Approaches to prevent acute bacterial meningitis in developing countries.

Approaches to prevent acute bacterial meningitis in developing countries* P.F. Wright1 Endemic acute bacterial meningitis of childhood appears to be neglected as a cause of morbidity and mortality in developing countries, probably because it has been overshadowed by the dramatic epidemics of meningococcal disease in sub-Saharan Africa. The available data based on reviews of hospitalized patients suggest that endemic meningitis is mostly a disease of young infants, Streptococcus pneumoniae and Haemophilus influenzae type b being the most important etiologic agents. The epidemiological pattern appears to be different in developing countries, compared with northern Europe or the USA, and closely resembles the early age of onset and high incidence of meningitis observed among the native American populations in Alaska. The mortality from meningitis appears to be much higher in developing countries than in industrialized countries. The availability of vaccines against the pneumococcus and haemophilus, particularly those in which the bacterial polysaccharide is conjugated to a protein, promises protection against systemic bacterial infection from these organisms. The assessment of the efficacy ofsuch vaccines will have to include a close examination ofmeningitis as an outcome. It is suggested that before such vaccines become available careful clinical and epidemiological studies ofmeningitis will help both to define the impactofthis disease andhow to design an intervention strategy. Introduction Acute bacterial meningitis is a pyogenic infection of the meninges due to colonization of the nasopharnyx and invasion of the nasal mucosa, followed by bac- teraemia and entry of bacteria into the cerebrospinal fluid and meninges, most often at the choroid plexus. The illness is usually seen in the first two years of life but can occur at any age. The organisms primarily responsible, which have been well characterized, are Haemophilus influenzae type b, Streptococcus pneu- moniae, and Neisseria meningitidis. Although other bacteria may be implicated, particularly in neonatal meningitis, most cases in both industrialized and developing countries are due to these three organisms. However, the differential diagnosis of the clinical syndrome of meningitis in developing countries has a broader spectrum of etiologies, including malaria, tuberculous meningitis, and a wide range of viral encephalitides, than in the more developed countries. The diagnosis of meningitis is dependent on not only the physician's skill and motivation, but also the performance of a lumbar puncture and the laboratory's capability to examine and culture the cerebrospinal fluid (CSF). A definite etiological diagnosis can be difficult, if not impossible, if the patient had previously * A resum6 of this article appears on pages 484-485. 'Professor of Pediatrics, Vanderbilt University, Nashville, TN 37232, USA. Former Consultant, Expanded Programme on Immu- nization, World Health Organization, Geneva, Switzerland. Reprint No. 5002 been treated with antibiotics. Epidemiological assess- ment of the impact of meningitis further demands that all such cases within a given population base be ascertained, which, to the best of our knowledge, has never been carried out in a developing country. However, the present review of the available fragment- ary data defines the impact of epidemic and endemic childhood meningitis in the developing world. Special attention is given to meningitis caused by Haemophilus influenzae type b (Hib), especially in developing coun- tries, because promising new vaccines, consisting ofthe polysaccharide coat of Hib coupled with a protein carrier, could protect even during the first six months of life against systemic infections due to Haemophilus. With these bacterial pathogens, meningitis is only one possible manifestation because the bacteraemia could lead to pneumonia and other focal infections. Epidemiology Epidemic meningitis The historical literature on meningitis in developing countries mostly describes the dramatic epidemics of meningococcal A disease across sub-Saharan Africa (1). Although the epidemiology has been well defined, the reasons for the characteristic epidemic patterns are not fully understood (1). Prevention of meningococcal disease has been attempted by the administration of meningococcal A and C polysaccharide vaccines, and Bulletin of the World Health Organization, 67 (5): 479-486 (1989) © World Health Organization 1989 479 P.F. Wright there is evidence supporting the efficacy of group A vaccine from a number of African countries (3) and the group C component (4) from Brazil in children over 24 months of age. The relative infrequency ofepidemics and the necessity to give vaccine at an age not included in the immunization schedule of the Expanded Pro- gramme on Immunization (EPI) have hampered the development of a uniform vaccination policy. No country is at present regularly vaccinating children with meningococcal vaccines. An apparent sparing of children under two years of age in epidemics and a duration of protection of at least four years in older children suggest that vaccination of children at two- year intervals and adults every 4-6 years might be a reasonable policy (5). Evidence that a single dose of chloramphenicol is curative means that the disease, if recognized early, may be relatively easily treated (6). Endemic childhood meningitis Owing to the emphasis on the epidemics of meningo- coccal disease, the impact of endemic childhood bacterial meningitis in developing countries has been overlooked. In the developed world the impact of childhood meningitis is well documented and appreci- ated. In North America, Hib accounts for the majority of meningitis cases (Table 1), which in England and France are frequently due to meningococcus (types B and C). The importance of the latter in Europe, compared with the USA, is unexplained epidemiolog- ically. As regards Hib infection in industrialized coun- tries, sections of the American population have been identified with enhanced susceptibility at an early age; they include American Indians, Eskimos, children with sickle-cell disease, and children in day-care centres. Both the incidence and age distribution of the disease within these groups differ from that in the remainder of the United States population. Another feature of Hib meningitis in Europe was that a higher proportion of cases occurred in children between 18 months and 5 years of age. The age distributions from some developing countries are also summarized in Table 2. The available data on mortality from Hib menin- gitis (Table 2) for the USA and Europe are population based; for the developing countries they represent a very limited sampling of university hospitals, where there may be some bias since only the very sick children were brought to these hospitals. In spite of these limitations, the figures reflect a very high mor- tality associated with meningitis in the developing countries; in contrast, proper management of Hib meningitis cases (by intensive care and rapid initiation of parenteral antibiotic therapy) has helped the indus- trialized countries to achieve the current low mortality figures. Not included in this assessment are the neuro- logical sequelae of meningitis with delayed mortality, which pose a substantial burden for families trying to care for these children. In several developing countries meningitis has been documented to account for a substantial number Table 1: Causes of bacterial meningitis, Including H. Influenza. type b (Hib), In children les than 5 years of age Population and country Percentage of cases caused by: Hib Pneumococcus Meningococcus Other organisms European England (17)' 37 12 48 3 France (18) 41 19 37 3 North American USA (19) 70 17 10 3 African Senegalb 41 39 10 10 Nigeria (20) 40 39 8 13 Cameroon (7) 42 43 3 12 African (meningitis belt) Egypt (21) 25 36 20 19 Ethiopia (22) 41 22 25 12 Asian Papua New Guinea (8) 49 43 5 3 South American Chile (23) 63 29 6 2 ' Figures in parentheses indicate the reference to the study. Trevoux, C. Etude statistique de 1052 cas de meningite purulente observde Dakar. Thesis, University of Lyon, 1972. 480 Approaches to prevent acute bacterial meningitis In developing countries Table 2: Age distribution of Hib meningitis and mortality In various populations Population and country Age group Mortality (%) < 1 year 1-2 years 2-5 years 5-15 years European Finland (24)' 25 35 34 6 3 Sweden (25) 31 23 36 10 1.4 England (17) 30 25 36 9 5.7 North American USA (19) 53 30 11 6 6 Alaska (non-native) (26) 50 30 18 2 Not available Alaska (native) (26) 67 24 6 3 6 Navaho (Indian) (27) 77 - 22- 1 8 African Senegalb 75 14 11 0 33 Nigeria (20) 84 8 4 8- 26 Cameroon (7) 55 25 17 3 24 Asian Papua New Guinea (8) 92 8 0 0 30 South American Chile (23) 82 13 4 1 11 ' Figures in parentheses indicate the reference to the study. b See footnote b in Table 1. ofin-hospital deaths, e.g., 8.5% in the paediatric unit in Yaounde (Cameroon) and 15% in Goroka Hospital in the Papua New Guinea highlands (7, 8). In summary, four different epidemiological and clinical patterns of Hib systemic disease are evident. (1) In northern European countries Hib infection occurs at a late age and with a low incidence; epiglottitis is an important component of the disease spectrum, accounting for 23% of the total Hib disease in Finland (9). (2) In the USA and in non-native Alaskan popu- lations there is an intermediate pattern ofHib infection with earlier onset than in Europe but with low mortality. (3) In other populations in the Americas, as shown by data from Chile, Jamaica, and the Eskimo and Navaho populations, there is very early onset and a high incidence of Hib infection. (4) In some African and south-east Asian coun- tries (Cameroon, Nigeria, Papua New Guinea and Senegal), there is very early onset of Hib disease with a high incidence and striking mortality; however, pneumococcal disease is relatively more prominent together with other organisms, chiefly Gram-negative enteric bacteria including salmonella species; epiglot- titis is rarely recognized clinically. The data on the incidence of meningitis are fragmentary in the developing countries (Table 3), the reported cases being mostly hospital based. However, none of the 147 meningococcal meningitis patients in a rural area of the Gambia was seen in a hospital (2); the two estimates from Senegal and Papua New Guinea are based on patients presenting from popula- tions known to be served by hospitals. No one has attempted the very difficult task of identifying cases that are not admitted for care. The available epidemiological facts, particularly the very young age of onset, would suggest that the Table 3: Attack rate of Hib meningitis (cases per year per 1000 children less than 5 years of age) Population and country Rate European England (17)' 0.11 Finland (24) 0.32 American USA (19) 0.19-0.63 Alaska (non-native) (26) 0.69 Alaska (native) (26) 2.8 Navaho (Indian) (27) 1.7 African Senegalb 0.36 Asian Papua New Guinea (8) 2.6 a Figures in parentheses indicate the reference to the study. See footnote b in Table 1. P.F. Wright incidence and mortality will be found to resemble or exceed those in identified high-risk groups such as native Alaskans. If one assumes that there are 2.0 cases per thousand children per year under five years of age with a mortality of 30%, this leads to a worldwide estimate ofone million cases ofHib meningitis per year in developing countries with 300 000 deaths and 200 000 children with neurological damage. In planning interventions the impact of menin- gitis and other systemic Hib disease (epiglottitis, septicaemia, septic arthritis, cellulitis and pneumonia) should be much more precisely defined. In Papua New Guinea, H. influenzae species were the most common organisms recovered from the lung aspirates and blood of children with pneumonia (10). Although not all of these were type b, approximately 20% of the cultures that grew significant bacteria were Hib, suggesting that other forms of systemic disease should be taken into consideration in assessing the effective- ness of Hib and other polysaccharide-conjugate vac- cines. Prevention Polysaccharlde vaccine development There was early recognition that the age-related incidence of Hib meningitis rose following the initial decay of maternal antibody and then declined in the second year of life with the gradual acquisition of natural antibody through pharyngeal carriage, sys- temic infection or exposure to cross-reacting antigens. The concept that circulating antibody was the critical determinant of the host's defence against Hib gained further credence from animal experiments, from sparing of disease in patients on replacement gamma-globulin for immune deficiency, and from use of prophylactic gamma-globulin (11). The critical antigen was identi- fied early in the understanding of Hib to be the capsular polysaccharide-polyribosyl ribitol phos- phate, PRP. This antigen shares the characteristic of many carbohydrate antigens of eliciting a T-cell independent immune response in which there is delay in the infant's ability to mount a response, which is often of the IgM phenotype and does not boost on repeat immunization. Polysaccharide vaccines that were developed for a number of encapsulated bacteria have been evalu- ated in adults and children, the findings amply confirm-- ing the difficulties in their use as immunogens but with encouraging signs. Pneumococcal polysaccharides, particularly the types common as pathogens, are typically poorly immunogenic in young children; in spite of this, a study from Papua New Guinea sugges- ted efficacy in prevention of deaths from respiratory illness (12). This study must be repeated since logical interpretation of the data is difficult without evidence of induction of an antibody response. Pneumococcal vaccines are, however, licensed in the USA for use in the elderly to prevent pneumonia. As described earlier, meningococcal A and C vaccines are efficacious in children and may have a role to play in epidemic and endemic childhood meningococcal disease. However, meningococcal B vaccine which has N-acetylneura- minic acid (as its capsular substance) is poorly anti- genic in all ages. Finally, a large trial with Hib in Finland demon- strated efficacy ofthe vaccine in children over approxi- mately 24 months old (13). Based primarily on the efficacy in Finland, the polysaccharide Hib vaccine was licensed for use in the USA in 1985 in children at 24 months of age. After introduction of the vaccine, surveillance was started to determine the vaccine's efficacy, estimates ofwhich vary in different studies but appear to be less than optimal, perhaps in the order of 50% (14). No benefit would be expected from such unconjugated polysaccharide vaccines in developing countries because their use is restricted to children over two years of age. Following the lack of immediate success of poly- saccharide vaccines in the prevention of childhood illness, efforts were begun to couple the polysaccharide to a protein carrier, thus changing the nature of the immune response to a T-cell dependent response which can be invoked at an earlier age. The choice of protein carrier, coupling agent, and formulation of the polysaccharide have all been varied in an effort to achieve optimal immunogenicity. The resulting vac- cines have been evaluated in a number of phase-I and phase-2 safety and immunogenicity trials, and an Hib polysaccharide-diphtheria toxoid vaccine has under- gone field trials in Finland and Alaska. In Finland the vaccine was given at 3, 4, 6, and 14 months to approximately 30000 infants, 80% of them having a serum antibody response after the third dose. After 9 months of follow-up, the vaccine efficacy was 87% with confidence limits of 50-96% (15). The Alaskan trial has been completed and is being analysed; the initial results are far less promising than those from Finland (J.I. Ward, personal communication, 1988). In December 1987, the Hib-diphtheria toxoid vaccine was licensed for use in the USA. Other Hib vaccines have used as conjugates a nontoxic variant (CRM197) of diphtheria toxin, the outer membrane proteins of N. meningitidis, and tetanus toxoid. It appears that these different structures are not likely to behave in an identical fashion as immunogens. Thus, among the currently evaluated series of vaccines against Hib infection, only one conjugate appears to have high efficacy, based on the Finnish data. However, the epidemiology of Hib disease in Finland is very different from that in the developing Approaches to prevent acute bacterial meningitis In developing countries countries where the available data suggest that 80% of recognized Hib illness occurs in the first year of life. In every population surveyed, the first two months of life are almost completely spared of disease and the peak incidence of disease remains in the second six months of life. Immunization would therefore have to be effective by six months of age. Analysis of a second important trial from Alaska will be completed shortly. It will be of more relevance to developing countries because ofthe similar early age of onset of disease. Further information is expected from post-licensing surveillance in the USA which should establish the efficacy of the Hib-conjugate vaccines for children older than 18 months; if these vaccines prove successful in this age group, they will undoubtedly be evaluated for inclusion in the primary series of immunizations in the first six months of life. Preliminary strategies With this background one can now consider how to demonstrate whether the conjugated Hib vaccine has a sufficiently high efficacy and cost-benefit ratio in developing countries to deserve consideration for inclusion in the EPI. The following clinical studies are suggested that should lead towards that goal. (1) Definition of maternal antibody levels of Hib antibody and the patterns of antibody decay and acquisition in the first two years of life If the pattern of acquisition of disease is reflected in the serum antibody levels, the height of maternal serum antibody levels might be expected to reflect the age of onset of susceptibility of the infant. With early onset of disease one would expect either that maternal levels are low, placental transfer is ineffective, and/or antibody decay is more rapid, or that exposure is so intense that a moderate level of passive antibody can be overcome. The subsequent rate of acquisition of antibody should be a reflection of the intensity of exposure. Significant questions exist about the duration of maternal anti- body in developing countries and capability of the infant in this setting to respond to antigenic challenge. Studies on 50 children, with serum samples every two months for the first six months of life and every four months thereafter until two years of age, at each offour geographically diverse sites would help to answer these questions. For each of these children a matched maternal serum sample should be drawn at the time of the birth of the child. Comparative data are published from the USA and from the Eskimo population. (2) Comparison of vaccine safety and immunogenicity in developing countries Another step is to do immunogenicity trials with Hib-conjugate vaccines given in a schedule that is consistent with incorporation into EPI. Safety ofthese vaccines is not a major issue, but their immunogenicity must be established in developing countries. This is important not only for the PRP vaccines mentioned above, but as a prototype for vaccines against other encapsulated bacteria. Age and interval between doses may be very important with these vaccines. It must be demonstrated that they can be combined with diphtheria-tetanus-pertussis (DTP) vaccines without loss of potency and without giving excessive amounts of carrier protein, particularly if there is already a vaccine component such as diphtheria or tetanus. Duration of the antibody will also be a significant concern with these vaccines. A primary series must provide protection until the initiation of a strong response to natural exposure. Vaccine and control groups (with 50 children in each) for studies at two to three representative sites throughout the developing world might be sufficient to establish immunogenicity. The determination of anti- body levels should be done in a reference laboratory in a developed country to ensure standardization of the results. Similar trials from the USA would provide comparative data. It must be recognized that the immunogenicity of different conjugated polysaccharides may vary. Al- though relevant data will be generated in the USA and elsewhere as to the optimal preparation for use in industrialized countries, these findings will have to be replicated in developing countries. The Alaskan data appear to confirm the need for this approach. (3) Definition ofthe age-specific attack rate ofmeningitis Sites would have to be chosen providing a high quality of clinical care and bacteriology services to a defined population base. Prospectively all cases of meningitis would be identified for a one-year period. Simultan- eously a retrospective one-year review would give interesting comparative information. The identifica- tion of cases that presented to the hospital would have to be coupled with an effort to assign a cause of death to those children who do not reach medical care. Meningitis can be a fulminant disease with death occurring within 24 hours of the onset of symptoms. Three possible categories of meningitis are (1) bacteriologically proved, (2) antigen-positive with evidence ofCSF changes, and (3) clinically diagnosed. A recent study has demonstrated good sensitivity and specificity of the antigen detection method in Ethiopia (16) but this could not substitute for a lumbar puncture or for good bacteriological techniques. Given that 20% of the population in a developing country is under 5 years old and that an estimated rate ofHib meningitis in this age group might be 2 per 1000 483 P.F. Wright per year, a total population of 38 000 people would have to be under surveillance for one year to establish that this was a real rate + 1 per 1000. We believe it is not unrealistic that such information might be obtain- able from selected centres in the developing world. There is also the question ofwhether by interview- ing parents shortly after a child's death an accurate cause of death can be established by their description of, for example, a febrile illness with neurological signs such as stiff neck or seizures. As virtually all untreated cases of meningitis are fatal, these records combined with the clinic and hospital records might give a rela- tively complete case ascertainment. Such investiga- tions might be conducted initially in a region free of malaria which may be the major confounding cause of sudden febrile illness leading to death. A counter argument for performing at least one study in central Africa is that sickle-cell disease and other haemoglo- binopathies clearly increase the risk of systemic infec- tion with encapsulated bacteria. If a cause of death survey could be validated as a tool for examining meningitis and other broad categories of disease it could prove very useful. This question is being approached in the Gambia and in Egypt. It is worth noting, however, that in England 26 out of 94 fatal cases of meningitis were undiagnosed until autopsy (17). Establishment of the disease burden imposed by meningitis will be essential for a consideration of vaccine usage. It would also be essential to any estimate of the size needed to conduct a trial aimed at establishing vaccine efficacy in a developing country. Obviously such a trial should attempt to document other systemic Hib, meningococcal and pneumococcal disease at the same time, particularly through the use of blood cultures. The establishment of bacterial etiology will be more difficult with pneumonia than with sepsis and meningitis. (4) Assessment of vaccine costs and stability Before further exploration of the vaccine its true cost would have to be established and assessed in terms of the results available from the above three investiga- tions. The scale ofdemand might make it an affordable vaccine after recouping the initial development costs in the industrialized countries. (5) Efficacy trial An efficacy trial with the optimal vaccine preparation will have to be carried out in a developing country to determine if antibody can be generated in time to prevent the very early onset of disease in this setting. One approach would be to use the conjugated Hib vaccine as a test group in a pneumococcal vaccine study. In addition to efficacy against meningitis, the Hib vaccine might lead to fewer pneumonia cases, approximately 10% of which may be caused by Hib. As the two vaccines would potentially decrease both total meningitis and respiratory illnesses, specific etiologies would have to be very carefully identified. This is the approach originally used in Finland in the meningococcus A and Hib vaccines study. It appears that any investigation to assess the impact of encapsu- lated bacteria, even if aimed primarily at acute respira- tory disease, should incorporate a careful examination of systemic disease into the protocol because of the impact of the disease and the unequivocal nature ofthe endpoint. The scope of such a trial would be large. In Finland 60 000 children were enrolled and followed for an average of nine months to establish efficacy. Their rate of illness is about one-tenth of that anticipated in a developing country, where the surveillance would almost certainly be less efficient. (6) Justification for the approach The above approach to determine the role of a particu- lar vaccine in immunization programmes in develop- ing countries includes a progression of investigations that must be done-some quite simple before an informed decision can be made on its use, for which the current epidemiological and clinical justifications are inadequate. Acknowledgement The author thanks Brian Greenwood, Richard Moxon, Heikki Petola and Joel Ward for their thoughfful comments during the preparation of this manuscript. Resume Prevention de la meningite bacterlenne algue dans les pays en developpement La meningite bacterienne aigue est une infection pyogenique des meninges resultant de la colonisa- tion du rhinopharynx et de l'envahissement de la muqueuse nasale, suivie du passage de la bacterie dans le sang, le liquide cephalo-rachidien et les meninges. Les principaux microorganismes res- ponsables sont Haemophilus influenzae type b, Streptococcus pneumoniae et Neisseria meningi- tidis. Le diagnostic de la meningite est fonde sur l'identification de la bacterie en laboratoire apres ponction lombaire. 11 est rendu plus difficile par une antibiotherapie prealable. Le veritable impact de la 484 Approaches to prevent acute bacterial meningitis In developing countries meningite ne peut Otre d6termin6 que si tous les cas survenus dans une population donnee sont detectes. Le present article porte principalement sur la meningite a Haemophilus, car il est possible de proteger les nourrissons a l'aide d'un vaccin dans lequel le polysaccharide bact6rien est conjugue a une proteine. Le tableau epid6miologique de cette forme de m6ningite differe d'un pays a I'autre: en Europe, son incidence est moindre et elle survient a un Age plus avance qu'aux Etats-Unis oiu, dans certaines populations (notamment les Esquimaux), la maladie est particulierement frequente dans les 6 a 12 premiers mois de la vie. Les rares donnees recueillies dans les pays en developpement donnent a penser que la situation est comparable a celle que l'on observe chez les Esquimaux. Un essai mene en Finlande a etabli l'efficacite d'un vaccin conjugue anti-Haemophilus dans ce pays. Un essai analogue en Alaska a ete moins convaincant. II est cependant possible que les vaccins actuels soient plus antigeniques que ceux qui ont 6te utilises dans ces essais. II est propose d'entreprendre une serie d'etudes pour evaluer l'importance de la m6ningite dans les pays en d6veloppement et etablir les bases d'une strategie d'intervention. Ces 6tudes consisteraient &: 1) d6finir les taux d'anticorps maternels diriges contre les Haemophilus et les mecanismes de degradation et d'acquisition des anticorps au cours des deux premieres annees de la vie, 2) comparer l'innocuit6 et l'immunog6nicite des vaccins dans les pays en developpement, 3) definir le taux d'atteinte de la m6ningite en fonction de l'age, 4) evaluer le cout et la stabilite d'un vaccin potentiel, et 5) mettre au point un essai d'efficacit6 des vaccins bacteriens dont les crit6res seraient le nombre de cas de m6ningite et d'infection bacterienne ainsi que l'impact du vaccin sur les maladies respiratoires aigues. References 1. Lapeyssonnle, L. La meningite cerebrospinale en Afrique. Bulletin of the World Health Organization, 28: 3-114 (1963). 2. Grenwood, B.M. et al. Factors influencing susceptibil- ity to meningococcal disease during an epidemic in the Gambia, West Africa. Journal of infection, 14: 167-184 (1987). 3. Greenwood, B.M. Selective primary health care: strate- gies for control of disease in the developing world. Xil. Acute bacterial meningitis. Reviews of infectious dis- eases, 6: 374-389 (1984). 4. Taunay, A.E. at al. [Assessment of the protection conferred by anti-group C meningococcal polysaccha- ride vaccine to 6-36-month-old children.] Revista do Instituto Adolfo Lutz, 38: 77-82 (1978) (in Portuguese). 5. Mohammed, I. et al. Control of epidemic meningococcal meningitis by mass vaccination. Journal of infection, 9: 190-196 (1984). 6. Vlmont-Vicary, P. & Rogerle, F. Epidemie de meningite cdrdbrospinale 4 Neisseria meningitidis dans la rdgion sanitaire de Ruhengeri (Rwanda). Medicine tropicale, 43: 155-161 (1983). 7. Benard-Bonln, A.C. & Ekoe, T. Les meningites purulentes de l'enfant a Yaounde: aspects epidemiologiques et prog- nostiques. Annales de la Socidt6 belge de Medecine tropicale, 65: 59-68 (1985). 8. Graotn, M. et al. The aetiology of purulent meningitis in highland children: a bacteriologic study. Papua New Guinea medical joumal, 28: 233-240 (1984). 9. Pola, H. & Virtanen, M. Systemic Haemophilus influenzae infection in Finland. Clinical pediatrics, 23: 275-278 (1984). 10. Shann, F. et al. Aetiology of pneumonia in children in Goroka hospital, Papua New Guinea. Lancet 2: 537-541 (1984). 11. Santoham, M. tal. Prevention of Haemophilusinfluenzae type b infections in high-risk infants treated with bacterial polysaccharide immune globulin. New England joumal of medicine, 317: 923-29 (1987). 12. Riley, l.D. et al. Pneumococcal vaccine prevents death from acute lower-respiratory-tract infections in Papua New Guinean children. Lancet 2: 877-881 (1986). 13. PdbIa, H. etal. Prevention of Haemophilusinfluenzaetype b bacteremic infectfons with the capsular polysaccharide vaccine. New Englandjournal of medicine, 310:1561-1566 (1984). 14. Immunizaton Practices Advsory Commilee. Update: Pre- vention of Haemophilus influezaetype B disease. Morbidity and mortality weekly report, 36: 529 (1987). 15. Eskola, J. et al. Efficacy of Haemophilus influenzae type b polysaccharide-diphtheria toxoid conjugate vaccine in infancy. New England journal of medicine, 317: 717-722 (1987). 16. Habte-Gabr, E. et al. Rapid etiologic diagnosis of pyogenic meningitis by coagglutination, latex aggluti- nation and immuno-osmophoresis of cerebrospinal fluid, serum and urine. Tropical and geographical medicine, 39: 137-143 (1987). 17. Goldacre, M.J. Acute bacterial meningitis in child- hood. Lancet, 1: 28-31 (1976). 18. Olivares, R. & Vacarle, M. Meningites bacteriennes de l'enfant 4 Paris de 1982 4 1986 epidemiologie et pronostic. Bulletin dpiddmiologique hebdomadaire (France), 47: 185-186 (1987). 19. Fraser, D.W. et al. Bacterial meningitis in Bernalillo county, New Mexico: a comparison with three other American populations. American journal of epidemio- logy, 100: 29-34 (1974). 20. Nottidge, V.A. Haemophilus influenzae meningitis: a 5-year study in Ibadan, Nigeria. Journal of infection, 11: 109-117 (1985). 21. Guirguls, N. et al. Bacterial meningitis in Egypt: analysis of CSF isolates from hospital patients in Cairo, 1977-78. Bulletin of the World Health Organiza- tion, 61: 517-524 (1983). 22. Hallemeskel, H. & Tafarl, N. Bacterial meningitis in childhood in an African city. Acta paediatrica scandina- vica, 67: 725-730 (1978). 485 P.F. Wright 23. Juliet, C. et al. [Bacterial meningitis in children: experience with 441 cases.] Revista medica de Chile, 111: 690-698 (1983) (in Spanish). 24. Valmarl, P. et al. Invasive Haemophilus influenzaeand meningococcal infections in Finland. Scandinavian journal of infectious diseases, 19: 19-27 (1987). 25. Claesson, B. et al. Incidence and prognosis of Haemo- philus influenzae meningitis in children in a Swedish region. Pediatric infectious disease, 3: 35-39 (1984). 26. Ward, J.l. et al. Invasive Haemophilus influenzae type b disease in Alaska: background epidemiology for a vaccine efficacy trial. Journal of infectious diseases, 153: 17-25 (1986). 27. Coulehan, J.L. et al. Bacterial meningitis in Navajo Indians. Public health reports, 91: 464-468 (1976). 486