Outcome of meningitis caused by Streptococcus pneumoniae and Haemophilus influenzae type b in children in The Gambia
Dr Tessa Goetghebuer, MRC Laboratories, Fajara, PO Box 273, Banjul, The Gambia. E-mail: email@example.com
Summary In developing countries, endemic childhood meningitis is a severe disease caused most commonly by Streptococcus pneumoniae or Haemophilus influenzae type b (Hib). Although many studies have shown that fatality rates associated with meningitis caused by these organisms are high in developing countries, little is known about the long-term outcome of survivors. The purpose of this study was to assess the importance of disabilities following pneumococcal and Hib meningitis in The Gambia. 257 children aged 0–12 years hospitalized between 1990 and 1995 with culture-proven S. pneumoniae (n = 134) or Hib (n = 123) meningitis were included retrospectively in the study. 48% of children with pneumococcal meningitis and 27% of children with Hib meningitis died whilst in hospital. Of the 160 survivors, 89 (55%) were followed up between September 1996 and October 1997. Of the children with pneumococcal meningitis that were traced, 58% had clinical sequelae; half of them had major disabilities preventing normal adaptation to social life. 38% of survivors of Hib meningitis had clinical sequelae, a quarter of whom had major disabilities. Major handicaps found were hearing loss, mental retardation, motor abnormalities and seizures. These data show that despite treatment with effective antibiotics, pneumococcal and Hib meningitis kill many Gambian children and leave many survivors with severe sequelae. Hib vaccination is now given routinely in The Gambia; an effective pneumococcal vaccine is needed.
Childhood bacterial meningitis is responsible for much morbidity and mortality in children throughout the world, especially in developing countries. In The Gambia, meningitis is responsible for about 1% of hospital admissions to a tertiary hospital ( Brewster & Greenwood 1993) and accounts for about 5% of deaths in infants ( Jaffar et al. 1997 ). The pathogens responsible for most cases of endemic meningitis in The Gambia, as in most other developing countries, are Streptococcus pneumoniae and Haemophilus influenzae type b (Hib), which are isolated from about 80% of the cases of culture-positive meningitis ( Palmer et al. 1999 ). Immediate case fatality rates of 37% to 55% for pneumococcal meningitis and 18% to 37% for Hib meningitis have been reported in previous Gambian studies ( Bijlmer et al. 1990 ; O'Dempsey et al. 1996b ; Palmer et al. 1999 ). However, the outcome of children who leave hospital is unknown. Sequelae of meningitis have been documented extensively in industrialized nations ( Sell 1987), but little information is available from developing countries. The aim of this study was to assess the prevalence and severity of late sequelae in Gambian children after Hib and pneumococcal meningitis.
Patients and methods
Children were included in the study if they had been admitted between January 1990 and December 1995 to one of the 3 hospitals in the Western Region of The Gambia with culture-proven S. pneumoniae or Hib meningitis. The hospitals were the Royal Victoria Hospital (RVH), a tertiary referral hospital in the capital Banjul; the Medical Research Council (MRC) Laboratories hospital Fajara, a secondary care institution which serves a predominantly peri-urban population; and Sibanor Mission Hospital, a secondary care hospital which serves a mainly rural population. For logistical reasons, only children resident in the western part of The Gambia were studied. From March 1993 until December 1995 an efficacy trial of an Hib vaccine was conducted in The Gambia ( Mulholland et al. 1997 ) and the incidence of Hib meningitis decreased during that period. However, as only one of the patients reported in this paper had been vaccinated, the findings reported here are unlikely to have been influenced by this trial. Two patients had second episodes of pneumococcal meningitis proven by cerebrospinal fluid (CSF) culture. In these cases, only the second episode was considered in the analysis. Hospital notes were studied and information recorded is shown in Table 1. The hospital notes for 41 children could not be found but their immediate survival status was ascertained from ward discharge books. Addresses were classified as either rural or urban according to the 1993 national census ( Anonymous 1993).
Documentation sheet used for follow up of children who had meningitis
Cultures of CSF were performed at the MRC Laboratories or the government microbiology laboratory at RVH. When samples were collected at the rural health centre of Sibanor, they were transported in trans-isolate medium at ambient temperature to the MRC Laboratories. Sediment from a centrifuged specimen of CSF was cultured on blood agar (BA), MacConkey agar (MA) and vitox-enriched chocolate agar (CA) plates. Plates were incubated for 24–48 h at 35 °C in an aerobic atmosphere (BA and MA), an anaerobic atmosphere (BA) or in an incubator at a gas concentration of 5% CO2 (CA). Isolates from cultures were identified by standard methods as described previously ( Adegbola et al. 1994 , 1996). All bacterial isolates were tested for in vitro antibiotic disc sensitivity.
Antibiotic treatment varied from patient to patient but in most cases chloramphenicol (100 mg/kg/day parenterally) was the first antibiotic administered. In infants less than one month old this was augmented with gentamycin (5–7.5 mg/kg/day). When S. pneumoniae was isolated, the treatment was changed to benzyl penicillin (400 000 IU/kg/day). Parenteral treatment was given in all cases for 10–14 days. Corticosteroids were used in about a quarter of the patients, but dosages and duration of administration differed from patient to patient.
Survivors to discharge were sought between September 1996 and October 1997. Children were examined either at home or at the MRC outpatient department by a physician or a medical student with the assistance of a field worker to interpret. Parents or guardians were interviewed and the children underwent a physical and neurological examination ( Table 1). Vision was assessed by asking a child to name a common small object at a distance of 5 m; in the youngest children behaviour was observed when different objects were presented. Hearing was assessed by asking a child to count the number of claps heard from a distance of 5 m. The youngest children were tested through behavioural audiometry, i.e. their response to the sound of a rattle behind the head or to a call from the mother. Psychomotor development was assessed using Denver developmental screening test criteria as indicators.
Hearing, visual and motor functions were categorized as normal, minor or major impairment. Psychomotor development was categorized as normal, minor delay, or major delay. A major impairment or delay required a disability interfering with normal social life. The above findings were summarized into three categories: major, minor or nondetectable sequelae. Blindness or deafness, hemiplegia, and major psychomotor delay were considered as major sequelae while partial impairment of vision or hearing, minor psychomotor delay and minor disability were considered as minor sequelae. Age at the start of walking was analysed separately for those who started walking after the episode of meningitis. A delay in walking was defined as failure to walk at the age of 18 months.
Weight-for-age Z-scores were calculated using the United States National Centre for Health Statistics (NCHS) standard for both weight on admission and weight recorded on follow-up. Comparisons of continuous variables were made using Student's t-test. Proportions were compared using the chi-squared test or Fisher's exact test, as appropriate. Adjusted rate ratios were calculated using Poisson regression. Analyses were performed using the Epi-Info and Stata software packages. The study was approved by the Gambian Government/ MRC Ethical Committee.
During the six-year study period, 257 children were admitted with either pneumococcal (134) or Hib meningitis (123) to one of the study hospitals: RVH (166), MRC (74) or Sibanor hospital (17). Characteristics of patients on admission are shown in Table 2. There were no significant differences between cases of pneumococcal or Hib meningitis regarding sex, place of residence, length of hospital stay, or weight for age. Pneumococcal meningitis was significantly more frequent during the dry season (from December to May) than during the rainy season (June to November); no seasonality was found for Hib meningitis. Age distribution differed significantly between the two groups. A quarter of the cases of pneumococcal meningitis occurred during the first month of life and another quarter occurred after the age of one year, while 82% of cases of Hib meningitis were encountered in children between 2 and 12 months of age.
Some features of study children with pneumococcal or Hib meningitis
One child aged 2 months and two children aged 4 months with pneumococcal meningitis relapsed. One relapsed 7 days after discharge; he died rapidly with signs of meningitis but no CSF was collected. Another child relapsed 8 days after discharge with purulent meningitis but negative CSF culture; he survived the second episode with major handicaps (mental delay, visual impairment and hemiplegia). The third child relapsed 22 days after discharge with culture-positive pneumococcal meningitis; he died 17 days after the second admission. These three relapses occurred despite adequate antibiotic treatment given for 15 days or more during the first episode of meningitis. One more child had a second episode of culture-positive pneumococcal meningitis 10 months after the first and survived without handicap.
Mortality in hospital was significantly higher among children with pneumococcal meningitis (48%) than among those with Hib meningitis (27%) (P = 0.004). The acute mortality rate did not vary significantly with age. Children with pneumococcal meningitis who died were significantly more malnourished than those who survived. The mean (SD) weight for age Z-score on admission of children with pneumococcal meningitis who died was – 1.8 (1.5), compared with – 0.5 (1.4) for children who survived (P < 0.001). This effect was not observed among children with Hib meningitis. Death in hospital usually occurred shortly after admission; the median duration (interquartile range) of hospitalization before death was 1 day (0, 3), while the median duration of hospitalization of survivors was 12 days (10, 47).
Late deaths and sequelae
Of the 160 survivors to discharge 89 children could be traced. Of these 40 had suffered pneumococcal meningitis (25% of survivors to discharge) and 49 had had Hib meningitis (31% of survivors to discharge). Children were traced more frequently in rural areas (follow-up rate = 81.5%), compared to urban areas (39.1%). Nine children who had had pneumococcal and 7 who had had Hib meningitis died after discharge from hospital. After adjustment for length of follow-up, the mortality rates were 83 and 34 per 1000 child years of follow-up for children who had had pneumococcal or Hib meningitis, respectively (rate ratio = 2.43; 95%CI 0.81–7.67). For cases of pneumococcal meningitis, the median duration (interquartile range) of hospital stay was significantly longer for those who died after discharge: 19 days (15, 35) vs. 12 (10, 16) for long-term survivors (P = 0.006). Eight of the children who died (4 in each group) had sequelae at the time of discharge and their deaths were most probably related directly to their meningitis. Three additional children, all with Hib disease, recovered completely from meningitis but died from an acute febrile disease, possibly malaria. The cause of death was not known for the remaining 5 children.
The sequelae observed in children who survived are shown in Table 3. Five children were known by reliable sources to be alive but had moved outside the study area or were travelling temporarily. Weight for age at follow-up was significantly lower in the patients with pneumococcal meningitis than in Hib meningitis patients, after adjustment for length and season of follow-up. 58% and 38% of long-term survivors of pneumococcal and Hib meningitis, respectively, had sequelae. The occurrence of seizures at any time after discharge from hospital was more common after Hib meningitis. In contrast, all other handicaps were encountered more frequently after pneumococcal meningitis. Poisson regression was used to assess the effects of various factors on the risk of adverse outcome (defined as death or recovery with severe sequelae). The type of causative organism, season (dry or wet), weight for age on admission, age, sex, residence (urban or rural), and length of hospital stay were analysed. Only the type of causative organism was statistically significantly associated with adverse outcome (rate ratio for S. pneumoniae vs. Hib: 2.49; 95%CI 1.07, 5.76). Admission during the dry season was also associated with adverse outcome (rate ratio: 2.02; 95%CI 0.75, 5.44), but did not reach statistical significance.
Findings at late follow-up of children with pneumococcal or Hib meningitis discharged from hospital. Percentages calculated after excluding missing data
The results of this study of 257 children with acute bacterial meningitis confirm the high mortality of this condition noted previously in The Gambia as in other developing countries ( Baird et al. 1976 ; Grimwood et al. 1996 ; Greenwood 1984; Bohr et al. 1985 ; Bijlmer et al. 1990 ; Baraff et al. 1993 ; Palmer et al. 1999 ). Early deaths were almost twice as frequent in children with pneumococcal as in those who had Hib meningitis. The reasons of the higher mortality from pneumococcal meningitis are not known but it is a consistent finding throughout the world. High bacterial load and persistence of bacteria in the CSF for many days despite appropriate antibiotic therapy may be a factor ( Greenwood et al. 1986 ). This phenomenon occurred in our study in 3 children with pneumococcal meningitis who relapsed. The consistently high mortality noted in African patients with pneumococcal meningitis and an increased risk of invasive pneumococcal disease in black compared with white Americans raises the possibility that genetic factors may be involved ( Baraff et al. 1993 ; O'Dempsey et al. 1996a ). Poor nutritional status, which we found to be a risk factor for death from pneumococcal meningitis, may also be a contributory factor ( Baraff et al. 1993 ).
Little is known about what happens to children in developing countries who leave hospital after surviving an episode of acute bacterial meningitis. The results of our study are disquieting. 10% of children who survived an episode of pneumococcal meningitis died during the following months from a disease that was probably related to their initial illness. The length of hospital stay was significantly longer for those who died than for the survivors, and many of the fatal cases had sequelae before they died. Furthermore, only 33% of the children who survived S. pneumoniae meningitis and 53% of those who survived Hib meningitis were without sequelae at a later follow-up, even when only relatively insensitive clinical measures of neurological disease were used. A wide range of sequelae was found including impairment of vision, deafness, motor disabilities and psychomotor delay. Early sequelae of meningitis such as deafness can resolve during subsequent months. However, as we examined children 11–90 months after their illness, it is likely that the sequelae detected were permanent ( Sell 1983; Pomeroy et al. 1990 ). Assessment of mental development is difficult in developing countries because of problems with language and lack of appropriate test instruments. We did not investigate neuropsychological sequelae, but had such studies been done the proportion of unscathed survivors would likely have been reduced even further ( Bohr et al. 1985 ). Similarly, had it been possible to use a more sensitive test of auditory function, the number of partially deaf children would almost certainly have been greater. Even minor degrees of hearing impairment can have an adverse effect on learning ( Fortnum 1992) but are unlikely to be detected in many developing country situations.
One limitation of this study is that follow-up of only about 50% of hospital survivors was obtained, as addresses were not recorded for some children and colleagues had moved away; mobility is high in the peri-urban areas around the capital Banjul. However, the follow-up rate was significantly higher in rural areas, and therefore the findings in children from the countryside, which were not significantly different from those in urban areas, lend additional weight to the overall study findings.
Can anything be done to reduce the high prevalence of sequelae in children who survive acute bacterial meningitis? A number of risk factors for a poor outcome following acute bacterial meningitis have been described in industrialized and developing countries. These include the young age, a delay in starting treatment, presence of positive cultures of samples other than CSF, the occurrence of seizures and or malnutrition ( Lindberg et al. 1977 ; Hailemeskel & Tafari 1978; Akpede et al. 1999 ). Age was not identified as a risk factor for adverse outcome in our study. A delay in the initiation of treatment is difficult to overcome in situations where health facilities are few and transport limited. However, in the western half of The Gambia, hospitals are located within a distance of about 50 km of most of the population and transport facilities are relatively good. This is reflected in the fact that the average duration of symptoms prior to admission was 2.8 days for S. pneumoniae and 4.8 days for Hib meningitis. Improving the ability of peripheral health care staff to recognize the signs of meningitis and emphasizing the need to refer such children immediately to hospital, through educational measures such as the programme for the Integrated Management of the Sick Child ( WHO 1998a), might help to improve outcome. Most early deaths occur within 48 h of admission to hospital despite good nursing and parenteral treatment with an effective antibiotic. The efficacy of chloramphenicol alone in meningitis was shown in studies from Nigeria and Papua New Guinea ( Whittle et al. 1973 ; Shann et al. 1985 ). Isolates of S. pneumoniae remain sensitive to penicillin, and chloramphenicol resistance is uncommon (8% in a recent study, Adegbola et al. unpublished data). Few Hib isolates from CSF were resistant to ampicillin (0.3%) or chloramphenicol (0.5%) ( Adegbola et al. 1996 ). It is therefore unlikely that resistance of the bacteria contributed to the observed bad outcome. It has been shown that dexamethasone reduces the incidence of deafness in children with Hib meningitis in industrialized countries ( Lebel et al. 1988 ) and there is some evidence that this drug reduced mortality and sequelae from meningitis in children in Brazil ( Macaluso et al. 1996 ) and mortality from pneumococcal meningitis in Egypt ( Girgis et al. 1989 ). Few children in the present study received corticosteroids and it can be argued that the incidence of sequelae might have been reduced had this been done. There is a need for a large-scale randomised trial of the effects of steroids on the outcome of bacterial meningitis in children in developing countries ( WHO 1998b). Because dexamethasone is expensive, glycerol is being advocated as a possible cheaper alternative ( Kilpi et al. 1995 ). Vaccination offers a way forward. A Hib conjugate vaccine has been dramatically effective in reducing the incidence of Hib meningitis in The Gambia and should now be incorporated into the EPI programme of most developing countries ( Mulholland et al. 1997 ; Adegbola et al. 1999 ). Unfortunately, financial constraints are making this difficult to achieve. Good progress is being made in the development of pneumococcal conjugate vaccines ( Mbelle et al. 1997 ) and such vaccines are needed urgently if the drastic consequences of pneumococcal meningitis in children in developing countries are to be prevented.
In conclusion, Hib and pneumococcal meningitis are responsible for high mortality rates in developing countries, and many of the survivors are affected with sequelae. As improved case management of established meningitis is likely to have a limited effect, increased efforts for prevention by vaccination are the hope for the future.
We acknowledge the help of Jennifer Horovicz, Dr Stephen Hilton, Dr Stanley Usen, and Dr Steven Obaro in tracing patients. Amy Jobateh and Pam Njai helped with retrieving patient notes; Dr Gisela Schneider, Dr Ayo Palmer, Dr Mariatou Jallow and Dr Tumani Corrah were in charge of the clinical treatment of the study patients. We thank the field workers Malick John, Foday Drammeh, Saihou Mendy, Saihou Badgie and Fatou Bah for tracing patients and helping in translation. We are very grateful also to the parents of the children who agreed to be interviewed and to the children who were examined for the purpose of the study. We also acknowledge Jane Rowley and Arnaud Marchant for critical review of the manuscript. The study was supported by Ciba Geigy and Glaxo Wellcome through their Gambian elective student program.