Students from 15 to 19 years of age who were attending school or college full- or part-time (but not at university) were recruited from centers in 8 geographic regions throughout the United Kingdom (Table 1) as previously described (19). The study was approved by the Trent MultiCentre Research Ethics Committee. Culture-positive data were not available for the London center in 1999; consequently, this center was not included in the analysis for this study.

The local consultant in communicable disease control (public health) asked schools and colleges in their health authority area to participate in the study, with the aim of selecting a sample of schools that broadly represented the social diversity of that population. Each center trained staff to take pharyngeal swabs according to a standard protocol. The swabbing teams visited schools and colleges from October to December for 3 successive years. In year 1 of the study (1999), swabbing took place immediately before MenC vaccination. All students 15–19 years of age in the last 2 school years before university were eligible for the study. After obtaining signed informed consent from the student (or parent/guardian), swabbing teams took a pharyngeal swab, and the student completed a short questionnaire assessing risk factors for carriage. Participants were questioned about age, sex, home postal code, school year, number of persons and rooms in household (to derive persons/room), sharing of bedroom, previous vaccination with meningococcal polysaccharide vaccine, current and recent antimicrobial drug use, active smoking, passive smoking at home, number of days in the last week they had visited a pub or club, and number of people they had intimately kissed in the last week.

Using data from the 1991 census, the postal code of home residence was used to link each person to an electoral ward and its Carstairs deprivation score (20) (http://www.mimas.ac.uk). A higher score reflects a greater level of social deprivation.

The following information was requested about schools and colleges: type of establishment, selective or nonselective entry, independent or state funded, single sex or coeducational, day pupils with or without boarders, and school size (small [<200 pupils], medium [200–499 pupils], or large [>500 pupils]).

Here we present the results from the first year of the study; swabs were collected in November and December 1999. The results represent meningococcal carriage just before MenC immunization.

Swabs were plated onto selective medium either directly or within 6 hours and incubated in CO₂ at 37°C. Colonies resembling meningococci were identified by conventional tests, and oxidase-positive, gram-negative diplococci were frozen and stored in duplicate as putative meningococci at –70°C. Plates negative after 24 h were reincubated and examined again after 48 h. All isolates from England and Wales were sent to the Meningococcal Reference Laboratory, Manchester, for typing and subtyping. Scottish isolates were sent to the Scottish Pneumococcal and Meningococcal Reference Laboratory, Stobhill, Glasgow. A sample was considered positive if N. meningitidis was confirmed by the reference laboratory. The duplicate isolate was examined if a viable Neisseria sp. was not obtained from the initial isolate.

Data from questionnaires were entered twice and validated by using Epi Info version 6.0 (21). Data inconsistencies were found and corrected when possible, and efforts were made to clarify incomplete dates of birth. Individual-level risk factors for meningococcal carriage were initially analyzed in single-variable models by using logistic regression in the package Stata 8.0 (StataCorp, College Station, TX, USA). School-level risk factors were initially analyzed individually within a multilevel model with students at level 1 and schools at level 2. All risk factors with p<0.1 were then included in a multilevel logistic regression model for a multivariable analysis; again, individual students were at level 1, and schools were at level 2 of the model with explanatory variables at both levels. Center was regarded as having a fixed effect at the school level. A further analysis regarding center as a third level was undertaken but gave similar results, and the 2-level model is presented (the 3-level model is available on request). Each variable included was tested for significance by using the Wald test. Interactions between significant variables were also investigated. The multilevel analysis was carried out in MLwiN (22), and the gllamm command (23) in Stata 8.0 was used.

Meningococcal carriage increased with age (Table 2, Figure 1). Some social and behavioral factors (level 1 factors) had a strong positive association with meningococcal carriage, namely, cigarette smoking, exposure to passive smoke at home, intimate kissing of >1 persons, and attendance at pubs or clubs in the previous week (Table 2). Weak evidence was found of an association with Carstairs score and number of persons per room. Current or recent antimicrobial drug use was negatively associated with meningococcal carriage. School year was associated with carriage; however, this variable was highly correlated with age and was not considered in the multivariable model. No association was found between carriage and sex, sharing a bedroom, and previous meningococcal polysaccharide vaccination. School-level analysis (Table 3) showed variation in frequency of carriage between centers (7.7%–23.7%, p<0.001). Associations with school type and school size were also highly significant. The presence of boarders, the source of funding, and the gender mix showed no significant association.

In the multivariable analysis, strong associations were found with cigarette smoking, intimate kissing, pub or club patronage, and antimicrobial drug use (Table 4). The association with pub or club attendance showed a clear dose-response relationship. The association with passive smoking was not as strong but remained significant. The rise in carriage by age was much reduced after controlling for these other factors (Figure 1, Table 4). The 15-year-olds had the highest adjusted carriage, but the numbers in this age group were relatively small, and the significant trend in age is attributable to the rise in carriage from 16 years to 18 or 19 years. Associations with Carstairs score and persons per room were no longer significant. Of the level 2 factors, no school characteristics were linked to carriage, and only the association with center remained significant.

The analysis of interactions showed evidence of small, but significant interactions between smoking and kissing (p = 0.005) and also between smoking and pub or club attendance (p = 0.003). Investigation of the relationship between these 3 variables showed that crude carriage rates varied from 7.8% to 32.8% (Figure 2). The interaction effect appears to be due to relatively high carriage (20.5%) in teenagers whose only risk factor among these 3 is smoking. This analysis was repeated by calculating odds ratios from the multivariable analysis for all combinations of the 3 variables, and the pattern was similar to that seen with the crude carriage rates.