Articles

5 versus 10 days of treatment with ceftriaxone for bacterial

meningitis in children: a double-blind randomised
equivalence study
Elizabeth Molyneux, Shaikh Qamaruddin Nizami, Samir Saha, Khanh Truong Huu, Matloob Azam, Zulqar Ahmad Bhutta, Ramadan Zaki,
Martin Willi Weber, Shamim Ahmad Qazi, for the CSF 5 Study Group*

Summary
Background Bacterial meningitis is an important cause of morbidity and mortality in developing countries, but the Lancet 2011; 377: 183745
duration of treatment is not well established. We aimed to compare the ecacy of 5 and 10 days of parenteral Published Online
ceftriaxone for the treatment of bacterial meningitis in children. May 26, 2011
DOI:10.1016/S0140-
6736(11)60580-1
Methods We did a multicountry, double-blind, placebo-controlled, randomised equivalence study of 5 versus
See Comment page 1809
10 days of treatment with ceftriaxone in children aged 2 months to 12 years with purulent meningitis caused by
*See webappendix for study
Streptococcus pneumoniae, Haemophilus inuenzae type B, or Neisseria meningitidis. Our study was done in ten contributors
paediatric referral hospitals in Bangladesh, Egypt, Malawi, Pakistan, and Vietnam. We randomly assigned
University of Malawi Medical
children who were stable after 5 days of treatment, through site-balanced computer-generated allocation lists, to School Department of
receive a further 5 days of ceftriaxone or placebo. Patients, their guardians, and sta were masked to study-group Paediatrics, Queen
allocation. Our primary outcomes were bacteriological failure or relapse. Our analysis was per protocol. This Elizabeth Central Hospital,
Blantyre, Malawi
study is registered with the International Standard Randomised Controlled Trial Number Register, (Prof E Molyneux FRCPCH);
number ISRCTN38717320. Division of Maternal and Child
Health, Aga Khan University,
Findings We included 1004 of 1027 children randomly assigned to study groups in our analyses; 496 received treatment Karachi, Pakistan
(Prof S Q Nizami FCPS,
with ceftriaxone for 5 days, and 508 for 10 days. In the 5-day treatment group, two children (one infected with HIV) Prof Z A Bhutta FRCPCH);
had a relapse; there were no relapses in the 10-day treatment group and there were no bacteriological failures in either Department of Microbiogy,
study group. Side-eects of antibiotic treatment were minor and similar in both groups. Dhaka Shishu Hospital, Dhaka,
Bangladesh (Prof S Saha PhD);
Children Hospital No 1,
Interpretation In children beyond the neonatal age-group with purulent meningitis caused by S pneumoniae, H inuenzae Ho Chi Minh City, Vietnam
type b, or N meningitidis who are stable by day 5 of ceftriaxone treatment, the antibiotic can be safely discontinued. (Prof K T Huu MD); Department
of Paediatrics, Wah Medical
Funding United States Agency for International Development. College, Islamabad, Pakistan
(Prof M Azam FRCPCH);
Department of Paediatrics,
Introduction limiting the usefulness of this antibiotic for empirical Abbasia Fever Hospital, Cairo,
Bacterial meningitis is an important cause of morbidity treatment of meningitis, and in Africa resistance is Egypt (Prof R Zaki MD); and
and mortality in children in developing countries.1,2 increasing against both antibiotics.1215 In most regions, Department of Child and
Adolescent Health and
Prospective hospital-based and population-based studies fewer than 1% of bacterial strains are highly resistant to Development, World Health
in Latin America, Africa and, most recently, Asia have both penicillin and third-generation cephalosporins.16,17 Organization, Geneva,
shown that Haemophilus inuenzae and Streptococcus Third-generation cephalosporins given to children Switzerland
(M W Weber DrMedHabil,
pneumoniae are the leading causes of bacterial meningitis with bacterial meningitis result in rapid sterilisation of
S A Qazi MD)
in young children.27 Worldwide, H inuenzae type b the cerebrospinal uid in almost all patients.18,19
Correspondence to:
causes about 173 000 cases of meningitis every year Ceftriaxone is a highly eective antibiotic for treatment Dr Shamim Ahmad Qazi,
in children younger than 5 years (uncertainty of meningitis, and it is the recommended treatment in Department of Child and
range 85 300226 000) and an additional 103 000 cases many countries, but the optimum duration of treatment Adolescent Health and
are caused by S pneumoniae (51 000131 000).8,9 is uncertain and recommendations have varied.20 Development, World Health
Organization, 20 Avenue Appia,
The mortality from meningitis is close to 100% in The duration of treatment of bacterial infection in Geneva 27, 1211, Switzerland
untreated individuals and is still up to 40% in children general, and of meningitis specically, is not based on [email protected]
who receive appropriate antibiotic treatment in solid evidence.11,21 For meningitis, the duration of
developing countries.10,11 Most of these fatalities are within antibiotic treatment is long and often continued in
72 h of admission to hospitals, and a high proportion of children with neurological sequelae, even though viable
survivors have neurological sequelae.26 bacteria are no longer present, because the treating
Early identication and prompt antibiotic treatment are physician fears stopping prematurely. This prolonged
essential for reducing mortality and morbidity. Standard treatment leads to unnecessary costs and potential side-
treatment for meningitis has been chloramphenicol, eects related to admission to hospital, so it is not in
often in combination with penicillin. However, pneumo- the interest of the patient, the family, or the health
coccal resistance to penicillin is increasing worldwide, system. The recommended duration of treatment is

www.thelancet.com Vol 377 May 28, 2011 1837

 Articles

710 days for meningitis due to H inuenzae type b, Methods

1014 days for S pneumoniae meningitis, and 7 days for Participants
Neisseria meningitidis meningitis.22,23 Between September, 2001, and December, 2006, we did a
Antibiotic treatment courses as short as 35 days are multicountry, double-blind, placebo-controlled, random-
eective in children and adults with meningitis due to ised equivalence study of 5 versus 10 days of treatment
N meningitidis.2427 Some studies with small sample sizes with ceftriaxone in children aged 2 months to 12 years
have shown that shorter courses of antimicrobial with purulent meningitis caused by S pneumoniae,
treatment also cure bacterial meningitis due to H inuenzae type b, or N meningitidis. Our study was done
H inuenzae and S pneumoniae without increased rates in ten paediatric referral hospitals in Bangladesh, Egypt,
of relapse or neurological sequelae.2836 An eective Malawi, Pakistan, and Vietnam. These hospitals serve
shorter antibiotic treatment regimen for serious large populations and each admits more than 150 children
childhood infections such as bacterial meningitis would with bacterial meningitis every year. Overall, H inuenzae,
make the use of more expensive antibiotics feasible and S pneumoniae, and N meningitidis are the most common
cost eective. We aimed to compare 5 with 10 days of causes. Admitting clinicians at each hospital did lumbar
treatment with parenteral ceftriaxone for bacterial punctures on children in whom bacterial meningitis was
meningitis in children. suspected clinically or in whom it needed to be excluded.
On admission all these children had a complete history
taken and were fully examined and weighed. The lumbar
Panel 1: Study criteria for enrolment and diagnosis puncture was done and, if meningitis was suspected,
Exclusion criteria at enrolment intravenous access was established; blood samples were
Children aged 2 months or younger or bodyweight 3 kg or less taken for full blood count, malaria parasites (thick lm),
Pre-existing neurosurgical conditions plasma glucose, electrolytes (if possible), and blood
Seizure disorders, cerebral palsy, degenerative neurological disorders culture; and treatment was started with ceftriaxone. All
Cranial fractures children underwent a second lumbar puncture 48 h after
Known immunodeciency states, symptomatic AIDS starting treatment (after two doses of intravenous
Active viral infections ceftriaxone). If this second sample of cerebrospinal uid
Known hypersensitivity to cephalosporins still contained bacteria, the child was excluded from
Cyanotic congenital heart disease random assignment. Supportive management of fever,
Previous random assignment in this study (child could not be randomly assigned on convulsions, uid management, and feeding were in
more than one occasion) accordance with local treatment guidelines. Children alive
Inaccessibility for follow-up on day 5 and those who were clinically stable or improving
Previous use of any parenteral antibiotics for 24 h before admission were considered for inclusion in our study. Panel 1 lists
the exclusion criteria for the study and diagnostic criteria
Exclusion criteria for random assignment at day 5 for bacterial meningitis.
Children were not randomly assigned to treatment groups if they met any criteria listed We obtained written informed consent from parents or
above or those listed below legal guardians. Separate consent was obtained for HIV
Meningitis due to any bacteria other than Haemophilus inuenzae, Streptococcus testing with full counselling before and after. Our study
pneumoniae, and Neisseria meningitidis was approved by the local institutional ethical review
Persistence of or growth of an organism from cerebrospinal uid taken 4872 h boards of the participating institutions in countries and
after admission WHO. Our study was monitored by an independent data
Isolated organism resistant to ceftriaxone and safety monitoring board (DSMB). All deaths and
Serious adverse reaction to the drug given serious adverse events were reported to the DSMB. The
Presence of pyogenic brain abscess, subdural empyema, or intracranial suppurative DSMB met yearly to review the safety aspects of the study.
thrombophlebitis
Child developing another infection during admission to hospital that needed another Procedures
injectable antibiotic Randomisation was done by site in permuted blocks of
Diagnostic criteria for bacterial meningitis (any one of the four) variable length of two, four, and six by a one-to-one ratio
Positive cerebrospinal-uid culture or latex agglutination for H inuenzae, to the 5-day or 10-day treatment groups. The list was
S pneumoniae, and N meningitidis prepared and kept at WHO headquarters, Geneva,
More than 10 white blood cells per mL of cerebrospinal uid, with positive blood Switzerland. At each site, the treatment allocations were
culture for the same organisms by day 3 of admission in sealed opaque envelopes, each with a unique study
Cerebrospinal-uid culture and latex agglutination both negative, but more than number held by a designated person. This person was
100 white blood cells per mL with more than 50% granulocytes and cerebrospinal-uid not involved in the clinical care of the patients and
glucose less than 50% of blood glucose or less than 166 mmol/L prepared the study drugs daily for each patient. The
More than 1000 white blood cells per mL of cerebrospinal uid with syringe containing the drug was masked by an opaque
75% polymorphonuclear cells sticker (because ceftriaxone has a faint yellow tinge and
the placebo was colourless). The clinicians giving the

1838 www.thelancet.com Vol 377 May 28, 2011

 Articles

treatment, nurses, and patients were unaware of the recorded and patients were managed in accordance with
contents of each syringe provided for each patient. clinical need. Patients who did not keep their
Parenteral ceftriaxone (80100 mg/kg bodyweight) was appointments were traced at home and encouraged to
given once daily for 5 days before random assignment to return for follow-up. If children were not found they
the study groups. On day 5 after the start of treatment, were declared lost to follow-up.
enrolment criteria were reviewed, and eligible patients Hearing was assessed by trained sta with otoacoustic
were randomly assigned to receive either ceftriaxone emissions tests and at some centres by auditory brain-
or placebo. stem response. All children were subsequently tested
Our primary endpoint was the rate of bacteriological with age-appropriate hearing tests on the two scheduled
failure dened as positive cultures of cerebrospinal uid follow-up visits (days 40 and 190). Hearing loss was
or blood on days 640 for the originally identied classied in accordance with the International
organism or for H inuenzae, S pneumoniae, or Classication of Impairments, Activities and Partici-
N meningitidis on days 640 in a patient from whom the pation37 as mild (2640 dB), moderate (4160 dB), severe
rst sample of cerebrospinal uid had been culture (6180 dB), and profound (>80 dB), and those identied
negative. Our secondary endpoints were death and with moderate to severe hearing loss were referred to
hearing, neurological, or visual sequelae at any time local audiological services for management. Children
during the study. were screened by age-appropriate tests for high-grade
All study sta were trained in study procedures and visual impairment, and those suspected of having any
documentation. Patients were assessed daily and a study impairment were referred for an ophthalmological
form was completed for each day, at discharge on day 10, opinion. Neurological and developmental status was
and at each follow-up visit at which all children were assessed and recorded by trained clinicians. The age and
assessed for hearing, visual, developmental, and stages questionnaire38 was used to assess development
neurological outcomes. After discharge, patients were and the score was used to quantify delayed development
reviewed on day 40 and day 190 after enrolment into our in personal and social interactions, ne motor
study, when full physical, neurological, and hearing movements, gross motor movements, and competency
assessments were done. Unscheduled visits were in the use of language.

1027 participants randomly assigned

to study groups

23 excluded
10 completely missing data or not assigned
8 had other organisms identied in
cerebrospinal uid or blood, such as
Escherichia coli or Staphylococcus aureus
5 opened assignment envelopes in error

496 in 5-day treatment group 508 in 10-day treatment group

1 death (day 9) 1 death (day 6)
4 left against medical advice 4 left against medical advice
17 had therapeutic failure 16 had therapeutic failures
(2 died between day 10 and 40) (2 died between day 10 and 40)
3 protocol violations 2 had diagnosis changed
2 had diagnosis changed
469 completed therapy day 10 485 completed therapy day 10

9 deaths after day 10 6 deaths after day 10

16 lost to follow-up 23 lost to follow-up

444 completed follow-up day 40 456 completed follow-up day 40

10 deaths after day 40 11 deaths after day 40

22 lost to follow-up 37 lost to follow-up

422 completed follow-up day 190* 417 completed follow-up day 190*

Figure: Trial prole

*Of the participants who were lost to follow-up at day 40, 19 patients (ten from the 5-day group and nine from 10-day group) were followed up at day 190;
one patient from the 10-day group died after day 40.

www.thelancet.com Vol 377 May 28, 2011 1839

 Articles

After random assignment, if the child developed a new

Panel 2: Details of children not randomly assigned to study groups fever (>38C), or had persistent convulsions or a
2000 children recruited on day 0* deteriorating coma score, we judged that the study
180 did not have bacterial meningitis treatment had failed and changed it or added to it in
68 had blood in their cerebrospinal uid accordance with hospital guidelines. We made eorts to
42 had viral meningitis identify any intracranial complications, by cranial
26 had encephalitis ultrasound where possible and CT scan when available.
23 had tuberculous meningitis We classied children who returned with signs of
20 had epilepsy meningitis within 2 weeks of discharge as having relapsed,
One had candida meningitis and treated them in accordance with hospital guidelines.
269 died before they were assigned to study groups If a parent withdrew consent, or the protocol was violated,
1551 children with bacterial meningitis were alive on day 5 we withdrew the child from the study but continued to
39 did not give consent treat them with unmasked ceftriaxone to complete a 10-day
18 left against medical advice course. Four monitoring visits, either by WHO internal
42 lived outside the study area monitors or independent external monitors, were made to
167 received antibiotics before admission each site to ensure protocol compliance, data management,
47 were ill for more than 7 days before admission and good clinical practice.
13 had previous cerebral palsy or were deaf Samples (blood, cerebrospinal uid, or both) were
16 had a neurosurgical disability obtained aseptically before giving the antibiotic. Samples
Three had a subdural abscess were labelled and immediately sent to the laboratory.
15 had other infections that needed other antibiotics or continued ceftriaxone Blood specimens were inoculated in trypticase soy broth
(four with cellulitis, one with septic arthritis, one with empyema, nine with with 025% sodium polyethanol sulphonate and
pneumonia) incubated at 37C for 7 days. Inoculated broths were
Three had bacterial persistence subcultured on blood and chocolate agar plates on
Two had cyanotic heart disease days 2, 3, and 7. Specimens of cerebrospinal uid were
80 had other bacterial causes (70 had non-typhoidal Salmonella spp, ve had cultured directly onto blood and chocolate agar plates
-haemolytic streptococci, one had Escherichia coli, two had Salmonella typhi, and and incubated at 37C for 72 h. Specimens of cerebrospinal
two had Acinetobacter spp) uid that were negative by culture were tested for
73 were judged by the treating physician as too sick to assign to a study group antigens of ve common organisms (H inuenzae type b,
22 had persistent or dicult to control seizures and fever on day 5 S pneumoniae, N meningitidis, group B streptococci, and
28 had persistence of fever on day 5 Escherichia coli) with latex-agglutination reagents in
19 had persistent seizures with no fever on day 5 accordance with manufacturer instructions. The latex-
Four had deteriorating level of consciousness (two with seizures) agglutination test was done only if the gram stain and
Six had no reason given for exclusion by treating physician culture of cerebrospinal uid were negative by day 2.
1027 were randomly assigned to treatment groups Isolates were identied with standard procedures.39
23 were excluded after assignment Susceptibility testing was done with antimicrobial drugs
1004 were included in the per-protocol analysis with E-test after the standard procedure described by the
Clinical Laboratory Standard Institute.40 For HIV, serum
*Although persistence of convulsions or fever on day 5 were not study exclusion criteria, the treating physician (who could be
samples were tested by at least two tests: Serodia-HIV
dierent than the investigator) felt the child should not be randomly assigned to study groups.
particle agglutination (Fujirebio Inc, Tokyo, Japan),
HIVSPOT (Genelabs Diagnostics, Singapore),
Determine-HIV (Abbott Laboratories, Abbott Park, IL,
Number of Dates of study accrual USA), and Capillus-HIV (Cambridge Diagnostics,
children Galway, Ireland). Discordant tests were conrmed either
enrolled (%)
by a third test or an in-house HIV PCR. Children younger
Bangladesh (Dhaka Shishu Hospital, Dhaka) 175 (17%) November, 2002, to December, 2006 than 15 months with a positive antibody test were
Egypt (Abbasia Fever Hospital, Cairo) 50 (5%) January, 2005, to December, 2006 conrmed with the HIV PCR.
Malawi (Queen Elizabeth Central Hospital, Blantyre) 366 (37%) July, 2002, to December, 2006
Pakistan (Liaquat University Hospital, Hyderabad; 153 (15%) September, 2001, to December, 2006 Statistical analysis
Civil Hospital, Karachi; Department of Paediatrics,
We calculated the study sample size on the basis of
Aga Khan University, Karachi; National Institute of
Child Health, Karachi) an a-priori denition of equivalence of up to
Pakistan (Children Hospital, Pakistan Institute of 96 (10%) September, 2001, to December, 2006 15% dierence of proven bacteriological failure up to
Medical Sciences, Islamabad; Rawalpindi General day 40. The investigators tolerance was for an increase
Hospital, Rawalpindi) from 05% to 20% of bacteriological proven treatment
Vietnam (Childrens Hospital N1, Ho Chi Minh City) 164 (16%) September, 2001, to December, 2006 failures as a statement of equivalence. We estimated
the sample size to be 1400 patients, 700 in each group
Table 1: Enrolment by study site for random assignment on day 5
(one-sided of 005, power of 80%). We did two interim

1840 www.thelancet.com Vol 377 May 28, 2011

 Articles

safety and ecacy analyses, the rst after one-third of

5-day treatment group 10-day treatment group
patients had been enrolled and the second after two-
thirds had been enrolled. Number of boys 285 (58%) 280 (55%)

Our analysis was per protocol, which is suitable for Age (months) 386 (425); 16 (661) 378 (416); 17 (661)
equivalence trials. We stratied baseline characteristics Presenting features
by treatment groups to look for dierences in the severity Irritability 287 (58%) 288 (57%)
and characteristics of presenting illnesses. We sum- Duration of illness before admission 38 (38); 3 (25) 35 (23); 3 (24)
(days)
marised study variable with proportions, mean, median,
Antibiotic use in previous 48 h 128 (26%) 138 (27%)
and IQRs. We tested for equivalence between the two
Drowsiness 229 (46%) 252 (50%)
treatments by calculating the risk dierence and two-
Vomiting 291 (59%) 289 (57%)
sided 95% CI for the primary and secondary outcomes.
We used SPSS (version 16) and Stata (version 11) for Poor feeding 335 (68%) 340 (67%)

analysis. This study is registered with the International Seizures 262 (53%) 266 (52%)
Standard Randomised Controlled Trial Number Register, Weight for age less than 3 Z score 37 (8%) 47 (9%)
number ISRCTN38717320. Temperature (C) 383 10 384 09
Systolic blood pressure (mm Hg; 971 (152); 100 (90107) 977 (146); 100 (90110)
n=963)
Role of the funding source
Glasgow coma score (n=454) 126 (28); 13 (1115) 129 (25); 14 (1115)
The sponsor of the study had no role in study design,
Blantyre score (n=615) 41 (12); 5 (35) 41 (12); 5 (35)
data collection, data analysis, data interpretation, or
Todd-Herson meningitis score (n=669) 37 (18); 3 (255) 36 (19); 3 (25)
writing of the report. SAQ and MWW, as employees of
WHO, were involved in study design, collating analyses Laboratory data

and interpretation of data, and writing of the report. Glucose in cerebrospinal uid (mg/dL)* 326 (533); 15 (0429) 361 (578); 18 (050)

The writing committee had full access to all the data in Protein in cerebrospinal uid (mg/dL) 2140 (1640); 200 (98300) 2075 (1791); 200 (90300)
the study, and the principal investigators, SAQ and Haemoglobin <6 g/dL 23 (5%) 41 (8%)
MWW, had nal responsibility for the decision to Haemoglobin 69 g/dL 201 (41%) 190 (37%)
submit for publication. Haemoglobin >9 g/dL 263 (53%) 265 (52%)
Total leucocyte count per L (n=968) 17 861 (18 593); 16 314 (11 279);
13 700 (947520 000) 14 000 (997520 225)
Results
Blood glucose <60 mg/dL 58 (12%) 52 (10%)
The gure shows the trial prole. Panel 2 lists the details
Blood glucose 6079 mg/dL 73 (15%) 66 (13%)
of children not randomly assigned to study groups.
Table 1 shows the accrual of patients by site and table 2 Blood glucose 80 mg/dL 341 (69%) 361 (71%)

lists the patients baseline characteristics. 20 enrolled Cerebrospinal uid culture positive 227 (46%) 240 (47%)

children had a positive second culture of cerebrospinal Cerebrospinal uid 155 (31%) 143 (28%)
latex-agglutination test positive
uid. 17 were caused by Salmonella spp, two by
Cerebrospinal uid gram stain 188 (38%) 188 (37%)
S pneumoniae (who died before day 5), and one by (gram-negative bacillus, gram-positive
H inuenzae type b. Mean age at presentation was cocci, gram-negative cocci)
38 months. On average children were ill for 4 days Blood culture positive 100 (20%) 100 (20%)
before enrolment. Bacteria identied by either means 334 (67%) 340 (67%)
Table 3 lists the outcomes of our study. There were no (signicant clinically)
bacteriological failures with either 5-day or 10-day treat- Infected with HIV 51 (10%) 66 (13%)
ment. The proportion of children that completed
Data are n (%) or mean (SD); median (IQR). *Glucose in cerebrospinal uid in one centre was checked by urine dipstick
treatment was similar between the two groups (gure). method. Positive for Haemophilus inuenzae, Streptococcus pneumoniae, and Neisseria meningitides. HIV-positive
Treating physicians judged that treatment had failed in patients only reported from Malawi.
33 children, equally divided in the two therapeutic
Table 2: Baseline comparison between study groups
groups. Four children were diagnosed with tuberculous
meningitis. After discharge, 23 children had another
episode of meningitis (table 3), of which two were had been admitted. There was no statistical dierence in
classied as relapse (one child was infected with HIV) study outcomes between the two groups whether as a
since the episodes were caused by the same organism whole or when analysed separately by bacterial cause
within 2 weeks of discharge. Of 919 children seen on (table 3). In children with pneumococcal meningitis,
either day 40 or day 190, 140 had moderate or severe survival with sequelae was more common in the 10-day
hearing loss, 14 had visual loss, and 51 other neurological than the 5-day group (p=059; table 3). Of the 21 children
sequelae (table 3). 16 children had both neurological and with pneumococcal meningitis who died, 15 were
moderate to severe hearing loss. 41 children died within infected with HIV: ten in the 5-day group and ve in the
the 190 day study period (webappendix p 5), including 10-day group (p=06). See Online for webappendix
two during admission to hospital, but only 15 deaths By day 40 after discharge, 39 children had been lost to
were related to the episode of meningitis for which they follow-up, nine children had died in the 5-day group, and

www.thelancet.com Vol 377 May 28, 2011 1841

 Articles

5-day 10-day Total Risk dierence (%; 5-day 10-day Total

treatment treatment 95% CI) treatment treatment
group group group group
(n=496) (n=508)
Positive culture of cerebrospinal 227 (46%) 240 (47%) 467 (47%)
Overall outcomes for all children uid
Therapy successfully completed (10 days) 469 (95%) 485 (96%) 954 092 (36 to 18) Haemophilus inuenzae type b 85 (17%) 85 (17%) 170 (17%)
Antibiotic therapy modied after random 17 (3%) 16 (3%) 33 03 (19 to 25) Streptococcus pneumoniae 115 (23%) 140 (28%) 255 (25%)
assignment or therapy failure Neisseria meningitidis 27 (5%) 15 (3%) 42 (4%)
Changed diagnosis (to tuberculous meningitis) 2 (0%) 2 (0%) 4 0009 (07 to 07) Latex agglutination test of 155 (31%) 143 (28%) 298 (30%)
Adverse events to the study drug 0 0 0 cerebrospinal uid
Bacteriological failures 0 0 0 H inuenzae type b 79 (15%) 72 (14%) 151 (15%)
Another episode of meningitis 8 (2%) 13 (3%) 21 095 (27 to 82) S pneumoniae 49 (10%) 56 (11%) 105 (11%)
Relapse of meningitis 2 (0%) 0 2 04 (015 to 096) N meningitidis 27 (5%) 15 (3%) 42 (4%)
Deaths related to meningitis only* 9 (2%) 6 (1%) 15 063 (087 to 21) Gram stain of cerebrospinal uid
Deaths due to any reason after cure (until 22 (4%) 19 (4%) 41 069 (18 to 31) Gram-negative bacilli 55 (11%) 51 (10%) 106 (11%)
follow-up at 6 months after enrolment) Gram-positive cocci 110 (22%) 123 (24%) 233 (23%)
Survival with sequelae 129 (26%) 138 (27%) 267 12 (66 to 43) Gram-negative cocci 23 (5%) 14 (3%) 37 (4%)
Hearing loss 105 (21%) 106 (21 %) 211 03 (47 to 53) Blood culture
Moderate or severe unilateral 22 (4%) 23 (5%) 45 009 (27 to 25) Positive for any study organism 100 (20%) 100 (20%) 200 (20%)
Moderate or severe bilateral 50 (10%) 45 (9%) 95 12 (24 to 48) Negative for any study 363 (73%) 376 (74%) 739 (74%)
Visual loss 4 (1%) 10 (2%) 14 12 (26 to 03) organism
Neurological sequelae including motor decit, 21 (4%) 30 (6%) 51 17 (043 to 103) H inuenzae type b 43 (9%) 37 (7%) 80 (8%)
nerve palsies S pneumoniae 49 (10%) 55 (11%) 104 (10%)
Motor decit 20 (4%) 29 (6%) 49 16 (43 to 098) N meningitidis 8 (2%) 8 (2%) 16 (2%)
Cranial nerve palsy 4 (1%) 4 (1%) 8 002 (11 to 11) Organism identied by any test* 334 (67%) 340 (67%) 674 (67%)
Afebrile seizures 3 (1%) 5 (1%) 8 04 (15 to 07) H inuenzae type b 134 (27%) 132 (26%) 266 (27%)
Hydrocephalus 2 (0%) 6 (1%) 8 08 (19 to 03) S pneumoniae 154 (31%) 181 (36%) 335 (33%)
Developmental delay 25 (5%) 33 (7%) 58 15 (43 to 14) N meningitidis 46 (9%) 27 (5%) 73 (7%)
Study outcomes by cause
Streptococcus pneumoniae n=154 n=181 n=335 *Positive by cerebrospinal uid, blood culture, or latex agglutination test, or
gram-negative diplococcic on gram stain.
Death due to any reason after cure (all deaths) 12 9 21 28 (25 to 81)
Survival with sequelae (all) 54 72 126 47 (151 to 57) Table 4: Organisms identied at the time of enrolment by treatment
Antibiotic therapy modied after random 2 3 5 04 (29 to 22)
assignment or therapy failure
causative bacteria were identied by culture of
Haemophilus inuenzae n=134 n=132 n=266
cerebrospinal uid, latex-agglutination test, gram stain,
Death due to any reason after cure (all deaths) 4 5 9 08 (51 to 35)
or blood culture in 674 cases (table 4). S pneumoniae was
Survival with sequelae (all) 33 31 64 11 (91 to 114)
the most common organism identied.
Antibiotic therapy modied after random 6 5 11 06 (41 to 55)
assignment or therapy failure
The DSMB recommended stopping the study after the
Neisseria meningitidis n=46 n=27 n=73
second interim analysis, because a dierence was
Death due to any reason after cure (all deaths) 1 0 1 22 (20 to 64)
unlikely to be found between the treatment groups within
the projected sample size. At stoppage more than
Survival with sequelae (all) 6 3 9 19 (134 to 173)
1000 children had been enrolled. A preplanned sample
Antibiotic therapy modied after random 1 0 1 21 (20 to 63)
assignment or therapy failure size calculation had shown that this would provide
No cause identied n=162 n=168 n=330 equivalence limits of 47% for death and 0524% for
Death due to any reason after cure (all) 5 5 10 01 (36 to 38) bacterial failures.
Survival with sequelae (all) 36 32 68 32 (56 to 119) Dexamethasone was given at some sites: 437 of
Antibiotic therapy modied after random 8 8 16 02 (45 to 48) 1004 children received dexamethasone on the rst day of
assignment or therapy failure antibiotic treatment; 428 received it for 2 days, 292 for
3 days, and 140 for 4 days. It was routine practice to give
*Decision about the association with meningitis was made after all deaths were reviewed by the data and safety
monitoring board. Six in the 5-day group and one in the 10-day group were infected with HIV. Of the 41 deaths,
dexamethasone at all sites except Malawi; however, even
25 were of children infected with HIV: 12 in the 5-day group and 13 in the 10-day group. All of children infected with HIV where it was routine practice not all patients received it.
were in Malawi, eight of them with another episode of meningitis. 209 of 496 children in the 5-day group received dexa-
Table 3: Outcomes by study group
methasone compared with 228 of 508 in the 10-day group.
Previous administration of dexamethasone did not aect
six died in the 10-day group (gure). By day 190, a the outcome in the patients overall or in either study
further 59 were lost to follow-up, ten children died in the group. Side-eects of antibiotic treatment were minor
5-day group, and 11 died in the 10-day group. The and similar in both groups.

1842 www.thelancet.com Vol 377 May 28, 2011

 Articles

Discussion
Our ndings show that further antibiotic treatment is Panel 3: Research in context
not needed in children beyond the neonatal age-group Systematic review
with purulent bacterial meningitis caused or presumably Before this study, we searched Medline and Embase for studies
caused by one of the three major pathogens, and in comparing short-course with long-course antimicrobial
whom the clinical condition is stable or improving by day treatment for childhood bacterial meningitis. Our search terms
5 of treatment with ceftriaxone. Both death and long- were bacterial, paediatric, pediatric, children,
term sequelae are the consequence of inammatory and childhood, short course, therapy, and antibiotic. We
ischaemic neural damage,41 which peak before or in the limited the search to publications in English, but not by date.
early stages of antibiotic treatment. Early diagnosis and We also searched reference lists of articles identied by this
prompt use of antibiotics are probably more important in strategy and this was updated during the write-up of our
reducing the mortality and morbidity than an extended study. Previous studies were small and did not give a
antibiotic treatment regimen. Studies of short versus conclusive answer. Several were limited to one cause, dierent
long duration of treatment with antibiotics have shown antibiotics were used, and outcome measures varied. A recent
no dierence in outcome (panel 3). systematic review42 summarised ve trials with a total of less
Most clinicians are particularly concerned about than 400 patients and found no dierence between short and
adequate treatment of meningitis caused by S pneumoniae, long treatment duration, but suered from the variability of
a notoriously dicult pathogen to eliminate. It is notable treatment approaches.
that the outcome of the subgroup of cases with proven
pneumococcal meningitis is similar in the two treatment Findings
groups in our study. Our study was powered to answer the question of short-
Hearing decit is one of the most common compli- course versus long-course treatment with ceftriaxone in
cations of bacterial meningitis and can vary from 5% to childhood bacterial meningitis. It was done where bacterial
36% in survivors.4347 Decit was similar between our meningitis is common, included the most important causes,
study groups during and by the end of our study. We and also children infected with HIV. Our results are conclusive
expected this nding because we believe that the hearing and conrm previous small study ndings.
decit is due to cochlear damage caused by the infection
and the hosts inammatory response, and by the time of
random assignment on day 5, this damage had already The overall outcome of meningitis in our study is far
been done. Similarly, neurological decits including from good, even with optimum antimicrobial treatment
developmental delay, quadriplegia, and isolated cranial- and much better care than is routinely available in the
nerve palsy, or any of these in combination, happened in participating countries. Many children died in the rst
a similar proportion of children in the two groups: 21 in few days before random assignment, or left hospital
the 5-day group and 30 in the 10-day group (p=029). with sequelae. To prevent this burden of disease,
At the design stage of our study, we decided to restrict preventive vaccines are vital and their development and
our trial to ceftriaxone, even though combination treatment deployment are of crucial importance. The H inuenzae
with chloramphenicol and penicillin were still the rst-line type b vaccine, as part of the pentavalent vaccine, was
treatment recommended by WHO.1 The emergence of introduced into the Malawian expanded programme for
pneumococci and H inuenzae increasingly resistant to immunisation in February, 2002. The number of cases
conventional rst-line treatment is making the combination of meningitis caused by H inuenzae type b fell
of chloramphenicol and penicillin less useful in meningitis substantially and, by 2005, the incidence rates fell
treatment. Ceftriaxone has several advantages that make it from 2040 per 100 000 to almost zero.48 Nevertheless,
the rst-line treatment of meningitis in most developed prompt access to good quality care is important for every
countries. In developing countries, the cost was the infected child. This means improving health systems
principal disadvantage but this has reduced in recent years a dicult and daunting task.
since the expiration of the drugs patent. There are We did our trial in children presenting with purulent
substantial advantages to a shorter course of in-hospital meningitis presumably caused by S pneumoniae,
treatment. Indirect costs of admission to hospital are H inuenzae type b, and N meningitidis, and could nd
important constraints for poor families. The maintenance evidence for these organisms in 67% of our cases. We
of intravenous access over a 10-day period is dicult and is excluded organisms other than the three main bacterial
a burden on the scarce hospital resources available in causes of meningitis from our trial, in particular non-
many developing countries. Risk of nosocomial infections typhoid salmonellae, which might become more
and spread of hepatitis B and other infections during prominent as other causes decline. Invasive infections
admission to hospital are major risks. Short hospital stays caused by non-typhoid salmonellae are common in
reduce in-patient numbers, bed occupancy, and total malarial parts of Africa. In such settings a gram stain of
management cost. Early discharge to home means less cerebrospinal uid is helpful in identifying the presence
loss of family income because of loss of working hours. of bacteria. Non-typhoid salmonellae are adapted to

www.thelancet.com Vol 377 May 28, 2011 1843

 Articles

survival in the intracellular environment, which makes and Chandiram, M Nand Lal (Hyderabad, Pakistan). We thank
them dicult to eradicate and means that they need Medochemie Limited (Cyprus) for the donation, free of charge, of active
and placebo ceftriaxone. The donor did not have any say in development
longer treatment courses than usual. A third-generation of the protocol, interpretation of results, or the conclusions.
cephalosporin with a quinolone (eg, ceftriaxone and
References
ciprooxacin) are recommended to be given for 1 WHO. World Health Statistics 2006. Geneva: World Health
46 weeks.49 Clinicians advise that cerebrospinal uid Organization, 2006.
should be re-examined at the end of treatment to ensure 2 Ramakrishnan M, Ulland A, Sreinhardt LC, Moisi JC, Were F,
Levine OS. Sequelae due to bacterial meninigitis among African
bacterial clearance.49,50 For this reason we do not suggest children: a systematic literature review. BMC Med 2009; 7: 47.
that a 5-day short course would be adequate for meningitis 3 Lehmann D, Yeka W, Rongap T, et al. Aetiology and clinical signs of
caused by these organisms. bacterial meningitis in children admitted to Goroka Base Hospital,
Papua New Guinea, 19891992. Ann Trop Paediatr 1999; 19: 2132.
Our study had several limitations. First, hearing
4 Palmer A, Weber M, Bojang K, McKay T, Adegbola RA. Acute
assessment was done with dierent methods at the bacterial meningitis in The Gambia. J Trop Pediatr 1999; 45: 5153.
various sites. For this reason we included only moderate 5 Molyneux E, Walsh A, Phiri A, Molyneux M. Acute bacterial
to severe hearing loss in our nal analysis, to avoid meningitis in children admitted to the Queen Elizabeth Central
Hospital, Blantyre, Malawi in 199697. Trop Med Int Health 1998;
inclusion of conductive (mild) hearing loss unrelated to 3: 61018.
the meningitis event. Second, some centres gave adjuvant 6 WHO. Antimicrobial and support therapy for meninigitis
steroids for the rst 48 h of antibiotic treatment; however, in children. Geneva: World Health Organization, 1998.
this did not aect the outcome. Third, treating physicians 7 Hussey G, Schaaf H, Hanslo D, et al. Epidemiology of post-neonatal
bacterial meningitis in Cape Town children. S Afr Med J 1997;
decided not to allow some of the children to be randomly 87: 5156.
assigned on day 5, most of whom had seizures that were 8 OBrien KL, Wolfson LJ, Watt JP, et al. Burden of disease caused
dicult to control, continuing fever, or deteriorating by Streptococcus pneumoniae in children younger than 5 years:
global estimates. Lancet 2009; 374: 893902.
consciousness. It is probable that most of these children 9 Watt JP, Wolfson LJ, OBrien KL, et al. Burden of disease caused
had sequelae rather than continuing infection, but since by Haemophilus inuenzae in children younger than 5 years:
they were not assigned to treatment groups, we cannot global estimates. Lancet 2009; 374: 903911.
10 Murray CJ, Lopez AD. Mortality by cause for eight regions of the
conclude that a short antibiotic course would have been world: Global Burden of Disease Study. Lancet 1997; 349: 126976.
sucient for these children. For conclusions to be 11 Goetghebuer T, West TE, Wermenbol V, et al. Outcome of meningitis
applicable to typical settings in a developing country caused by Streptococcus pneumoniae and Haemophilus inuenzae type b
in children in The Gambia. Trop Med Int Health 2000; 5: 20713.
where good culture facilities are not available, and the
12 Goto H, Shimada K, Ikemoto H, Oguri P. Antimicorbial
dierentiation between aseptic sequelae and possible susceptibility of pathogens isolated from more than 10,000 patients
untreated infection cannot be made, we would have with infectious respiratory diseases: a 25 year longitudinal study.
preferred to include all children presenting with clinical J Infect Chemother 2009; 15: 34760.
13 Velasquez PA, Parussolo L, Cardoso CL, Togmin MC, Garcia LB.
meningitis. However, such a study would have been High prevalence of children colonized with penicillin-resistant
unethical. So, our study was done in circumstances as Streptococcus pneumoniae in public day-care centers. J Pediatr (Rio J)
close as possible to routine clinical settings. Finally, our 2009; 85: 51622 (in Portuguese).
14 Kisakye A, Makumbi I, Nansera D, et al. Surveillance for Streptococcus
study sample size was insucient for separate assessment pneumoniae meningitis in children aged <5 years: implications for
by causative organism. Few children had persisting immunization in Uganda. Clin Infect Dis 2009; 48 (suppl 2): S15361.
organisms on the second lumbar puncture. Equally less 15 Mudhune S, Wamae M, Network Surveillance Pneumococcal
Disease in the East African Region. Report on invasive disease
than 8% of children were not assigned because they were and meningitis due to Haemophilus inuenzae and
judged to be too ill. Other organisms were rarely found, Streptococcus pneumonia from the Network for Surveillance of
with the exception of Salmonella spp in Malawi, which Pneumococcal Disease in the East African Region. Clin Infect Dis
2009; 48 (suppl 2): S14752.
raises the issue of this organism for parts of sub-Saharan 16 Klugman KP. Pneumococcal resistance to the third generation
Africa. In conclusion, a 5-day course of ceftriaxone is not cephalosporins: clinical, laboratory and molecular aspects.
only clinically eective but also cost ecient, and we Int J Antimicrob Agents 1994; 4: 6367.
recommend this regimen for uncomplicated cases of 17 ONeill E, Humphreys H, Phillips J, Smyth EG. Third-generation
cephalosporin resistance among gram-negative bacilli causing
meningitis in children, although our sample size was meningitis in neurosurgical patients: signicant challenges in
inadequate to answer this by individual bacteria. ensuring eective antibiotic therapy. J Antimicrob Chemother 2006;
57: 35659.
Contributors 18 Martin E, Koup JR, Paravicini U, Stoeckel K. Pharmacokinetics of
SAQ conceived the project. SAQ and MWW were involved in designing ceftriaxone in neonates and infants with meningitis. J Pediatr 1984;
the study. SQN and SAQ supervised all analyses. EM, SAQ, MWW, and 105: 47581.
SQN wrote the paper with input from all the authors. All authors 19 Girgis NI, Abu el Ella AH, Farid Z, Haberberger RL, Galal FS,
reviewed and approved the nal draft. Woody JN. Intramuscular ceftriaxone versus
Conicts of interest ampicillin-chloramphenicol in childhood bacterial
meningitis. Scand J Infect Dis 1988; 20: 61317.
We declare that we have no conicts of interest.
20 Radetsky M. Duration of treatment in bacterial meningitis:
Acknowledgments a historical inquiry. Pediatr Infect Dis J 1990; 9: 29.
We acknowledge the involvement of Rafeza Khanam, Arifur, 21 Lambert HP. Dont keep taking the tablets? Lancet 1999; 354: 94345.
Maksuda Islam, Fahmida Zabeen, Jalal Ahmed, Amala Badiya, 22 Visintin C, Mugglestone MA, Fields EJ, et al. Management of
Rehana Begum (Bangladesh); Goenke Poerksen, Enitan Carroll, bacterial meningitis and meningococcal septicaemia in children and
Sarah White, Mavuto Mukaka, Amos Phiri, Lorna Wilson (Malawi); young people: summary of NICE guidance. BMJ 2010; 340: c3209.

1844 www.thelancet.com Vol 377 May 28, 2011

 Articles

23 Feigin RD, Pearlman E. Bacterial meningitis beyond the neonatal 36 Martin E, Hohl P, Guggi T, Kayser FH, Fernex M. Short course
period. In: Feigin RD, Cherry JD, eds. Textbook of Pediatric single daily ceftriaxone monotherapy for acute bacterial meningitis
Infectious Diseases, 3rd edn. Philadelphia, PA: WB Saunders Co, in children: results from a Swiss multicenter study. Part 1: Clinical
1992: 40030. results. Infection 1990; 18: 7077.
24 Nathan N, Borel T, Djibo A, et al. Ceftriaxone as eective as 37 Simeonsson RJ, Lollar D, Hollowell M. Revision of the International
long-acting chloramphenicol in short-course treatment of Classication of Impairments, Disabilities, and Handicaps:
meningococcal meningitis during epidemics: a randomised developmental issues. J Clin Epidemiol 2000; 53: 11324.
non-inferiority study. Lancet 2005; 366: 30813. 38 Squires J, Bricker D, Potter LW. Revision of a parent-completed
25 Marhoum el Filali K, Noun M, Chakib A, Zahraoui M, Himmich H. developmental screening tool: ages and stages questionnaires.
Ceftriaxone versus penicillin G in short-term treatment of J Pediatr Psychol 1997; 22: 31328.
meningococcal meningitis in adults. Eur J Clin Microbiol Infect Dis 39 WHO. Chapter 5: Bacteriological mediaGuidelines on Standard
1993; 12: 76668. Operating Procedures for Microbiology. Geneva: World Health
26 Auvergnat JC, Le Tallec JY, Marchou B, Massip P, Carriere JP, Organization, 2006. http://www.searo.who.int/en/Section10/
Armengaud M. Shortened antibiotic therapy of menningococcal Section17/Section53/Section482_1782.htm (accessed April 4, 2011).
meningitis: 5-day administration of ceftriaxone. Pathol Biol (Paris) 40 Wikler MA, Hindler JF, Cockerill FR, et al. Performance standards
1988; 36: 73537 (in French). for antimicrobial disk susceptibility tests; approved standardtenth
27 Viladrich PF, Pallares R, Ariza J, Ru G, Gudiol F. Four days of edition. Wayne, PA: Clinical Laboratory Standard Institute, 2009.
penicillin therapy for meningococcal meningitis. Arch Intern Med 41 Trunkel AR, Scheld WM. Pathogenesis and pathophysiology
1986; 146: 238082. of bacterial meningitis. Clin Microbiol Rev 1993; 6: 11836.
28 Lin TY, Chrane DF, Nelson JD, McCracken GH Jr. Seven days of 42 Karageorgopoulos DE, Valkimadi PE, Kapaskelis A, Rafailidis PI,
ceftriaxone therapy is as eective as ten days treatment for bacterial Falagas ME. Short versus long duration of antibiotic therapy for
meningitis. JAMA 1985; 253: 355963. bacterial meningitis: a meta-analysis of randomised controlled trials
29 Martin E, Hohl P, Guggi T, Kayser FH, Fernex M. Short course in children. Arch Dis Child 2009; 94: 60714.
single daily ceftriaxone monotherapy for acute bacterial meningitis 43 Kutz JW, Simon LM, Chennupati SK, Giannoni CM, Manolidis S.
in children: results of a Swiss multicenter study. Part I: Clinical Clinical predictors for hearing loss in children with bacterial
results. Infection 1990; 18: 7077. meningitis. Arch Otolaryngol Head Neck Surg 2006; 132: 94145.
30 Roine I, Ledermann W, Foncea LM, Ban A, Cohen J, Peltola H. 44 Forsyth H, Kalumbi F, Mphaka E, et al. Hearing loss in Malawian
Randomized trial of four vs seven days of ceftriaxone treatment children after bacterial meningitis: incidence and risk factors.
for bacterial meningitis in children with rapid initial recovery. Audiol Med 2004; 2: 10007.
Pediatr Infect Dis J 2000; 19: 21922. 45 Qazi SA, Khan MA, Mughal N, et al. Dexamethasone and bacterial
31 Singhi P, Kaushal M, Singhi S, Ray P. Seven days vs 10 days meningitis in Pakistan. Arch Dis Child 1996; 75: 48288.
ceftriaxone therapy in bacterial meningitis. J Trop Pediatr 2002; 46 Fortnum HM. Hearing impairment after bacterial meningitis:
48: 27379. a review. Arch Dis Child 1992; 67: 112833.
32 Sholtz H, Hofmann T, Noack R, Edwards DJ, Stoeckel K. 47 Eisenhut M, Meehan T. Risk factors for hearing loss in bacterial
Prospective comparison of ceftriaxone and cefotaxime for meningitis: delay in treatment and clinical manifestations.
the short-term treatment of bacterial meningitis in children. Audiol Med 2002; 11: 8697.
Chemotherapy 1998; 44: 14247.
48 Daza P, Banda R, Misoya K, et al. The impact of routine infant
33 Lutsar I, Gontmachr A, Narska M, et al. Five days of antibacterial immunization with Haemophilus inuenzae type b conjugate
therapy for bacterial meningitis in children? Infection 1995; vaccine in Malawi, a country with high human immunodeciency
23: 11318. virus prevalence. Vaccine 2006; 24: 623239.
34 Lutsar I, Gontmachr A, Narska M, et al. Five days of antibacterial 49 Price EH, de Louvois J, Workman MR. Antibiotics for Salmonella
therapy for bacterial meningitis in children? Infection 1995; meningitis in children. J Antimicrob Chemother 2006; 46: 65355.
23: 11318.
50 AAP. Salmonella infections. In: Peter G, ed. Report of the
35 Jadavji T, Biggar WD, Gold R, Prober CG. Sequelae of acute Committee on Infectious Diseases, 25th edn. Elk Grove Village, IL:
bacterial meningitis in children treated for seven days. Pediatrics American Academy of Pediatrics, 2000.
1986; 78: 2125.

www.thelancet.com Vol 377 May 28, 2011 1845

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.