Which contacts of patients with meningococcal disease carry the pathogenic strain of Neisseria meningitidis? A population based study

The Telemark project

The detailed organisation of this project has been described previously.5 After isolation of meningococci from a patient specimen the local health officer is alerted, who then collects throat samples from members of the patient's household before initiating penicillin treatment to those under 15 years of age.17 Parents accompanying the patient to hospital are often sampled by the hospital staff.

Based on information from people who know the patient a list of close contacts is drawn up; throat samples are collected from these contacts on the same or next day. Simultaneously, the local community is informed about the disease in open meetings. When a pathogenic strain is found, rifampicin prophylaxis (600 mg twice daily for two days; children <12 years of age 10 mg/kg) is given and throat samples collected from the contact's household and kissing contacts (secondary contacts).

For each contact a standard questionnaire is completed for personal data, type of contact with the patient, and symptoms of respiratory disease.

Collection of throat samples — Both tonsils and the posterior pharyngeal surface were sampled with a cotton swab, plated immediately on GC agar base (Mast Diagnostics, Merseyside, UK) supplemented with haematin, 1% IsoVitalex (BBL, Cockeysville, MD,USA), vancomycin (3 mg/l), and colistin (7.5 mg/l), and incubated at 37°C in 10% carbon dioxide within two hours of sampling. Wherever possible, sampling was done by two of the authors (BK and YT).

Identification of the pathogenic strain — Contacts carrying a meningococcus with a chromosomal DNA fingerprint identical to that of the patient isolate were identified as previously described18 and were defined as carrying the pathogenic strain

Statistical methods

Confidence intervals for the prevalence of the pathogenic strain were calculated as follows:

The parameter of interest is the probability, p, of carrying the pathogenic strain. m is the number of patients diagnosed. Assume that patient i has n_ij contacts in class j, and that each of them has probability p_ij of carrying a pathogenic strain. The number, X_ij, of the n_ij contacts carrying the pathogenic strain is a binomial (n_ijp_ij; i=1,2,…m; j= I, II, III). Then p_j, the probability of carriage for each of the three contact classes, is estimated by the weighted average of the p_ij's: which has a statistical mean p_j and a statistical variance estimated by Since the n_ij's and X_ij's are small, the p^_ij's are estimated by Thus an approximate 95% confidence interval for p_j is given by

Results

Disease characteristics —From 1 November 1987 to 1 December 1996 there were 48 cases (cases 3-50) of meningococcal disease in the county of Telemark, Norway, verified bacteriologically (table 1). Thirty isolates were serogroup B, 14 were serogroup C, three were serogroup Y, and one was serogroup W135. Twenty four of the patients were under 4 years of age. The remaining cases were distributed in the age groups 5-12 years (six cases), 13-18 years (nine), 19-60 years (five), and >60 years (four).

Classification of contacts into groups and classes —We collected throat specimens from 1535 close contacts (primary contacts) of the patients, and from 78 secondary contacts who were household members or kissing contacts of primary contacts found to carry the pathogenic strain. All contacts approached consented to sampling. The primary contacts were divided into 16 contact groups and further organised into three classes according to the degree of contact with the patient (tables 2 to 4). Class 1 consists of household members and kissing contacts, the groups of contacts assumed to have the closest contact with the patient. Class 2 contacts are considered to have closer contact with the patient than class 3 contacts. Secondary contacts were placed in a separate group, group 18 (table 5).

Carrier rate —Among 1535 primary contacts, 234 meningococcal carriers were found. Of these, 42 carried the pathogenic strain. Thirty six of the 145 class 1 contacts carried meningococci. The pathogenic strain was found in 18 (12.4%: 95% confidence interval 5.5% to 19.3%) of these contacts: 6/37 (16.2%) mothers, 5/37 (13.5%) fathers, 4/29 (13.8%) brothers, 1/21 (4.8%) sisters, 1/18 (5.5%) other household members, and 1/3 (33.3%) kissing contacts. Of 576 class 2 contacts, 105 (18.2%) carried meningococci. The pathogenic strain was found in 11 (1.9%: 0.9% to 3.4%). Of 814 class 3 contacts, 93 (11.4%) carried meningococci. The pathogenic strain was found in 13 (1.6%: 0.14% to 3.1%) of these contacts. Of 78 secondary contacts, 20 (25.6%) carried meningococci. The pathogenic strain was found in four (5.1%) of these contacts.

The pathogenic strain was found in the primary contacts of 27 of the 48 patients, the number of carriers varying between 1 and 6 (table 5). The pathogenic strain was found more often in contacts in the 5-12 years age group than in the other age groups (table 1).

Discussion

In most countries the use of chemoprophylaxis is recommended to prevent secondary disease in close contacts of patients with meningococcal disease.11^–14 In a few other countries, including Norway, chemoprophylaxis is not recommended, but household members under 15 years of age are treated with penicillin orally.17 Neither approach has been evaluated in controlled studies.

When a case of meningococcal disease occurs, many people may fulfill accepted criteria for receiving chemoprophylaxis,11^–14 and chemoprophylaxis may be prescribed in excess of what is needed.14 High consumption of chemoprophylactic agents may select bacterial resistance, which in meningococci may be associated with virulence.19 Chemoprophylactic agents may also kill non-virulent meningococci and other bacteria that stimulate an immune response against the meningococci.20 Chemoprophylaxis should therefore be restricted to those who are likely to carry the pathogenic strain.

Our study shows that only 42 (2.7%) of 1535 close contacts carried the pathogenic strain of N meningitidis. General use of chemoprophylaxis in all these contacts therefore seems excessive. Sensitive and rapid techniques for identification of the pathogenic strain 8 9 allow targeting of chemoprophylaxis to carriers, but have not yet been widely applied as most laboratories lack the technology and resources to perform these tests. In most cases, therefore, the decision of whether to give chemoprophylaxis must be made on the basis of closeness of contact with the patient.

Our study shows that the risk of carriage of the pathogenic strain is highest (12.4%, 95% confidence interval 5.5% to 19.3%) in household members and kissing contacts. Household members have a high relative risk of meningococcal disease (1000-40001^–3) and the use of chemoprophylaxis in this group therefore seems well justified. Contacts outside this group, most of whom qualified for chemoprophylaxis according to accepted criteria,11^–14 had a considerably lower prevalence of carriage of the pathogenic strain (class 2, 1.9%, 0.9% to 3.4%; class 3, 1.6%, 0.14% to 3.1%). This is higher than the 0.7% prevalence in the general population, during periods of low disease incidence, 18 21 but not dramatically so. Should these contacts receive chemoprophylaxis? Our results do not support this practice. However, the relative risk of meningococcal disease is over 1000 for household members, although our results indicate that the carriage rate (12.4%) is only 18 times higher than that found in the general population (0.7%) in other studies. The relative risk of meningococcal disease is therefore not a simple function of the prevalence of the pathogenic strain. Another way to view the problem would be to ask whether the prevalence of the pathogenic strain approaches that needed to initiate epidemic disease. It has been suggested that a high rate of carriage is a prerequisite for epidemic disease,22 but the threshold is not known, and in any case the prevalence will vary from case to case. We therefore feel that the choice of whether to give chemoprophylaxis to contacts outside the patient's household and kissing contacts should be made on an individual basis, taking into account: other cases in the vicinity or other reasons to suspect an outbreak in the community; a high incidence of influenza or other respiratory infection that may predispose contacts to meningococcal disease and mask the symptoms of early infection; and other predisposing factors. If, however, an isolated case occurs in an otherwise healthy community, we believe that a conservative approach to chemoprophylaxis is justified.

Our study illustrates the need for better understanding of the relation between carrier rate and risk of secondary disease. A controlled study comparing different chemoprophylaxis strategies would in our view be of considerable help. The following strategies should be considered: (a) chemoprophylaxis according to standard recommendations, (b) chemoprophylaxis given only to household members and kissing contacts, and (c) chemoprophylaxis given to household members, kissing contacts, and other close contacts who are found to carry the pathogenic strain. Cost benefit analysis and studies of the prevalence of the pathogenic strain would enhance the value of such a study.

It has been argued that throat swabbing underestimates the true rate of meningococcal carriage.23 Low levels of bacteria in the sample, loss of viability under transport, and variable sampling techniques can all influence the measured carrier rate, but sampling and transport are probably the more important factors.24 We have addressed this problem by attempting to confine sampling to two well trained members of our staff, by plating samples directly after collection, and by sampling the throat, which has been reported to have 100% sensitivity relative to other sampling sites.25 A minority of the samples were collected by hospital and clinic staff and transported before plating. These are almost exclusively samples collected from members of the patient's household—that is, the group where we found the highest rates of carriage. Serious underestimation of carriage might be expected to lead to secondary cases among the contacts we sampled. None of the 1535 primary and 78 secondary contacts contracted meningococcal disease. We therefore do not think that sampling problems seriously affect this study.