- Top of page
- Materials and methods
Background Over the last decade Streptococcus pneumoniae has emerged as the most common bacterial pathogen for meningitis in all age groups, beyond the neonatal period.
Objective To determine the epidemiological and clinical characteristics; and risk factors for mortality of pneumoccocal meningitis in children in a developing transitional country.
Materials and methods A retrospective study that included patients <15 years of age admitted at the Instituto de Medicina Tropical of Paraguay, from January 1990 until December 2003 with the diagnosis of bacterial meningitis caused by S. pneumoniae. Clinical and laboratory data were collected and analysed in order to identify risk factors associated with morbidity and mortality outcomes of this infection.
Results Seventy-two patients (between the ages of 35 days and 14 years) were identified. Forty-two per cent of patients had seizures prior to or at the time of admission, 36% were admitted in a comatose state, and 19% with shock. Mortality was 33% (24/72), and 18% of the survivors (11/60) developed severe sequelae. Upon admission, the following variables were strongly correlated with mortality: age <12 months (P = 0.007), the presence of seizures (P = 0.0001) or development of seizures 48 h after admission (P = 0.01), a cerebrospinal fluid (CSF) glucose level of <10 mg/dl (P = 0.01), CSF albumin >200 mg/dl (P = 0.0003), an absolute blood neutrophil count <2000/mm3 (P = 0.006) and a haemoglobin value of <9 g/dl (P = 0.0001).
Conclusions This study confirms the high morbidity and mortality associated with S. pneumoniae meningitis in Paraguay. Certain clinical parameters and laboratory findings in blood and CSF at the time of admission could be used as predictors for mortality or severe sequelae among survivors.
- Top of page
- Materials and methods
Bacterial meningitis is one of the most severe forms of infection of the central nervous system, which can progress rapidly and result in death or permanent sequelae of the patient (Feigin et al. 1992). After the introduction of the vaccine against Haemophilus influenzae type b (HiB), Streptococcus pneumoniae has emerged as the main etiologic pathogen for meningitis in all age groups (except in neonates) in countries that have implemented routine anti-HiB vaccination (Schuchat et al. 1997; Peltola 2000). The development of broad-spectrum antibiotics with higher activity against various microorganisms, including S. pneumoniae, and the introduction of improved supportive care in intensive care units, has led to the transformation of bacterial meningitis from an incurable to a curable disease (Weiss et al. 1967; Baird et al. 1976). However, morbidity and mortality of meningitis due to S. pneumoniae still remain unacceptably high (Arditi et al. 1998; Fiore et al. 2000; Buckingham et al. 2001; Kellner et al. 2002).
Although many studies regarding pneumococcal meningitis have been published, the majority of them have been carried out in developed countries. There is scarce information about this entity in developing countries, and less from countries with developing transitional economies, defined as countries with lower-middle income economies of the World Bank list (Gomez-Barreto et al. 1999; Asturias et al. 2003). The identification of risk factors for death or severe sequelae in patients with pneumococcal meningitis at the time of admission or developed during the course of the disease provides invaluable information for clinicans. Furthermore, the identification of predictors of unfavourable outcomes in our population may facilitate maximizing therapeutic efforts in these patients.
- Top of page
- Materials and methods
Seventy-seven children with pneumococcal meningitis were identified during the study period. Streptococcus pneumoniae was isolated in CSF or blood culture in 55 cases (71%), whereas 22 patients (29%) were diagnosed through observation of gram (+) diplococcus in CSF and/or through a positive test for S. pneumoniae antigen in a sample of CSF.
Five patientes were excluded from the analysis due the presence of one pre-disposing condition [previous episodes of bacterial meningitis, congenital hydrocephalus, cyanotic congenital cardiopathy, ethmoidal fistula and cranioencephalic trauma (one patient for each condition)] that could affect the interpretation of the results, mainly the evaluation of sequelae. The mean age of cases was 48 ± 37 months, with a range of 35 days to 15 years. More than the half of the cases (58%) ocurred under the age of 24 months, and 29% of the patients were older than 5 years. Only three cases were observed in children less than 2 months of age. There was no predominance of sex (53% males; 47% females). Hospital admission occurred after a mean duration of symptoms and signs of 4.1 ± 3.4 days. Classical symptoms of bacterial meningitis such as fever, vomiting and headache were observed in 95%, 66% and 32% of cases, respectively. Eighteen patients exhibited ≥ grade I malnutrition, 15 (26% of the total group) of whom were severely malnourished (Gomez grade II or III). Seizures were present in 30 patients (46%) prior to or at admission, and among 27.6% of patients who survived >48 h (18/65). Twenty-six children (27%) were admitted in severe coma. Shock was present on admission in 14 cases (19%).
Analysis of CSF upon admission revealed the presence of severe hypoglycorrhachia (<10 mg/dl), pleocytosis >1000 cells/μl and albumin levels >200 mg/dl in 46 (64%), 28 (39%) and 41 patients (57%), respectively. Eight patients (11%) were admitted with a blood neutrophil level of <2000/mm3 and moderate to severe anemia [haemoglobin (Hb) <9 g/dl] was found in 26 cases (36%).
There were 52 strains of S. pneumoniae available for susceptibility study. Fifty-one (98%) were susceptible to penicillin. Streptococcus pneumoniae resistant to penicillin was isolated in one case. No strain was resistant to cefotaxime by Kirby-Bauer disk susceptibility test.
All patients were treated with cefotaxime (n = 61) or penicillin (n = 11), with a mean of 12 days (range: 10–28 days). In three cases vancomycin was added to treatment (all in the last 12 months of the study).
Of the 72 cases studied 24 (33%) died. Of the survivors (48 patients, 67%), 37 (77%) were discharged without or with mild sequelae (favourable outcome), and 11 (23%) had severe sequelae. The mean hospital stay for survivors was 12.3 ± 8.7 days (range: 10–41 days).
Table 1 compares clinical characteristics of patients who died or survived with severe sequelae (unfavourable outcome), with patients who had favourable outcomes. Although the duration of symptoms and/or signs before admission was similar in both groups (3.7 ± 2.9 vs. 4.5 ± 3.9 days, respectively) as well as the presence of malnutrition, the mean age of the patients with unfavourable outcome was significantly lower (30.7 ± 48.5 months vs. 69.3 ± 61.2 months, respectively, P < 0.05). Similarly, patients with unfavourable outcomes were admitted with conditions that imply a higher severity of the disease, such as coma or shock (P = 0.1), seizures (60%vs. 24%; P < 0.01) and more commonly experienced seizures >48 h after admission (50%vs. 11%; P < 0.01).
Table 1. Clinical and laboratorial characteristics of patients with pneumococcal meningitis in children who died or had severe sequelae in Asuncion, Paraguay 1990–2003
|Variable||Dead or with severe sequelae (n = 35)||Healthy or with minor sequelae (n = 37)|
|Age (months)||30.7 ± 48.5*||69.3 ± 61.2|
|Duration of symptoms/signs before admission (days)||3.7 ± 2.9||4.5 ± 3.9|
|Number of patients with grade I malnutrition or more||12 (34%)||6 (16%)|
|Seizures at admission||21 (60%)**||9 (24%)|
|Severe coma at admission||17 (49%)***||9 (24%)|
|Shock at admission||10 (29%)||4 (11%)|
|Seizures >48 h after admission||14/28 (50%)*||4 (11%)|
| Haematologic data|
| Hb (g/dl)||8.9 ± 1.9*||10.8 ± 2.1|
| White cells count (mm3)||13 281 ± 10 758*||19 175 ± 110 148|
| Pleocitosis (mm3)||1124 ± 2511||1574 ± 1419|
| Glucose (mg/dl)||8.4 ± 15.3**||25.4 ± 25.6|
| Albumin (mg/dl)||364 ± 255*||239 ± 209|
Analysis of laboratory data disclosed interesting findings between both groups of patients (Table 1). Upon admission, patients with unfavourable outcomes had lower levels of Hb (8.9 ± 1.9 g/dl vs. 10.8 ± 2.1 g/dl; P < 0.05) and WBC (13 281 ± 10 758 mm3vs. 19 174 ± 10 148/mm3; P < 0.05) than the children who were discharged with mild or no sequelae. Furthermore, findings in CSF were identified as predictors of patient's outcome. Thus, mean CSF glucose levels in patients with unfavourable outcomes were significantly lower (8.4 ± 15.3 g/dl vs. 25.4 ± 25.6 g/dl; P < 0.01), and the albumin levels in CSF were significantly higher (364 ± 255 mg/dl vs. 239 ± 209 mg/dl; P < 0.05); however, there was no significant difference in the degree of pleocytosis between both groups.
Table 2 illustrates clinical and laboratory patient data, which were analysed in order to determine which factors identified a priori predicted the prognosis of pneumococcal meningitis. Among the clinical factors, age <12 months (OR: 5.77; 95% CI: 1.82–18.92; P = 0.0007), seizures at admission (OR: 4.66; 95% CI: 1.53–14.63; P = 0.002) or after 48 h post-hospitalization (OR: 8.25; 95% CI: 2.03–39.21; P = 0.0005) and severe coma (OR: 2.94; 95% CI: 0.97–9.11; P = 0.03) proved to be stronger predictors of unfavourable outcome. In contrast, the presence of shock at admission and clinical symptoms >48 h before hospitalization were not identified as predictors of unfavourable outcomes.
Table 2. Pneumococcal meningitis in children: risk factors for mortality or severe sequelae in Asuncion, Paraguay 1990–2003
|Risk factors||Dead or with severe sequelae (n = 35)||Heathy or with minor sequelae (n = 37)|| OR (95% CI)*||P value|
|Age <12 months||25||13||5.77 (1.82–18.92)||0.0007|
|Severe coma at admission||17||9||2.94 (0.97–9.11)||0.03|
|Shock at admission||10||4||3.30 (0.81–15.87)||0.06|
|Seizures at admission||21||9||4.66 (1.53–14.63)||0.002|
|Symptoms/signs >48 h before admission||24||26||0.92 (0.30–2.83)||0.87|
|Previous antibiotic therapy||9||14||0.56 (0.18–1.74)||0.27|
|Weight deficit for age >10%, <25%||3||4||0.77 (0.11–4.98)||0.74|
|Weight deficit for age >25%||6||5||1.32 (0.29–6.09)||0.67|
|Seizures 48 hrs after admission||14/28*||4||8.25 (2.03–39.21)||0.0005|
|Fever >5 days after treatment||9||6||1.78 (0.48–6.91)||0.32|
|Pleocytosis in CSF <1000 cells/mm3||26||18||3.05 (1.02–9.39)||0.02|
|Albumin in CSF >200 mg/dl||27||14||5.54 (1.78–17.92)||0.0008|
|Glucose in CSF <10 mg/dl||29||17||5.68 (1.72–20.37)||0.001|
|Neutrophils in blood <2000 cells/mm3||7||1||9.00 (1.03–416.5)||0.019|
|Hb <9 g/dl||20||6||6.28 (2.06–24.86)||0.0003|
When the death variable was analysed alone (Table 3), aside from age <12 months (P = 0.007), presence of coma (P = 0.005) and seizures at admission (P = 0.0001); conditions significantly associated with mortality also included: CSF pleocytosis <1000/mm3 (P = 0.05), albumin levels in CSF >200 mg/dl (P = 0.0003), and hypoglycorrhachia <10 mg/dl (P = 0.01). In addition, a Hb <9 g/dl (P = 0.0001) and an absolute blood neutrophil count <2000/mm3 (P = 0.0006) were strongly associated with increased mortality.
Table 3. Pneumococcal meningitis in children: risk factors for mortality
|Risk factors||Dead (n = 24)||Survivors (n = 48)|| OR (95% CI)*||P value|
|Age <12 months||17||18||4.04 (1.26–13.68)||0.007|
|Severe coma at admission||14||12||4.20 (1.31–13.55)||0.005|
|Shock at admission||7||7||2.41 (0.61–9.36)||0.14|
|Seizures at admission||18||12||9.00 (2.58–33.42)||0.0001|
|Symptoms/signs >48 h before admission||20||30||3.00 (0.81–13.81)||0.07|
|Previous antibiotic therapy||7||16||1.57 (0.21–10.16)||0.72|
|Weight deficit for age >10%||3||4||1.84 (0.39–8.22)||0.57|
|Weight deficit for age >25%||5||6||4.90 (1.10–21.59)||0.35|
|Seizures 48 h after admission||7/17*||6||4.90 (1.10–21.59)||0.01|
|Fever >5 days after treatment||4||10||0.76 (0.15–3.08)||0.67|
|Pleocitosis in CSF <1000 cells/mm3||19||27||2.95 (0.86–11.67)||0.05|
|Albumin in CSF >200 mg/dl||22||23||11.95 (2.41–112.58)||0.0003|
|Glucose in CSF <10 mg/dl||20||26||4.23 (1.15–19.23)||0.01|
|Neutrophils in blood <2000 cells/mm3||7||1||19.35 (2.13–889.94)||0.0006|
|Hb <9 g/dl||16||10||7.60 (2.24–26.44)||0.0001|
- Top of page
- Materials and methods
Worldwide S. pneumoniae is the third most common etiologic cause of meningitis in children and the first cause of meningitis in adults (Peltola 1999). However, in countries that have already implemented universal vaccination against HiB, S. pneumoniae is the main etiologic agent of bacterial meningitis in all age groups (Peltola 1999), with a calculated annual incidence of 2–11 cases per 100 000 children under the age of 24 months (Eskola et al. 1992; Kaplan et al. 1998).
The purpose of this study was to examine the characteristics of pneumoccocal meningitis in a major referral institute for infectious diseases in Paraguay, a representative Latin American country with a transitional economy. In this study, we found that the age distribution of cases presented a bimodal pattern. Although cases were most commonly identified in children less than 24 months of age (58%), a new increment was seen in the group of children older than 5 years (29%). A similar distribution pattern was recently observed in USA. Schuchat et al. (1997) found an increased incidence of bacterial meningitis in the group of children <24 months, with a decline beyond this age and a new gradual increment towards adolescence in an active surveillance study of invasive pneumococcal infection. In a similar manner, more than 70% of patients with pneumococcal meningitis identified by Kaplan et al. (1998) over a 3-year surveillance period (1996–1999) and by Buckingham et al. (2001) were younger than 24 months of age.
Patients with unfavourable outcomes (died or discharged with severe sequelae) had some clinical and laboratory features that differ from those of patients discharged with favourable outcomes (with or without minor sequelae). Patients with unfavourable outcomes were significantly younger (mean age 30.7 vs. 69.3 months, P < 0.05), more frequently had seizures at time of admission (60%vs. 24%, P < 0.01), or after 48 h of treatment (50%vs. 11%, P < 0.05), and had a tendency to be admitted more frequently in a severe comatose state (P < 0.1). These findings are consistent with previous reports (Weiss et al. 1967; Baird et al. 1976; Kornelisse et al. 1995; Arditi et al. 1998; Fiore et al. 2000; Goetghebuer et al. 2000; Saez-Llorens & McCracken 2003). Similarly, as it occurs in cases of meningitis due to other bacterial pathogens such as HiB, patients who died or had severe sequelae were younger and had a higher frequency of seizures and shock than patients with favourable outcomes (Herson & Todd 1977; Basualdo & Arbo 2004).
Regarding CSF findings, patients with unfavourable outcomes exhibited a clearly lower mean glucose concentration (8.4 vs. 25.4 mg/dl, P < 0.01) and significantly higher albumin levels (364 vs. 239 mg/dl, P < 0.05) than patients with favourable outcomes. Since the grade of hypoglucorrachia and elevated album levels reflect the magnitude of inflammation that occurs at the subarachnoid space, these findings are not surprising and suggest that the intensity of the inflammatory cascade in meningitis results in more deleterious neurological morbidity (Saez-Llorens & McCracken 2003; van der Flier et al. 2003).
Our study focused on determining the risk factors for mortality in cases of pneumococcal meningitis as well as identifying severe sequelae among survivors. Of the different factors studied, the most specific prognostic factors in patients who died or had severe sequelae at discharge were: age <12 months, seizures before/at admission or its presence after 48 h of hospitalization, and severe coma. The fact that young age is a risk factor for poor outcome could be due to immaturity of the immune system during the first months of life (Kovarik & Siegrist 1998). In addition, a delay in establishing a diagnosis of bacterial meningitis at younger ages, when classic manifestations of this infection are not classically present, could be considered a contributing pathogenic factor (Herson & Todd 1977). Furthermore, the presence of seizures during the course of bacterial meningitis can result in significant neuronal damage during the inflammatory process. Necropsy studies in patients who die or have severe sequelae due to meningitis frequently show an impaired vascular flow in different areas of the brain secondary to vasculitis. They also show cerebral edema and neuronal necrosis (Saez-Llorens & McCracken 2003; van der Flier et al. 2003).
Various prognostic factors were identified when the death variable was independently analysed. In this case, aside from age <12 months, and the presence of seizures prior to/at admission, some laboratory characteristics at admission displayed stronger associations with mortality, such as: hypoglycorrhachia of <10 mg/dl, pleocytosis<1000/mm3, absolute blood neutrophil count <2000, and a Hb<9 g/dl. Several previous studies have pointed out the value of a pleocytosis of <1000/mm3 in CSF, a low glucose level in CSF and elevated album levels as prognostic factors in bacterial meningitis (Weiss et al. 1967; Fiore et al. 2000; Goetghebuer et al. 2000; Buckingham et al. 2001). In addition, they have all been confirmed as risk factors in this study. However, the association of a low blood neutrophil count with higher mortality in pneumococcal meningitis has not been previously reported. Given that neutrophils represent the first line of defense against bacterial pathogens (Malech & Gallin 1987), it is not surprising that the presence of an inadequate leukocytic response could produce a lower capacity for microorganism clearance, which results in the subsequent severity of infection.
A correlation of unfavourable prognosis with a low Hb level could relate in part to the severity of infection, as has been observed in cases of infection due to HiB (Herson & Todd 1977). We can speculate that the higher the bacterial count, the higher the probability that the pneumococcal infection will be accompanied by a haemolytic component. This may be attributed to the fact that S. pneumoniae is a powerful activator of the complement cascade (Tuomanen et al. 1986) or simply to the low level of Hb levels could reflect a higher representation of children younger than 12 months in the group that died or remained with severe sequelae.
In summary, we were able to identify specific clinical and laboratory predictors for increased risk of mortality or development of severe neurologic sequelae among survivors of pneumococcal meningitis. These results may provide clinicians with the opportunity to incorporate these risk factors in their decision-making process in order to institute aggressive therapeutic measures in the group of patients with higher predicted risk of unfavourable outcomes. More importantly, we should recognize that improving case management of established bacterial meningitis patients is likely to have a limited effect in improving the outcome. Therefore, increased efforts for the institution of preventative vaccination programmes with the currently available pneumococcal conjugate vaccines are urgently needed in countries such as ours.