Neisseria meningitidis phenotypic markers and septicaemia, disease progress and case-fatality rate of meningococcal disease: a 20-year population-based historical follow-up study in a Danish county

The North Jutland Meningococcal Research Database contains data on patients who were admitted with MD to any hospital in the county from 1 January 1980 onwards (Sørensen et al., 1995, 1998). The patients were identified from the following sources: (i) the local and national Medical Offices of Health, to which notification is required, (ii) the County Hospital Discharge Registry, (iii) the regional Department of Clinical Microbiology at Aalborg Hospital and (iv) the National Neisseria Reference Laboratory, Statens Serum Institut, Copenhagen.

A diagnosis of MD was made when at least one of the following criteria was met: (i) isolation of meningococci from blood or cerebrospinal fluid (CSF) specimens; (ii) presence of Gram-negative diplococci in the CSF as determined by direct microscopy; (iii) seroconversion or positive meningococcal antibody test (MAT) (< 3 on a scale from 0 to 18) combined with clinical findings typical of MD (meningitis and/or septicaemia with typical haemorrhagic rash); or (iv) clinical findings typical of MD (meningitis and/or septicaemia with typical haemorrhagic rash) (Sørensen et al., 1995).

A positive blood culture was used as a surrogate for septicaemia and a positive CSF culture for meningococci. Duration of disease before admission was used as a surrogate for rapidity of disease progress. CFR was defined as the proportion of patients who died during their stay in hospital. Three patients who were dead on arrival at the hospital were included in the study. Hospital and microbiological records were reviewed to ensure that all patients fulfilled the criteria for MD. The following data were collected for each patient: age and sex, date and year of admission, duration of disease before admission (24 h or >24 h), specimen type (CSF and/or blood), bacteriological characteristics of the strain and outcome of disease (death or survival).

A 10-digit personal registration number is given to all residents in Denmark at birth or following immigration. This ID number is used in all health registries, thus making linkage simple and valid.

Microbiological methods.
Culture of N. meningitidis was performed at the Department of Clinical Microbiology, Aalborg Hospital. CSF specimens were plated on chocolate agar and 10 % horse blood agar (Media Department, Statens Serum Institut) and incubated in 5 % CO₂. Three different blood culture systems were used in succession, as described elsewhere (Madsen et al., 1999). Isolates were sent to the Neisseria Unit, Statens Serum Institut, Copenhagen, and stored at -196 °C. One isolate from each patient was included in the study. Each meningococcal strain was characterized by its serogroup, serotype, serosubtype (Frasch et al., 1985) and sulphonamide resistance.

Serogrouping was performed by co-agglutination using rabbit antisera coated on protein A-rich Staphylococcus aureus cells as carriers (Kronvall, 1972). From July 1999, a latex agglutination test (Slidex méningite-kit 5) was used for determination of serogroups A, B and C, as described by the manufacturer (bioMérieux). Serotyping and serosubtyping were performed by a whole-cell ELISA using mAbs (Lind & Berthelsen, 1998; Abdillahi & Poolman, 1987) provided by Rijksinstituut voor Volksgezondheid & Milieu (RIVM), Bilthoven, The Netherlands. In order to ensure comparability of results, strains obtained before 1990 were retyped with the panel of reagents used after 1990. Susceptibility to sulphonamide was determined by an agar dilution method and resistance was defined as a minimum inhibitory concentration (MIC) 8 mg l^-1 (Lind & Berthelsen, 1998). We defined the phenotype as the combination of serogroup, serotype and serosubtype.

Statistical analysis.
Age-standardized IRs were calculated by use of annual population data provided from Statistics Denmark, with the age distribution of the population of North Jutland County 198083 as reference.

Phenotypes were divided into three categories: B : 15 : P1.7,16, C : 2a : P1.2,5 and any other phenotype. The manifestation of MD was categorized in two ways: (i) septicaemia alone versus meningitis with or without septicaemia or (ii) septicaemia with or without meningitis versus meningitis alone. The rapidity of disease progress was evaluated on the basis of the duration of symptoms before admission ( 24 h and > 24 h). The study period was divided into five 4-year periods or into two 10-year periods. The proportions of the three phenotype categories and of cases with negative culture were determined for each 4-year period.

We obtained contingency tables for the phenotype and the main study variables and 95 % confidence intervals (CIs) for proportions were calculated. Contingency tables were also obtained for the 4-year period and the main study variables. Next, contingency tables were obtained for the main study variables and CFRs. In order to examine the association between phenotype and CFR and to examine the impact of calendar period, we fitted a logistic regression model including calendar period as five design variables and age as a continuous variable and calculated odds ratios (ORs) with 95 % CIs. The fit of the model was assessed using the HosmerLemeshow test (Hosmer & Lemeshow, 1989).

The impact of phenotypes on the CFR during the first and second decade was studied by comparing the prevalences of phenotypes, CFR as related to phenotype and the relative risk (RR) of case fatality for patients with B : 15 : P1,7,16 and C : 2a : P1.2,5 compared with any other phenotype. Finally, the difference in CFR between 198089 and 199099 was calculated for each phenotype category with 95 % CI.

The study was approved by the Danish Data Protection Agency (journal no. 1994-1200-236).

An isolate of N. meningitidis was obtained from blood, CSF or both in 320 cases (77 %). A strain was available for serological characterization from 315 cases (98 %). Serogroup B constituted 226 (72 %) of the strains, serogroup C 75 (24 %) and other serogroups 14 (4 %). The most frequent serogroup B phenotype was B : 15 : P1.7,16 (n = 100, 32 %) and the most frequent serogroup C phenotype was C : 2a : P1.2,5 (n = 31, 10 %). Apart from B : NT : P1.9 (n = 30, 10 %), all other phenotypes each constituted less than 3 % of the strains (Table 1). Resistance to sulphonamides was recorded for all B : 15 : P1.7,16 strains, for 68 % of the C : 2a : P1.2,5 strains and for 31 % of other strains. The following increases were observed from 198083 to 199699 in the 315 patients: 7 to 28 % in the proportion who were 30 years, 9 to 41 % in the proportion with septicaemia without meningitis and 43 to 69 % in the proportion with symptoms for 24 h before admission. The CFR increased from 1.8 to 17 %, with a peak of 21 % during 198891.

Table 1.Distribution of serogroups and phenotypes for invasive strains of N. meningitidis with a complete phenotype in North Jutland County, Denmark, during 198099 (n = 315) One isolate from blood or CSF was included per patient. NG, Not grouped; NT, not typed; N-ST, not subtyped.

Changes over time in the prevalence of cases with phenotypes B : 15 : P1.7,16, C : 2a : P1.2,5, other phenotypes and without culture-confirmation of the diagnosis are shown in Fig. 1. The number of patients with B : 15 : P1.7,16 strains doubled during 199295, thus contributing markedly to the increase in the total number of cases. Among culture-confirmed cases, the proportion of B : 15 : P1.7,16 strains increased from 29 % during 198083 to 50 % during 199699. The prevalence of cases with C : 2a : P1.2,5 strains reached a peak of 19 % during 198487 and decreased thereafter to 3.7 % during 199699.

The distribution of patient characteristics and clinical manifestations by phenotype of the invasive strain are shown in Table 2. The median age of patients with B : 15 : P1.7,16 strains was about 5 years higher than for other MD patients. Septicaemia without meningitis tended to be less common in cases with B : 15 : P1.7,16 and C : 2a : P1.2,5 strains. Conversely, a larger proportion of patients with these phenotypes had both meningitis and septicaemia. The CFR was 7.3, 9.5 and 24.3 % for patients with meningitis only, both meningitis and septicaemia and septicaemia only, respectively (data not shown). No association was found between phenotype and rapid disease progress. The CFR was 14.7 % for patients with duration of illness 24 h and 7.9 % for patients with duration of illness > 24 h (data not shown).

Table 2.Characteristics of 315 patients with culture-confirmed MD in North Jutland County, Denmark 198099 by phenotypic markers 95 % CIs for percentages are given in parentheses.

The CFR was higher for patients with B : 15 : P1.7,16 and C : 2a : P1.2,5 strains than for patients with other phenotypes. In addition, there was an association between calendar period and CFR, with an increased CFR during 198899, reaching a maximum during 198891 (Table 3).

Table 3.Crude and adjusted ORs of CFR related to phenotype and calendar period for 315 patients with culture-confirmed MD in North Jutland County, Denmark, during 198099 HosmerLemeshow test statistics: χ2 = 5.8, d.f. = 8, P = 0.7. Values for ORs have 95 % CIs in parentheses.

During 199099, B : 15 : P1.7,16 constituted a larger proportion of the strains than during the previous decade and the CFR for patients with this phenotype increased from 7.9 to 19.4 %. Despite a fourfold increase in the number of fatalities among patients with B : 15 : P1.7,16 strains during 199099, there was less variation of CFR between phenotype categories because the CFR also increased from 2.3 to 13.5 % for patients with other phenotypes. During 199099, C : 2a : P1.2,5 constituted a smaller proportion of the strains than during the previous decade. Thus, despite a high CFR (> 20 %) throughout the study period, this phenotype contributed less to the overall case fatality during the second decade than during the first (Table 4).

Table 4.Comparison of CFR and relative risk of case fatality by phenotype in patients with MD in North Jutland County, Denmark, during 198089 and 199099

The major strength of the present study derives from its population-based design, with complete registration of MD cases and complete follow-up during 20 years in a well-defined population with a uniformly organized health-care system (Sørensen et al., 1995). A limitation of our study was the limited statistical precision for subgroup analysis. Another limitation was the exclusion of 23 % of the patients because the diagnosis was not confirmed by culture. This may have led to selection bias of the overall association between phenotype and CFR. The direction will depend on whether the strength of the association in patients with culture-confirmed MD differs from the unknown association in excluded patients. We are aware that the heterogeneous group of phenotypes used as reference may include phenotypes with distinctly different associations with CFR. However, we did not detect any variation between the different phenotypes (data not shown), though the small numbers should be taken into consideration.

The specificity of the antibodies used for serotyping and serosubtyping is considered to be 100 % (Poolman et al., 1995). Thus, misclassification of strains that have been assigned a complete phenotype can be excluded with confidence. Non-typable and non-subtypable strains may differ from the fully typable strains in respects other than typability, which is why these strains were assigned to the reference group.

The categorization of duration of illness before admission was intended to distinguish a fulminant course, which is a well-known risk factor for fatal outcome, from the more protracted courses (Van Deuren et al., 2000; Sinclair et al., 1987). Our choice of a positive blood culture as a surrogate for septicaemia is in accordance with the definition of septicaemia, which includes a positive blood culture combined with overt clinical signs of infection (Effersøe & Nielsen, 1986). We deem the latter to have been verified by the fact that the patient was evaluated clinically before sampling of blood for culture. MD evolves to meningitis through haematogenous dissemination, and a positive CSF culture may therefore represent a distinct category of patients who have survived the initial bacteraemic stage and might thus have a better prognosis than patients with positive culture of blood only (Van Deuren et al., 2000). Disease progress and manifestation of disease (septicaemia alone, meningitis and septicaemia or meningitis alone) may be regarded as intermediate steps in the process of infection and, as such, they were not included as potential confounders in the logistic regression model.

During 198085, the mean annual IR of notified cases of MD in Denmark was 3.4 per 100 000 person-years at risk. In 1986, the IR increased by 60 % and, during 198688, the mean annual IR was 5.4 per 100 000 person-years at risk (Samuelsson et al., 1991). In North Jutland County, a similar increase occurred with a delay of about 5 years. Reliable data on CFR for all of Denmark are available from 1988 onwards and, during 198891, 199295 and 199699, the CFR was respectively 9.6, 8.1 and 7.8 % (for North Jutland CFRs, see Table 3). We have no explanation for this difference. Since 1993, B : 15 : P1.7,16 and C : 2a : P1.2,5 have respectively been confirmed to be the predominant phenotypes among serogroup B and C strains (L. Berthelsen, unpublished observations; Lind & Berthelsen, 1998). North Jutland County is an important holiday resort, with tourists from many European countries including many teenagers. This should provide opportunities for the introduction of novel meningococcal strains. However, we found meningococcal phenotypes to be stable over an extended period of time.

It is not clear whether the most relevant markers for the identification of virulent strains are phenotypic or genotypic. In a separate study (Jensen et al., 2002), we selected 181 strains for further characterization by MLEE and ribotyping. We identified two distinctive clone complexes among serogroup B strains (including 92 of the 100 B : 15 : P1.7,16 strains) and one distinctive clone complex among serogroup C strains (including 30 of the 31 C : 2a : P1.2,5 strains). Each of the three clone complexes was associated with an increased CFR (1320 %) compared with any other strain (CFR 3 %) (Jensen et al., 2002). However, the CFRs for phenotypes B : 15 : P1.7,16 and C : 2a : P1.2,5 tended to be higher than the CFRs for the three clonal complexes. This points to the phenotypes being relevant surrogate markers for determinants of virulence of meningococci.

In accordance with our finding, a high CFR was reported in patients with C : 2a : P1.2,5 strains in the Czech Republic and, in Norway, CFR was increased in patients with B : 15 or C : 2a strains (Krïzová et al., 1997; Iversen & Aavitsland, 1996). However, the Norwegian data were based on notified cases only and the analysis included only 105 patients admitted during one year, 1994. In contrast to the strains from the Czech Republic, we found that the CFR associated with C : 2a : P1.2,5 infection remained high (>20 %) over a 20-year period, whereas the CFR associated with B : 15 : P1.7,16 increased from an intermediate to a high level. The Norwegian study found no difference in clinical manifestations between serogroup B and serogroup C MD, but it is not clear whether this applied to B : 15 and C : 2a MD. Furthermore, the classification of septicaemia versus meningitis was based on clinical criteria and is thus not directly comparable with ours. In a previous Danish study, the increased CFR associated with B : 15 : P1.16 was related to a high proportion of patients with septicaemia, but it is noteworthy that, in septicaemic patients, the CFR for B : 15 : P1.16 did not differ from that of other phenotypes (Samuelsson et al., 1992). Furthermore, the latter study related to an outbreak situation, whereas our study covered a long period of high endemicity. This setting resembled the situation in Norway, where B : 15 : P1.7,16 (belonging to the ET-5 complex) has been predominant for more than two decades. In contrast to other European countries, in which increases in IR have been related to C : 2a : P1.2,5 (Krïzová et al., 1997; Kremastinou et al., 1999; Connolly & Noah, 1999), the occurrence of this phenotype actually decreased in North Jutland County during 199099.

The increased prevalence of B : 15 : P1.7,16 cases during 199299 could be a reflection of increased virulence or waning population immunity. Taha et al. (2001) recently described a mutation in the DNA region encoding one of the epitopes determining the dual serosubtype in B : 15 : P1.7,16 meningococci. As a result, the epitope was not recognized by the bactericidal serum antibodies that were prevalent in the population (Taha et al., 2001). Thus, underlying genetic changes might explain a sudden decrease in population immunity against B : 15 : P1.7,16 despite frequent occurrence for more than a decade. Theoretically, evasion of bactericidal antibodies, as described above, may lead not only to increased incidence of disease but also to increased CFR among patients infected with the mutated strain. However, we did not detect any genotypic change of B : 15 : P1.7,16 strains that correlates with the increases in IR and CFR (Jensen et al., 2002).

The lack of association in the present study between phenotype and septicaemia alone and between phenotype and rapid disease progress is surprising (Van Deuren et al., 2000; Sinclair et al., 1987); it suggests that these factors may be related to host factors rather than to the pathogen itself. Our finding of an association between calendar period and prognosis, which was independent of meningococcal phenotype and age of the patient, indicates that one or more hitherto unknown factors contributed to the observed trend in CFR. The increases in the proportions of patients with septicaemia alone and of patients with short pre-admission duration of disease are likely to have contributed to the increased CFR, but remain in themselves unexplained. This underlines the need for further research into the pathogenesis of MD, including the possible impact of meningococcal phenotypic and genotypic characteristics.