The majority of cases of bacterial meningitis were boys and 57% of those in ... meningitis was 19%, permanent neurological sequelae were seen in 26% of ...
Annals of Tropical Paediatrics ( 1990) 10, 203-210
Endemic bacterial meningitis in Sudanese children: aetiology, clinical findings, treatment and short-term outcome M.A. M. SALIH, A. I. EL HAG, H. SID AHMED*, M. BUSHARA**, I. YASIN***, M. I. A. OMER, Y. HOFVANDERt & P. OLCENt Departments of Paediatrics and Child Health and* Microbiology, Faculty of Medicine, University of Khartoum; **Department of ENT and*** University Paediatric Ward ( C 2 ) , Khartoum Teaching Hospital, Sudan; t International Child Health Unit, Department of Pediatrics, University Hospital, Uppsala; tDepartment of Clinical Microbiology and Immunology, Orebro Medical Center Hospital, Orebro, Sweden (Received 30 October 1989)
Summary. During the period April1985 to November 1986 (18 months), 196 children (of age > 1 month) admitted to the Children's Emergency Hospital in Khartoum, Sudan, with clinical suspicion of meningitis /meningoencephalitis were followed up prospectively. Bacterial meningitis was diagnosed by culture, direct microscopy and/ or antigen -detecting assays (co-agglutination and enzyme immunoassay) in 44 infants (25 Haemophilus infiuenzae type b, 8 Neisseria meningitidis, 7 Streptococcus pneumoniae, 3 enterobacteria and one mixed infection), aseptic meningitis in 52, cerebral malaria in 4 and febrile convulsions in 96. The majority of cases of bacterial meningitis were boys and 57 % of those in whom H. infiuenzae was the commonest isolate were less than 1 year old. The presenting signs and symptoms are described as well as the transient and permanent short-term sequelae. The total mortality from bacterial meningitis was 19 % , permanent neurological sequelae were seen in 26 % of survivors . Prospective follow-up, including audiometry, of 35 children 1-2 months after discharge showed that 11 % had hemiplegia and 20% had hearing impairment. The potential impact of vaccination against invasive H. infiuenzae infections is discussed.
Introduction Studies on bacterial mening1t1s in Sudan, including those related to children, have been focused on meningococcal (MC) meningitis. 1- 4 Being within the African meningococcal belt, Sudan has experienced repeated epidemics throughout its history followed by years of relative quiescence. 5 With the success of field trials of group A meningococcal polysaccharide vaccine and its implementation in 1974, eventual control of epidemics of MC disease seemed to be possible, provided wider coverage was
achieved and vaccination programmes were sustained. 3' 6 However, no prospective study has been done to reveal the endemic (i.e. inter-epidemic) situation of meningococcal meningitis in Sudan. Also, the role of the other two important organisms that cause the disease (Haemophilus infiuenzae (HI) and Streptococcus pneumoniae (PNC)), has not been ascertained. Such a study seems to be warranted in view of reports of an overall increasing incidence of HI meningitis, 7 its variability with race 8 and the recent advent of a vaccine that can prevent 90% of invasive HI type b disease in children over the age
Reprint requests to: Dr Mustafa Abdalla M. Salih, Department of Paediatrics and Child Health, Faculty of Medicine, PO Box 102, Khartoum, Sudan.
204
M.A. M. Salih et al.
of 18 months. 9 Additional aims were to learn to recognize the clinical signs and symptoms in children on admission as well as the complications. Materials and methods
Patients The Children's Emergency Hospital (CEH) serves as the largest referral emergency centre for childhood diseases in the province of Khartoum, and was therefore chosen to serve as the base for our study. Children admitted to CEH with a provisional diagnosis of meningitis/meningoencephalitis, Saturday through Wednesday, between April1985 and November 1986 were studied prospectively. For reasons concerning the proper handling of samples by the laboratory, admissions on Thursdays and Fridays were excluded. Infants less than 1 month of age were excluded because pyogenic meningitis in this age-group can be considered to be an entity on its own, reflecting either the maternal flora or the environment in which the child was born. Children were cared for in the CEH and in the isolation ward of Khartoum Teaching Hospital, to which they were transferred after 2- 4 days for a minimum of 10 days . Their antibiotic treatment throughout the period of hospitalization consisted of i.v. benzylpenicillin (300 IU /kg/day) in combination with chloramphenicol (75-1 00 mg/kg/ day) given 6-hourly. Chloramphenicol was given orally on the 3rd day after admission or when the child's condition permitted. Chemotherapy was altered when necessary in the light of sensitivity tests, whereas supportive measures (e.g. dexamethasone and anticonvulsants) were used as indicated. On admission, details of signs and symptoms relevant to meningitis were entered for each child on a specially designed form and also details of daily surveillance of complications and sequelae. The nutritional status of each child was evaluated according to the Wellcome classification using NCHS growth charts. 10' 11
The state of consciousness was allocated to one of the four following categoriesY stage !-drowsiness; stage II- light coma (the patient is unarousable even with painful stimuli but can moan or make semipurposeful avoidance movements); stage III- deep c;:oma (failure of response to or decerebrate posturing resulting from painful stimuli); stageiV- patient is flaccid and apnoeic. Assessment of the development of subdural effusion and/or hydrocephalus depended on facilities available, i.e. routine transillumination test and monitoring the head circumference. Skull X -rays and ultrasound (in those with an open anterior fontanelle) were done when either of these complications was suspected. Each child was examined about 1 month after admission and findings were registered on a specially designed sheet. He/she was then referred for audiological assessment to one of the authors (M. Bushara). The hearing test was selected according to the child's ability to cooperate. Behavioural audiometry was used for young children and conventional audiometric techniques for older ones. Impedance studies were performed on all children. Home visits were arranged for families who failed to bring their children for follow-up. As a result, 35 children had audiological assessment at between 1 and 2 months after admission. Methods Each cerebrospinal fluid (CSF) was examined for cells and bacteria after Gram staining and was cultured on conventional culture media. In the presence of a white blood cell count (WBC) of > 10 x 10 6 / 1, CSF was examined for bacterial antigens by a commercial coagglutination (CoA) kit (Phadebact CSF Test, Pharmacia, Uppsala, Sweden). A portion was also stored in liquid nitrogen for 7- 14 months before being transported to Sweden where it was retested using an enzyme immunoassay (EIA) technique (Pharmacia Meningitis EIA-Test). Details of these laboratory tests are reported elsewhere. 13 The
Bacterial meningitis in Sudanese children diagnosis of bacterial meningitis was based on one or more of the following results in testing CSF: (a) positive culture; (b) positive direct microscopy with Gram stain; (c) presence of bacterial antigens detected by CoA or EIAtest. Associate Professor Anneka Ehrnst and Professor Marianne Forsgren at the Stockholm City Central Microbiological Laboratory kindly examined the CSF and acute and convalescent plasma samples from children with suspected viral meningitis/meningoencephalitis. Virus isolations were done on both CSF and acute plasma. The serological assays covered influenza A, mumps, Herpes simplex, varicella, measles and CMV. Statistics For the comparison of two proportions, Fisher's exact test was used, with calculation of the probabilities in both tails of the distribution. Only p values < 0.05 were considered to indicate significance.
R esults A total of 196 children were admitted during the study period (18 months) with meningitis or meningoencephalitis as a provisional diagnosis . Of these, 44 had bacterial meningitis and 52 had clinical and laboratory features consistent with aseptic meningitis. A virus was isolated in three of these (one ECHO 16 and two mumps) and the serological tests were suggestive of a recent primary viral infection in two additional cases (one measles and one CMV). The final diagnosis in 96 cases was febrile convulsions and cerebral malaria in four. The bacterial cases consisted of28 boys and 16 girls (male:female= 1.8:1). Their ages ranged from 3 months to 14 years (Fig. 1). Twenty-six (59%) were aged 1-11 months, two (5 % ) were 12-17 months, 8 (18 %) were 1.5-5 years and 8 (18 % ) were older. Aetiological bacteria During the 18 months of the study, HI was the most frequently isolated pathogen, causing meningitis in 25 (57 % ) of the 44 cases, 21
205
of whom were less than 18 months of age (Table I, Fig. 1). All strains examined were type b . MC meningitis was the second most common strain, affecting eight children (18 % ). Seven (16%) had PNC meningitis. The three MC strains available for seragrouping were all group A and sensitive to sulphadiazine. Other organisms infected four children, all of whom were less than 12 months of age (Table I). Of the pre-school children ( < 5 years) with meningitis, a higher proportion were infected by HI than by MC orPNC ( p=0.002). One infant whose CSF grew Salmonella spp. was removed by his family against advice about 1 hour after lumbar puncture and therefore was thereafter excluded from the study. Presenting symptoms and signs As shown in Table II, neck rigidity, convulsions and anorexia were the symptoms most commonly reported on admission, presenting in 86 % , 77 % and 67 % of the children, respectively. There was no correlation between convulsions prior to admission and age or type of bacteria. About half of the children showed symptoms of acute respiratory tract infection, but this had no significant association with the organism causing meningitis. Clouding of consciousness was reported in 18 (42 % ). Twenty-one (49 % ) children were malnourished; 16 (37 %) of them were underweight and 5 (12 % ) were marasmic. The numbers of children showing one or more of the cardinal signs of meningitis are shown in Table III. The axillary temperature ranged between 37.5 and 39.0°C in 32 (74 % ) children, six (14 % ) had a higher temperature and 5 (12 % ) were afebrile. Of the 19 (44 % ) who had disturbed sensorium, 13 (30 % ) showed stage I, 4 (9 % ) stage II and 2 (5 % ) stage III coma. No significant association was found between age and either disturbed consciousness on admission or frequency of neck rigidity. Sixteen (37 % ) of the children were dehydrated, the dehydration
206
M.A. M. Salih et al.
20
llii!] H. influenzae
IS
"'c:
~
-~
~
CIS
Q..
fiJ
N. meningitidis
§I
~- pneumoniae
10
"""'0 0 c:
5
I 1- 11
12-17
18-59
>59
age-group (months) FIG. 1. Age distribution of 44 Sudanese children with meningitis due to H. infiuenzae, N. meningitidis and S. pneumoniae, April1985-November 1986. I. Aetiology of bacterial meningitis in 44 Sudanese children(> 1 month), Apri11985-November 1986
TABLE II. Symptons of meningitis in 43 Sudanese children ( > 1 month) with bacterial meningitis, April 1985-November 1986
being mild in 11 (26%), moderate in 4 (9%) and severe in 1. Skin rash was seen in a child with MC meningitis. No photophobia was recorded in the study population. Lip herpes was seen
in two children with Salmonella and MC meningitis. Neurological abnormalities on admission consisted of palsy of the 3rd cranial nerve in
TABLE
Bacterial meningitis in Sudanese children
207
TABLE III. Signs of meningitis on admission in 43 Sudanese children over the age of 1 month, April 1985-November 1986
TABLE IV. Transient complications and permanent sequelaeofbacterial meningitis among 35 surviving children ( > 1 month) in Sudan, Aprii1985-November 1986
Sign
Transient complications
Fever Neck stiffness Kernig's sign Bulging fontanelle* Coma (stages I-IV)**
Number (%) 38 (88) 38 (88) 25 (58) 13 (54) 19 (44)
* In 24 children with open anterior fontanelle; **for definition see Materials and methods section.
one child and the 6th in another. Five had hemiplegia, associated with Jacksonian seizure in one patient with PNC meningitis. Quadriplegia was observed in one. The latter two children and those with cranial palsies died within 24 hours of admission.
Symptoms and signs during hospital stay Eight children died, giving a total case fatality rate of 18.6 %. HI was fatal in four children (16%) and PNC meningitis in two. One child died of a mixed infection (E. coli+ E. cloacae+Pseudomonas spp.) and another with sickle cell disease died of Salmonella meningitis. In five children, death occurred less than 48 hours after admission to hospital. There was no association between mortality and any of the following: sex of the child; delayed (more than 48 hours) initiation of therapy following symptoms of meningeal irritation or other general symptoms of infection; history of convulsions; nutritional status of the child; presence of antibacterial activity in plasma or CSF taken on admission. However, of the 19 who presented with abnormal consciousness, seven died, a significant association ( p = 0.01). Transient and permanent complications among survivors are shown in Table IV. Seizures were the most frequent transient complication. They were focal with secondary generalization and complicated by hemiplegia in one child. In another three they
Seizure Bronchopneumonia Subdural effusion Cranial nerve palsy (N III ) Permanent sequelae Hearing impairment Hemiplegia and hearing impairment Hemiplegia
Number (% ) 4 (11 ) 3 (8 .6) (2.9) (2.9) Number (% ) 5 (14) 2 (5.7) 2 (5.7)
consisted of recurrent episodes of Jacksonian fits that were present on admission. Subdural effusion was detected in one child and resolved spontaneously. Hearing loss was the most frequent permanent complication and affected 20 % of the survivors. It was profound in three, symmetrical in two and unilateral in one. Of the four children who were discharged with hemiplegia, two had hearing impairment. Late commencement of treatment ( > 48 hours) was not associated with the development of hearing loss. Furthermore, no association was observed between the development of permanent sequelae and sex, age, causative organism, duration of specific symptoms of meningitis before antibiotic treatment, history of convulsions, nutritional status, level of consciousness on admission or the presence of antimicrobial activity in pre-treatment plasma or CSF samples.
Discussion The three organisms commonly associated with childhood bacterial meningitis (HI, MC and PNC) accounted for 91% of bacterial infections. HI was the commonest pathogen, affecting mainly the under-Ss, as has been reported elsewhere in Africa, Scandinavia and the USA. 14 - 18 However, its peak incidence was in infants < 12 months, an observation consistent with other reports
208
M.A. M. Salih et al.
from East, West and Southern Africa and Scandinavia. 8 ' 15 ' 17 ' 19- 21 The proportion of meningitis cases caused by MC in our series ( 18 % ) is in agreement with published reports from Ethiopia 15 and South Africa, 21 but is lower for PNC meningitis (16 % ) than in the latter two series (39 % and 40 % , respectively). In Ibadan, Nigeria (outside the meningitis belt), PNC was the most frequently isolated organism (34 % ) compared to HI (28 % ) and MC (6 % ). 19 The relatively small yield of virus identification in the 52 who were considered to be cases of aseptic meningitis suggests the possibility that some of them might have been cases of partially treated bacterial meningitis. However, two methods of bacterial antigen detection (CoA and EIA-test) were used in this study to sharpen the diagnostic yield, the latter of which was found to be reliable in situations where pre-treatment with antibiotics was likely. 13 The percentages of children in our study who were without fever, coma or neck stiffness on admission were 12%, 56 % and 12%, respectively. In a series from Sweden, Salwen and associates reported an incidence of 3 %, 17 % and 20 %, respectively, for the absence of these signs. 17 Gieseler et al. reported 16 of 1064 children (1.5 %) with bacterial meningitis beyond the neonatal period who had no signs of meningeal irritation during the entire period of hospitalization. 18 A high degree of suspicion and an early diagnostic lumbar puncture remain the cornerstones of management of bacterial meningitis.22 The overall case fatality rate of 18.6 % in this report is significantly higher than the range of 2.1-7.8 % in larger series reported from Europe and the United States, 16- 18 and is comparable to the mortality rate for meningitis in Swedish children (19 % ) between 1956 and 1965. 17 Similar studies from Africa of bacteriologically proven cases give an overall mortality rate of 22- 27 % . 15 , 21 In these reports and others from Sudan, the case fatality rate was found to be low (4.4- 22 % ) in groups A, Band C MC meningitis compared
to other organisms, 20 ' 21 ' 24 ' 25 mortality due to HI meningitis was moderate and ranged between 19 % and 26 % , 15 ' 19' 2 1 and PNC had the highest case fatality rate of between 39 % and 55 % . 15 ' 21 ' 26 The presence of malnutrition did not affect survival adversely in this study, c;ontrary to reports from Ethiopia and South Africa, 15 ' 2 1 nor was the presence of antimicrobial activity in the plasma and/or CSF statistically significant. However, one must bear in mind the sample size in our study before drawing firm conclusions . Estimates of deafness as a result of meningitis depend on the type of study (prospective or retrospective), age of the patient, causative organism and audiological methods used for assessment. 27 Although an incidence as high as 20 % has been reported, 28 the overall incidence in Europe has been estimated to be 12-13 % .29 We are not aware of any study in a developing country where hearing was prospectively and systematically evaluated in childhood bacterial meningitis. An incidence of 20 % in survivors is high and might have been higher if audiological testing included auditory brainstem-evoked responses, which allows more objective assessment in the younger age-groups .28 ' 30 However, the incidence of deafness in this report is similar to that of a report from Nigeria where bacterial meningitis accounted for 21 % of 186 children with profound deafness in whom an associated cause could be ascertained. 3 1 Salwen and associates reported a significant association between deafness and the late introduction of therapy (more than 48 hours) in their study of childhood bacterial meningitis over a period of 25 years in Sweden. 17 However, the present report and others did not find such an association. 32 It is apparent that hearing loss can occur early in the course of illness, even before signs of meningeal irritation have developed.28 On the other hand, Lebel et al. showed that dexamethasone when given as adjuvant therapy for bacterial meningitis significantly reduced the incidence of moderate to profound bilateral sensorineural hearing impairment. 33 Such an effect is difficult to evaluate in this study
Bacterial meningitis in Sudanese children given its design and the small numbers involved. It is evident from our study that in endemic situations in Sudan H. infiuenzae is the commonest cause of bacterial meningitis with a high incidence of morbidity and mortality. The majority of susceptible children· are below the age (18 months) when pure H. infiuenzae capsular polysaccharide (PRP) vaccine was found to be protective. 34 Recent clinical trials with a PRP-diphtheria toxoid conjugate vaccine have been encouraging. 35 The vaccine was proved to be both safe and immunogenic when given in three doses subcutaneously along with DPT immunization, starting at 2 months of age. Sudanese children were found to have a satisfactory serological response to combined vaccines, including meningococcal A and C polysaccharide vaccine, given as early as 3 months of age. 36 An immunogenic HI vaccine, compatible with primary immunization schedules, would therefore be expected to have a significant impact on childhood mortality and morbidity in Sudan and other developing countries. 37 Acknowledgements The investigation was supported by grants from the Maud and Birger Gustavsson Fund. We wish to thank colleagues and technicians at the Central Microbiological Laboratory in Stockholm, Sweden for their virological diagnostic help and Mrs Theodora Fredriksson of the International Child Heath Unit, Department of Paediatrics, University Hospital, Uppsala, Sweden for all the help offered throughout the project. References Hassan MM, Hasab-El RA. Non tuberculous meningitis in Sudanese children. J Trop Med Hyg 1969; 72:19-21. 2 Orner MIA, Orner HO, El Khidir A, Hassan MM. Treatment of meningococcal meningitis in children in Khartoum. Indian Pediatr 1975; 12:387-91. 3 Sippel JE, Sid Ahmed H, Abu Ahmed H, Mikhail IA. Cerebrospinal meningitis in the Sudan in 1980. Sudan Med J 1982; 18:39-46.
209
4 Sippel JE, Sid Ahmed H, Mikhail IA. Epidemiologic studies on meningococcal meningitis in the Sudan. Med Trop 1983; 43:39-41 . 5 Bayoumi A. The history of Sudan health services. Nairobi: Kenya Literature Bureau 1979; 206-21. 6 Erwa HH, Haseeb MA, Idriss AA, Lapeyssonnie L, Sanborn WR, Sippel JE. A serogroup A meningococcal vaccine: studies in Sudan to combat cerebrospinal meningitis caused by Neisseria meningitidis group A. Bull WHO 1973; 49:301-5. 7 Smith EW, Haynes RE. Changing incidence of Haemophilus infiuenzae meningitis. Pediatrics 1972; 50:723-7. 8 Broom CV. Epidemiology of Haemophilus infiuenzae type b infections in the United States. Pediatr Infect Dis 1987; 6:779-82. 9 Peltola H, Kayhty H, Sivonen A, Makela PH. Haemophilus infiuenzae type b capsular polysaccharide vaccine in children: a double-blind field study of 100 000 vaccinees 3 months to 5 years of age in Finland. Pediatrics 1977; 60:730-7. 10 Anonymous. Classification of infantile malnutrition. Lancet 1970; ii: 302. 11 Nutritional Center for Health and Statistics Growth Charts Monthly Vital Statistics Report 25 (suppl) No. HRA 76-1120. Rockville MD,Health Resources Administration 1976. 12 Huttenlocher PR. The comatose child. In: Behrman RE, Vaughan VC, eds . Nelson Textbook of Pediatrics. Philadelphia: WB Saunders 1983; 1557-9. 13 Salih MAM, Ahmed HS, Hofvander Y, Daniels son D, Olcen P. Rapid diagnosis of bacterial meningitis by an enzyme immunoassay of cerebrospinal fluid. Epidemiol Infect. In press. 14 Brown KGE. Meningitis in Queen Elizabeth Central Hospital, Blantyre, Malawi. East Afr Med J 1975; 52:376-85. 15 Hailemeskel H, Tafari N. Bacterial meningitis in childhood in an African city. Acta Paediatr Scand 1978; 67:725-30. 16 Spanjaard L, Bol P, de Jong MCJW, Zanen HC. Bacterial meningitis in 366 children in The Netherlands, 1982-1983. Epidemiology and antibiotic therapy. Tijdschr Kindergeneeskd 1986; 54: 1-8. 17 Sal wen KM, Vikerfors T, Olcen P . Increased incidence of childhood bacterial meningitis. Scand J Infect Dis 1987; 19:1-11. 18 Gieseler PJ, Nelson KE, LevinS, Reddi KT, Moses VK. Community-acquired purulent meningitis: a review of 1316 cases during the antibiotic era, 1954-1976. Rev Infect Dis 1980; 2:725-45. 19 Nottidge VA. Haemophilus infiuenzae meningitis: a 5-year study in Ibadan, Nigeria. J Infect 1985; 11: 109-17. 20 Rosen EU, Davis MD. Nutritional status of children with bacterial meningitis. S Afr Med J 1980; 58: 1004-6.
210
M.A. M. Salih et al.
21 Mulla MI, Mossajee I, Rubidge CJ, Moosa A. Nutritional status of children with pyogenic meningitis. J Trop Pediatr 1984; 30:303-6. 22 Harper JR, Lorber J, Hillas-Smith G, Bower BD, Eykyn SJ. Timing of lumbar puncture in severe childhood meningitis. Br Med J 1985; 291:651-2. 23 Ryder CS, Beatty DW, Heese HdeV. Group B meningococcal infection in children during an epidemic in Cape Town, South Africa. Ann Trop Paediatr 1987; 1:47-53. 24 Nottidge VA. Meningococcal meningitis in childhood. A five-year study on meningococcal meningitis in Ibadan, Southern Nigeria. J Infect 1983; 7: 39-45. 25 Evans-Jones LG, Whittle HC, Onyewotu II, Egler LJ, Greenwood BM. Comparative study of group A and group C meningococcal infection. Arch Dis Child 1977; 52:320-3. 26 Berkowitz FE. Pneumococcal bacteraemia-a study of 75 black children. Ann Trop Paediatr 1981; 4: 229-35. 27 Brookhouser PE, Auslander MC, Meskan ME. The pattern and stability of post meningitic hearing loss in children. Laryngoscope 1988; 98:940-8. 28 Baldwin RL, Sweiber RB, Freind DM. Meningitis and sensorineural hearing loss. Laryngoscope 1985; 95:802-5. 29 Bellman SC. Hearing disorders in children. Br Med Bull1987; 43:966-82.
30 Vienny H, Despland PA, Lutschg J, Deonna T, Dutoit-Marco ML, Granev C. Early diagnosis and evolution of deafness in childhood bacterial meningitis. A study using brainstem auditory evoked potentials. Pediatrics 1984; 73:579-86. 31 Obiako MN. Profound childhood deafness in Nigeria: a three year survey. Ear Hear 1987; 8:74-7. 32 Dodge PR, Davis H, Feigin RD, et a!. Prospective evaluation of hearing impairment as sequelae of acute bacterial meningitis. N Eng! J Med 1984; 311: 869-74. 33 Lebel MH, Freij BJ, Syrogiannopoulos GA, eta!. Dexamethasone therapy for bacterial meningitis. N Eng! J Med 1988; 319:964-71. 34 Parke JC. Capsular polysaccharide of Haemophilus infiuenzae type b as a vaccine. Pediatr Infect Dis 1987; 6:795-8. 35 Lepow M. Clinical trials of the Haemophilus infiuenzae type b capsular polysacharide-diphtheria toxoid conjugate vaccine. Pediatr Infect Dis 1987; 6: 804-7. 36 Orner MIA, El Dawla G, Nicolas A, Roumiantzeff R, Lapeyssonie L. Humoral antibody responses to biannual multiantigen vaccination: report of a field trial on children in Sudan. Ann Trap Paediatr 1986; 4:243-8. 37 Robbins JB, Schneerson R. Haemophilus infiuenzae type b: the search for a vaccine. Pediatr Infec Dis 1987; 6:791-4.
