Meningitis is one of the most severe infections in infants and children with an incidence of 25 per 100,000 children in Africa [1] . This infection is associated with high rates of acute complications, death and the risk of long-term morbidity. Its outcome was generally fatal before the advent of antibiotics whose discovery, along with those of vaccines, improved on the incidence and prognosis of the disease [2] [3] .

Although morbidity, mortality and occurrence of sequelae have diminished, these continue to be important in the pediatric population of developing countries, and their severity is highest in infants and children [3] . In 2012, in the course of the surveillance of meningitis in Africa the World Health Organization (WHO) [4] identified 22,000 meningitis cases in 14 countries in the meningitis belt. In Cameroon, several studies have been done on this subject [5] -[8] . Case fatality rates ranging from 8% to 27.44% and rates of appearance of sequelae ranging from 4% - 20% have been reported in studies conducted in different parts of the country [5] - [8] . In order to improve the prognosis of bacterial meningitis in infants and children in our environment, we decided to conduct this study whose aim was to determine the etiology and describe the prognosis in infants and children aged 2 months to 15 years who had been admitted for purulent meningitis at the Yaounde Gyneco-obstetric and Pediatric Hospital.

This was a descriptive cross-sectional study, with retrospective data collection for the period from 1 January 2009 to 31 December 2013. Sampling was consecutive. All patients aged 2 months to 15 years admitted in the pediatric unit with the diagnosis of bacterial meningitis based on the clinical and bacteriological criteria, through identification of the pathogen in CSF culture or soluble antigen in the CSF. We excluded patients in whom the diagnosis of meningitis was established but the pathogen was not identified. We selected all patients with meningitis who fulfilled our inclusion criteria through the registers of the pediatric unit. Data collected included: gender, age, weight of the child, the clinical signs of the child at the time of admission, the findings of the physical examination, the results of the cytology and bacteriology CSF analysis, CSF soluble antigen test, acute complications (death, altered state of consciousness, status epilepticus), sequelae at the time of discharge (psychomotor regression, motor deficit, hydrocephalus). The results were analyzed using SPSS 18.0 and Microsoft Office Excel 2007. The statistical test Chi-square was used to determine associations between variables. The significance threshold was set as P < 0.05.

The incidence of bacterial meningitis in our study remained high with 1.54% of admissions. This figure could be underestimated because we excluded patients in whom the pathogen was not identified. The sex ratio of 1.2 was close to Faye [9] in Ivory Coast who found a sex ratio of 1.3 in 2003. Meanwhile Sile [6] in northern Cameroon found a sex ratio of 1.6 in 1999. We did not find an explanation for this male predominance. The average age of our patients was 32.1 months with a standard deviation of 39.4, close to the average age of most authors [5] [10] [11] . Clinically, the study noted that the major symptoms to suggest meningitis in infants and children were fever, seizures and vomiting in respectively 98.8%, 44.4% and 28.7% of patients (Table 1).

In our study, Haemophilus influenzae was the most common pathogen, found in 39.2% of the patients, followed by Streptococcus pneumoniae and Neisseria meningitidis with respectively 31.6 and 10.5% of the patients (Table 2). In a study done in the same unit between 2004 and 2009 Haemophilus influenzae was the most commonly identified pathogen followed by Streptococcus pneumoniae and Neisseria meningitidis, respectively, in 40.3%, 34.2% and 5.4% [5] . Our results are also similar to those reported in Senegal by Cisse [12] who found Haemophilus influenzae (42.3%) followed by Streptococcus pneumoniae (31.9%) and Neisseria meningitidis (11.2%). Almuneef [13] in Saudi Arabia also identified Haemophilus influenzae as the first pathogen in bacterial meningitis in persons younger than 5 years followed by pneumococcus and meningococcus. Other authors [14] [15] in Mozambique and Angola also had, with some variation at the relative frequencies, globally the same results as our study with Haemophilus influenzae as the first cause of meningitis before the age of 15 years. How-

^*Intracranial hypertension.

ever other studies [7] [8] [16] done locally in Cameroon and other African studies [1] [17] [18] reported, with some variation in the relative frequencies, a predominance of Streptococcus pneumoniae, followed by Haemophilus influenzae and Neisseria meningitis. Massenet [19] in a study in the northern part of Cameroon found a high prevalence of Neisseria meningitidis with 70.2%, followed by 19.5% with Streptococcus pneumoniae and Haemophilus influenzae with 10.3%. This predominance of Neisseria meningitidis in the study of Massenet is explained by the fact that the study was conducted in the northern part of the country which lies in the African meningitis belt. The predominance of Haemophilus in our study may be related to selection bias because we have included only the patients who had a positive culture or positive soluble antigen. However the high incidence of meningitis caused by Haemophilus influenzae may also be related to the lack of vaccination in most of these patients, because since the vaccine was introduced in 2008, a decrease in the incidence of Haemophilus influenzae meningitis was expected. This is unlike the study of Wall [17] in Malawi that showed a significant decline in Haemophilus influenzae meningitis after the introduction of the vaccine.

The mortality rate observed in our study was 18.5%. This rate is close to the mortality rate of 21.8% presented by Gervaix [8] in Cameroon in 2012 and the 20.3% of deaths found by Gomes [20] . It is less than the mortality rate reported in other studies [7] [10] in Cameroon which were respectively 27.4% and 29.8%. This mortality rate is also lower than the 24% found by Roca [14] in Mozambique. The most vulnerable age group is that of two months to one year, which represented 43.7% of deaths and 38.8% of sequelae. The vulnerability of children under one year was also found by other authors [7] [8] .

Streptococcus pneumoniae was the pathogen responsible for most of the deaths and sequelae in our study as confirmed by several other studies [7] [8] [21] - [23] . Seventeen patients had at least one sequela at the time of discharge representing of 9.9%. This rate is similar to that found by Moufdi [24] who reported 10.2% of sequelae at time of discharge. This frequency of sequelae is probably underestimated because of the absence of an evaluation long after the acute episode. In evaluating these patients later, these sequelae would certainly be more severe.

The poor prognostic factors were age < 12 months, pneumococcal infection, delayed care > 3 days, the pre- sence of a coma, and signs of intracranial hypertension. Pelkonen [15] in Angola found as poor prognostic fac- tors coma and delayed care, while Gervaix [8] found as poor prognostic factors: age < 2 years and pneumo- coccal infection. Farag [25] found age < 12 months, and Kirimi [26] in Turkey had coma as a factor of poor prognosis.

The incidence of acute bacterial meningitis remains high in our context. The main causative pathogens are Haemophilus influenzae, Streptococcus pneumoniae and Neisseria meningitidis. The mortality and acute sequelae remain high with poor prognostic factors such as age less than 12 months, delays in care, pneumococcal meningitis, coma, brain abscess, and intracranial hypertension. A focus should be placed on strengthening the routine immunization on vaccine-preventable diseases of infants and children against Haemophilus influenzae, pneumococcus and meningococcus.