Prognostic factors and determinants of fatal outcome due to bacterial meningitis in the Lazio region of Italy, 1996–2000

Article Outline

• Summary
• Introduction
• Methods
  • Case definition
  • Control definition
  • Factors analyzed and data sources
  • Data analysis
• Results
• Discussion
• Conclusions
• Acknowledgements
• References
• Copyright

Summary 

Objectives

To estimate case fatality rates (CFR) of bacterial meningitis and analyze factors associated with mortality due to bacterial meningitis in the Italian region of Lazio.

Methods

Patients reported with bacterial meningitis during the period 1996–2000, who died within 30 days from hospitalization (cases), were compared with survivors (controls) for factors related to healthcare. Age, gender, residence, bacterial agent, co-morbidities, and signs of disease severity were also analyzed in the final model. Healthcare factors were analyzed using current surveillance databases.

Results

Disease severity (OR=8.84; 95% CI=3.35–23.34) and age >44 years (OR=4.59; 95% CI=2.01–10.48) were the risk factors most strongly associated with death, while treatment in an infectious diseases ward was a protective factor, although modified by patient residence and by co-morbidities.

Conclusions

This protective effect was possibly due to differences in treatment protocols between the infectious diseases ward and other wards. The protective effect was found to be stronger for residents of Rome, suggesting delayed access to infectious diseases wards for non-residents. The difference in risk of dying from meningitis at younger ages than that found in other studies should be further evaluated, using information on bacteria serogroups and antibiotic susceptibility.

Keywords: Bacterial meningitis, Fatal outcome determinants, Healthcare indicators

Back to Article Outline

Introduction 

Bacterial meningitis is a public health problem, as despite progress in antibiotic therapy, mortality rates have not decreased in recent decades.¹, ² Age is the strongest predictor of bacterial meningitis mortality.³, ⁴, ⁵, ⁶ Other factors influencing mortality are the etiological agent, sepsis, the time elapsed to sterilize CSF cultures,³, ⁷, ⁸ and other clinical features such as altered mental status, hypotension, and seizures.², ⁴, ⁶

Co-morbidities such as diabetes, cancer,⁵ and chronic debilitating diseases³, ⁴ have been reported to be negative prognostic factors for bacterial meningitis, but one study has failed to show the association of these diseases with fatal outcomes.⁶ Mortality rate studies have shown the importance of co-morbidities in causing death in patients with infectious diseases. Other authors have maintained that co-existing chronic conditions contribute to most deaths occurring in patients with bacterial infections when the underlying cause was coded as an infectious disease.⁹

A few papers have studied factors related to the management of meningitis patients, such as type of treating hospital¹⁰ and antimicrobial therapy. The timing of appropriate therapy and pre-admission use of antibiotics emerged as the most important factors associated with fatal outcome.², ⁶, ⁸ Although no randomized trials have been performed on the effect of pre-hospital antibiotic treatment, recommendations have been implemented suggesting immediate treatment with benzylpenicillin for any suspected case of bacterial meningitis in adults.¹¹

We conducted a study of risk factors for death in patients with bacterial meningitis, carried out in the Lazio region of Italy from 1996 to 2000, using surveillance databases.

Back to Article Outline

Methods 

A case–control study was designed that enrolled all patients reported to the surveillance system with bacterial meningitis from 1996 to 2000. All meningococcal diseases were included, both meningitis and sepsis, according to surveillance definitions; meningitis due to Mycobacterium tuberculosis was excluded, as well as patients with bacterial meningitis from outside the region.

Case definition 

Death due to bacterial meningitis was defined as a confirmed or probable case of bacterial meningitis reported to the Lazio surveillance system from 1996 to 2000, registered as deceased at the General Registry Office, and who died during the hospitalization in which meningitis was diagnosed, or at home within 30 days from hospital admission with bacterial meningitis listed as cause of death.

A confirmed case of meningitis was defined as a patient with clinical signs such as fever and/or headache and/or neck stiffness with a positive culture or antigen latex agglutination test of cerebrospinal fluid (CSF) or blood. Gram-negative results from microscopic examination were also accepted as meningococcal disease. A probable case of bacterial meningitis was defined as a patient with the above-mentioned clinical signs and polymorphonuclear pleocytosis in the CSF (1000/mm³ leukocytes or more), without bacterial growth in blood or CSF cultures. We excluded meningitis cases that were not confirmed by the laboratory.

We defined the meningitis as nosocomial when the patient had already been hospitalized for other causes in the 1–25 days before the onset of symptoms; if the patient was not hospitalized during this period, we defined the disease as community meningitis.

In the Hospital Discharge Abstract Registry (HDAR), the index hospitalization was defined as the admission during which the meningitis was diagnosed. Information on intra-hospital deaths was obtained, after considering patient transfers. The Nominative Cause Mortality Registry (NCMR) was used to detect deaths from bacterial meningitis outside the hospital. The International Classification of Diseases, Ninth Revision (ICD-IX) codes we used were 027.0, 036.0, 036.1, 036.2, 320.0, 320.1, 320.2, 320.3, 320.7, 320.8, and 320.9.

Control definition 

We selected as controls all the confirmed or probable cases of bacterial meningitis who were discharged alive following the hospitalization in which meningitis was diagnosed and survived more than 30 days after hospital admission. Controls had to meet the same criteria as cases to be included: they had to be residents in the Lazio region and had to be reported to the surveillance system during the same period of time.

Factors analyzed and data sources 

The risk factors we studied related to patient management. We did not hypothesize which of them had the greatest impact.

Among those related to the hospital structure we analyzed hospital classification, presence of an infectious diseases ward, and hospital location. Hospital classification was based on the total number of diagnosis-related groups (DRGs) in one year and the number of surgical and medical specialty services offered by each (Regional Law No. 2069/1999), class 1 being those with the highest number of different DRGs discharged and the most specialty services offered.

Among factors related to the quality of healthcare, we analyzed timing of the hospitalization, patient transfer during the illness, and the characteristics of hospitals involved. The timing of the hospitalization was defined as the time elapsed between the onset of symptoms, as reported on the notification form, and the first hospitalization. Patients who developed meningitis during a hospitalization for other causes were assigned zero as time elapsed from the onset of symptoms.

We selected the following factors as possible confounders of the association between healthcare received and mortality due to bacterial meningitis: bacterial agent, disease severity, year of notification, residence, and patient characteristics such as gender, age, and co-morbidities.

Co-morbidities analyzed included diagnoses of cirrhosis cancer, acquired immunodeficiency (HIV-AIDS), immunodeficiency due to absence or extremely low levels of serum globulins, autoimmune diseases, chronic renal failure, congenital or hypoplastic neutropenia, diabetes, and hemorrhagic diathesis only in meningococcal disease. Both index and previous hospitalizations were used to detect co-morbidities.

Disease severity was reported only from the index hospitalization, and was defined by the presence of shock and/or focal neurological signs, such as seizures, and/or pneumonia if meningitis was due to Streptococcus pneumoniae.

Data analysis 

The association between death and each factor was estimated through odds ratios (ORs) and 95% confidence intervals (CIs). Unconditional logistic regression was carried out, which focused on healthcare-related factors as exposure variables and assessed the modifying and the confounding effect of factors related to infection and patient characteristics such as age, gender, residence, bacterial agent, co-morbidities, and disease severity. The multivariable analysis included age and gender along with any other variable associated with a p value of 0.1 or less. The timing of hospitalization was excluded from the model because of the high percentage of missing values.

Some of the variables were reclassified into groups: age into three groups: 0–4, 5–44, and ≥45 years; residence into two: inside and outside Rome; bacterial agent into Neisseria meningitidis, Streptococcus pneumoniae, other bacteria, and not identified; hospital class into two groups, the first included the 1st and 2nd classes, and the second group included the 3rd and 4th.

We used the likelihood ratio test (LRT) to select the final models. Finally, separate models were built excluding disease severity, the strongest predictor of fatal outcome, to test if it interacted with the other determinants. Stata software (Version 7, Stata Corporation, Texas) was used for data analysis.

Back to Article Outline

Results 

Six hundred and two cases of bacterial meningitis were reported in the Lazio region from 1996 to 2000; 525 were resident in the region (population 5209486), 77 were resident outside the region. Of the 525 resident cases, 404 were confirmed and 121 were probable cases. There were 98 patients who died from bacterial meningitis according to our case definition, 95 of whom died in hospital and three at home. The information about in-hospital death was obtained from the HDAR in 94 cases, and from notifications in one case for which the index hospitalization was not found. Among the three cases that died at home, only two had been previously hospitalized. The index hospitalization was identified for 466 (89%) patients, 96 of whom had died. Information about hospitalizations for an additional 54 cases was obtained from notifications.

Univariable analysis (Table 1) showed that the risk of dying increased with age, doubling the ORs in each age group. No important differences were detected between males and females. The risk of dying was higher when the disease was due to Listeria (OR=3.04; 95% CI=1.09–8.46) or Streptococcus pneumoniae (OR=2.45; 95% CI=1.23–4.88), when shock was among the clinical signs (OR=11.84; 95% CI=3.68–38.10), and when co-morbidities were present (OR=3.49; 95% CI=1.97–6.19). Among the associated healthcare-related factors, being treated in a hospital with an infectious diseases ward, both at the first (OR=0.43;95% CI=0.28–0.69) and last hospitalization (OR=0.12; 95% CI=0.06–0.22), had protective effects; transfer after the first admission increased the risk (OR=1.49; 95% CI=0.95–2.35), especially when both were Roman hospitals in the highest class. The timing of hospitalization was not associated with risk of death, but the percentage of missing information is very high (41.8%) for this variable. None of the 14 patients with nosocomial infections (two had been hospitalized 65 and 83 days before) died. When only community meningitis cases were analyzed the estimate of fatal outcome decreased from 0.72 to 0.58 (OR=0.58; 95% CI=0.28–1.23) in patients hospitalized 1–3 days after the onset of symptoms compared with those admitted the same day.

Table 1. Factors associated with death in meningitis cases, Lazio, Italy, 1996–2000

	Patients					OR	95% CI
	Total	Dead		Alive
		n	%	n	%
Factor of infection and patient characteristics
Total	525	98		427
Gender
Female	238	42	42.9	196	45.9	1
Male	287	56	57.1	231	54.1	1.13	0.73–1.76
MI	0
Age group (years)
0–4	164	12	12.2	152	35.6	1
5–24	93	9	9.2	84	19.7	1.36	0.55–3.35
25–44	87	14	14.3	73	17.1	2.43	1.07–5.52
45–64	97	26	26.5	71	16.6	4.64	2.21–9.72
>64	83	37	37.8	46	10.8	10.19	4.91–21.14
MI	1	0		1	0.2
Residence
Rome	302	64	65.3	238	55.7	1
Lazio, outside Rome	223	34	34.7	189	44.3	0.67	0.42–1.06
107–20000 inhabitants	108	14	14.3	94	22.0	0.55	0.30–1.04
>20000 inhabitants	115	20	20.4	95	22.3	0.78	0.45–1.36
Notification year
1996	105	19	19.4	86	20.1	1
1997	86	14	14.3	72	16.9	0.65	0.34–1.26
1998	103	15	15.3	88	20.6	0.58	0.28–1.18
1999	120	22	22.4	98	23.0	0.51	0.25–1.01
2000	111	28	28.6	83	19.4	0.67	0.35–1.25
Bacterial agent
Neisseria meningitidis	97	13	13.3	84	19.7	1
Haemophilus influenzae	72	5	5.1	67	15.7	0.48	0.16–1.42
Streptococcus pneumoniae	142	39	39.8	103	24.1	2.45	1.23–4.88
Streptococcus	28	8	8.2	20	4.7	2.58	0.94–7.07
Staphylococcus	14	4	4.1	10	2.3	2.58	0.71–9.47
Listeria	25	8	8.2	17	4.0	3.04	1.09–8.46
Other bacteria	26	3	3.1	23	5.4	0.84	0.22–3.21
Not isolated	121	18	18.4	103	24.1	1.13	0.52–2.44
Focal neurological signs without co-morbidities
No	458	93	94.9	365	85.5	1
Yes	8	3	3.1	5	1.2	2.35	0.55–10.03
MI	59	2	2.0	57	13.3
Shock without co-morbidities
No	451	85	86.7	366	85.7	1
Yes	15	11	11.2	4	0.9	11.84	3.68–38.10
MI	59	2	2.0	57	13.3
Pneumonia without co-morbiditiesa
No	125	37	94.9	88	85.4	1
Yes	5	2	5.1	3	2.9	1.59	0.25–9.88
MI	12	0		12	11.7
Any serious disease sign without co-morbidities
No	438	80	81.6	358	83.8	1
Yes	28	16	16.3	12	2.8	5.97	2.72–13.10
MI	59	2	2.0	57	13.3
Co-morbiditiesb
No	418	71	72.5	347	81.3	1
Yes	60	25	25.5	35	8.2	3.49	1.97–6.19
MI	47	2	2.0	45	10.5
Healthcare management factors
Total	525	98		427
Etiological diagnosis
Yes	402	79	80.6	323	75.6	1
No	123	19	19.4	104	24.4	0.75	0.43–1.29
MI	0
Classification of first hospital
1st class	283	46	46.9	237	55.5	1
2nd class	166	35	35.7	131	30.7	1.38	0.84–2.24
3rd class	49	12	12.2	37	8.7	1.67	0.81–3.45
4th class	22	5	5.1	17	4.0	1.51	0.53–4.31
MI	5	0		5	1.2
Site of first admission
Rome	372	74	75.5	298	69.8	1
Lazio, outside Rome	148	24	24.5	124	29.0	0.78	0.47–1.29
MI	5	0		5	1.2
Infectious diseases service in the first hospital
No	164	46	46.9	118	27.6
Yes	356	52	53.1	304	71.2	0.43	0.28–0.69
MI	5	0		5	1.2
Classification of last hospital
1st class	333	61	62.2	272	63.7	1
2nd class	168	30	30.6	138	32.3	0.97	0.60–1.57
3rd class	7	3	3.1	4	0.9	3.34	0.73–15.33
4th class	2	2	2.0	0		NC
MI	15	2	2.0	13	3.0
Site of last admission
Rome	424	83	84.7	341	79.9	1
Lazio, outside Rome	86	13	13.3	73	17.1	0.73	0.39–1.38
MI	15	2	2.0	13	3.0
Infectious diseases service in the last hospital
No	43	26	26.5	17	4.0	1
Yes	467	70	71.4	397	93.0	0.12	0.06–0.22
MI	15	2	2.0	13	3.0
Hospital transfer
No	355	59	60.2	296	69.3	1
Yes	170	39	39.8	131	30.7	1.49	0.95–2.35
Kind of hospital transferc
No transfer	355	59	60.2	296	69.3	1
From class 2 to class 1	60	13	13.3	47	11.0	1.39	0.71–2.72
From class 1 to class 1 in Rome	73	20	20.4	53	12.4	1.89	1.05–3.40
From class 1 outside Rome to class 1 inside Rome	23	2	2.0	21	4.9	0.48	0.11–2.09
MI	14	4	4.1	10	2.3
Timing of hospitalizationd
Within 24h	93	18	18.4	75	17.6	1
1–3 days	170	25	25.5	145	34.0	0.72	0.37–1.40
4–30 days	65	14	14.3	51	11.9	1.14	0.52–2.50
MI	197	41	41.8	156	36.5

MI, missing information. NC, not calculable.

The multivariable analysis (Table 2) showed that disease severity was the most important factor associated with fatal outcome (OR=8.84; 95% CI=3.35–23.34). Other risk factors were being over 44 years of age (OR=4.59; 95% CI=2.01–10.48) and being admitted to a lower ranked treating hospital, which had a strong though not statistically significant association (OR=8.06; 95% CI=0.36–181.4). Hospitalization in an infectious diseases ward remained an important protective factor against death due to bacterial meningitis, but only when the patient did not have co-morbidities and resided in Rome (OR=0.05; 95% CI=0.02–0.18). When the patient resided in Rome but presented co-morbidities (OR=0.40; 95% CI=0.06–2.62) or when he did not present co-morbidities but resided outside Rome (OR=0.53; 95% CI=0.04–6.70) the effect was still protective though weaker and not statistically significant. Finally, being treated in an infectious diseases ward increased the risk of death in patients who both resided outside Rome and presented serious co-morbidities (OR=3.85; 95% CI=0.31–48.63) (Table 2).

Table 2. Adjusted odds ratios of death in patients with bacterial meningitis, Lazio, Italy, 1996–2000 (number of observations=455)

Factor	ORa	95% CI
Age
5–44 years	1.55	0.63–3.79
≥45 years	4.59	2.01–10.48
Streptococcus pneumoniae in CSF or blood	1.44	0.60–3.42
Bacteria in CSF or blood other than Neisseria meningitidis and Streptococcus pneumoniae	1.09	0.43–2.74
Presence of serious disease signs	8.84	3.35–23.34
Treatment in infectious diseases wards
If resident in Rome and:
co-morbidities no	0.05	0.02–0.18
co-morbidities yes	0.40	0.06–2.62
If resident outside Rome and:
co-morbidities no	0.53	0.04–6.70
co-morbidities yes	3.85	0.31–48.63
Site of last admission outside Rome	0.67	0.26–1.72
Second class last hospital	8.06	0.36–181.4

The factors associated with meningitis death did not change when we excluded disease severity from the model (Table 3), nor did the meaning of their association. The same results were observed even when the analysis was restricted to patients with less severe meningitis, but the risk of death was higher for patients over 44 years of age and when co-morbidities were present (Table 4).

Table 3. Adjusted odds ratios of death in patients with bacterial meningitis, Lazio, Italy, 1996–2000 (number of observations=467)

Factor	ORa	95% CI
Age
5–44 years	1.41	0.60–3.29
≥45 years	4.08	1.89–8.83
Streptococcus pneumoniae in CSF or blood	1.66	0.73–3.77
Bacteria in CSF or blood other than Neisseria meningitidis and Streptococcus pneumoniae	1.09	0.45–2.61
Treated in infectious disease wards
If resident in Rome and:
co-morbidities no	0.05	0.02–0.17
co-morbidities yes	0.35	0.05–2.31
If resident outside Rome and:
co-morbidities no	0.80	0.07–8.74
co-morbidities yes	5.03	0.39–64.10
Site of last admission outside Rome	0.56	0.23–1.37
Second class of last hospital	6.23	0.33–116.5

Table 4. Adjusted odds ratios of death in patients with bacterial meningitis and without serious disease signs, Lazio, Italy, 1996–2000 (number of observations=429)

Factor	ORa	95% CI
Age
5–44 years	1.72	0.65–4.55
≥45 years	5.19	2.12–12.71
Streptococcus pneumoniae in CSF or blood	1.65	0.65–4.18
Bacteria in CSF or blood other than Neisseria meningitidis and Streptococcus pneumoniae	1.18	0.45–3.12
Treated in infectious disease wards
If resident in Rome and:
co-morbidities no	0.05	0.02–0.18
co-morbidities yes	0.38	0.06–2.47
If resident outside Rome and:
co-morbidities no	0.64	0.05–8.27
co-morbidities yes	4.50	0.36–56.61
Site of last admission outside Rome	0.68	0.27–1.77
Second class of last hospital	9.44	0.40–220.7

Back to Article Outline

Discussion 

The results indicate that being treated in an infectious diseases ward was the strongest determinant of surviving bacterial meningitis. The protective association is remarkable even after adjusting for risk factors related to infection and host characteristics, although only in patients who resided in Rome and who did not present co-morbidities. A possible implication of this result is that treatment was suboptimal in non-infectious diseases areas. The association between outcome and hospital classification strengthened this result; the probability of death from meningitis was higher when the treating hospital was in the lower rated group, none of which had infectious diseases services. A previous paper reported that specialized treatment services in hospital can serve as a protective factor for fatal outcomes in meningococcal meningitis; the authors hypothesized that the association was due to differences in treatment protocols.¹⁰ Furthermore, lower case fatality rates from meningococcal disease have been reported in association with admitting all children with meningitis to one pediatric hospital equipped with an intensive care unit.¹²

The protective effect of infectious diseases wards was lower when co-morbidities were present, as expected, but in non-Roman residents affected by co-morbidities this effect was no longer protective, and actually was associated with higher mortality. A reverse causation may have occurred here; many of these patients were transferred to specialized hospitals in Rome when clinical conditions were severe, but the transfer seems to have occurred too late. Other observations support this hypothesis, in fact, hospital transfer was associated with fatal outcome in the univariable analysis, and the probability of death observed in hospitals outside Rome was lower than in hospitals in Rome, though the former include fewer infectious diseases wards. Although our interpretation of these results should be tested more extensively, they suggest the possibility that treatment at hospitals specializing in the treatment of meningitis has an impact on survival, and access to these hospitals is more likely for residents of Rome than for residents outside Rome.

It has been maintained in the medical literature that the later the meningitis is diagnosed and treated, the greater is the possibility of adverse outcomes. Although this assumption is widely supported, there are papers that report conflicting results. Timing of antibiotic therapy was associated with adverse outcomes, both death and neurological deficit, only in patients who started treatment in advanced clinical stages of the disease;², ⁸ the duration of symptoms before therapy did not have an impact in another study;⁴ finally, children with very short courses of the disease had a worse clinical picture than those with longer courses, characterized by seizures and coma,¹³ and a fulminating form of meningitis showed a strong association with death.⁶ One possible conclusion is that the most severe forms of meningitis, rapid admission and therapy notwithstanding, show a high risk of fatal outcome, while meningitis characterized by less severe signs could improve with rapid diagnosis and care. We analyzed the timing of hospitalization as a proxy of these factors and did not find any association. Although the high percentage of missing information could influence this result, the risk of fatal outcome was lower in patients who presented symptoms for more than one day before admission, and was even lower when analysis was restricted to community meningitis.

Finally, we could not study the pre-admission antibiotic therapy because this information is not reported in any database. This intervention has been reported as an independent predictor of favorable prognosis for meningococcal disease with a risk estimate of 0.09(p value=0.003)⁶ and is currently recommended by a consensus statement for any case of suspected bacterial meningitis in adults.¹¹

Among factors linked with infection and host characteristics, the strongest predictor of fatal outcome was the presence of serious disease signs, as expected, based on the literature.², ³, ⁴, ⁵, ⁶, ⁸ Although we found a risk of dying as high as 11.84 for shock, other studies have reported values up to 18.1 and 39.7.⁶, ⁸ For focal neurological signs (seizures) and pneumonia, we found weaker associations as well, consistent with previous papers,², ⁵ but with lower estimates of risk. The use of information from the dataset instead of from clinical charts could account for this difference.

Another important risk factor of dying from meningitis was being 45 years of age or older. Older age is a known risk-factor for meningitis death but in most studies the threshold has been 60 years.¹, ⁴, ⁵, ⁶ Only one study showed higher risk in patients over 40 years of age, but it analyzed meningococcal diseases during outbreaks.¹⁰ Our results should be evaluated further using clinical data with information on bacteria serogroups and antibiotic susceptibility. In a previous study¹⁴ we estimated a much higher percentage (78%) of N. meningitidis strains intermediately resistant to penicillin in our region than in other countries (3% in the USA and 40% in Spain).

Co-morbidities were selected according to risk factors of adverse clinical outcomes in patients with bacterial meningitis of other studies;², ⁴, ⁵, ⁶, ⁸ we used a validated co-morbidity index that allowed us to include underlying diseases or conditions that might alter the risk of short-term mortality.¹⁵ Unfortunately, since information was gathered from databases, meningitis cases occurring in the first year had a lower probability of being classified by the presence of co-morbidities than those cases reported in the second year of the study period based on previous hospitalizations.

S. pneumoniae was the most common cause of bacterial meningitis in our region and it was associated with a slightly higher risk of death. A similar epidemiologic picture was reported in other countries¹⁶ and in studies focused on community-acquired bacterial meningitis.² Where nosocomial diseases are prevalent, Gram-negative bacillary meningitis is reported more frequently than pneumococcal meningitis.¹, ⁵ This could indicate that hospital meningitis was under-represented among our cases.

Our study presents possible limitations. First of all they concern the case definition. The definition of bacterial meningitis included only officially reported cases, and may be an underestimate. Twenty-six additional cases of bacterial meningitis were not reported to the surveillance system, and therefore were not included in the analysis. This lack of sensitivity was due to fatalities: out of 26 cases not reported, 15 were registered with the cause of death as bacterial meningitis, and 11 died in hospital with either a primary or secondary diagnosis of bacterial meningitis. Although the same inclusion criteria were used in other studies,¹⁷, ¹⁸ the differences in reporting deaths could have introduced a bias in analyzing factors related with fatal outcome in our study.

The definition of a bacterial meningitis death is even more disputable; out of the 98 deaths reported, 45 had a different underlying cause of death listed, and 23 in-hospital deaths did not have bacterial meningitis listed as either the primary or secondary diagnosis. Our definition included a time criterion; the choice of a 30-day time span was intended to detect deaths due to acute meningitis or its immediate complications. Although the same criteria have been used in other studies to define bacterial meningitis death,¹, ¹⁸ other authors have suggested a 14-day time span to discriminate between deaths attributable to meningitis and those to other causes,⁴ and even a 7-day time span has been used by others to define bacterial meningitis death.⁶ We used a 30-day time period only for patients who had died at home; the survival time was longer than 30 days in 13 cases who died in hospital, ranging from 32 to 156 days. This choice could have introduced a bias, because patients who died from complications of meningitis may have been included more frequently if they died in hospital.

We did not use volume of patients treated as an indicator of healthcare quality, though it is used frequently for chronic diseases. The number of meningitis admissions does not depend on anything but demand, strictly connected to the characteristics of infection and transmission of bacteria.

Back to Article Outline

Conclusions 

Being treated in infectious diseases wards was a protective factor for patients with bacterial meningitis who did not present co-morbidities. This protective effect was not shown for residents outside Rome, possibly due to delayed access to infectious diseases wards for these patients. Further confirmation of this hypothesis could lead to better care management of meningitis cases and possibly improve prognosis.

We also found a higher probability of dying from bacterial meningitis when specific clinical conditions were present, known as predictors of fatal outcome, for older patients or those with concurrent co-morbid diseases. The coherence of these results with those of previous studies allows us to be more confident in our conclusions about healthcare-related factors in the treatment of bacterial meningitis.

Back to Article Outline

Acknowledgements 

We thank Claudia Cimaglia, Paola Colais, and Claudia Marino for support in processing data from the Hospital Discharge databases, and Margaret Becker for the English revision.

Conflict of interest: No conflict of interest to declare.

Back to Article Outline

References 

1. Durand ML, Calderwood SB, Weber DJ, Miller SI, Southwick S, Caviness VS, et al. Acute bacterial meningitis in adults. A review of 493 episodes. N Engl J Med. 1993;328:21–28
   • View In Article
   • MEDLINE
   • CrossRef
2. Aronin SI, Peduzzi P, Quagliariello VJ. Community-acquired bacterial meningitis: risk stratification for adverse clinical outcome and effect of antibiotic timing. Ann Intern Med. 1998;129:862–869
   • View In Article
   • MEDLINE
3. Kastenbauer S, Pfister H-W. Pneumococcal meningitis in adults. Spectrum of complications and prognostic factors in a series of 87 cases. Brain. 2003;126:1015–1025
   • View In Article
   • MEDLINE
   • CrossRef
4. McMillan DA, Lin CY, Aronin SI, Quagliariello VJ. Community-acquired bacterial meningitis in adults: categorization of causes and timing of death. Clin Infect Dis. 2001;33:969–975
   • View In Article
   • MEDLINE
   • CrossRef
5. Tang L-M, Chen S-T, Hsu W-C, Lyu R-K. Acute bacterial meningitis in adults: a hospital-based epidemiological study. Q J Med. 1999;92:719–725
   • View In Article
6. Barquet N, Domingo P, Caylà JA, Gonzales J, Rodrigo C, Fernandez-Viladrich P, et al. Prognostic factors in meningococcal disease. Development of a bedside predictive model and scoring system. JAMA. 1997;278:491–496
   • View In Article
   • MEDLINE
7. Sàez-Llorens X, McCracken GH. Bacterial meningitis in children. Lancet. 2003;361:2139–2148
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (136 KB)
   • CrossRef
8. Lu C-H, Huang C-R, Chang W-R, Chang C-J, Cheng B-C, Lee P-Y, et al. Community-acquired bacterial meningitis in adults: the epidemiology, timing of appropriate antimicrobial therapy, and prognostic factors. Clin Neurol Neurosurg. 2002;104:352–358
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (113 KB)
   • CrossRef
9. Bi P, Parton KA, Whitby M. Co-existing conditions for deaths from infectious and parasitic diseases in Australia. Int J Infect Dis. 2004;8:121–125
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (94 KB)
   • CrossRef
10. El Bushra HE, Hassan NMM, Al-Hamdan NA, Al-Jeffri MH, Turkistani AM, Al-Jumaily A, et al. Determinants of case fatality rates of meningococcal disease during outbreaks in Makkah, Saudi Arabia, 1987–97. Epidemiol Infect. 2000;125:555–560
    • View In Article
    • MEDLINE
    • CrossRef
11. Begg N, Cartwright KA, Cohen J, Kaczmarski EB, Innes JA, Leen CLS, et al. Consensus statement on diagnosis, investigation, treatment and prevention of acute bacterial meningitis in immunocompetent adults. J Infect. 1999;39:1–15
    • View In Article
    • MEDLINE
    • CrossRef
12. Booy R, Habibi P, Nadel S, de Munter C, Britto J, Morrison A, et al. Reduction in case fatality rate from meningococcal disease associated with improved healthcare delivery. Arch Dis Child. 2001;85:386–390
    • View In Article
    • CrossRef
13. Kilpi T, Anttila M, Kallio MJ, Peltola H. Severity of childhood bacterial meningitis and duration of illness before diagnosis. Lancet. 1991;338:406–409
    • View In Article
    • Abstract
    • CrossRef
14. Faustini A, Fabrizi E, Sangalli M, Bordi E, Cipriani P, Fiscarelli E, et al. Role of laboratories in population-based surveillance of invasive diseases in Lazio, Italy, 1998–2000. Eur J Clin Microbiol Infect Dis. 2002;21:824–826
    • View In Article
    • MEDLINE
    • CrossRef
15. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373–383
    • View In Article
16. Schuchat A, Robinson K, Wenger JD, Harrison LH, Farley M, Reingold AL, et al. Bacterial meningitis in the United States in 1995. N Engl J Med. 1997;337:970–976
    • View In Article
    • MEDLINE
    • CrossRef
17. Goldacre MJ, Roberts SE, Yates D. Case fatality rates for meningococcal disease in an English population, 1963–98: database study. BMJ. 2003;327:596–597
    • View In Article
    • CrossRef
18. Moura AS, Pablos-Mendez A, Layton M, Weiss D. Epidemiology of meningococcal disease, New York city, 1989–2000. Emerg Infect Dis. 2003;9:355–361
    • View In Article
    • MEDLINE