Background. Accurate etiological diagnosis of meningitis in developing countries is needed, to improve clinical care and to optimize disease-prevention strategies. Cerebrospinal fluid (CSF) culture and latex agglutination testing are currently the standard diagnostic methods but lack sensitivity.
Methods. We prospectively assessed the utility of an immunochromatographic test (ICT) of pneumococcal antigen (NOW Streptococcus pneumoniae Antigen Test; Binax), compared with culture, in 5 countries that are conducting bacterial meningitis surveillance in Africa and Asia. Most CSF samples were collected from patients aged 1–59 months.
Results. A total of 1173 CSF samples from suspected meningitis cases were included. The ICT results were positive for 68 (99%) of the 69 culture-confirmed pneumococcal meningitis cases and negative for 124 (99%) of 125 culture-confirmed bacterial meningitis cases caused by other pathogens. By use of culture and latex agglutination testing alone, pneumococci were detected in samples from 7.4% of patients in Asia and 15.6% in Africa. The ICT increased pneumococcal detection, resulting in similar identification rates across sites, ranging from 16.2% in Nigeria to 20% in Bangladesh. ICT detection in specimens from culture-negative cases varied according to region (8.5% in Africa vs. 18.8% in Asia; P<.001), prior antibiotic use (24.2% with prior antibiotic use vs. 12.2% without; P<.001), and WBC count (9.0% for WBC count of 10–99 cells/mL, 22.1% for 100–999 cells/mL, and 25.4% for ⩾1000 cells/mL; P<.001 by test for trend).
Conclusions. The ICT provided substantial benefit over the latex agglutination test and culture at Asian sites but not at African sites. With the addition of the ICT, the proportion of meningitis cases attributable to pneumococci was determined to be similar in Asia and Africa. These results suggest that previous studies have underestimated the proportion of pediatric bacterial meningitis cases caused by pneumococci.
Streptococcus pneumoniae (or pneumococcus) is a major cause of bacterial meningitis in the developing world [1–4], including recently described epidemic disease in sub-Saharan Africa [5, 6]. Safe and effective vaccines to prevent pneumococcal meningitis exist [7–10], and many developing countries will introduce the vaccines in the near future. As part of vaccine introduction, countries will monitor the impact of the vaccine on various outcomes, including meningitis incidence. Additionally, in the African meningitis belt, patients with suspected acute bacterial meningitis are given treatment with a single dose of oily chloramphenicol (a long-acting preparation of chloramphenicol in an oil suspension, recommended by the World Health Organization as first-line treatment for presumptive bacterial meningitis in the African meningitis belt during the epidemic meningitis season), under the assumption that their illness is caused by meningococcus . This treatment is suboptimal for pneumococcal meningitis. Accurate etiological tests appropriate for situations in developing countries will assist in measuring the impact of vaccines and in improving clinical care.
Culture-based identification of the etiological agents causing acute bacterial meningitis in developing countries may have poor sensitivity, for a variety of reasons . These include prior treatment with antibiotics, specimen contamination, lack of transport media, delays in transportation from outlying areas, inconsistent availability of laboratory supplies, and challenges in the training and motivating of laboratory staff . Antigen detection in CSF specimens provides a useful adjunct to culture-based diagnosis [14, 15]. It allows swifter feedback than does culture and often permits accurate etiological diagnosis among patients with meningitis who have received prior treatment with antibiotics. Latex agglutination (LA) testing is widely used, but the test must be interpreted by an experienced reader and, therefore, is not usually considered to be a “bedside” technique. Moreover, LA kits have a relatively short shelf life, particularly in tropical climates.
The NOW S. pneumoniae Antigen Test (Binax) is an in vitro rapid immunochromatographic test (ICT) for detection of pneumococcal antigen in the urine of adult patients with pneumonia and in the CSF of patients of all ages with meningitis. The tool has demonstrated high sensitivity and specificity in testing CSF samples from patients with culture-confirmed meningitis [16, 17]. Moreover, it was as sensitive as PCR in a study of children with purulent CSF in Bangladesh . In this study, we aimed to further evaluate the sensitivity and specificity of the ICT among patients with meningitis at 5 field sites in South Asia and Africa, to test whether the ICT consistently increased the number of pneumococcal cases detected among patients with meningitis in these settings and to assess its utility among patients with and patients without prior antimicrobial treatment.
Study population. The study was conducted at 5 sites with existing surveillance of acute bacterial meningitis: Centre Hospitalier Universitaire Yalgado Ouédraogo, Ouagadougou, Burkina Faso; Centre Médical avec Antenne Chirurgicale, Kaya district, Burkina Faso; Kilifi District Hospital, Kilifi, Kenya; University College Hospital, Ibadan, Nigeria; Dhaka Shishu Hospital, Dhaka, Bangladesh; and Aga Khan University, Karachi, Pakistan (table 1). At the time of the study, Haemophilus influenzae type b (Hib) vaccine was being used for routine infant vaccination in only Kenya (since 2001) and Burkina Faso (since January 2006).
The primary target population was children aged 1–59 months with a clinical suspicion of acute bacterial meningitis, although older persons were included as well. In particular, Burkina Faso actively recruited all patients with suspected acute bacterial meningitis, regardless of age.
Laboratory procedures. CSF was collected from all study subjects and was cultured using standard microbiological procedures . LA testing was performed according to the manufacturer's instructions using Pastorex kits (Bio-Rad) in Burkina Faso and Wellcogen S. pneumoniae and H. influenzae type b kits (Remel) in Bangladesh, Kenya, Nigeria, and Pakistan. Testing for prior antimicrobial treatment was conducted by plating a strain (Kocuria rhizophila; American Type Culture Collection 9341) sensitive to most common antibiotics on sheep blood agar plates, placing a disk with 10 µL of CSF on the bacterial lawn, and measuring the diameter of growth inhibition after overnight incubation at 37°C. Any zone of inhibition was considered to be indicative of the presence of antibiotics.
The ICT was performed on all CSF specimens with a WBC count ⩾10 cells/mL or a positive culture result. The ICT consists of a hinged device in which rabbit anti-pneumococcal antibody is adsorbed onto a nitrocellulose membrane (the sample line), and goat antirabbit IgG is adsorbed onto the same membrane as a second stripe (the control line). A second set of rabbit anti-pneumococcal antibodies are conjugated to gold particles dried onto an inert fibrous support. The technician performing the assay dips a swab into the CSF sample and inserts it into the test device, adds a citrate buffer to facilitate antigen flow, and closes the device. If pneumococcal antigen is present in the specimen, it binds to the gold-conjugated rabbit antibodies, and the resulting complex is captured by the immobilized rabbit IgG stripe, forming the sample line. In addition, immobilized goat antirabbit IgG captures excess conjugated rabbit antibody, forming the control line. Results are read visually after 15 min, with the appearance of only the control line indicating a negative test result and the appearance of both the control and sample lines indicating a positive test result.
Case definitions. Suspected acute bacterial meningitis was defined by fever (temperature, ⩾38°C) and ⩾1 of the following meningeal signs: convulsions, bulging fontanelle (in children aged <12 months), stiff neck, poor sucking, irritability, prostration, lethargy, or petechial or purpural rash. CSF samples with a WBC count of 10–99 cells/mL and nonturbid appearance were considered to be abnormal but nonpurulent. Purulent CSF was defined by visual turbidity or a WBC count ⩾100 cells/mL.
Data entry and analysis. Data were entered locally, were deidentified, and were sent to the Pneumococcal Vaccines Accelerated Development and Introduction Plan (PneumoADIP) at the Johns Hopkins University for analysis. In this article, we report data from patients who had an ICT performed. We calculated the proportion of specimens testing positive by culture or LA, according to age, prior antibiotic use, and study site, and the proportion of specimens testing positive by ICT, according to age, results of culture and LA testing, prior antibiotic use, WBC count, and study site. We used χ2tests to compare proportions across strata and set a cutoff of P=.05 for statistical significance. All analyses were conducted using Stata, version 9.2 (Stata Corporation).
Ethical issues. CSF was collected as part of routine care; therefore, countries did not require informed consent for the study. Nevertheless, informed consent was obtained from all participants in Nigeria, and an information sheet explaining the purpose of the study was provided to patients or their guardians in Burkina Faso. The study was approved by local ethics review committees at all sites and by the Johns Hopkins Institutional Review Board. The Binax NOW kits were provided to the PneumoADIP at a reduced cost by Binax; otherwise, this company did not provide study support, and no staff from the company had access to study data or to advance copies of the manuscript.
Patient demographic and clinical characteristics. Recruitment began on 1 March 2006 in Bangladesh and 1 July 2006 at all other sites and continued through 30 June 2007. All patients with suspected bacterial meningitis and WBC counts ⩾10 cells/mL or a positive CSF culture result were included in the study, except those for whom the quantity of CSF remaining after all routine tests were completed was insufficient for ICT testing (211 patients in Bangladesh, 5 in Burkina Faso, 7 in Nigeria, and 37 in Pakistan). We enrolled 1173 patients, of whom 859 were aged <5 years: 415 patients in Bangladesh (358 [86%] aged <5 years), 354 in Burkina Faso (121 [34%] aged <5 years), 78 in Kenya (68 [87%] aged <5 years), 37 in Nigeria (all 37 aged <5 years), and 289 in Pakistan (275 [95%] aged <5 years). Infants (i.e., patients aged <1 year) represented 65%-73% of the children aged <5 years enrolled in Bangladesh, Kenya, and Nigeria, whereas they represented 34% in Burkina Faso and 52% in Pakistan. The percentage of male enrolled patients was 61% overall, 55% in Burkina Faso, 58% in Pakistan, 62% in Kenya, 60% in Nigeria, and 69% in Bangladesh. Enrollment followed no clear seasonal patterns at any of the sites, except in Burkina Faso, which lies within the African meningitis belt and, consistent with historical patterns, experienced a sharp increase in cases from December 2006 through April 2007, with peak recruitment in March.
In Burkina Faso, a majority (61%) of patients had a WBC count ⩾1000 cells/mL, whereas at the 4 other sites, most patients had WBC counts <100 cells/mL (figure 1). The percentage of CSF specimens that were purulent ranged from 44.6% in Bangladesh to 81.6% in Burkina Faso. The distribution of WBC counts was similar across age groups at all sites (data not shown).
The prevalence of prior antibiotic use, as detected by CSF testing for antimicrobial activity, varied from 12% in Pakistan to 53% in Bangladesh and ranged from 30% to 35% at the 3 African sites. The prevalence of prior antibiotic use did not vary by age group in Burkina Faso, Nigeria, or Pakistan. It was significantly higher among infants than among older children in Kenya (39% vs. 11%; P=.02) and Bangladesh (60% vs. 36%; P<.001). The percentage with prior antibiotic use increased with increasing WBC count in Bangladesh (48% for WBC count of 10–99 cells/mL; 53% for 100–999 cells/mL; and 69% for ⩾1000 cells/mL; P=.02); no significant trends were found at other sites.
CSF culture results. Of the 1168 CSF cultures, 209 (17.9%) were positive for a bacterial organism: 19 (9.1%) for H. influenzae , 69 (33.0%) for S. pneumoniae , 106 (50.7%) for Neisseria meningitidis (all but 1 were from Burkina Faso), and 15 (7.2%) for other pathogens (table 2). Overall, culture isolation rates were 11.1% among infants (range for individual sites, 3.1%–41.5%), 16.6% among children aged 1–4 years (range, 2.3%–57.5%), 30.6% among children aged 5–14 years (range, 0.0%–53.2%), and 30.9% among patients aged ⩾15 years (Burkina Faso only). Yields were highest in Burkina Faso, where elevated rates of meningococcal meningitis were observed from December through April 2007, coinciding with the epidemic meningitis season. In Africa, 39.9% of CSF specimens yielded an etiological agent by culture, compared with 3.2% in Asia. Pneumococci were identified by culture in 13.0% of CSF specimens from Africa and 1.1% of CSF specimens from Asia (13.3% and 1.3% among children aged <5 years, respectively; P<.001) (table 3).
LA test results. The addition of LA testing greatly increased detection of Hib and pneumococcus in both Asian countries but had limited benefit over culture in the African countries (table 2). With the addition of LA testing, pneumococci were identified in 16.0% of samples from patients in Africa and 6.3% in Asia (16.4% and 6.5% among children aged <5 years, respectively; P<.001) (table 3).
ICT versus culture. The percentage of patients with a positive ICT result ranged from 16.2% to 20%, depending on the geographic site considered (P=.96, by test for homogeneity). With CSF culture as the reference standard, the ICT was positive for 98.6% (95% CI, 92%–100%) of patients with a culture result positive for pneumococcus and was negative for 99.3% (95% CI, 96%–100%) of patients with a culture result positive for another pathogen. The ICT greatly increased the detection of pneumococci, compared with use of culture alone, at Asian sites but not at African sites (P<.001) (figure 2). Differences between regions were strongest among patients with an elevated WBC count (⩾100 cells/mL) or without prior antibiotic use.
The ICT identified pneumococci in 15.8% (95% CI, 13.5%-18.2%) of culture-negative CSF specimens; this percentage varied by prior antibiotic use (24.2% with prior antibiotics vs. 12.2% without; P<.001), region (8.5% in Africa vs. 18.8% in Asia; P<.001), and WBC count (9.0% for WBC count of 10–99 cells/mL, 22.1% for 100–999 cells/mL, and 25.4% for ⩾1000 cells/mL; P<.001, by test for trend).
Regional differences persisted after stratification by prior antibiotic use and WBC count. Among patients with negative culture results and no prior antibiotic use, the ICT identified pneumococci in 1.1% of children aged <5 years in Africa and 15.1% in Asia (P<.001). Comparable values for patients aged ⩾5 years were 5.5% and 34.6%, respectively (P<.001). Similar trends were seen among patients with negative culture results and prior antibiotic use, although the regional differences were not statistically significant for any age group (percentage with positive ICT results among all ages, 19.4% in Africa vs. 26.4% in Asia; P=.19). Among patients with negative culture results and WBC counts of 10–99 cells/mL, the ICT results were positive for 5.1% in Africa and 10.1% in Asia (P=.10), compared with 9.7% in Africa and 26.4% in Asia for patients with WBC counts of 100–999 cells/mL (P=.01) and 11.7% in Africa and 39.8% in Asia for patients with WBC counts ⩾1000 cells/mL (P<.001). These values did not vary by age group.
Although the ICT identified pneumococci in an increasing percentage of culture-negative CSF specimens as WBC count increased, the absolute number of newly identified pneumococcal infections did not vary by WBC count, because of the greater number of children with lower WBC counts: 46 pneumococcal meningitis cases were detected among the group with WBC counts of 10–99 cells/mL, compared with 54 cases among the group with WBC counts of 100100-999 cells/999 cells/mL and 51 cases among the group with WBC counts ⩾1000 cells/mL.
ICT versus LA. The ICT result was positive in 99% (95% CI, 94%–100%) of cases with an LA test positive for pneumococcus and was negative in 100% (95% CI, 98%–100%) of cases with an LA test positive for Hib, meningococcus, or other pathogens. Among children aged <5 years, 0% (95% CI, 0%–5%) of LA-negative cases tested positive by ICT in Africa, compared with 14.6% (95% CI, 11.6%–17.9%) in Asia (P=.001). Among older children (aged 5–14 years), these values were 3.7% (1 of 27) in Africa and 14.3% (9 of 63) in Asia (P=.14). Too few adults were enrolled in Asia for meaningful comparisons within this age group.
Similar to the comparisons of ICT with culture, differences between regions were observed in comparisons of ICT with LA, irrespective of prior antibiotic use or WBC count. Among patients given prior treatment with antibiotics, 5.6% with a negative result of LA tested positive by ICT in Africa, compared with 17.8% in Asia (P=.07); among patients who did not receive prior treatment with antibiotics, these values were 6% and 14%, respectively (P=.001). Among patients with CSF WBC counts of 10–99 cells/mL, 0% of those in Africa with a negative result of LA had a positive ICT result, compared with 9% in Asia (P<.001); among patients with WBC counts of 100–999 cells/mL, these values were 10.3% and 19.5%, respectively (P=.24), whereas among patients with WBC counts ⩾1000 cells/mL, these values were 5.0% and 35.0%, respectively (P<.001). The percentage of LA-negative specimens that tested positive by ICT increased with increasing WBC count (7.6% for WBC count of 10–99 cells/mL; 18% for 100–999 cells/mL; and 23.0% for ⩾1000 cells/mL; P<.001, by test for trend). There were no differences in case detection by ICT among patients with negative results of LA, according to prior antibiotic use (15.5% for prior antibiotic use vs. 11.4% for no prior antibiotic use; P=.15).
When compared with culture and the LA test, the Binax NOW ICT was >99% sensitive for the diagnosis of pneumococcal meningitis and yielded no false-positive results for Hib or meningococcal cases in these 5 African and Asian settings. It identified additional pneumococcal cases among patients with negative CSF culture and LA test results, primarily at Asian sites. Previous studies in Asia have reported a confirmed etiology—including pneumococcus—for a substantially lower proportion of suspected bacterial meningitis cases than have previous studies in Africa [1, 21–27]. However, in our study, we found that the addition of the ICT resulted in identification of a similar proportion of meningitis cases due to pneumococci at all sites, regardless of location in Asia or Africa. This occurred despite differences in patient population characteristics, such as age distribution, prior antibiotic use, and WBC count in CSF, and in environmental characteristics, such as Hib vaccine use, presence of meningococcal epidemics, prevalence of other meningitis-like syndromes, and HIV prevalence. These findings indicate that, in Asia, where the LA test and culture failed to identify the majority of pneumococcal meningitis cases, the ICT has a role in evaluations of disease burden and vaccine impact, as well as in clinical management of meningitis. In Africa, the ICT may also prove valuable in settings without adequate staff or laboratory capacity to perform culture and the LA test.
The proportion of culture-negative specimens, LA-negative specimens, and all specimens with a positive ICT result increased with increasing WBC count in CSF. This supports the contention that the ICT is identifying true instances of pneumococcus and is not yielding false-positive results, which is further supported by negative ICT results for culture- and LA-confirmed non-pneumococcal cases. Despite the higher percentage of CSF samples with pneumococci identified among patients with higher WBC counts, our study found that the total number of pneumococcal cases identified remained similar for all WBC count groups evaluated (those with WBC counts ⩾10 cells/mL). One implication of this finding is that studies focusing only on purulent meningitis, usually defined by a WBC count ⩾100 cells/mL, will miss a substantial proportion of the pneumococcal disease burden.
Compared with culture alone, use of the ICT led to increases in the proportion of specimens testing positive for pneumococcus in patients with prior antibiotic treatment that were greater than the increases in patients without prior antibiotic treatment, regardless of region. Patients who have received antibiotics prior to lumbar puncture are less likely to have viable organisms in their CSF for isolation by culture, which enhances the value of non-culture based tools, such as the LA test and the ICT, for this group.
At the African sites, the ICT provided little benefit over culture, and the small advantage disappeared when the LA test was added. The increased utility of the ICT in Asia suggests an underperformance of culture and the LA test in that region. There are several possible explanations for this finding. Our study was conducted at 2 urban centers of Asia where children may access care, including presentation to a hospital and receipt of a lumbar puncture, earlier in the course of disease than may children at the 3 African sites. The prevalence of prior antibiotic use did not differ by site. Nevertheless, differences in the availability, timing, type, and dosing of antibiotics and patient compliance with treatment may lead to higher levels of antibiotics in CSF samples among patients in Asia; the current study could not evaluate this hypothesis. These differences in access to care and patterns of antibiotic use could lead to lower antigenic loads in Asia, limiting the ability of culture or the LA test to detect pneumococcus. Regardless of the reason, it appears that, at sites with good laboratory facilities and highly trained microbiologists, a majority of cases can be identified currently by conventional methods alone in Africa, whereas more-sensitive tools such as the ICT are needed in Asia. These findings may not hold in nonresearch settings with limited laboratory facilities and poorly trained staff. Results also may change over time as meningitis epidemiology changes.
Epidemiological studies in Africa have consistently documented high incidence rates of laboratory-confirmed bacterial meningitis [1, 6, 27–30], but most studies conducted in Asia have shown very low rates [21, 22, 31, 32]. Our results suggest that the differences in bacterial meningitis incidence rates between Asia and Africa—as determined by traditional surveillance approaches—may result in part from differences in detection by the current standard methods. In contrast to surveillance studies in Asia, vaccine-probe studies (i.e., the use of vaccine in a randomized, controlled trial to evaluate disease burden rather than vaccine efficacy) conducted in Indonesia  and Bangladesh  found very high incidences of bacterial meningitis. For example, the study in Indonesia, by use of the vaccine-probe study design, estimated a bacterial meningitis incidence 10-fold higher than that estimated using results of the LA test and culture. Our study sheds light on these results. We recommend that the ICT be used as an adjunct to culture in pneumococcal meningitis surveillance projects in Asia, to ensure optimal case identification. In Africa, the ICT may be useful in settings with poor laboratory capacity or very high levels of antibiotic use. However, the ICT cannot replace the LA test in either of these settings, because commercially available LA kits test for Hib and meningococcus in addition to pneumococcus, which is particularly useful in the African meningitis belt.
We thank Faisal Imram, Désiré Ilboudo, Maxime Kienou, Régina Idohou, Shazia Azeem, and Kirimi Anampiu for help with data collection and data management.
Financial support. PneumoADIP at Johns Hopkins University, which is funded in full by the GAVI Alliance and The Vaccine Fund.
Supplement sponsorship. This article was published as part of a supplement entitled “Coordinated Surveillance and Detection of Pneumococcal and Hib Disease in Developing Countries,” sponsored by the GAVI Alliance's PneumoADIP of Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland.
Potential conflicts of interests. B.D.G., B.-M.N.-L., L.S., and J.M. state that the Association pour l'aide à la médecine preventive (AMP) receives a substantial proportion of financing for all its activities from Sanofi Pasteur, a manufacturer of pneumococcal polysaccharide vaccines. B.D.G., B.-M.N.L., L.S., and J.M. are affiliated with AMP, which was also part of the PneumoADIP surveillance network. B.D.G. and J.M. have received meeting honoraria from GlaxoSmithKline, a manufacturer or an investigational pneumococcal conjugate vaccine. All other authors: no conflicts.
Figures and Tables
- meningitis, bacterial
- developing countries
- diagnostic techniques and procedures
- latex fixation tests
- leukocyte count
- meningitis, pneumococcal
- pneumococcal infections
- cerebrospinal fluid
- pathogenic organism
- disease prevention
- surveillance, medical
- immunochromatographic test
- streptococcus pneumoniae antigen