Association of respiratory tract infection symptoms and air humidity with meningococcal carriage in Burkina Faso

Transcription

1 Tropical Medicine and International Health doi: /j x volume 13 no 12 pp december 2008 Association of respiratory tract infection symptoms and air humidity with meningococcal carriage in Burkina Faso Judith E. Mueller 1, Seydou Yaro 2, Yoann Madec 3, Paulin K. Somda 2,Régina S. Idohou 1, Berthe-Marie Njanpop Lafourcade 1, Aly Drabo 2, Zekiba Tarnagda 4, Lassana Sangaré 5, Yves Traoré 6, Arnaud Fontanet 3 and Bradford D. Gessner 1 1 Agence de Médecine Préventive, Paris, France 2 Centre Muraz, Bobo-Dioulasso, Burkina Faso 3 Unité d Epidémiologie des Maladies Emergentes, Institut Pasteur, Paris, France 4 Institut de Recherche en Science de la Santé, Bobo-Dioulasso, Burkina Faso 5 Centre Hospitalier Universitaire Yalgado Ouédraogo, Ouagadougou, Burkina Faso 6 Université de Ouagadougou, Ouagadougou, Burkina Faso Summary objectives To evaluate risk factors for meningococcal carriage and carriage acquisition in the African meningitis belt, comparing epidemic serogroup A (NmA) to non-epidemic serogroups. methods During the non-epidemic meningitis season of 2003, pharyngeal swabs were taken at five monthly visits in a representative population sample (N = 488) of Bobo-Dioulasso, Burkina Faso (age 4 29 years) and analysed by culture. Standardized questionnaires were administered. In 2006, a similar study was performed in 624 individuals (age 1 39 years) during an NmA meningitis epidemic. We evaluated serogroup-specific risk factors for carriage, carriage acquisition and clearance using multivariate logistic and Poisson regression, and a Cox proportional hazard model. results The prevalence of NmA carriage (current or recent pharyngitis or rhinitis) was 16% (31%) vs. 0% (9%) in the epidemic vs. the hyperendemic setting. During the epidemic situation, NmA carriage was significantly associated with recent sore throat (adjusted odds ratio (OR), 3.41) and current rhinitis (OR 2.65). During the non-epidemic meningitis season in 2003, air humidity (20 39% and 40%, compared to <20%) during the month before swabbing was significantly and positively associated with carriage acquisition of non-groupable meningococci (OR 2.18 and 1.55) and inversely with carriage clearance (hazard ratio 0.61 and 0.27, respectively). conclusion Respiratory tract infections may increase meningococcal carriage, and thus contribute to epidemic risk, in addition to seasonality in the meningitis belt. Humid climate may favour carriage of unencapsulated meningococci. These findings may help identifying interventions against epidemic and hyperendemic meningococcal meningitis due to non-vaccine serogroups. keywords Neisseria meningitidis, Africa, carrier state, risk factors, disease outbreaks, respiratory tract infections, weather Introduction Meningococcal meningitis causes a substantial burden of disease in sub-saharan Africa. The meningitis belt is characterised by high incidence of hyperendemic meningococcal disease during the dry season (about December through May) and by periodic large epidemics, mainly due to serogroup A (NmA) (Greenwood 1999). Serogroups W135 (NmW135) and X (NmX) also cause occasional epidemics in the region (Boisier et al. 2007; World Health Organization 2008). Recently, a conjugate polysaccharide meningococcal vaccine against serogroup A has been developed for sub- Saharan countries (LaForce et al. 2007). This vaccine is expected to prevent meningitis epidemics not only through direct protection of the vaccinee, but also indirectly by reducing the risk for acquisition and transmission of pharyngeal carriage. To predict and evaluate the impact of conjugate vaccine immunization strategies, and to develop complementary, possibly non-serogroup-specific preventive strategies oriented against carriage, risk factors for meningococcal carriage during the meningitis season in ª 2008 Blackwell Publishing Ltd 1543

2 sub-saharan Africa need to be better known. Sequence type 2859, the genotype of a major part of NmA currently identified in the meningitis belt (Traoré et al. 2006; Caugant & Nicolas 2007), seems to be rarely carried outside of epidemics (Mueller et al. 2006; Raghunathan et al. 2006). We conducted a carriage study during an NmA epidemic in western Burkina Faso in 2006 that offered a rare opportunity to study prevalence and risk factors for NmA carriage. Carriage of unencapsulated, and thus non-groupable meningococci (NmNG) probably induces immunity against non-capsular meningococcal structures and therefore against encapsulated meningococci (Kremastinou et al. 1999). NmNG carriage prevalence varies greatly between different studies in sub-saharan Africa (Mueller et al. 2006; Leimkugel et al. 2007) and was related to seasonality in a previous carriage study in Bobo-Dioulasso during 2003 (Mueller et al. 2007). We used these data to evaluate factors associated with acquisition and clearance of NmNG carriage in this previous study. Methods Carriage study during hyperendemic period in 2003: longitudinal study During February to June 2003 [a meningitis season without occurrence of an epidemic (Traoré et al. 2006)], we conducted a longitudinal meningococcal carriage study among a random sample of healthy residents of urban Bobo-Dioulasso (Burkina Faso), with examinations at five monthly clinic visits. Ten of 22 neighbourhoods of Bobo- Dioulasso were randomly selected, as were 25 household communities (compounds) in each neighbourhood, and one subject aged 4 14 years and years each from each compound. Epidemiological and microbiological methods were reported in detail elsewhere (Mueller et al. 2006, 2007). At all five visits, swabs were taken from the posterior pharyngeal wall via the mouth, and analyzed by established bacteriological methods (Poolman & Abdillahi 1988; Tenover et al. 1995; Nicolas et al. 1997; Maiden et al. 1998; World Health Organization 1999; Taha 2000).We also administered questionnaires to all participants, recording sociodemographic information, personal and family medical history related to meningitis, meningococcal vaccination status, active and passive smoking, kitchen smoke exposure, professional activity, crowding in household and literacy (if aged 15 years or older). Children under 9 years of age had weight and height measured in a standardized way with a simple medical scale and measuring board. Children below the third percentile of weight for height were considered to have malnutrition (Ge & Chang 2001). For the study period, the Bobo-Dioulasso airport meteorological service provided daily meteorological data including minimum temperature, relative air humidity and wind speed. For each visit, average values for preceding periods were calculated (previous week, 2 weeks and 1 month) using various cut-offs for categorisation or continuous variables. These meteorological variables were separately used in models, and only the best-fitting model is presented. Carriage study during epidemic in 2006: cross-sectional study During February and March 2006, the rural health centers Lena and Kofila, situated approximately 60 km east of Bobo-Dioulasso, reported a meningitis epidemic. Using exhaustive PCR-based surveillance, we estimated average and peak weekly incidence rates of 84 and 247 per for NmA meningitis during March. A mass vaccination campaign with meningococcal group A C polysaccharide vaccine was conducted in the 2- to 29-yearold population with high coverage during March 12 15, From the approximately 10 villages attending the two health centers, we selected three (Konkourouna, Kofila and Lena) representing a total population of about Proportionally to population size, trained study workers conducted two-stage cluster sampling in each village. In Konkourouna and Kofila, starting from the village center, pathway directions were randomly selected, along which they included every third compound. In Lena, where a demographic information system existed, compounds were randomly selected from the compound list. In each selected compound one individual in each of four age groups (1 4, 5 9, and years) was randomly selected, asked to provide informed consent, and invited to come to the study visit the following day at a central place of the village. Procedures of recruitment and bacteriology were the same as in the 2003 study, while minor modifications occurred on the questionnaire. For <10-year-old children, mid-upper arm circumference (MUAC) and height were measured in a standardized way using a simple measuring tape and a measuring board. Malnutrition was defined as a MUAC-for-height below two standard deviations (Mei et al. 1997). During the study, meteorological data were collected at the Bobo-Dioulasso airport. Statistical analysis We used data from both studies to evaluate risk factors for serogroup-specific carriage. Controls were participants that did not carry the respective serogroup, or, for alternative analyses, who did not carry any meningococcus. We used 1544 ª 2008 Blackwell Publishing Ltd

3 logistic regression models, which included random effects for individual and compound effects, thereby taking into account the cluster sampling design. Serogroup-specific acquisition rates were estimated using a random effect Poisson model. For evaluation of risk factors for NmNG carriage acquisition, acquisition was defined as isolation of a nongroupable meningococcus from a participant at a particular visit, given that at the previous visit, the participant did not have meningococcal carriage of the same pheno- and genotype. Incidence of carriage acquisition was estimated using a random effect Poisson model; two random effects were introduced in the model to account for the correlation between one subject s measurements (individual effect) and the correlation between subjects from the same compound (compound effect). The time at risk was the time interval between two consecutive visits, or half of this interval in case of carriage acquisition, as carriage acquisition was assumed to occur at the midpoint between the visit without and the visit with carriage. Moreover, after an observed carriage episode at a given visit, the participant was assumed to be again at risk for carriage acquisition from 2 weeks after the visit on, according to the previously estimated carriage episode duration of 30 days (Mueller et al. 2007). Risk factors for NmNG carriage clearance at visits after the acquisition event were analysed using a Cox proportional hazard model. In all analyses, visits were not considered if a participant reported relevant antibiotic treatment within the previous 4 weeks. All available variables were evaluated first for association with carriage in univariate models, of which, with the exception of age, sex and symptoms of respiratory infection, only variables contributing at a significance level of P < 0.2 are shown in Tables 1 and 3 (complete tables can be obtained from authors). Variables contributing at a significance level of P < 0.2 were included in a full multivariate model, from which factors were eliminated by stepwise backward selection, using likelihood ratio tests, until only variables contributing at a significance level of P < 0.1 remained. Statistical analyses were performed on STATA version 8.0 and 9.0, using the GLLAMM module (available at (Rabe-Hesketh et al. 2002, 2005) for multilevel modelling. Ethical approval The study protocols were approved by the Centre Muraz ethics committee (2003), and the Burkina Faso National Ethics Committee (2006). Participants or their guardians (for those aged less than 18 years) provided written informed consent. Results Carriage study during hyperendemic period in 2003 During the longitudinal study in 2003, meningococcal carriage was observed at 152 of the 2327 study visits. Meningococcal isolates were attributed to NmW135 (N = 28, 18%), X (N = 5, 3%), Y (N = 3, 2%) and 116 (76%) were non-serogroupable, auto- or poly-agglutinable (NmNG). NmNG carriage prevalence increased from 1.6% in February to 8.6% in May June (Figure 1), while encapsulated meningococci showed a constant prevalence of around 1.5%. Carriage acquisition rates per 100 personmonths were estimated as 1.51 (95% confidence interval [CI] ) for any Nm, 0.27 ( ) for NmW135 and 1.34 ( ) for NmNG, respectively. The monthly average of minimal air humidity increased from 16% in February to 55% in June (overall range 4 72%) (Figure 1). The median minimal temperature was 25.0 C (range C), and median wind speed was 4 m s (range 1 6). Given the absence of substantial variation in wind speed, we did not evaluate it in the models. Carriage of any encapsulated meningococcus (W135, X or Y) was associated with male sex (OR 3.04 [95%-CI ], P = 0.037) and age 10 to 19 years [compared to 4 9 years: OR 0.29 ( )]. During the person-months of follow-up, 81 carriage acquisition events of NmNG were observed. In multivariate analysis, being a student was protective, while higher air humidity during the 2 weeks before assessment was associated with increased risk of carriage acquisition (Table 2). At least one follow-up visit was observed for 70 NmNG acquisition events, of which 50 eventually were cleared. Adjusting for age and neighbourhood, higher air humidity during the month prior to carriage assessment was associated with reduced clearance of NmNG carriage: 20 39%, hazard ratio (HR) 0.61 (95%-CI ); and 40%, HR 0.27 ( ). In all analyses, no significant effect modification was found and using as controls participants who did not carry any meningococcus did not change the estimates. Carriage study during epidemic in 2006 During March 13 28, 2006 we examined 624 healthy residents. 134 (22%) of the 617 persons without antibiotic treatment were meningococcal carriers. Serogroup-specific carriage prevalence was 15% for NmA (N = 95), 6% for NmY (N = 35), and 0.6% for NmNG (N = 4). No isolates were found for serogroups B, C, W135 or X. Median minimal air humidity in Bobo-Dioulasso during the study period was 7.5% (range 3 27%), median minimal temperature was 25 C (23 27 C) and the median wind speed 3m s (2 5 m s). In multivariate analysis, age groups ª 2008 Blackwell Publishing Ltd 1545

4 Observation time (person-months) (total = ) Number of carriage acquisition events Crude incidence rate ratio (95% CI) Age (years) ( ) ( ) Sex Female Male ( ) Student No Yes ( ) Commercial or market profession No Yes ( ) 0.16 Illiteracy (if 15 years) No Yes ( ) Localisation of kitchen Outside Inside ( ) Previous self-reported respiratory infection No Yes ( ) 0.64 Recent sore throat No Yes ( ) Pharyngitisà No Yes No estimate Rhinitisà No Yes ( ) 0.53 Coughà No Yes ( ) 0.77 Tonsillitisà No Yes No estimate Month in which observation period started February March ( ) April ( ) May ( ) Minimal air humidity (average during 2 weeks prior to visit) <20% % ( ) 40% ( ) Global P-value Table 1 Characteristics of 488 study participants examined at 5 visits, contributed person time and incidence rate for carriage acquisition of non-groupable meningococci. Bobo-Dioulasso, February through June 2003 Crude incidence rate ratio as estimated in two-level model. Self-reported information; at first visit: 2 months prior to visit, at following visits: since last visit. àobserved during swabbing. 1 3 years and years had the lowest odds for NmA carriage, and residence in Kofila and five or more persons sharing the bedroom were also inversely associated with current NmA carriage. Recent sore throat, observed current rhinitis, recent meningitis in the family and male sex were positively associated with NmA carriage 1546 ª 2008 Blackwell Publishing Ltd

5 Carriage prevalence (%) incidence rate (/10 5 ) Carriage NmNG Air humidity Weekly incidence Oct Dec Feb Apr Jun Calendar week and month Figure 1 Meningococcal carriage prevalence in 488 individuals at 5 monthly study visits between February and June; notified meningitis weekly incidence rates in the sanitary districts Secteur 15 and Secteur 22; and minimal relative air humidity (weekly average) measured at local airport. Bobo-Dioulasso, For the second through fourth study visit, participants were randomly assigned to two groups that were examined in 2-week-intervals. Table 2 Effect estimates for risk factors of carriage acquisition of non-groupable meningococci. Bobo-Dioulasso, February through June 2003 IRR (95% CI) (Table 4). If any previous or current symptoms combined or previous or current pharyngitis and rhinitis combined replaced previous sore throat and current rhinitis in the final model, the respective effect estimates were, OR 2.12 (95%-CI 1.25, 3.58), P = 0.005; and OR 3.41 (1.98, 5.87), P < Similar to symptoms of pharyngitis and rhinitis (Tables 1 and 3), prevalence of any previous or current symptoms combined or previous or current pharyngitis and rhinitis combined was significantly higher in March 2006 (44.37 and 31.27%, respectively) than in March 2003 (22.69% and 9.24%,respectively) (both differences P < 0.001) Air humidity (%) Global P-value Student No Yes 0.49 ( ) Kitchen localisation Outside Inside 0.63 ( ) Minimal air humidity (average during 2 weeks prior to visit) <20% % 2.18 ( ) 40% 1.55 ( ) Parsimonious multivariate models (variables P < 0.10). Variable does not remain in final model with selection P < NmY carriage was associated with daily exposure to cigarette smoke, number of persons sharing dinner and participation in a meeting of >10 persons during the previous week (Table 4). Discussion During an NmA meningitis outbreak in Burkina Faso, we found that in addition to several risk factors previously documented in Africa and the industrialized world (Hassan-King et al. 1979; Blakebrough et al. 1982; Yazdankhah & Caugant 2004), previous or current symptoms of upper respiratory tract infection were associated with carriage of the outbreak strain. Previous authors have described an association between meningococcal disease outbreaks and symptoms of respiratory tract infection, especially during influenza outbreaks (Young et al. 1972; Cartwright et al. 1991; Huber et al. 1992; Makras et al. 2001). A study during an NmA epidemic in Chad reported an association between NmA carriage and self-reported recent fever, but not recent sore throat or cough (Moore et al. 1990). In the same study, carriage of respiratory viruses and mycoplasma was associated with NmA meningitis, and to a lesser extent with NmA carriage. A study after an epidemic in Burkina Faso reported that coughing, nasal congestion or sore throat during the preceding month were associated with carriage of the NmW135 outbreak strain among persons 15 years of age (Raghunathan et al. 2006). Some authors have proposed that viral infection may weaken pre-existing mucosal or general immunity, thereby increasing the risk for invasive disease (Young et al. 1972; Moore et al. 1990; Cartwright et al. 1991; Alonso & Taha 2003). Moreover, as our results suggest, upper respiratory tract infections may increase transmission and mucosal adhesion in the pharynx of bacteria like meningococci (Young et al. 1972; Alonso & Taha 2003; Jacoby et al. 2007). The observed relative risk of 3 may seem too small to make respiratory infections alone an explanation of the Nm epidemic. However, on the individual level, the association may have been underestimated, as recall of sore throat may be low and meningococcal colonization may be rapidly cleared through pre-existing immunity. In the population, the occurrence of an upper respiratory infection outbreak in 2006 (as indicated by the higher prevalence of respiratory infection symptoms) may increase sufficiently the frequency of NmA transmission and carriage, such that a particular individual in the community need not have a respiratory infection to be subject to a higher risk of NmA carriage acquisition and invasive disease. Under specific conditions given during the dry ª 2008 Blackwell Publishing Ltd 1547

6 Table 3 Characteristics of 617 study participants by meningococcal carriage. Burkina Faso, March 2006 NmA (n = 95) NmY (n = 35) n n (%) carriers OR (95% CI) Global P-value n (%) carriers OR (95% CI) Global P-value Age (8) 0.35 ( ) (3) 0.72 ( ) (18) 1 9 (5) (19) 1.03 ( ) 8 (5) 1.14 ( ) (20) 1.12 ( ) 10 (12) 2.84 ( ) (9) 0.45 ( ) 3 (4) 0.86 ( ) Sex Female (11) (5) Male (20) 2.24 ( ) 17 (6) 1.09 ( Cigarette smoking No (15) (5) Yes 13 3 (23) 1.79 ( ) 2 (15) 3.15 ( ) Student No (19) (6) Yes (14) 1.46 ( ) 8 (5) 0.86 ( ) Meningococcal vaccination (A C) during No (15) (6) Yes (20) 1.52 ( ) 6 (5) 0.82 ( ) Participation in meeting >10 persons during the previous week No (16) (5) Yes (12) 0.79 ( ) 9 (9) 1.98 ( ) Recent respiratory infection No (14) (6) Yes (28) 2.51 ( ) 2 (4) 0.76 ( ) Recent sore throat No (13) 1 < (5) Yes (33) 3.33 ( ) 6 (7) 1.35 ( ) Cough during the previous 2 months No (13) (6) Yes (17) 1.34 ( ) 16 (5) 0.87 ( ) Recent flu symptoms No (16) (6) Yes 75 8 (11) 0.58 ( ) 4 (5) 0.94 ( ) Pharyngitisà No (15) (6) NE Yes (28) 2.40 ( ) 0 Rhinitisà No (14) (5) Yes (23) 1.79 ( ) 6 (6) 1.25 ( ) Coughà No (14) (6) Yes (28) 2.42 ( ) 4 (6) 1.07 ( ) Tonsillitisà No (15) (5) Yes (17) 1.13 ( ) 7 (9) 1.91 ( ) Meningitis case in compound during previous month No (12) 1 < (6) Yes (30) 3.34 ( ) 6 (5) 0.95 ( ) Number of persons sharing meal with participant (19) (2) (17) 0.81 ( ) 11 (5) 2.71 ( ) > (11) 0.47 ( ) 20 (8) 4.42 ( ) 1548 ª 2008 Blackwell Publishing Ltd

7 Table 3 (Continued) NmA (n = 95) NmY (n = 35) n n (%) carriers OR (95% CI) Global P-value n (%) carriers OR (95% CI) Global P-value 5 persons sharing bedroom No (17) (6) Yes (10) 0.50 ( ) 9 (6) 1.08 ( ) Number of persons living in compound < (30) (1) (13) 0.32 ( ) 21 (5) 4.50 ( ) (13) 0.31 ( ) 12 (8) 7.02 ( ) Kitchen localisation Outside (17) (5) Inside (12) 0.63 ( ) 10 (6) 1.02 ( ) Hours per day spent in kitchen smoke < (30) (7) (15) 0.38 ( ) 24 (6) 0.88 ( ) > (11) 0.25 ( ) 6 (4) 0.48 ( ) Daily exposure to cigarette smoke No (14) (5) Yes (21) 1.87 ( ) 11 (10) 2.28 ( ) Village Konkourouna (23) 1 < (3) Kofila (6) 0.18 ( ) 15 (7) 2.71 ( ) Lena (19) 0.75 ( ) 16 (5) 1.86 ( ) NE, no estimate due to empty cells. Univariate logistic regression models. Due to missing data, not all N in categories sum up to 617. Self-reported information during previous 2 months. àobserved during swabbing. season, this carriage surge may lead to a meningitis epidemic. Thus, respiratory infection outbreaks may be a trigger for a surge in meningococcal carriage leading to an epidemic. In addition, there may be increased risk for invasive meningococcal disease during viral infection. Several agents are known to occasionally cause epidemics of respiratory infections, including mycoplasma, respiratory syncytial virus, influenza virus, adenoviruses, and others (CDC Manual 2000). Although their peak incidence in the tropical zone is reported during the wet season, some outbreaks may occur during the dry season under specific conditions. Our study does not allow any conclusion as to whether the observed effect exists specifically for one pathogen or for respiratory viruses in general. Symptoms of respiratory tract infection in our study could have been caused by nasopharyngeal NmA colonisation (Lapeyssonnie 1963). However, symptoms during the month before carriage assessment were associated with carriage, which points to a separate disease entity. We did not include a control group in a non-epidemic condition, which would be required to show that the observed association is specific for carriage of outbreak strains. However, NmY carriage in our study was not associated with respiratory tract symptoms, and the results from the 2003 carriage study in nearby Bobo-Dioulasso outside an epidemic suggest that respiratory tract infection symptoms are not associated with higher risk of meningococcal carriage in general. The association between epidemic meningococcal carriage and respiratory tract infection epidemics is not necessarily specific for NmA, given reports for serogroup B, C, and W135 meningococcal disease in the context of viral epidemics (Young et al. 1972; Cartwright et al. 1991; Huber et al. 1992; Raghunathan et al. 2006), but NmA may be most susceptible for stimuli of transmission and colonisation. While a given districts in the meningitis belt will experience hyperendemic meningitis incidence during each dry season, epidemics occur irregularly and are often limited to a few villages. Clonal waves of new meningococcal strains and low strain-specific seroprevalence have been hypothesized to increase epidemic risk (Leimkugel et al. 2007), but they fail to explain why epidemics in Africa occur in such a geographically focussed manner. Focal outbreaks of respiratory tract infections could help to explain this characteristic. ª 2008 Blackwell Publishing Ltd 1549

8 NmA NmY OR (95% CI) Global P-value OR (95% CI) Global P-value Table 4 Risk factors of serogroup A and Y meningococcal carriage during an NmA epidemic. Burkina Faso, March 2006 Age (years) ( ) ( ) ( ) ( ) Sex Female Male 2.41 ( ) Village Konkourouna Kofila 0.25 ( ) Lena 0.93 ( ) Daily exposure to cigarette smoke No Yes 2.51 ( ) 5 persons sharing the bedroom with participant No Yes 0.44 ( ) Number of persons sharing the (evening) meal with participant ( ) > ( ) Participation in meeting >10 persons during previous week No Yes 2.16 ( ) Recent sore throatà No 1 < Yes 3.41 ( ) Rhinitis No Yes 2.56 ( ) Meningitis case in compound during last month No Yes 2.53 ( ) Parsimonious multivariate models (variables P < 0.10). Variable does not remain in final model with selection P < àself-reported information during previous 2 months. Observed during swabbing. Longitudinal analyses of the 2003 data allowed quantifying the association of air humidity and carriage while taking into account other time-dependent variables and correlations induced by sampling design. It is possible that the doubling of NmNG acquisition risk with air humidity 40% compared to lower levels could be explained by some other unknown calendar-related factor, e.g., school vacation, but this factor would not explain the observed gradual increase in carriage or the association of higher air humidity with maintenance of carriage after acquisition as shown in the survival analysis. The correlation between climate and meningitis has been extensively studied, with evidence supporting a causal relation between low air humidity during the dry season and meningococcal meningitis epidemiology (Thompson et al. 2006). This may result from more frequent meningococcal transmission, as suggested by a study showing higher bacterial load in dry climates (Ghipponi et al. 1971), or more frequent invasion of bacteria that colonise pharyngeal mucosa damaged by dry climate (Greenwood 1999). In our study, higher air humidity seems to have favoured NmNG carriage, or conversely, lower air humidity reduced acquisition of 1550 ª 2008 Blackwell Publishing Ltd

9 NmNG carriage and favoured their clearance from the pharynx. One possible mechanism for this is that dryer air irritates the mucosa, leading to inflammation, better perfusion with blood and increased local presence of anti-meningococcal antibody. Unencapsulated meningococci may have lower resistance against antibody-mediated clearance from the mucosa than encapsulated ones, leading to a changed proportion of unencapsulated to encapsulated strains, even without competition between meningococcal strains. This could explain the lower NmNG carriage prevalence in dry climate, while virulent meningococci showed no variation over time. In summary, dry climate could change the risk of invasive meningococcal disease by increasing the ratio of encapsulated to unencapsulated Nm in carriage, which in turn would reduce the chance for harmless contact with the bacterium and subsequent natural immunity. While awaiting high-coverage and long-lasting immunization against all meningococcal serogroups with epidemic potential in Africa (currently serogroups A, W135, X), other interventions may exist to reduce hyperendemic meningitis during the dry season and the risk of meningococcal epidemics. These interventions may target to reduce harmful effects of dry air, such as humidification of rooms, vapour inhalation or nasal washes during the dry season. Furthermore, if influenza virus was involved in the observed respiratory tract infection outbreaks, influenza vaccine may have a role and should be evaluated in a controlled trial in the meningitis belt. In Burkina Faso, we found evidence that supports a relation between respiratory tract infections and epidemic carriage of meningococci. Our data are also consistent with a causal association between higher air humidity, carriage of non-virulent meningococci and lower meningitis incidence, although further studies are needed for confirmation. After introduction of preventive group A conjugate vaccine in the meningitis belt, non-vaccine prevention strategies may exist that could decrease the risk of meningococcal meningitis due to other serogroups. Acknowledgements We thank the participants, the study field teams, and the Meteorological Service at the Bobo-Dioulasso airport, as well as the sanitary authorities of Secteur 15 and 22, Hauts-Bassins region and Burkina Faso. Special thanks to Dr Sita Kroman who conducted the study monitoring. The studies were funded by the Institute Pasteur, Paris, and the Bill and Melinda Gates Foundation (2003), and Sanofi Pasteur (2006). JEM, RSI, BMNL and BDG work for AMP, which receives substantial financial support from Sanofi Pasteur. Other authors do not have any association that might pose a conflict of interest. References Alonso JM & Taha MK (2003) Viroses respiratoires et surinfection bactériennes invasives. Archives de Pédiatrie 10, Blakebrough IS, Greenwood BM, Whittle HC, Bradley AK & Gilles HM (1982) The epidemiology of infections due to Neisseria meningitidis and Neisseria lactamica in a northern Nigerian community. The Journal of Infectious Diseases 146, Boisier P, Nicolas P, Djibo S, et al. (2007) Meningococcal meningitis: unprecedented incidence of serogroup X-related cases in 2006 in Niger. Clinical Infectious Diseases 44, Cartwright KA, Jones DM, Smith AJ, Stuart JM, Kaczmarski EB & Palmer SR (1991) Influenza A and meningococcal disease. Lancet 338, Caugant DA & Nicolas P (2007) Molecular surveillance of meningococcal meningitis in Africa. Vaccine 25(Suppl. 1), A8 A11. Control of Communicable Diseases Manual (2000) Control of Communicable Diseases Manual (ed. J Chin). American Public Health Association, Washington. Ge KY & Chang SY (2001) Definition and measurement of child malnutrition. Biomedical and Environmental Sciences 4, Ghipponi P, Darrigol J, Skalova R & Cvjetanovic B (1971) Study of bacterial air pollution in an arid region of Africa affected by cerebrospinal meningitis. Bulletin of the World Health Organization 45, Greenwood B (1999) Meningococcal meningitis in Africa. Transactions of the Royal Society of Tropical Medicine and Hygiene 93, Hassan-King M, Greenwood BM & Whittle HC (1979) An epidemic of meningococcal infection at Zaria, northern Nigeria. 3. Meningococcal carriage. Transactions of the Royal Society of Tropical Medicine and Hygiene 73, Huber B, Watier L, Garnerin P & Richardson S (1992) Meningococcal disease and influenza-like syndrome: a new approach to an old question. The Journal of Infectious Diseases 166, Jacoby P, Watson K, Bowman J, et al. & on behalf of the Kalgoorlie Otitis Media Research Project Team (2007) Modelling the co-occurrence of Streptococcus pneumoniae with other bacterial and viral pathogens in the upper respiratory tract. Vaccine 25, Kremastinou J, Tzanakaki G, Pagalis A, Theodondou M, Weir DM & Blackwell CC (1999) Detection of IgG and IgM to meningococcal outer membrane proteins in relation to carriage of Neisseria meningitidis or Neisseria lactamica. FEMS Immunology and Medical Microbiology 24, LaForce FM, Konde K, Viviani S & Préziosi MP (2007) The Meningitis Vaccine Project. Vaccine 25(Suppl. 1), A97 A100. ª 2008 Blackwell Publishing Ltd 1551

10 Lapeyssonnie L (1963) La méningite cérébro-spinale en Afrique. Bulletin of the World Health Organization 28(Suppl. 1), Leimkugel J, Hodgson A, Forgor, et al. (2007) Clonal waves of Neisseria colonisation and disease in the African meningitis belt: eight-year longitudinal study in northern Ghana. PLoS Medicine 4, e101.doi: /journal.pmed Maiden MCJ, Bygraves JA, Feil E, et al. (1998) Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proceedings of the National Academy of Sciences of the United States of America 95, Makras P, Alexiou-Daniel S, Antoniadis A & Hatzigerogiou D (2001) Outbreak of meningococcal disease after an influenza B epidemic at a Hellenic Air Force Recruit Training Center. Clinical Infectious Diseases 33, Mei Z, Grummer-Strawn LM, de Onis M & Yip R (1997) The development of a MUAC-for-height reference, including a comparison to other nutritional status screening indicators. Bulletin of the World Health Organization 75, Moore PS, Hierholzer J, DeWitt W, et al. (1990) Respiratory viruses and mycoplasma as cofactors for epidemic group A meningococcal meningitis. The Journal of the American Medical Association 264, Mueller JE, Yaro S, Traoré Y, et al. (2006) Neisseria meningitidis: carriage and immunity in Burkina Faso, The Journal of Infectious Diseases 193, Mueller JE, Sangaré L, Njanpop-Lafourcade BM, et al. (2007) Molecular characteristics and epidemiology of carried meningococci in Burkina Faso, Emerging Infectious Disease Journal 13, Nicolas P, Parzy D & Martet G (1997) Pulsed-field gel electrophoresis analysis of clonal relationships among Neisseria meningitidis A strains from different outbreaks. European Journal of Clinical Microbiology and Infectious Diseases 16, Poolman JT & Abdillahi H (1988) Outer membrane protein serosubtyping of Neisseria meningitidis. European Journal of Clinical Microbiology and Infectious Diseases 7, Rabe-Hesketh S, Skrondal A & Pickles A (2002) Reliable estimation of generalized linear mixed models using adaptive quadrature. The Stata Journal 2, Rabe-Hesketh S, Skrondal A & Pickles A (2005) Maximum likelihood estimation of limited and discrete dependent variable models with nested random effects. Journal of Econometrics 128, Raghunathan PL, Jones JD, Tiendrebéogo SR, et al. (2006) Predictors of immunity after a major serogroup W-135 meningococcal disease epidemic, Burkina Faso, The Journal of Infectious Diseases 193, Taha MK (2000) Simultaneous approach for nonculture PCR-based identification and serogroup prediction of Neisseria meningitidis. Journal of Clinical Microbiology 38, Tenover FC, Arbeit RD, Goering RV, et al. (1995) Interpreting chromosomal DNA restriction patterns produced by pulsed-field gel electrophoresis: criteria for bacterial strain typing. Journal of Clinical Microbiology 33, Thompson MC, Molesworth AM, Djingarey MH, Yameogo KR, Belanger F & Cuevas LE (2006) Potential of environmental models to predict meningitis epidemics in Africa. Tropical Medicine and International Health 11, Traoré Y, Njanpop-Lafourcade BM, Adjogble KLS, et al. (2006) The rise and fall of epidemic Neisseria meningitidis serogroup W135 meningitis in Burkina Faso, Clinical Infectious Diseases 43, World Health Organization (1999) Laboratory methods for the diagnosis of meningitis caused by Neisseria meningitidis, Streptococcus pneumoniae, and Haemophilus influenzae: WHO Communicable Disease Surveillance and Response. (WHO CDS CSR EDC 99.7). World Health Organization, Geneva. World Health Organization (2008) Epidemic and Pandemic Alert and Response (EPR). Meningococcal disease. (cited 2008 April 4) Available from meningococcal_disease/en/index.html Yazdankhah SP & Caugant DA (2004) Neisseria meningitidis: an overview of the carriage state. Journal of Medical Microbiology 53, Young SL, LaForce FM, Head JJ, Feeley JC & Bennett JV (1972) A simultaneous outbreak of meningococcal and influenza infections. The New England Journal of Medicine 287, 5 9. Corresponding Author Dr Judith E. Mueller, Agence de Médecine Préventive, rue du Dr Roux, Paris cedex 15, France. Tel.: ; Fax: ; jmueller@aamp.org 1552 ª 2008 Blackwell Publishing Ltd