Cerebrospinal meningitis is a major cause of mortality and morbidity in both children and adults. In Ghana, cases of meningitis outbreaks are mostly reported in the northern part of the country and has been attributed to the low humidity in that area. In the years 1996-1997, the three northern regions of Ghana for instance recorded 18,703 cases of meningitis out of which 1,356 lost their lives during an outbreak. [30, 31]. Similar cases of meningitis have been reported in other parts of Ghana [7, 24, 32] however hospital based information on the different aetiological agents, antimicrobial susceptibility patterns and the seasonality is limited.
In our study the prevalence of confirmed and probable meningitis were 3.3% and 2.1% respectively. Our result is similar to other surveillance (based on cultures) of meningitis in other parts of Africa [33, 34]. The prevalence however appears low compared to some reports from the northern part of Ghana. This is because our reports were from all patients suspected of meningitis while reports from northern Ghana were mostly obtained during outbreaks. We however believe the prevalence could have been higher if techniques using polymerase chain reactions were applied in testing the samples [35, 36]. Another reason that could contribute to the low meningitis prevalence is the possible use of antibiotics before hospital admission, a common practice in many developing countries [35, 37]. Even though we have no records on the number of patients on antibiotics before admission, our observation show that most of the admitted cases referred from other primary healthcare facilities had been managed with antibiotics before referral. Future studies are however needed to confirm this.
The most prevalent bacteria was S. pneumoniae occurring mostly among patients less than 18 years of age. Neonates accounted for 8.7% and children from one month to less than 5 years accounted for 23%. Our result is similar to previous studies done by Holliman et al.,  where they found majority of S. pneumoniae infections occurred in children and young patients. Adult pneumococcal meningitis which accounted for 41.8% of infections was equally significantly identified as reported in other studies . A possible reason for the predominant prevalence of pneumococcal meningitis could be as a result of auto-infection from colonizing bacteria in the nasopharynx. Previous studies have shown the prevalence of nasopharyngeal colonization with streptococcal pneumoniae to be 51.4%  in Ghana. The auto-transmission from the nasopharynx into the meninges is therefore highly possible especially during the dry season when cracks and injuries tend to occur in the nasopharynx. Countries that introduced pneumococcal conjugate vaccines have however reported a reduction in the cases of pneumococcal meningitis .
We found one case of H. influenzae over the three year period. This low prevalence is quite remarkable and could be explained by the introduction of Hib vaccines in Ghana in 2002. The impact of Hib vaccine on the reduction of meningitis has been reported by Renner et al . Other developing countries such as Turkey have also reported low Hib meningitis prevalence . It is however possible that the infected child may not have been vaccinated against H. influenzae. Though the Hib immunization status of the child with the Hib meningitis could not be determined it is possible he might have been born in a rural area where access to healthcare and vaccination may be difficult. It is also possible that the under reporting of meningitis cases might have contributed to the low numbers. This is because some developed countries like the USA which introduced vaccines against Hib meningitis decades ago still reports prevalence of 6.7% .
The contribution of other bacteria pathogens to meningitis apart from N. meningitidis, S. pneumoniae and Haemophilus influenzae has not been widely reported in many developing countries. The present study identified E.coli, Salmonella species, Pseudomonas species, Klebsiella species, Staphylococcus aureus and Enterobacter species as contributing to 12% of all bacterial isolates among children and infants. Some studies in developing countries have similarly reported these observations [43–45]. Contrary to these findings however, children and neonates in developed countries have been reported to have predominance of Group B streptococcus and Lysteria monocytogenes. The reasons for the non-identification of Lysteria monocytogenes in our laboratory could be due to less attention given to the laboratory diagnosis of this pathogen. This is because Lysteria monocytogenes resembles diphtheroids and could easily be regarded as contaminants. Furthermore reagents such as esculin and hippurate hydrolysis or CAMP test (Christie, Atkins and Munch-Peterson) used for confirming Lysteria monocytogenes are not readily available in our laboratory.
Of interest in our study is the identification of Cryptococcus neoformans, occurring mostly in patients between the ages of 18 and 50 years. Previous studies in Ghana (Kumasi) did not identify Cryptococcus neoformans in meningitis patients perhaps due to the small number of study patients . The identification of Cryptococcus neoformans is however not surprising since the number of people living with HIV/AIDS continue to increase even though the national HIV prevalence had reduced from 1.8% in 2008 to 1.5% in 2010. The contribution of Cryptococcus neoformans to meningitis has similarly been reported to mostly occur in HIV endemic African countries [47–49].
All isolates tested with ceftriaxone were 100% susceptible with the exception of Staphylococcus aureus. Eighty three percent (83%) of S. pneumoniae isolates were susceptible to chloramphenicol whereas 98.9% were susceptible to penicillin. Our study found no evidence of high level resistance against chloramphenicol and penicillin. Previous studies in Ghana have however documented pneumococcal penicillin resistance rates varying from 8%-31% [7, 8, 50] and chloramphenicol resistance rates of 5-20.6% . The difference in the resistance patterns could be due to the methodology used. The sensitivity patterns of our isolates were done using only the Kirby-Bauer method which could underestimate the resistance levels. On the other hand, the susceptibility of S. pneumoniae, N. meningitidis and H. influenzae to ceftriaxone has been mostly reported to be 100% [24, 39, 50].
Appropriate treatment for meningitis depends on the local antimicrobial susceptibility patterns. In Ghana, the hospital antibiotic policy recommends penicillin and chloramphenicol as the first choice of meningitis treatment and ceftriaxone is considered as alternative . This study therefore emphasizes the need for clinicians to continuously rely on ceftriaxone as the best choice of drug for the treatment of meningitis.
Seasonal distribution of confirmed meningitis cases were noted to increase from December through January and peaked in February of 2008 (figure 1). The next highest peak of confirmed meningitis also occurred in February of 2010. This trend is not surprising as the dry season reaches its peak during these months and thus provides the conditions for destruction of the mucosal defences thus making individuals more susceptible to meningitis [22, 52]. These conditions may however not entirely promote meningitis since the trend in 2009 was quite different. More studies on seasonal variation is therefore needed to understand the exact seasons that significantly promote meningitis.