Seasonal upsurge of pneumococcal meningitis in the Central African Republic

A high incidence of bacterial meningitis was observed in the Central African Republic (CAR) from December 2015 to May 2017 in three hospitals in the northwest of the country that are within the African meningitis belt. The majority of cases were caused by Streptococcus pneumoniae (249/328; 75.9%), which occurred disproportionately during the dry season (November-April) with a high case-fatality ratio of 41.6% (95% confidence interval [CI] 33.0, 50.8%). High rates of bacterial meningitis during the dry season in the meningitis belt have typically been caused by Neisseria meningitidis (meningococcal meningitis), and our observations suggest that the risk of contracting S. pneumoniae (pneumococcal) meningitis is increased by the same environmental factors. Cases of meningococcal meningitis (67/328; 20.4%) observed over the same period were predominantly group W and had a lower case fatality rate of 9.6% (95% CI 3.6, 21.8%). Due to conflict and difficulties in accessing medical facilities, it is likely that the reported cases represented only a small proportion of the overall burden. Nationwide vaccination campaigns in the CAR against meningitis have been limited to the use of MenAfriVac, which targets only meningococcal meningitis group A. We therefore highlight the need for expanded vaccine coverage to prevent additional causes of seasonal outbreaks.


Introduction
The northern districts of the Central African Republic (CAR) form part of the meningitis belt, a broad swathe of sub-Saharan Africa where the incidence of bacterial meningitis typically peaks during the dry season 1 . The mechanism for the increase in cases of bacterial meningitis has been postulated to be a result of damage to host mucosal defenses by the extreme environmental conditions of the dry season in this region, resulting in an increased rate of conversion from asymptomatic carriage to invasive disease 2 . The CAR is among the world's least developed nations and large areas of the country remain unstable following conflict between ethnic groups in 2013 and 2014. Consequently reliable medical data is scarce.
While outbreaks in the meningitis belt, including the CAR, have historically been caused by meningococcus group A 3,4 , national vaccination campaigns with MenAfriVac, which targets exclusively this group, has resulted in a shift in the epidemiology of bacterial meningitis. A surveillance study in ten African countries found that since MenAfriVac was introduced in certain countries in 2010, the proportion of confirmed cases due to meningococcus group A dropped substantially, while the proportion attributable to meningococcus group W and pneumococcus increased over the same period 5 . The CAR began nationwide vaccination with MenAfriVac in 2016, though coverage levels have not been assessed.
Here we report on findings from routinely collected data on meningitis patients in health facilities supported by the medical organisation Médecins Sans Frontières (MSF) in Bossangoa, Ouham Prefecture from 1st December 2015 to 31st May 2017, along with additional data from MSF supported hospitals in Batangafo, Ouham Prefecture and Paoua, Ouham-Pendé Prefecture within the same period. These towns are all located in the northwest of CAR close to the Chadian border and are separated by linear distances of 130-200km (see map, Figure 1).

Patients
Data were collected prospectively from three reporting hospitals (shown in Figure 1) from December 2015 to May 2017 as part of routine communicable disease surveillance. Our confirmed case criteria and clinical management of patients followed the guidelines from the World Health Organization 6 . Our analysis included all confirmed cases with bacterial meningitis in the three hospitals over the observed periods. Data on suspected and negative cases were collected in Paoua using established definitions 5 . Patient outcomes were recorded in Bossangoa and Batangafo along with socio-demographic characteristics (age, sex, residence) in all regions. Patient outcomes and incidence rates have previously been reported from Paoua Hospital over a subset of our observation period 7 .

Amendments from Version 1
This version of the manuscript was revised in light of comments from the three reviewers. We include information on suspected and negative cases in one hospital where data was available. We include information on serotyping from a small number of S. pneumoniae isolates. We have also restructured the manuscript, with an expanded Introduction and have added a Discussion section.

Statistical analysis
Patient data were recorded using Microsoft Excel and statistical analysis was performed in R (version 3.5.1). Logistic regression and a chi-squared test for counts were performed. Statistical significance was defined as p<0.05 or non-overlapping 95% confidence intervals.

Ethical Statement
Médecins Sans Frontières (MSF) is able to operate in countries such as the Central African Republic only with the support of national and local authorities and through continued dialogue with the beneficiary communities.
The high incidence of meningitis cases during this period was known to the communities and MSF gave health advice and provided medical services during the outbreak. Communities in the CAR were made aware that MSF collects data as part of its medical operations, including for research purposes. This research fulfilled the exemption criteria set by the MSF Ethical Review Board (ERB) for a posteriori analyses of routinely collected clinical data and thus did not require MSF ERB review. Data collected from patients was anonymized, though patient consent was not sought retrospectively. This study was conducted with permission from the Medical Director Sidney Wong (MSF-Operational Centre Amsterdam).

Overview of cases
In Bossangoa Hospital, 139 cases of confirmed bacterial meningitis were reported over 18 months (Dec 2015 -end May 2017). The median case age was 16 years (Interquartile range [IQR] 7, 32) and 52.2% of patients were male. The majority of confirmed cases were caused by Streptococcus pneumoniae (pneumococcal meningitis); 114/139 (82.0%), rather by than Neisseria meningitidis (meningococcal meningitis) and this was also observed in the majority of confirmed meningitis cases in Paoua Hospital over 17 months (93/106, 87.7%) and Batangafo Hospital over 12 months (42/83, 50.6%), see Table 1 and Figure 2 for cases of pneumococcal meningitis by reporting Hospital. Across all sites, the majority of confirmed cases of bacterial meningitis were caused by S. pneumoniae (249/328; 75.9%) followed by N. meningitidis (67/328; 20.4%) and Haemophilus influenzae (12/328; 3.7%); see Figure 3 for cases by causative agent over time.
There were 42 suspected cases of bacterial meningitis in Paoua Hospital over 17 months and 409 individuals that tested negative. The proportion of suspected cases relative to the total number of suspected, negative and confirmed cases (42/557; 7.5%) is low compared to other outbreaks of pneumococcal meningitis 9 , suggesting that the majority of patients arriving with a clinical suspicion of bacterial meningitis were laboratory tested. The proportion of patients testing positive from latex agglutination test of cerebrospinal fluid (106/515; 20.6%) is comparable to a contemporaneous outbreak of pneumococcal meningitis in Ghana (23.8%) 9 .
The temporal distribution of cases over the observed period was non-random, with the majority occurring in the dry season of November-April. Of all cases of bacterial meningitis, 175/221 Table 1. Summary of the patients diagnosed as positive for bacterial meningitis from three hospitals in the northwest Central African Republic. The number of patients with reported sex and case fatality ratios may be lower than the total numbers due to missing data. IQR, interquartile range; CFR, case fatality ratios; CI, confidence interval; pneumococcal, pneumococcal meningitis; meningococcal, meningococcal meningitis.   (79.2%) were in the dry season months in 2016, and 128/161 (79.5%) of pneumococcal meningitis cases (Figure 2 and Figure 3). A chi-squared (χ 2 ) test for count data showed the difference between the number of cases in dry and wet season months to differ significantly for all cases of bacterial meningitis by any causative agent (χ 2 = 75, p = 2.2 ×10 −16 , degrees of freedom [df] = 1), and for cases of pneumococcal meningitis (χ 2 = 56, p = 7.0 × 10 −14 , df = 1). Species confirmation by PCR was consistent with the latex agglutination test for the subset of confirmed cases. Six isolates of S. pneumoniae from Paoua Hospital were confirmed by PCR as serotype 1, however the majority of isolates were not serotyped.

Patient outcomes
Medical staff in MSF-supported facilities reported that patients typically presented with advanced symptoms (Glasgow Coma Scale 1-4) and that they had often sought the advice of traditional healers beforehand, though data on clinical symptoms and previous treatment at admission were not systematically recorded. Patient outcomes were recorded in Bossangoa and Batangafo ( A higher CFR was also found in adults with pneumococcal meningitis in Burkina Faso from 2002-05 11 , though age did not predict mortality in Northern Nigeria from 1971-76 12 . This may reflect statistical control for clinical severity in 8, and the variation in mortality we observe may be confounded by differences in healthcare-seeking behaviour between children and adults, whereby adults present later to healthcare facilities when their symptoms are more advanced 13 . The effect of sex on the odds of mortality was not significant (OR for males = 1.39, p = 0.39).

Incidence rates
The World Health Organization lacks a formal definition for outbreaks of pneumococcal meningitis 14 ; however, the 'alert threshold' is defined by the CAR MoH as 3 cases per 100,000 per week and 'epidemic threshold' as 10 cases per 100,000 per week. Bossangoa reported 11 cases of pneumococcal meningitis in the 9th epidemiological week of 2017; given an estimated target population of 350,000 this meets the alert threshold (3.1 cases per 100,000 per week; population estimate based on 2003 census).

Meningococcal meningitis
Amongst the cases of meningococcal meningitis (67, 19.5% of total confirmed cases) the CFR estimates were significantly lower in both Bossangoa, 6.67% (95% CI 0.3, 40.0%) and Batangafo, 10.8% (95% CI 3.5, 40.0%) than those reported for pneumococcal meningitis (see Table 1). Reliable estimates of the underlying population in the CAR are lacking which makes it challenging to calculate per-capita incidence rates. It has been estimated that one-fifth of the population of the CAR has been displaced internally or in neighbouring countries following civil conflict in 2014 23 . Therefore our estimates of the per-capita incidence should be treated cautiously as they likely underestimate the true rate. Improved reporting of suspected cases from Batangafo and Bossangoa would enhance epidemiological surveillance, as a threshold of 10 suspected cases per 100,000 per week has been suggested as an appropriate "epidemic threshold" and outbreak definitions using suspected cases is also employed for meningococcal meningitis 5,22 .
Further uncertainty surrounds the proportion of true cases that report to our medical facilities given the ongoing instability and the difficulties faced by patients from remote rural areas in accessing care. In February 2016 MSF medical staff made an exploratory visit to the community of Kouki in Ouham Prefecture (population 600) as a response to a rise in pneumococcal cases in Bossangoa Hospital. Community health workers reported 17 deaths with symptoms suggestive of bacterial meningitis over the previous 2 months. None of these fatal cases had reported to a healthcare facility, raising the possibility that the cases we observed represent only the 'tip of the iceberg' during the seasonal peak. Given the relatively low attack rate of S. pneumoniae, whereby bacteria in carriage become invasive and cause disease, it is likely that there is high underlying prevalence of S. pneumoniae carriage in the community 24 .
In conclusion it appears that the northern region of the Central African Republic experienced an outbreak of pneumococcal meningitis, most likely serotype 1, over the observed period, with a similar seasonal pattern to meningococcal meningitis. A comprehensive follow-up of cases in the community was not possible due to security constraints. Despite a MenAfriVac campaign conducted in 2017 meningococcal meningitis is still present although with predominantly non-A groups circulating, namely W. Our analysis is limited by incomplete PCR confirmation and serotyping for S. pneumoniae, however our findings suggest that increasing PCV13 coverage in routine vaccination programmes would be beneficial in preventing future seasonal outbreaks of pneumococcal meningitis.

Grant information
This research was conducted as part of routine MSF surveillance, therefore data collection and the provision of medical care was funded by MSF. T.C. performed the analysis while an employee of the Mahidol Oxford Tropical Medicine Research Unit, funded by the Wellcome Trust (106698).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

2.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

5.
Methods/Patients: How many patients where excluded? Why were suspected cases excluded? These cases were excluded because they had a negative result or because cerebrospinal fluid analysis could not be performed? Details about this, should be provided at the beginning of the Result section.
Statistical analysis: Authors stated that "Patient data were recorded using Microsoft Excel". They should provide these data in their Excel format. Visibly, this is not the case at this link: . https://dx.doi.org/10.6084/m9.figshare.7210367 Results/Discussion: It seems to me more interesting to separate these two parts.
Conclusion: The limitations of this study ("A comprehensive follow-up of cases in the community….. Our analysis is limited by incomplete PCR confirmation and a lack of serotyping……"could be moved at the end of the discussion.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly This study represents an important contribution to understand the current epidemiology situation of bacterial meningitis in the African Belt in the context of One serogroup targeted vaccine. Pneumococcal meningitis continues to kill people and this contribution could help in designing potential new vaccination strategies for meningitis control.
The limit of this study consists in the lack of information about the serotypes of Streptococcus pneumonia. Indeed, authors conclude from their results that PCV vaccines can help to control pneumo meningitis control but they did not provide serotypes information for cases in comparison to serotypes included in PCV vaccine. The study deserves to be published without this information about serotypes but authors should discuss this lack of serotype information as a limitation and explain that it should be a next step of the study to be able to inform the potential role of PCV vaccination in the meningitis control.
The paper is simply and clearly written. Introduction could be more developed to provide readers with a brief overview of what is known about pneumo meningitis in the African belt. The point is that this information is lacking so far and that makes this study highly relevant and informative.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Ecology of infectious diseases, Meningitis ecology and dynamics, Epidemiology, I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 20 Mar 2019 , Médecins Sans Frontières Holland, Bangui, Central African Republic Tom C We thank Dr Broutin for her helpful remarks.
We thank Dr Broutin for her helpful remarks.
We have added details on serotyping. Six pneumococcal CSF isolates have been typed as serotype 1 by PCR. While this represents a small proportion of the total isolates, it confirms previous research in the Central African Republic that found serotype 1 to be the dominant strain in meningitis cases in Bangui. An outbreak that occurred contemporaneously in S. pneumoniae Ghana in 2015/6 was also predominantly pneumococcus serotype 1. This suggests that PCV10/13 would be effective against the outbreak strains of in CAR.

S. pneumoniae
Reviewer Expertise: Epidemiology I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 20 Mar 2019 , Médecins Sans Frontières Holland, Bangui, Central African Republic Tom C In response to the helpful comments by Prof. Trotter we have included information on suspected and negative cases (this was possible in only one of the reporting hospitals), and we have expanded the introduction and discussion section to place the outbreak in CAR into a broader context. We include all of the suggested references. We discuss the contemporaneous outbreak of pneumococcal meningitis in Ghana in the discussion. We also include brief information on serotype (six pneumococcus isolates were typed by PCR as serotype 1). Overall we consider the quality and usefulness of the article has improved substantially as a result of the comments by Prof Trotter and the other reviewers.

None
Competing Interests: