Background
Bacterial meningitis (BM) can be a life-threatening emergency if not properly diagnosed and managed [
1]. Over 1.2 million cases of BM are estimated to occur worldwide annually [
2]. Incidence and case-fatality rates vary by region, country, pathogen and age group [
3]. Case-fatality rate can be as high as 70% in untreated patients and 1 in 5 survivors may be left with permanent sequelae [
3]. Therefore, BM is considered as one of the most feared childhood diseases. Consequently, the WHO developed recommendations for detecting BM epidemics in highly endemic countries in Africa (
Alert and
Epidemic definitions) [
4,
5].
The epidemiology of BM has changed dramatically over the last 20 years, primarily as a result of the introduction of conjugate vaccines [
1] against the commonest meningeal pathogens;
Neisseria meningitidis,
Streptococcus pneumoniae and
Haemophilus influenzae [
2]. Worldwide, the incidence of meningitis due to
N. meningitidis is highest in the meningitis belt; a region of sub-Saharan Africa described as hyper-endemic and epidemics occurring during the dry season (December to June) [
5,
6]. Statistically, the incidence rate is 10 to 100 cases per 10
5 populations punctuated by explosive epidemics in 8 to 12 year cycles with incidence rates that can be greater than 10
3 cases per 10
5 populations [
5,
6]. Across the meningitis belt, at least 350 million people are at risk for meningitis during these annual epidemics [
6]. The climato-geographic location renders Sudan at permanent risk. A total of 15,595 cases including 1670 deaths due to
N. meningitidis were reported from 4 countries in the meningitis belt in 2007 [
7]. One of these countries was Sudan where 6946 cases with 430 deaths were reported from 9 out of 10 Southern states in the former Republic of Sudan [
7]. About 1.1 million meningococcal vaccine doses were released in response to the outbreak in South Sudan targeting people in the affected areas in mass vaccination campaigns [
7]. Mass vaccinations can lead to herd immunity resulting in dramatic reduction in infection rates among populations at risk [
8]. The massive return of displaced Southerners to their homelands after the separation; where most areas fall in the meningitis belt, should have definitely affected the intensity and distribution of meningitis in the North; where minimum areas are considered to be within the doomed region.
Meningitis due to
S. pneumoniae occurs most commonly in the very young and the very old, with an estimated incidence rate of 17 cases per 10
5 in children less than 5 years, and case fatality rates that exceed 73% in some areas [
9].
H. influenzae type b (Hib) is a major cause of infant and childhood meningitis [
10]. Rates are highest in children less than 5 years reaching 31 cases per 10
5 populations [
10]. In young children, the case-fatality rate of Hib meningitis is generally higher than that for meningococcal meningitis [
10]. Because of vaccination [
11,
12] the burden of Hib meningitis is dramatically decreased in most industrialized countries and has been virtually eliminated as a public health problem [
13]. We anticipate a comparable situation since Sudan has introduced Hib vaccines earlier than most developing countries and has maintained immunization coverage of 93% [
13].
Distinguishing BM is often difficult [
14] therefore several highly accurate decision support tools have been developed and validated to guide decision making and limit unnecessary hospital admissions and prolonged antibiotic use [
14‐
16]. In Sudan, as in many other countries around the world, children who are suspected and provisionally diagnosed with meningitis are routinely admitted to hospitals and administered broad-spectrum antibiotics irrespective of culture and laboratory confirmation. Proper diagnosis of infectious CNS syndromes and the subsequent ability to distinguish BM is vital. This study aims at critically evaluating the hospital diagnosis of BM using internationally validated clinical prediction rules. We also aimed at studying the clinical parameters and identifying the type and frequency of bacterial aetiologies.
Methods
Study design and ethical considerations
This cross-sectional hospital-based study was conducted at a large central pediatric reference hospital in Khartoum, Sudan, during 10 months period (October to July, 2010). All febrile (>37 °C) attendees age 1 day to 15 years who were suspected of meningitis during the study period were included. Independent and dependent variables (demographic, clinical and conventional laboratory data and final outcome) were collected simultaneously in a pre-designed structured data sheet and kept anonymous. Ethical clearance was obtained from the Ethical Committee Board of Al-Neelain University. Permission to collect data was granted from hospital authorities. Patients were not contacted directly; data were obtained from hospital files.
Study population and specimens
Hospital case definition for suspected meningitis is: “sudden onset of fever, headache, stiff neck, episodes of seizure before or during admission and/or other symptoms as; nausea, vomiting, photophobia, altered mental status and coma”. In newborns and young children: “General signs of being unwell as irritability, vomiting, poor feeding and/or bulging fontanelle”. A confirmed case is: “a clinically compatible case that is laboratory confirmed by Gram stain and culture”.
History of the current illness was evaluated upon admission by pediatricians and/or house-officers. Patients were inquired about self-medication with antibiotics 1 week prior to lumbar punctures (LP) and contact with individuals with similar illness. Patients were examined clinically to evaluate onset of the classic symptoms: fever, headache, neck-stiffness, vomiting, bulging fontanelle, chills, seizures, altered mental status, skin rash, petechiae and coma. Demographic data included age and sex. Information recorded during hospital stay were: types of antibiotics administered, duration of hospital stay, final diagnosis and outcome. Total volumes of 3 ml CSF were obtained from each patient via lumbar or ventricular puncture by hospital medical staff.
Laboratory investigations
Color and turbidity of CSF specimens were macroscopically inspected immediately upon withdrawal. Microscopical examination was performed on wet preparations and on Gram stained smears from the sediment. White cell count was performed in the non-centrifuged portions of specimens (diluted with an isotonic 0.1% toluidine blue, 1 in 2) using modified Fuchs-Rosenthal ruled counting chamber. CSF cell count was reported as cells/mm3. When no white cells were seen, the count was reported as <5 cells/mm3. Percentages of polymorphs and lymphocytes were estimated.
GlucoseOxidase/Peroxidase® and Biuret® methods were used to measure CSF glucose level and CSF total proteins immediately upon collection using manual colorimetric methods and reagents (Biosystems®) following the supplier’s protocols.
Irrespective of Gram stain result, all CSF specimens were cultured on Blood, Chocolate and MacConkey agar media. Inoculated cultures were incubated aerobically at 35 °C overnight. Blood and Chocolate agar cultures were incubated in CO
2 enriched atmosphere and kept for 72 h. Bacterial growth was examined daily and suitable biological and serological tests were applied to identify the aetiologic agent following the recommendations of Gray & Fedorko [
17] and Vandepitte et al [
18].
Secondary analysis using published clinical prediction rules
Data collected throughout this study were used to distinguish BM according to the criteria listed in the
Bacterial Meningitis Score (BMS) [
14] which was later validated in the era of widespread pneumococcal vaccination [
16] and the
International Classification of Diseases–Clinical Modification Code 320.9 (ICD-CMC320.9) for BM [
15]. These criteria aim at identifying infectious meningitis and distinguishing the bacterial condition in children with CSF pleocytosis. Following the criteria, meningitis suspected cases were re-classified as BM based primarily on CSF pleocytosis and the presence of at least one of the following indicators; seizures on or before presentation, other clinical signs of meningeal irritation, CSF absolute neutrophil count, CSF Gram stain and CSF protein concentration. Cases showing CSF pleocytosis were considered as
Infectious Meningitis, those with one or more criteria of the BMS were considered as
High Risk for BM, cases with in vitro culture confirmed bacterial aetiology were considered as
Proven Bacterial Meningitis which was further specified as
Meningococcal Meningitis when the isolated bacteria was
Neisseria meningitidis or as
Streptococcal Meningitis when it was
Streptococcus pneumoniae.
Statistical analysis
The statistical package program SPSS version 21 was used. All numerical variables were organized into categories for better interpretation. Age obtained in days was converted into groups of neonates (1 to 29 days), infants (1 to 11 months), toddlers (1 to 5 years), children (6 to 10 years) and teenagers (11 to 15 years). All categorical variables were expressed in frequencies and percentages; high statistical significance (p < 0.001) for frequencies distribution was detected for each variable using the General Trend Analysis. Numerical variables were described using measures of central tendency and of dispersion. When Shapiro-Wilk Test of Normality was conducted, all data were found to be non-normally distributed; accordingly, nonparametric statistical tests were performed. Comparisons between categorical variables were done using cross-tabulations. Inferential statistics for statistically significant differences under the 0.05 level was applied using either Pearson Chi-square test; when 0 cells in the contingency table have expected count < 5, or Fisher’s Exact test; when ≥1 cells in the table have expected count < 5. Nominal by Nominal Phi and Cramer’s V Correlation (Corre) along with its 95% confidence intervals (CI) were obtained to assess possible relationships between categorical variables.
Discussion
In reference to the data from hospital records, BM was assigned for 56% of the study population. This number seems to be exaggerated since neither an
Alert nor an
Epidemic [
5] of BM was announced by National and/or International authorities in Sudan at the time of the study. In fact, a much smaller figure of 1.5% was later identified as
Proven Bacterial Meningitis when codes and criteria were followed (Table
5). A total of 40 cases out of the study population were found to either have
Proven Infectious Meningitis or are
Highly Suspected to have the disease. Accordingly the proportion of those with proven infectious meningitis among our study population would be 10% (Table
5). We found this figure to be rather diluted because of the wide denominator i.e. provisional diagnosis of infectious CNS disease was too inclusive. This shortcoming is not only limited to hospitals in Khartoum since several articles from around the world reported almost similar findings. The proportion of children who were proved to have meningitis by laboratory confirmation was always around 50% or less among those who were highly suspected and subjected to LP. Studies from the developing world reported much less estimates that are closer to ours. A study conducted in pediatric hospitals in Iran has identified meningitis in 16.8% (65/387) [
19]. Countries falling in the meningitis belt reported small estimates as well. Laboratory confirmed cases among suspects during 5 years in Burkina Faso were 22% (4503/20,389) [
20], 24% (279/1176) during 6 years in Ghana [
21] and 26% (871/3306) in 20 years surveillance in Mali [
22]. Studies in developed countries reported different results. Amarilyo et al [
23] reported 54% (58/108) among meningitis suspected children. Other studies conducted by Dubos et al [
24], Oostenbrink et al [
25], Bonsu et al [
26] and Oostenbrink et al [
27] in children who underwent LP have reported comparable findings (≈ 40% prevalence of meningitis). The corresponding low value we obtained from hospital records could be attributed to lack of strict adherence to meningitis criteria, a situation that appears to be common in hospitals in developing countries. We identified only 1.5% as proven BM which was closely in agreement with published reports where only 3.7% (
n = 122) were identified in the study of Nigrovic et al [
16] among 3295 children with CSF pleocytosis. Among highly suspected cases in our study, however, a higher figure of 15% was identified as BM. This was in good agreement with previous published reports. Dubos et al [
28] and Amarilyo et al [
23] identified 12% (20/166) and 10.3% (6/58), respectively, as BM among suspected patients in pediatric emergency units. Similarly, Bonsu et al [
26] and Oostenbrink et al [
27] identified a maximum of 20%. In contrast, a national Polish survey identified bacterial aetiology in 40% (980/2475) of cases with neuroinfections in 1 year [
29]; most probably a result of an outbreak in Poland.
When it comes to diagnosing a debilitating and potentially fatal illness in a pediatric population of a country located in the meningitis belt, health care officers in Sudan often tend not to dismiss a poorly suspected case. Fear of misdiagnosis or late treatment result in over diagnosis that could lead to an increased hospitalization costs as evident by the extended periods of hospitalizations (Table
3). Over diagnosis can also lead to the emergence of drug-resistant strains due to antibiotics overuse, and hence the accuracy of the national health registry becomes questionable. Therefore, the ability to monitor simple seasonal rise in disease incidence, identifying an actual epidemic and implementing the appropriate control measures will consequently be affected.
Pneumococcal and meningococcal conditions were equally identified in 7.5% among those with proven infectious meningitis in this study. A recent meta-analysis [
30] covering 56 studies identified
S. pneumoniae and
N. meningitidis as the predominant pathogens of BM in children of all ages in all regions. In previous studies, pneumococcal disease was usually identified in relatively small number of cases. Dubos [
28] reported 5% (9/166)
S. pneumoniae meningitis among patients with CSF pleocytosis. Even though in a meningitis outbreak situation, only 6% (
n = 149) was reported as
S. pneumoniae among 40% with bacterial aetiology [
29]. Never the less, the aforementioned meta-analysis reported
S. pneumoniae as the most common cause of BM in children ranging from 22.5% in Europe and 41.1% in Africa [
30]. A reduction in meningococcal meningitis occurrence was reported in Burkina Faso [
20] and Ghana [
21], where Dubos [
28] and Turczyńska [
29] identified
N. meningitidis in 5% (9/166) and 9% (220/2475) respectively, closely comparable to our findings of 7.5%. On the other hand, studies in Mali reported 44% meningococcal meningitis and 31% pneumococcal meningitis [
22].
All our patients were admitted in the high risky season that is described climatologically as dry compared to about 65% of all cases that were recorded in the Ghanaian study in this season [
21]. This period constitutes the peak of meningococcal disease occurrence, unlike the pneumococcal occurrence that varies seasonally [
31,
32]. In this study, all meningococcal patients were less than 5 years, an age group that has the highest reported incidence rate [
30,
33]. Pneumococcal meningitis occurs most commonly in the very young and the very old [
9], where all cases we have studied were infants aged 1 to 11 months. Fortunately, all of them have recovered and discharged despite reports on high case-fatality rates [
9].
Diagnostic signs for BM in young children are unclear; they do not often exhibit the general symptoms and may only be irritable and look unwell [
34]. All cases with proven BM in this study presented with high fever and seizures. Other classic symptoms, namely, neck stiffness, chills and bulging fontanelle, were seen in one third of the children only. Bulging fontanelle and admission in summer were the only parameters that showed significant statistical difference and strong association. An inflammation in the brain or the meninges can cause a bulging fontanelle [
34]. Amongst the several aetiologies bacteria and viruses are the most common and these occur more frequently in summer [
34]. The fact that all 37 (20%) patients who suffered from a bulging fontanelle (Table
1) presented in summer strongly suggests a microbial aetiology. However, only 2 (33.3%) of our 6 BM confirmed cases presented the symptom. In the study of Amarilyo et al [
23], bulging fontanelle was present in 50% of patients with meningitis and had a positive predictive value of only 38%. Several studies have reported atypical clinical findings in young patients with BM of either meningococcal or pneumococcal aetiology [
23,
35]. In fact, a guideline describing these anomalies was established [
36]. Amarilyo [
23] recommended that these clinical indicators should not be the sole determinants for referral to further diagnostic testing and LP.
Typical findings on CSF analysis [
37,
38] were seen in all cases with
S. pneumoniae and in only one third with
N. meningitidis, while another third showed normal CSF picture. Normal CSF cellular count in those with positive CSF culture can be demonstrated, however, rarely [
39]. Garges et al [
40] concluded that BM in babies frequently occurs in the presence of normal CSF parameters, including WBC count [
40].
Hib was not identified in this study probably as the result of the introduction of Hib vaccine to the pediatric population in Sudan which was first started in 1976 and later on in 2001 [
13]. Following critical review of accessible publications, this finding is in conformity. The study from Burkina Faso reported
H. influenzae in a small proportion of 2% [
20] and Ghana did not identify the bacterium amongst other aetiologies of BM in children [
21]. Further, Strange [
41] affirmed that
H. influenzae meningitis in children has become so rare that a case could hardly be seen. The study of Dubos [
28] identified only one case with Hib meningitis (0.6%) among 166 BM suspected patients and the prospective French nationwide survey of Bingen et al [
42] detected
H. influenzae in only 2.5% (27/1084). Amongst published articles within our reach were those from Mali which stated that the country has suffered high morbidity and mortality of BM and high occurrence of
H. influenzae meningitis [
22,
43]. Hib conjugate vaccine was first introduced to children in Mali in 2005 to 2007 [
43]. In 20 years surveillance (1996–2016) Mali reported
H. influenzae in 23% [
22] and a post vaccination evaluation [
43] identified a decrease of 86% of Hib among high risk infants of 6 to 7 months old and an overall annual incidence drop of 74% by the second year. All the aforementioned studies, including those from Mali, and many others along with ours confirm the notion that Hib meningitis is about to be eliminated. It is worth mentioning that the Federal Ministry of Health in Sudan has been demonstrating strong commitment and good adherence to vaccination programmes provided by UNICEF and WHO [
13]. The observed absence of Hib infections and the great reduction in conditions of BM caused by all other aetiologies are closely comparable to that reported by Schuchat [
12] and fairly places Sudan in a better position regarding the control of such disease compared to other African countries.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.