Antibiotic therapy for bacterial meningitis in children in developing countries.

Antibiotic therapy for bacterial meningitis in children in developing countries P. Kumar1 & I.C. Verma2 We carried out a study to investigate the effectiveness of chloramphenicol alone as a treatment for bacterial meningitis. A total of 70 consecutive children aged >3 months with bacterial meningitis, who had been admitted to the paediatric hospital of the All India Institute of Medical Sciences, were randomized to receive chloramphenicol alone or chloramphenicol + penicillin. The two groups were matched with each other. Treatment failure occurred with three (9°%) patients in the chloramphenicol-alone group and with four (12.1%) patients in the combination therapy group (P >0.05). The mean duration of intravenous therapy, the number of intravenous cannulae used per patient, and the incidence of thrombophlebitis were significantly higher for the group that received the combination therapy. Also, the cost of using chloramphenicol + penicillin was four times higher than that of chloramphenicol alone. Hence, chloramphenicol alone was as effective as chloramphenicol + penicillin and much cheaper and more convenient to use. Introduction Bacterial meningitis is an important cause of mor- bidity and mortality in children in developing countries. In a recent multicentre survey in India, bacterial meningitis constituted 1.5% of admis- sions to paediatric wards, and the mean case fatality rate was 16% (1). The frequency of sequelae in developing countries is probably greater than the 8-40% reported for developed countries (2). The prognosis can, however, be improved by early administration of appropriate therapy; to be effective, this has to be provided in rural hospitals. Although the third-generation cephalosporins are currently the preferred therapy for bacterial meningitis in developed countries, their use is pro- hibitively expensive for developing countries. There, the most commonly used drugs for this purpose are penicillin and chloramphenicol, administered together, to cover the three commonest organisms that cause meningitis in children aged >3 months of age: Neisseria meningitidis, Haemophilus influenzae, and Streptococcus pneumoniae. Penicillin and chloramphenicol are cheap and readily available; however, the pharmacological antagonism between penicillin and chloramphenicol has been known since the in-vitro interaction was first described over 40 years ago by Jawetz et al. (3). This antagonism is of 1 Senior Resident, Department of Pediatrics, All India Institute of Medical Sciences, New Delhi, India. At present: Senior Resident, Department of Pediatrics, Post-graduate Institute of Medical Education and Research, Chandigarh, India. 2 Professor, Department of Pediatrics, All India Institute of Medi- cal Sciences, Ansari Nagar, New Delhi-1 10 029, India. Requests for reprints should be sent to Professor Verma. Reprint No. 5372 special importance for infections, such as bacterial meningitis, that necessitate early, rapid killing of bacteria for recovery. Administration of penicillin alone would be inadequate because H. influenzae is only moderately sensitive to it and S. pneumoniae often has reduced sensitivity (4). On the other hand, chloramphenicol is bactericidal for the three common- est meningeal pathogens (5, 6). The advantage of using chloramphenicol alone is that an intravenous line is required for a shorter period and there is no threat of thrombophlebitis, which occurs when penicillin is used. Use of chlor- amphenicol alone carries less risk of sepsis through the intravenous route, results in less discomfort and enhanced mobility for the child, and is easier for physicians to administer. This treatment can be conveniently carried out in primary health care centres, where securing and maintaining venous access in children may itself be a problem. The present study was carried out to investigate whether chloramphenicol alone is as effective as the combination penicillin + chloramphenicol for the treatment of bacterial meningitis in children. Materials and methods A prospective, randomized controlled trial was per- formed on children over 3 months of age who had been admitted to the paediatric hospital of the All India Institute of Medical Sciences with a diagnosis of bacterial meningitis. The study was carried out over a period of 19 months (January 1988 to August 1989). Lumbar punctures were performed on patients with clinically suspected bacterial meningitis. The following investigations were carried out on the samples of cerebrospinal fluid (CSF): total and differ- Bulletin of the World Health Organization, 71 (2): 183-188 (1993) © World Health Organization 1993 183 P. Kumar & I.C. Verma ential cell counts; Gram staining; latex agglutination tests; and determination of protein and glucose levels. Samples were also sent to the institute's department of microbiology for culture and tested for sensitivity. The blood glucose levels were estimated immediately prior to performing the lumbar punc- ture. Total and differential leukocyte counts were performed for all the patients on admission and re- peated after a week of therapy. Children were included in the study if they had a CSF smear that was positive for bacteria on Gram staining or their CSF was turbid with >500 poly- morphs per gl. Informed consent was obtained from the parents of each child included in the study and the protocol was approved by the hospital ethics committee. Selection of antibiotics A computer-generated list of random numbers was used to prepare a set of sealed numbered envelopes. After each child had been admitted to the trial, the appropriate envelope was opened to determine the treatment to be given. Children randomized to receive chloramphenicol alone were administered the drug intravenously at a dose of 100 mg.kg-1.day-1 in four divided doses. Those who received chloramphenicol + penicillin were given 100 mg.kg-1.day-1 of chloramphenicol and 300 000-400 000 IU.kg-l.day-I of crystalline penicillin in 6-hourly doses intravenously after their sensitivity to penicillin had been tested. Chloram- phenicol was given orally after 3-5-days' treatment if there was an improvement in sensorium and no vomiting. The antibiotics were prescribed for a total of 10-14 days. Progress and recovery The condition of each child at admission was record- ed on a pre-designed proforma. Daily changes in temperature, blood pressure, vomiting, seizures, focal deficits, etc. were recorded. Sensorium was graded using the Glasgow coma scale for children aged >5 years and using the Adelaide paediatric coma scale for under-5-year-olds (7, 8). Repeat samples of CSF were taken 36-72 hours after admis- sion to the study. The treatment was changed after 36-72 hours if a child's clinical condition worsened or did not improve and there was a deterioration or non- improvement in the CSF findings. Treatment failure was defined as the death of a patient or a change in the treatment for the above reasons. The data were analysed using Student's t-test, x2-test (with Yates's correction where applicable), and Fisher's exact test. The total leukocyte and CSF cell counts, as well as CSF protein and glucose levels were analysed after logarithmic transforma- tion. Results A total of 70 consecutive children aged >3 months with bacterial meningitis were enrolled in the study. Three cases were excluded for the following reasons: two (one in each therapy group) were later found to have a tubercular etiology, one (in the chlorampheni- col alone group) was inadvertently also given peni- cillin after 24 hours. The data for one patient in the combination therapy group were excluded from the statistical analysis because the child concemed died of irreversible shock within 4 hours of admission. Hence, there were 66 cases (33 in each group) in the study. There was no significant difference between the two treatment groups with respect to distribution of age, sex, malnutrition, symptoms and signs, duration of symptoms, shock, focal deficits, degree of uncon- sciousness, total leukocyte counts, or history of pre-admission antibiotic therapy (Table 1). Table 1: Initial characteristics of the children in the two treatment groups Treatment group Chloramphenicol (n= 33) Mean age ± SD (years) No. aged <1 year Male: female No. malnourished (all grades)a Symptoms and signs Fever Headache Vomiting Altered sensorium Seizures Rash Ear discharge Mean duration of symptoms + SD (days) No. who took pretreatment antibiotics Mean Glasgow coma score ± SD (n= 24) No. with shock No. with focal deficits Mean TLC ± SDC 6.07 + 3.52 7 2.3 22 (66.6) b 33 (100) 19 (57.5) 27 (81.8) 17 (51.5) 10 (30.3) 9 (27.2) 3 (9) 2.0 + 1.03 8 (24.2) 11.2 ± 2.6 4 2 15 928 ± 1.54 Penicillin + chloramphenicol (n= 33) 5.57 ± 4.17 8 2.0 21 (63.6) 32 (96.9) 15 (45.4) 22 (66.6) 15 (45.4) 11 (33.3) 8 (24.2) 4 (12.1) 2.1 ± 1.07 9 (27.2) 11.9 ± 2.3 3 2 15 440 ± 1.7 a As classified by the Indian Academy of Pediatrics, ref. 9. b Figures in parentheses are percentages. c TLC = Total leukocyte count. WHO Bulletin OMS. Vol 71 1993184 Antibiotic therapy for bacterial meningitis Table 2: Mean cerebrospinal fluid (CSF) characteristics of the study children at diagnosis and after 36-72-hours' treatment At diagnosis After 36-72 hours CHL groupa CP + CHL groupb CHL group CP + CHL group No. of cells per gl 1419 ± 3.36 1257 ± 3.26 285 ± 2.95 360.8 ± 2.70 % of polymorphs 88.7 ± 6.99 88.6 ± 9.22 48.3 + 15.3 47.2 ± 14.2 Protein level (mg/dl) 148.2 ± 2.05 151.3 ± 1.95 67.7 ± 1.48 75.3 ± 1.37 CSF glucose/ 28.6 ± 2.20 31.0 ± 1.82 58.6 ± 6.93 58.0 ± 7.14 blood glucose (%) a CHL = Chloramphenicol. b CP = Crystalline penicillin. The mean number of cells in the CSF, propor- tion of polymorphs, and CSF protein and glucose levels were also comparable for the two groups (Table 2). The cell count ranged from 525 per ,ul to 16 000 per ,ul, while the protein level varied from 53 mg/dl to 1000 mg/dl. The CSF glucose level as a proportion of the blood glucose ranged from 0 to 69%. The etiology was established for 33 patients (50%); the various causative organisms are shown in Table 3. All the meningococci were uniformly sensi- tive to both chloramphenicol and penicillin. Cultures of S. pneumoniae from one case were partly sensitive to chloramphenicol, while all other strains were sen- sitive to both antibiotics. One strain of H. influenzae was resistant to penicillin, while one was partly sen- sitive to chloramphenicol. There were three deaths (4.5%) - all in the chloramphenicol + penicillin group. The death that occurred within 4 hours of admission was from Waterhouse-Friderichsen syndrome and hence was unrelated to the antibiotic therapy. In one fatal case the causative agent was Klebsiella pneumoniae, and the strain involved was resistant to both chloram- phenicol and penicillin. The treatment was changed after 36-72 hours for three patients in the chloramphenicol-alone group Table 3: Distribution of in the study children causative organisms identified No. of children: CHL groupa CP+CHL groupb Causative organism (n = 33) (n = 33) Neisseria meningitidis 8 8 Streptococcus pneumoniae 5 7 Haemophilus influenzae 2 1 Other 1 c 1 d a CHL = Chloramphenicol. b CP = Crystalline penicillin. ca-Haemolytic streptococci. d Klebsiella pneumoniae. and for two patients in the chloramphenicol + penicillin group. Of the three therapy changes in the chloramphenicol-alone group, one involved a child who was infected with a-haemolytic streptococci that were partly sensitive to chloramphenicol and resistant to penicillin. Although this child's clinical condition and CSF both improved, ampicillin was also administered, based on culture/sensitivity find- ings. Hence, treatment failure (deaths + change of treatment) was recorded for three patients (9%) in the chloramphenicol-alone group and for four patients (12%) in the chloramphenicol + penicillin group (P >0.05). The repeat samples of CSF collect- ed from all the patients 36-72 hours after starting the treatment did not reveal any difference between the two groups in terms of the variables that were examined (Table 2). The mean duration of fever in the chlorampheni- col-alone group was 2.06 ± 0.63 days, compared with 2.38 ± 1.02 days in the chloramphenicol + penicillin group. The duration of neck rigidity was 3.75 ± 0.55 days in the chloramphenicol-alone group and 4.2 ± 0.75 days in the combination group. Neu- rological deficits at discharge occurred with three patients in the single drug group and with two patients in the combination group. Intravenous therapy was continued for 4.27 + 1.01 days in the chloramphenicol-alone group, while it was required for 10.3 ± 1.99 days in the chlor- amphenicol + penicillin group (P <0.01). The average number of intravenous cannulae used per patient was 1.89 ± 0.71 for the chloramphenicol group and 4.73 ± 1.99 for the combination group (P <0.01). Significant thrombophlebitis occurred in 17 patients (58.6%) in the combination group but only in one patient (3.3%) in the chloramphenicol- alone group (P <0.001). Drug fever occurred in three patients in the combination group and in one patient in the chloramphenicol-alone group. After a week of therapy, none of the patients had a total leukocyte count of less than 4000. WHO Bulletin OMS. Vol 71 1993 185 P. Kumar & I.C. Verma Discussion Jawetz reported the antagonism between chloram- phenicol and penicillin as early as 1951 (3). In the same year, Lepper & Dowling advocated further studies on the efficacy of chloramphenicol "with- out adjuvant penicillin" in the treatment of bac- terial meningitis (10). The successful use of chloram- phenicol alone to treat bacterial meningitis was re- ported first in 1953, with a mortality rate comparable to that for other drugs (11). Subsequently, no more clinical trials of this nature were reported for more than 30 years. How- ever, encouraged by the reports of Rahal & Simber- koff, in 1979, and of Weber, in 1983, demonstrating that chloramphenicol was a bactericidal drug for the three commonest meningeal pathogens (5, 6), Shann et al., in 1985, undertook a multicentre trial compar- ing chloramphenicol alone and chloramphenicol + penicillin for the treatment of bacterial meningitis and pneumonia (12, 13). Their results indicated that therapy with chloramphenicol alone was satisfactory. The major advantage of using chloramphenicol is that its absorption is optimal when administered orally (14); as soon as a child with meningitis stops vomiting, oral therapy can begin. Absorption of chloramphenicol following intramuscular injection is comparable to that resulting from intravenous admin- istration (15). Hence, the drug can be given to chil- dren intramuscularly, a route that is easily tolerated and causes only minimal pain; this is especially important when chloramphenicol is being used in rural areas. A long-acting oil suspension of chloram- phenicol given intramuscularly in one or two doses is also effective and convenient, as well as being the cheapest antibiotic for use in presumptive treatment of meningitis, especially during epidemics (16, 17). Therefore, administration of chloramphenicol alone to children with bacterial meningitis does not require intravenous use for long periods and the risk of thrombophlebitis is minimized. This approach makes less demands on the time of hospital staff and carries a smaller risk of nosocomial infection and of over- hydration. Use of chloramphenicol alone is.much cheaper than that of the combination chloramphenicol + penicillin. In our study, the average cost of treatment for 2 weeks (excluding hospital costs) for the combi- nation group was four times that for the chloram- phenicol-alone group. Although newer and more potent antibiotics have been developed to treat paediatric bacterial meningitis, their cost is usually prohibitive in developing countries; nevertheless, despite their use, the prognosis for bacterial meningi- tis has not changed appreciably over the last 20 years (18). Furthermore, chloramphenicol is effective also against those pneumococci that are relatively resis- tant to penicillin (4). Third-generation cephalo- sporins are not reliably effective against pneumo- cocci, which can constitute up to 21% of pathogens in paediatric patients (19). The infrequent association of chloramphenicol with aplastic anaemia represents the principal disad- vantage of the drug. This condition occurs in less than one case for every 50 000-100 000 courses of therapy (20), a risk that is not more than that of fatal anaphylaxis from penicillin. In any case, chloram- phenicol is used in combination therapy for bacterial meningitis and its use as a single agent at the same dosage would not increase the risk of this complica- tion. The best approach to minimizing the occurrence of aplastic anaemia associated with chloramphenicol would be to restrict its administration only for speci- fic indications, such as meningitis or typhoid fever, rather than to indiscriminately prescribe it as a broad-spectrum antibiotic. Malnourished children who are often deficient in pancreatic lipase may have impaired absorption of chloramphenicol because they are unable to hydrolyse it or can paradoxically have high levels of the drug because of impaired clearance (6, 21). However, the outcomes of treating malnourished children with chloramphenicol alone are not different from those for children who are not malnourished. Simultaneous administration of phenytoin and phenobarbital may reduce serum levels of chloram- phenicol owing to induction of hepatic microsomal enzymes (22); however, the clinical significance of this has not been established. Prober has reported a reduction in serum levels of chloramphenicol follow- ing administration of rifampicin to eradicate H. influenzae from the nasopharynx (23). The clinical implication of this finding would be to give rifampicin only after a course of chloramphenicol has been completed. In our study, the etiology was known for only 50% of cases. The rate of bacterial isolation was poor because direct plating of CSF could not be carried out at the bedside. Also, transport of the CSF samples to the laboratory was not always very prompt at night, when the majority of cases were admitted. However, our study addressed the question that is of most importance to clinicians: For a child with CSF findings that suggest bacterial meningitis, should the antibiotic therapy be chloramphenicol alone or chloramphenicol + penicillin? Among the children in the study whose etiology was known, the outcome in the two treatment groups was not differ- ent. We did not determine the levels of chloram- phenicol in serum and CSF. The monitoring of these levels is useful primarily to gauge whether adequate WHO Bulletin OMS. Vol 71 1993186 Antibiotic therapy for bacterial meningitis levels of the drug are present in the serum and CSF of cases of apparent clinical failure. In our study, the failure rate was similar for both treatment groups; hence, it is unlikely that the results would have been different had the drug levels in serum and CSF been determined. Chloramphenicol alone was therefore as effec- tive as chloramphenicol + penicillin for the treat- ment of bacterial meningitis in children aged .3 months. The combination therapy with penicillin offered no benefit; instead, it increased the cost of treatment, caused more thrombophlebitis, and required a longer period of intravenous therapy. Resume M6ningite bact6rienne chez l'enfant: antibioth6rapie dans les pays en d6veloppement Dans les pays en developpement, la m6ningite bacterienne constitue un probl6me de sante publique en raison des taux de letalite 6leves et de la forte incidence des sequelles qu'elle entral- ne. Chez les enfants de plus de 3 mois, les germes les plus communement rencontr6s sont Neisseria meningitidis, Haemophilus influenzae et Streptococcus pneumoniae. Le traitement initial empirique g6neralement utilise dans ces cas est une association de p6nicilline et de chloramph6nicol. Cependant, ces deux m6dicaments sont pharmaco- logiquement antagonistes. Nous avons effectue un essai contr6le pros- pectif randomise a New Delhi pour d6terminer si le chloramphenicol seul est aussi efficace que I'association chloramph6nicol + penicilline pour traiter la meningite bacterienne chez les enfants de plus de 3 mois. Les 66 participants ont 6t6 affectes au hasard a deux groupes de traitement de 33 enfants chacun. L'age moyen des enfants 6tait de 5,82 ± 3,75 ans et le rapport sex-ratio de 2,14:1. L'agent etiologique le plus souvent retrou- v6 a ete le m6ningocoque (16 enfants), suivi du pneumocoque (12), et d'H. influenzae (3). Les deux groupes ont ete bien apparies. Le taux de l6talit6 a ete de 4,5%. On a observe des 6checs th6rapeutiques (d6ces + modification du traitement) chez trois malades (9%) du groupe traite par le chloramphenicol seul et chez quatre malades (12,1%) du groupe chloramphenicol + p6nicilline (P >0,05). La dur6e de la fievre et de la raideur de la nuque, I'incidence des d6ficits en foyers et la disparition des troubles de la cons- cience ont ete les memes dans les deux groupes. La duree du traitement intraveineux a 6te de 4,27 + 1,01 jours dans le groupe des sujets traites par le chloramphenicol seul, alors qu'il a 6te de 10,3 ± 1,99 jours dans le groupe chloramphenicol + peni- cilline (P <0,01). On a observe une thrombophle- bite marquee chez 58,6% des sujets du groupe recevant l'association th6rapeutique, mais seuts 3,3% des enfants ne recevant que du chloram- phenicol en ont pr6sente une (P <0,001). L'absorption du chloramphenicol est maximale lorsqu'il est administre par voie orale, mais son absorption apres injection intramusculaire est comparable a celle apres injection intraveineuse. L'utilisation du chloramphenicol seul presente donc les avantages suivants: la voie intraveineuse n'est pas necessaire longtemps, il n'y a donc pas de probleme de thrombophlebite, il est moins onereux, demande moins de temps au personnel hospitalier, et provoque moins d'inconfort chez l'enfant. Ce schema th6rapeutique peut etre appli- qu6 sans inconvenient dans les centres de soins de sante primaires, ou la pose et le maintien de l'abord veineux chez l'enfant peut en lui-meme poser des problemes. References 1. Kabra, S.K. et al. Bacterial meningitis in India - an IJP survey. Indian journal of pediatrics, 58: 505-511 (1 991). 2. Sell, S.H. Long-term sequelae of bacterial meningi- tis in children. Pediatric infectious disease journal, 2: 90-93 (1983). 3. Jawetz, E. Studies on antibiotic synergism and antagonism. Archives of internal medicine, 87: 349-359 (1951). 4. Klein, J.O. et al. Report of the Task Force on Di- agnosis and Management of Meningitis. Pediatrics, 78(suppl.): 959-982 (1986). 5. Rahal, J.J. & Simberkoff, S.S. Bactericidal and bacteriostatic action of chloramphenicol against meningeal pathogens. Antimicrobial agents and chemotherapy, 16: 13-18 (1979). 6. Smith, A.L. & Weber, A. Pharmacology of chloram- phenicol. Pediatric clinics of North America, 30: 209-236 (1983). 7. Teasdale, G. & Jennett, B. Assessment of coma and impaired consciousness: a practical coma scale. Lancet, 2: 81-84 (1974). 8. Simpson, D. & Reilly, P. Paediatric coma scale. Lancet, 2: 450 (1982). 9. Nutrition Subcommittee of the Indian Academy of Pediatrics. Indian pediatrics, 9: 360-364 (1972). 10. Lepper, M.H. & Dowling, H.F. Treatment of pneu- mococcal meningitis with penicillin compared with penicillin plus aureomycin. Archives of internal medi- cine, 88: 489-494 (1951). 11. Deane, G.E. et al. Treatment of meningitis with WHO Bulletin OMS. Vol 71 1993 187 P. Kumar & I.C. Verma chloromycetin palmitate. Pediatrics, 11: 368-380 (1953). 12. Shann, F. et al. Chloramphenicol alone versus chloramphenicol plus penicillin for bacterial meningi- tis in children. Lancet, 2: 681-683 (1985). 13. Shann, F. et al. Chloramphenicol alone versus chloramphenicol plus penicillin for severe pneumo- nia in children. Lancet, 2: 684-685 (1985). 14. Yogev, R. et al. Pharmacokinetic comparison of IV and oral chloramphenicol in patients with Hemophil- us influenzae meningitis. Pediatrics, 67: 656-659 (1981). 15. Shann, F. et al. Absorption of chloramphenicol sodium succinate after intramuscular administration in children. New England journal of medicine, 313: 410-414 (1985). 16. Rey, M. et al. Traitement minute de la meningite cerebrospinale epidemique par injection intra- musculaire unique de chloramphenicol (suspension huileuse). Med. mal. infec., 6:120-124 (1976). 17. Puddicombe, J.B. et al. A field trial of a single intramuscular injection of long acting chlorampheni- col in the treatment of meningococcal meningitis. Transactions of the Royal Society of Tropical Medi- cine and Hygiene, 78: 399-403 (1984). 18. McCracken, G.H. Jr. Management of bacterial meningitis - current status and future prospects. American journal of medicine, 176: 215-221 (1984). 19. Jackson, M.A. et al. Relative penicillin-resistant pneumococcal infections in pediatric patients. Pedia- tric infectious disease journal, 3: 129-132 (1984). 20. Leikin, S.L. et al. Aplastic anemia due to chloram- phenicol. Clinical proceedings of the Children's Hospital (Washington), 17: 171-181 (1961). 21. Mehta, S. Chloramphenicol metabolism in children with protein-calorie malnutrition. American journal of clinical nutrition, 28: 997-1081 (1975). 22. Powell, D. Interaction of chloramphenicol, phenytoin and phenobarbital in a pediatric patient. Journal of pediatrics, 98: 1001-1003 (1981). 23. Prober, C.G. Effect of rifampicin on chloram- phenicol levels. New England journal of medicine, 312: 788-789 (1985). 188 WHO Bulletin OMS. Vol 71 1993