In the African meningitis belt, the mortality rate is an important health indicator. During the major epidemics of meningococcal meningitis which breaks out with a periodicity of 5-12 years, thousands of cases occur every year [1, 2]. While most of the mortality rates are estimated from hospital survey data, cases of bacterial meningococcal meningitis are also reported and managed by small healthcare centers with lack of laboratories and adequate staff. Only a few healthcare centers were able to use Rapid Diagnostic Tests (RDT) since 2007. Niger is one of the countries reporting the highest number of cases of bacterial meningitis [1-3]. In 2005, reported mortality due to meningococcal meningitis ranged from 4% in Mali to 26% in Benin [3]. Of note, these numbers are calculated on the basis of notified presumptive cases and are not laboratory-confirmed cases. The case fatality ratio (CFR) varies according to the etiological bacteria among the three bacteria classically diagnosed, as shown by a study done in Niger between 2003 and 2006 [4]. These ratios reached 5.5% for the serogroup A of N. meningitidis, 12% for the serogroup X or the serogroup W135 and 49.8% for S. pneumoniea. Since 2002, the meningococcal survey has been reinforced, the number of health care centers increased, the management of bacterial meningitidis is well known and the distribution of RDT scaled up, so it would be worthwhile to re-investigate the CFR.

The protocol of the study followed recommendations of the Association des Epidémiologistes de langue Française (ADELF) [5]. It was approved by the national ethic committee of Niger on March 9th 2011, N° 003/2011/Comité Consultatif National d’Ethique (CCNE).

We conducted a retrospective and prospective study on suspected cases of bacterial meningitis in cerebrospinal fluids (CSF) which reached the national laboratory of the Centre de Recherche Médicale et Sanitaire (CERMES), the reference center for meningitis in Niger, through health care centers that reported and managed these cases. The study extended from November 1st, 2006 to June 30th, 2011. The retrospective study concerned the period extending from November 1st, 2006 to March 9th, 2011 and the prospective study extended from March 9th to June 30th, 2011.

In practice, questionnaires accompanying the CSF specimen received by CERMES allowed us to identify the type of health facilities that managed the case and to verify or complete the outcomes of the disease.

The study concerned patients whose CSF specimen arrived at CERMES within the framework of the meningitis surveillance performed in Niger. This population represented a part of the entire cases of suspected meningitis notified to Niger Ministry of Health and the World Health Organization (WHO). The sending of CSF to the reference laboratory was voluntary, thus it was neither exhaustive nor randomized.

Patients included in the study consulted one of the health centers in the regions of Dosso, Maradi and Niamey, from November 1st 2006 to June 30th 2011 for suspicion of meningitis according to the standard case definition of acute meningitis [6]. The aetiological diagnosis was performed by Polymerase Chain Reaction (PCR).

Identification of the patients included the meningitis database of CERMES that contains information on meningitis since November 2002. It was done in strict collaboration with the Direction de la Surveillance et de la Riposte aux Epidémies (DSRE) and the national office of WHO in Niger.

Besides the collection of data on the outcome of the disease, questionnaire data was completed by a study technician on the field. The registers of the health facilities were consulted on the spot by the technician. The health facility’s workers were informed by the regional directors of public health. The information contained in the database is reported on a questionnaire as “publipostage”. The technician collected the missing data on the questionnaire during his field missions. This information completed those of CERMES meningitis database.

We reviewed the available data collected by the questionnaire accompanying the CSF obtained in the meningitis database of CERMES. These data were:

• -sociodemographic: name, surname, age, sex, place of residence of the case
• -relative to the meningitis: date of the beginning of the disease, date of consultation, bacterial or meningococcal serogroup.

These data completed by a field study are:

• -the type of health care facilities that managed the case on the date that the CSF was taken.
• -the outcome of the disease (death or survival) after the biological confirmation of the case.

The CFRs were calculated for all suspected cases whose CSF reached CERMES, for each of the 3 bacteria investigated: N. meningitidis, S. pneumoniae, H. influenzae (serogroups A, X and W135), and for suspected cases without any bacterial etiology.

Data was obtained from the meningitis database of CERMES which had been secured during the microbiological surveillance, since 2002. Access to the database was strictly enforced, according to accplicable laws.

Qualitative and quantitative variables were described with usual statistical parameters. Pearson’s chi-square test or the Fisher’s exact test, Kruskal-Wallis and ANOVA were used whenever appropriate. The Odds ratios (OR) of death were estimated with a confidence interval at 95%. Variables available for evaluation as potential risk factors included sex, age group, aetiological agent, health facilities and the localities. Multivariable logistic regression analysis was fitted to the data, with all variables that were significant at p < 0.05 on univariate analysis. Independence of data was assumed, because the vast majority of cases were considered to be unrelated. The criterion of significance was p < 0.05. Statistical analyses were performed using R2.15.0 software [7].

During the study period, we obtained information on the outcome of the disease for 2947 presumptive cases, whose CSF had been received at CERMES. Among the 2922 CSF which could be tested, etiology on one of the three bacteria routinely researched has been identified in 1063 cases (36.4 %). N. meningitidis, S. pneumoniae and H. influenzae were identified in 793 (26.9% of the total reception and 74.6% of the positive results), 195 (6.6% and 18.3%), and 75 (2.5% and 7.1%) CSF specimens, respectively. Other bacterial etiologies represented 5 cases (0.2%) and 1850 CSF specimens (62.8%) were tested negative. The conclusion was missing for 4 cases. The sex ratio was 1.4 men for 1 woman. The mean age was 8.8 years. The mean ages of H. influenzae, N. meningitidis and S. pneumoniae were 1.5, 10.3 and 12.6 years respectively (P<10^-5). Patients aged between 0 to 5 years were higher for H. influenzae 85.3%, followed by S. pneumoniae 35.9% and N. meningitidis 31.4%, p<0.001.

Among the 793 cases of meningococcal meningitis, 515 cases (64.9%) were due to serogroup A, 268 (33.8%) to serogroup W135, 6 cases (0.76%) to serogroup X and 4 cases due to indeterminate meningococcus. Serogroups A and W135 together represented 98.7% of meningococcal meningitis and were responsible for case fatality rates of 10.5% and 8.8% in 2010 and 2011.

Hospitals at district or regional level accounted for 1730 cases (58.7 %) and basic health facilities for 1113 cases (37.8 %). H. influenzae and S. pneumoniae cases were mostly managed by hospitals; 78.7% and 55.4% respectively, whereas 57.5% of N. meningitidis cases were managed by basic health care centers (Table 1).

Over the study period, the CFR reached 8% and 10% for serogroups A and W135, respectively. Among the 6 cases of N. meningitidis X, 3 died, giving a CFR of 50%. S. pneumoniae’s CFR was 31.8%. It was lower in basic health facilities than in hospital (23.17% versus 37.96%; p<0.001).

Figure 1. Age-specific CFR for the 1030 patients positive to S. pneumoniae, H. influenzae, N. meningitidis A and N. meningitidis W135, Niger, November 2006 to June 2011. The age denotes the time of diagnosis.

The CFR due to N. meningitidis decreased from 17.4% in 2007 to 6.6% in 2009. But in 2010, this rate peaked at 10.5% and decreased to 8% in 2011 and 2012. Pneumococcal rates varied between 23.5% in 2007 and 43.6% in 2008 and turned around 30% from 2009 to 2011. H. influenzae increased its CFR from 2007 to 2009 (30% to 66.7%), and decreased to 11.1% in 2010 (Table 1). The global CFR after meningococcal disease was 14.7% (156/1063) during the study period. It peaked at 18.1% in 2008.

The CFR due to H. influenzae was more important in the group age of 1-4 years. This agent constitutes the primary cause of death by meningitis in this group. S. pneumoniae leads to a high CFR at every age. The differences between groups were statistically significant from 22.2% in the 5 – 10 years group to 38.5% in the >=15 years group (p<0.001). N. meningitidis had the least important CFR in all age groups, in comparison with S. pneumoniae and H. influenzae. The serogroup W135 lethality exceeded the serogroup A one in patients aged 10 years or older (Fig.1).

After univariate analysis of various factors, we found that age, bacterial agent, type of health facilities and the region were associated with the risk of death. Patients having meningitis due to S. pneumoniae had a risk of death 5 times higher than those who had N. meningitidis as an etiological agent (Table 2). H. influenzae and S. pneumoniae together showed a high risk of death (OR=3, 95% CI, 0.3 to 6.7). The risk of death also increased with age. The fact of being taken care in a hospital showed a risk of death 10 times higher than in private hospital. There was no correlation between the age and the etiological agent. Multivariable analysis showed that age>=15 years, diagnosis of N. meningitidis A, N. meningitidis W135 and negative CSF were risk factors of death by meningitis.

The results of this study show that the meningitis CFR is still rising: 14.7% over the study period. It peaked at 18.1% in 2008. Meningitis due to S. pneumoniae and H. influenzae is the most lethal with a rate exceeding 30% during on the whole study period. The specific rates by etiological agents in this study are closer to those estimated by a previous study done in Niger between 2003 and 2006, for meningococcus, Haemophilus and negative CSF. The rates were 7.4% versus 8.9% in the present study for meningococcus, 30.5% versus 30.7% for the heamophilus, and 12.9% versus 13% for negatives. The rates estimated for the pneumococcus are different from those of this study, 49.8% against 31.8% in the present study [4]. This comparison should take into account the fact that the present study concerns only the regions of Niamey, Dosso and the city of Maradi which are strongly urbanized, while the study of 2003 – 2006 had been realized in the whole country, with an important rural component (70% of the population).

H. influenzae and S. pneumoniae cases were mostly managed by hospitals (78.7% and 55.4% respectively), whereas cases of N. meningitidis were managed by basic health care centers (57.5%). The health facilities taking care of meningitis are essentially the hospitals because the great majority of cases seen in the basic health facilities are referred to them, especially during the non-epidemic periods. During epidemic periods, the provision of oily chloramphenicol in the health centers allowed to take care of cases on the spot. Oily chloramphenicol is the most commonly used drug in basic health facilities as well as in hospitals [4]. So, a great part of cases due to S. pneumoniae or H. heamophilus were referred to the hospitals.

The present study finds a CFR of 8% for the N. meningitidis A against 5.5% for the period of 2003 – 2006. N. meningitidis W135 represented 4.7% of the N. meningitidis cases and the corresponding CFR reached 12.7% against 10% in the present study. During the period of this study, the distribution of the serogroups of meningococcus changed, the total number of cases with meningococcal meningitis fell and W135 became dominant from 2010 because of the anti-meningococcal A conjugate vaccine (MenAfriVac). Results from Burkina Faso suggest that the MenAfriVac has reduced serogroup A disease by nearly 100% [8]. This change in the serogroups distribution could explain the decrease of the N. meningitidis’CFR from 2010, 10.5% to 8.8% in 2011.

S. pneumoniae is consistently found with higher CFR in every year than that of other etiologies, but lower than in the precedent studies done in Niger: 49.8% [4], in Ghana: 44.4% [9] and according to a study which aimed to estimate the load of the meningitis disease in several African countries: an average of 45% [10]. The decrease of its CFR in comparison to that of the 2003 – 2006 period could be explained by a better care of the cases, due to the generalization of RDTs which allowed to eliminate the serogroups A and W135 of meningococcus. So, in spite of the lack of laboratory, the health workers can quickly specify the treatment. A favorable prognosis depends on initiating treatment early, and confirmatory test [11]. The CFR due to H. influenzae has not changed. For H. influenzae, N. meningitidis and other atiologies, the study of Peltola showed CFRs of 29.8 and 22% [10]. In an adult population of India, meningitis was associated with higher CFRs of 36.9%, when compared with pneumonia with a CFR of 20.8% [12].

This study found that age, the bacterial agent, the health facilities, and the region, as risk factors of death for meningitis, on univariate analysis. The risk increased with age. Xu and his coworkers also found that age group (15 – 24 years) and bacterial agent (Serogroup C) were associated with higher rates of death [13]. On multivariate analysis we found that age >=15 was a significant risk factor of death. The patients presenting pneumococcal meningitis had a risk of dying 5 times higher than those with N. meningitidis as an etiological agent. The reason could be due to the generalization of RDTs in the health facilities that could allow a better care of N. meningitidis cases. Indeed, RDTs are bedside diagnostic tools for detecting N. meningitidis polysaccharide antigens of serogroups A and W135, the two serogroups that are responsible for major epidemics in the African meningitis belt [14]. The development of RDTs for the detection of S. pneumoniae could bring hope for the decrease of its CFR before the introduction of the pneumococcal vaccine in the expanded programme on immunization in Niger. For this, a conjugate vaccine was suggested because of the high prevalence of the serotype 1 and the theoretical vaccine coverage across all age groups [15].

The risk was higher in hospitals because the most serious cases are referred there. The risk of death by pneumococcal and meningococcal meningitis increased with a constant rate from the age of 5 years according to an analysis made in four countries of the meningitis belt: Burkina Faso, Sénégal, Niger and Ghana [16].

We recognize that, due to lack of human and financial resources and other challenges, not all CSF samples reached the central level. Other reasons include geographical remoteness, which has an impact on the lumbar puncture procedure, the different concerns and unequal efforts put forth by the staff aimed at detecting suspected cases. Based on these reports, it is possible that a selection bias could have impacted our results, particularly in terms of CFR. Calculations based on confirmed cases probably underestimated the CFR because CSF samples were not available for all cases, and some of them (although fewer) were not possible to examine due to insufficient quantity or pronounced purulence of the fluid. Furthermore, some patients afflicted with full-blown bacterial meningitis could not attend healthcare centers in time, while other patients did not show up due to economic or cultural reasons (e.g use of traditional medicines), thus resulting in under-reporting. Another selection bias could result from patients who leave the health facilities centers early, against medical advice.

The hope lies in the fact that the CFR of S. pneumoniae, the most mortal bacteria, is in regression. This drop could become more marked with the introduction of the pneumococcal vaccine, as nonrandomized studies clearly suggest that vaccination in a susceptible population results in decrease in mortality and hospitalization due to pneumonia [17]. So, the vaccination against S. pneumoniae is necessary, especially for children, giving the CFR due to this pathogen. N. meningitidis A almost disappeared with the introduction of the MenAfriVac A conjugate vaccine. The serogroup W135 with a CFR higher than serogroup A can be prevented with the tetravalent vaccine. The re-emergence of the serogroup X is not excluded since there is no available vaccine.

Meningitis continues to be a problem in Niger and is associated with high case fatality, especially for S. pneumoniae and H. influenzae, in spite of treatment in the hospitals. The age >=15 years, the bacterial agent and the hospital as type of facilities having taken care of the case constituted the main risk factors of death by meningitis. Surveillance of meningitis disease remains very important to study the age distribution, the bacterial etiology and to understand the changing trends of patterns and serotype distribution. The information will allow improving the systems of prevention by guiding the choice of the vaccine and the target of the vaccination. RDTs should be developed for a rapid detection of S. pneumoniae too in order to perform the treatment.

The authors gratefully acknowledge Sanofi-Pasteur for the financial support, all the doctors and nurses who have sent CSF specimens and epidemiological forms to the CERMES and Dr. Shikha Tarang, Creighton University School of Dentistry, Dept. of Oral Biology, Omaha, NE 68178, USA, for English editing. We gratefully acknowledge CERMES units of biology, epidemiology and bacteriology.