METHODS

A total of 196 children admitted to the Royal Liverpool Children's Hospital with an illness suggestive of MCD between January 2000 and March 2001 (WB-Taqman) were studied and compared to 319 patients from December 1997 to March 1999 (S-Taqman). These included patients admitted to the paediatric intensive care unit from Liverpool and surrounding areas (South Cumbria to mid Wales). Both studies followed the same protocol and were approved by the local research and development and ethics committees. Consent was obtained from a legal guardian and, where appropriate, assent obtained from the patient by the research fellow.

All patients were seen at or soon after admission. History, and clinical and laboratory data from admission until discharge were collected using a proforma for all patients. Each patient was classified as probable or possible MCD at admission and then reclassified daily until discharge as confirmed, probable, or possible cases using criteria set out by Stuart and colleagues.⁷ Stuart defined probable cases as a clinical diagnosis of meningococcal meningitis or septicaemia without confirmation where the clinician managing the case considered that meningococcal disease was the most likely diagnosis. In the absence of other positive microbiological investigations or an alternative diagnosis, a feverish, ill patient with a petechial/purpuric rash was regarded as a probable case. Possible cases were defined as cases in which the managing clinician or research fellow considered diagnoses other than meningococcal disease were as likely. These included cases treated with antibiotics whose probable diagnosis was viral meningitis. The gold standard for case definition used in both cohorts was therefore a clinical diagnosis based on the above criteria.

Admission EDTA samples were collected from the majority of patients with suspected MCD and sent to the MRU, Manchester PHLS for meningococcal DNA detection using the Taqman PCR method. Where dedicated samples had not been collected for PCR analysis, admission blood sent for full blood count was retrieved and subsequently sent. For patients transferred to our intensive care unit from other hospitals, admission EDTA samples were sent for meningococcal PCR from the transferring hospital. Only patients where blood samples for meningococcal PCR had been obtained within 24 hours of admission were included (193 of 196). Those not included consisted of two patients with occult meningitis and one patient who had a PCR sent after 24 hours, MCD confirmed by blood cultures. For the latest cohort the following diagnostic tests were obtained (all patients versus MCD patients respectively); blood cultures (98.5 v 100%), rapid latex antigen (RLA) (84 v 85%), and throat swabs cultured on appropriate media (vancomycin, nystatin colistin media) (59 v 66%). For the earlier cohort, percentages of blood cultures, PCRs, and RLA obtained were 100%, 98%, and 87% respectively.⁴

RESULTS

Following the introduction of WB-Taqman, blood culture and PCR confirmed the diagnosis in 33% and 87% respectively of patients classified as probable cases. No patients were blood culture positive but PCR negative, compared to 14% of patients from the S-Taqman era. The PCR positive rate in patients either already on oral antibiotics (n = 8) or who had received preadmission benzylpenicillin (n = 6, 6%) remained unchanged at 93% while blood culture confirmation fell to 21%. In patients who had clinical signs of meningitis, there was a non-significant decrease in blood PCR confirmation to 78% (16/22). Only nine patients had an LP performed. N meningitidis was cultured from three cerebrospinal fluid (CSF) samples, from patients who had received either no antibiotics prior to LP or had an LP performed within two hours of their first dose of parenteral antibiotics. Seven CSF samples were sent for meningococcal PCR; six were positive, and in two patients this was the only confirmatory result. The single patient with meningitis who had a negative CSF PCR, clinically also had meningococcal septicaemia and was subsequently confirmed on blood PCR. Acute (n = 8) and convalescent serology (n = 4) were only sent in PCR negative patients. The proportion of diagnostic investigations found positive showed a considerable variation as illustrated in table 1⇓.

There were no differences between the number of cases confirmed by either RLA or blood cultures before or after WB-Taqman was introduced. There was a significant increase in cases confirmed by PCR after the introduction of WB-Taqman (47% v 87%, p < 0.001, table 2⇓).

There were no false positive PCR or blood culture results in either series as all patients had been classified as probable cases prior to confirmation. Classifying patients into the categories probable and possible was not influenced by any diagnostic test results but based on clinical and non-diagnostic laboratory data. Seven patients classified at admission as a possible cases, were reclassified as probable cases shortly after admission, as it became obvious in three that they were developing overt meningococcal disease, while in two patients, covert meningococcal meningitis was diagnosed when an LP was performed. A total of six patients classified as probable cases at admission were subsequently reclassified as alternative diagnoses following confirmatory test results. They included patients with Haemophilus influenzae (b) septicaemia, status epilepticus (meningitis ruled out), Campylobacter jejuni and rotavirus gastroenteritis, lobar pneumonia, and Henoch–Schönlein purpura. Two patients classified as probable cases at admission were soon after reclassified as possible cases, as they clinically did not have MCD. All other patients classified as probable were either confirmed as MCD (n = 84) or remained probable (n = 7) cases.

The false negative rate by PCR alone decreased to 13%, and with the addition of RLA testing and serology further decreased to 7%. This increased the case confirmation rate when all confirmatory test results were combined from 72% (118/166 S-Taqman) to 94% (91/98 WB-Taqman; p < 0.001). This gives sensitivities of 88% or 94% and specificities of 100% or 97% for WB-Taqman PCR alone or all combined confirmatory cases respectively. The 97% specificity, observed when all confirmatory tests post WB-Taqman were combined, resulted from three patients who had false positive N meningitidis B/Escherichia coli K1 RLA results. The first patient had an E coli urinary tract infection. The second had recurrent “septicaemic” episodes for which no cause was found; she did not have petechiae and PCR samples have been consistently negative for meningococci. The third had acute disseminated encephalomyelitis; no cause was found.

DISCUSSION

The most noticeable difference between the two studies is that following the introduction of the WB-Taqman ctrA method, the case confirmation rate has improved considerably. This is mainly a result of the Gentra whole blood DNA extraction method now used, compared to serum previously. The redesigned ctrA primers are more sensitive compared to those previously used and therefore have also contributed to the improved sensitivity.

The improved case confirmation rate seen after the introduction of WB-Taqman could be partly explained by bias between the two studies. The most obvious potential discrepancy would be observer bias—that is, how each research fellow classified patients into possible and probable cases. This is unlikely for a number of reasons. Both research fellows used the same criteria set out by Stuart and colleagues⁷ for classifying patients, and both fellows had a similar level of paediatric experience. In the latest study we have shown how we classified, or in a minority reclassified patients, giving reasons for doing so, thus making this process more transparent. Secondly, in post WB-Taqman there were no negative PCR results in patients who were blood culture positive, compared to 14% in the S-Taqman era, indicating a real improvement in PCR sensitivity. Lastly, the proportions of possible and probable cases in both studies have remained roughly the same (approximately 50:50).

This improved rate of confirmation means that PCR is the essential confirmatory investigation in MCD. The reason that it is not 100% sensitive is probably because it has a lower detection sensitivity of 10³ DNA copies (bacteria)/ml of blood. Patients with bacterial loads below this limit would not be detected. In patients where clinical signs of meningitis are more prominent than septicaemia, we found that bacterial load is significantly higher in CSF than in blood (personal observations). CSF samples from these patients, which are rarely sent, would therefore have a greater chance of giving a positive PCR result. A further confounding factor is that patients with no confirmed diagnosis, who are unwell with a temperature and petechiae (for which there is a long list of other potential diagnoses), are by default classified as probable meningococcal cases.

We recommend that RLA, blood cultures, and PCR be sent in all patients with probable MCD. PCR is especially important in patients who are on or who have received parenteral antibiotics prior to blood culture being sent. RLA testing has a low sensitivity, but as its name implies, can confirm a case quickly, which may aid public health decisions. However the only false positive results in our study occurred in three patients with positive RLA.

Blood cultures, generally available before PCR results, allow the organism to be serotyped and serosubtyped (important from a public health and epidemiological viewpoint), and also antibiotic sensitivity to be performed. We believe that serology need only be sent in PCR negative patients or in cases of conjugate C vaccine failure. Where PCR results will not be available prior to the patient's discharge from hospital, acute serum should be stored. If the PCR is negative, convalescent serum, approximately 3–6 weeks post illness, should be sent with the acute sample.

An attempt should be made to confirm all cases of probable MCD, enabling accurate disease surveillance, which is particularly important following introduction of the meningococcal C conjugate vaccine. This would permit the detection of the majority of vaccine failures and the occasional patients who have recurrent disease (in the WB-Taqman series, three patients, 3%). For both recurrent disease and in vaccine failures complement function, immunoglobulin levels and functional antibody assessment needs to be undertaken.

On average, 10% of the population carry N meningitidis in their nasopharynx,⁸ which may only be one quarter of true carriage rates.⁹ The relevance therefore of detecting a meningococcus in a throat swab from a patient with possible MCD needs to be interpreted in the context of their clinical disease.

Meningococcal PCR is supplied free of charge by the public health laboratory services at the MRU. The turnaround time for the screening test is 4–5 hours from receiving the sample. This means that a sample received at the MRU before 10 am on a working day will usually have a result before 5 pm that day. Therefore for the majority of samples, Taqman PCR results were available 24–48 hours after sampling.

The ideal study would have been evaluating both tests on the same samples, which was not possible here. Our comparison with “historical” controls slightly reduces the significance of the result. However, the highly significant difference observed between the cohorts means that the disparity is not a result of either chance or bias.