001; Table5).

Multivariate analysis showed that sex, recurrent seizures or convulsive status, changes in muscle tone, dyspnoea, and decreased peripheral blood WBC counts in the acute phase were independent risk factors for poor prognosis (all P < 0.05; Table6).

Studies have shown that SP often colonises the nasopharyngeal mucosa in humans, and PM is more likely to be induced in the locations of anatomical defects such as intracranial structural malformations, CSF leakage, host immune function deficiency or reduction, or underlying diseases [10]. In this study, about 30% of the children with PM had such predisposing factors, the most common of which were intracranial structural abnormalities and CSF leakage. This may have been related to improper brain development in the children. All 10 children with recurrent PM had intracranial anatomical defects, which may explain the recurrence of their diseases.

The clinical features of PM include fever, vomiting, convulsions, altered consciousness, and meningeal irritation. Studies have shown that altered consciousness, convulsions, and dyspnoea in children with BM upon admission are associated with death and poor prognosis [11]. In this study, all children had fever at the time of admission, with common clinical manifestations that included weak mental responses, vomiting, convulsions, and altered consciousness. These were also accompanied by signs of meningeal irritation, muscle tone changes, and dyspnoea. Multivariate analysis showed that recurrent seizures or convulsive status, muscle tone changes, and dyspnoea were independent predictors of poor prognosis in children with PM (P < 0.05), suggesting that clinicians should pay attention to early clinical manifestations in children, actively reduce intracranial pressure, control seizures, and improve clinical conditions such as dyspnoea to reduce the occurrence of poor PM prognosis.

Patients with PM often experience several clinical complications. Studies have shown that 71% of infants (< 1year old) with BM develop neurological complications, compared to 38% of children aged 1–5years and 10% of children aged 6–16years [12]. Infants are therefore at higher risk of developing neurological complications than older children. In this study, the median age of the children was 12months, and 63.12% experienced neurological complications. The top five were subdural effusion, hearing impairment, hydrocephalus, intracranial haemorrhage and encephalorrhagia. The highest incidence of neurological complications seen in this study was in the younger patients. However, our research unit is located in the National Children's Medical Centre, which accepts more severe paediatric cases. This may also explain the high incidence of complications observed in our study cohort.

Typical changes in the CSF during BM infection include increased protein and decreased glucose concentration. Vasilopoulou et al. [13] found that a decrease in CSF glucose concentration (< 40mg/dL) and an increase in protein concentration (> 1.00g/L) were high-risk factors for poor prognosis in children with BM. Previous studies from our group have shown that CSF protein > 1.00g/L and CSF glucose < 1.5mmol/L may indicate a poor prognosis in children with BM [14]. Here, the median CSF protein concentration was significantly higher in the poor prognosis group than in the good prognosis group (P < 0.001), while the median CSF glucose concentration was significantly lower in the poor prognosis group compared to the good prognosis group (P < 0.05). The balance between the inflow across the blood–brain barrier and outflow with CSF determines the CSF protein content [15]. Therefore, the increase in protein levels may be related to the degree of blood–brain barrier damage or circulation in the CSF.

Studies have shown that leukopenia is associated with disease progression and death [11]. In this study, the number of WBCs in the peripheral blood and CSF during the acute phase was lower in the poor prognosis group than in the good prognosis group (P < 0.001). Of the WBCs, a decrease in peripheral blood leukocyte count was determined to be an independent risk factor for poor PM prognosis. A decrease in WBCs in the peripheral blood and CSF may indicate an insufficient immune response [16], which can lead to further aggravation of the infection.

With the continuous development of vaccines and antibiotics, the treatment of PM in children has improved. However, reports of antibiotic resistance have been increasing. According to data from the Centers for Disease Control of the United States, ~ 30% of SP infections are resistant to one or more antibiotics, and ~ 1.2 million infections are caused by drug-resistant SP strains annually [17]. Among the strains tested in this study, erythromycin (97.01%), clindamycin (96.61%) and tetracycline (88.52%) were identified to be the least effective antibiotics for treating the PM cases of our patient cohort, followed by cefepime (66.29%) and penicillin (64.67%). No correlation was found between drug resistance and poor prognosis, suggesting that disease prognosis may be related to the pathogen and host as well as the interaction between the host and pathogen. However, most of the drug sensitivity results for the children in this study came from the patients’ medical records pulled from other hospitals which had some missing data. The detection methods and test drugs used in different medical institutions can also vary, which may have affected the results.

Advances in molecular biological technology have enabled the development of SP serotyping [18]. However, because most children included in this study received antibiotic treatments in other hospitals prior to admission, the positive culture rate in our hospital was low. The pathogen serotype was identified in only 52 of the children, among which 19F was the most common (19 strains, 36.54%). Studies have shown that the most common serotypes associated with invasive infections are 14, 4, 1, 6A, 6B, 3, 8, 7F, 23F, 18C, 19F, and 9V; and that pathogenic serotypes vary with age—with serotypes 6, 14, 18C, 19F, and 23F being the most common in children) [19, 20]. The use of the SP vaccine has significantly reduced the incidence of IPD caused by bacterial serotypes covered by the vaccine [21]. In this study, 92.31% of the serotypes detected were covered by the pneumococcal conjugate vaccine (PCV)13, but only 7% of the children enrolled were vaccinated with the SP vaccine. This finding highlights the importance of PCV vaccination in children regarding preventing PM.

Treatment delay is a known risk factors for neurological complications in BM; however, there is no unified definition of the delay duration. In one study, the incidence of neurological complications was lower (40%) in children with disease durations of < 48h than in those admitted after 48h [12]. In another study, late diagnosis (i.e., patients who were not treated with lumbar puncture or antibiotics at admission because BM was not considered) was associated with increased hospital mortality [22]. For patients with BM, broad-spectrum antibiotics should be administered early (within 1–3h) to reduce mortality [23]. In this study, the children who received their first dose of antibiotic treatment time > 3days following symptom onset had a worse prognosis than those who received their first dose within ≤ 3days of symptom onset (P = 0.027). Therefore, early antibiotic treatment is crucial for improving the prognosis of PM. In our study, the rate of empirical antibiotic therapy administered in accordance with the guidelines increased, from 56.99% in 2012–2016 to 64.94% in 2017–2021, indicating that the treatment that clinicians administer for this disease is becoming increasingly standardised.

The mortality rate of the children with PM in the present study was 15.83%. The mortality rate of children aged < 1year was significantly higher than that of children aged ≥ 1year (21.21% vs. 11.64%). Univariate analysis showed that age < 1year was significantly associated with poor prognosis. Additionally, Sadie et al. showed that young age (< 1year old) is an important factor affecting poor prognosis in case of PM [12]. Nyasulu et al. found that age < 1year old was an independent risk factor for death caused by PM [24]. This may be related to the underdeveloped blood–brain barrier and low autoimmune function in young children, who are more prone to invasive infections. Therefore, close attention should be paid, in clinical practices, to the diagnosis and treatment of such patients. Approximately 30% of the children in this study experienced long-term sequelae that caused significant burdens to their families and society. The most common of these including hearing impairment, dyskinesia, language impairment and dysgnosia. Early identification of patients with poor prognoses and active individualised treatment is particularly important in this patient demographic.

This study retrospectively analysed the data of children with PM in our centre over the past 10years. We summarised the clinical characteristics of children with PM and analysed the risk factors affecting poor prognosis, to lay a foundation for early identification and intervention in patients with poor prognoses. However, this study was also subject to certain key limitations worth noting. First, this was a single-centre retrospective study, so the results merit verification in the further prospective multi-centre studies. Second, some of the study subjects first visited other hospitals. Therefore, some of the acute laboratory tests and drug sensitivity results came from other institutions, meaning that the methods and standards used may have differed and thus impacted our results. At present, there is still a lack of research on the long-term prognosis of children with PM. More prospective multi-centre studies with large sample sizes are warranted to further clarify the risk factors for poor prognosis in cases of PM, for more active treatment strategies to be adopted to improve the prognoses of children with PM and reduce the burden of this disease on families and society.

IPD:

Invasive pneumococcal disease

SP:

Streptococcus pneumoniae

PM:

Pneumoniae meningitis

CSF:

Cerebrospinal fluid

WBC:

White blood cell

GOS:

Glasgow Outcome Scale

GOS-E:

GOS–Extended

IQR:

Interquartile range

MLST:

Multilocus sequence typing

BM:

Bacterial meningitis

PCV:

Pneumococcal conjugate vaccine

OR:

Odds ratio

CI:

Confidence intervals

Author notes

1. Yixuan Li and Binglin Jian contributed equally to this work.

Authors and Affiliations

1. Department of Infectious Diseases, Key Laboratory of Major Diseases in Children, Ministry of Education, Beijing Children’s Hospital, Capital Medical University, National Centre for Children’s Health, Beijing, 100045, China

   Yixuan Li,Binglin Jian,Xuekai Kang,Mengjie Zhang,Lingyun Guo,Xue Ning,Liang Zhu,Tianming Chen,Bing Hu,ShuPing Liu,Haijuan Xiao,Xin Guo,Wenya Feng,Zhenzhen Dou,Linlin Liu,Qinjing Li,Bing Liu&Gang Liu

2. Research Unit of Critical Infection in Children, Chinese Academy of Medical Sciences, Beijing, 2019RU016, China

   Yixuan Li,Binglin Jian,Xuekai Kang,Mengjie Zhang,Lingyun Guo,Xue Ning,Liang Zhu,Tianming Chen,Bing Hu,ShuPing Liu,Haijuan Xiao,Xin Guo,Wenya Feng,Zhenzhen Dou,Linlin Liu,Qinjing Li,Bing Liu&Gang Liu

Contributions

All of the authors had access to the full dataset (including the statistical reports and tables) and take responsibility for the integrity of the data and the accuracy of the data analysis. LG, GLY, HB, CTM, GX, FWY, XHJ, ZL, NX, LSP conceived and designed the study. ZMJ, KXK, JBL, LYX, DZZ, ZL, GLY, NX, LLL, LQJ, LB collected the data and designed the analysis. LG, LYX, JBL, KXK and interpreted the data. LYX and JBL wrote the first draft of the paper. GL reviewed and approved the final report.

Corresponding author

Correspondence to Gang Liu.

Ethics approval and consent to participate

This study was reviewed and approved by the Ethics Committee of Beijing Children’s Hospital Affiliated to Capital Medical University (IEC‑C‑008‑A08‑V.05.1). This is a retrospectively study, we obtained the data of patients from the Medical Records and Statistics Room and we analyzed the data anonymously; thus, informed consent was not required. The Ethics Committee of Beijing Children’s Hospital Affiliated to Capital Medical University waived the consent to participate for this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions