INTRODUCTION
There is interest in whether maternal and perinatal factors influence the occurrence of infectious disease later in life [Reference Yuan1–Reference Mahon4]. A recent study found that low birth weight and pre-term birth are associated with an increased risk of hospitalization for infections during childhood [Reference Yuan2]. It has also been reported that maternal smoking and low maternal body mass index pre-pregnancy increased the risk of subsequent hospitalization for infectious disease in children up to age 5 years [Reference Yuan1].
It is well documented that low birth weight predisposes to coliform meningitis in neonates [Reference Goldacre5], but less is known about whether perinatal and maternal factors influence the risk of meningitis in later infancy and childhood. Suggested risk factors for meningitis in children include smoking in the home, low social class, low birth weight, overcrowding and having older siblings [Reference Sorensen3, Reference Yusuf6–Reference Heyderman10].
The aims of this study were to investigate associations between maternal and perinatal risk factors and subsequent hospitalization for meningococcal, Haemophilus or enteroviral meningitis in the child. These were selected because we were interested in meningitis and they were the three most common causes of meningitis in the dataset used. We used record linkage data from a historical archived dataset in a large geographically defined English population.
METHODS
We used the specialist maternity dataset of the Oxford record linkage study (ORLS). The ORLS included inpatient hospital records, and records of hospital-based day-case care, in an area of South East England. Standard data collection in the ORLS was undertaken in two health districts from 1963 to 1999 (population 0·9 million in 1999) and, from 1975 to 1999, in a further four adjacent districts (total population 1·9 million). The characteristic that distinguished the ORLS from other data collection systems in England was that the data were record-linkable; and they are now in a fully linked archived data file. In other respects, the data are similar to those in routine hospital administration systems in many industrialized countries. In addition, a specialist data collection system for maternity – used as the basis of this study – covered all births in National Health Service (NHS) hospitals in Oxfordshire and West Berkshire from 1970 to 1989.
The maternity data from 1970 to 1989 were abstracted from maternity records by clerical staff who were trained by senior medical staff at the ORLS. The records of each mother and her baby were routinely linked by the use of unique numbers for each mother and each baby that were assigned to both the baby's and the mother's record at the time of birth. Detailed data collection on maternities in the ORLS area ceased in 1989 following government reforms to increase the uniformity of NHS data collection systems. Data collection for general hospital admissions in the ORLS area continued but, with further changes to NHS information systems in 1999, it is not possible to link pre- and post-1999 ORLS data.
We identified cases of meningococcal meningitis from the diagnoses recorded on hospital admission records, using International Classification of Diseases (ICD) codes 036·0 in the eighth and ninth revisions (ICD-8, ICD-9), and A39·0 in ICD-10. We identified cases of Haemophilus meningitis using ICD codes 320·0 in the 8th and 9th revisions and G00·0 in the 10th revision. We identified cases of enteroviral meningitis using ICD codes 045 in the eighth revision, 047 in the ninth revision and A87·0 and A87·9 codes in the tenth revision.
There were 253 060 pregnancies with a maternity record in the ORLS from 1970 to 1989. Of these, we excluded 985 that were recorded as having ended in abortion, 1560 stillbirths and 1567 neonatal deaths (none of the latter had a diagnostic code in the above list for meningitis). We also excluded 289 maternities in which the birth weight was recorded as <1000 g because many of these records had implausibly low values and/or considerable missing data for some of the other variables. After applying these exclusion criteria there were 248 659 children in the study.
Initial analyses were undertaken on fine groupings of the maternal and perinatal variables and, following scrutiny of whether the fine groups added anything material, broader groupings (as shown in the Tables) were selected for further analysis and presentation herein. Breastfeeding data were recorded in the original ORLS records as mother breastfeeding at discharge. The social class data were based on occupational social class, collected as the occupation of the head of the household (data on education or income was not available). Some of the data items were not collected through the whole period: notably, data on maternal smoking, breastfeeding and social class were only collected after 1975.
The significance of univariate associations was tested using χ 2 tests. When using logistic regression for multivariate modelling, all variables that were significant (P < 0·05) in the univariate analysis were included in the initial model and the variables that were not significant were removed. In further modelling, each variable that was not significant in univariate analysis was re-introduced, one at a time, into the model. The purpose of this was to test whether any variable, not significant in univariate analysis, became so when modelled with the other significant variables. We did not include maternal meningitis in the model because the number of cases of meningitis in both mother and child was too small to be meaningful to model. Year of the child's birth was included in all models to take account of different lengths of follow-up for children born in different years.
Approval for the research programme of the Unit of Health-Care Epidemiology using the anonymized dataset of the Oxford Record Linkage Study was obtained from the Central and South Bristol Research Ethics Committee (04/Q2006/176).
RESULTS
Of the 248 659 children in the dataset, 127 children had a subsequent admission for meningococcal meningitis, 160 for Haemophilus meningitis and 312 children for enteroviral meningitis. Table 1 shows the age and sex distribution of cases. There were no neonatal cases. In children with meningococcal meningitis, 33·1% were admitted aged <1 year, 26% were admitted aged 1–4 years and most of the rest were admitted aged ⩾15 years (25·2%). For those with Haemophilus, 51·3% were admitted aged <1 year, and 46·9% were admitted aged 1–4 years. For enteroviral meningitis, the spread of cases was much more evenly split between the age groups, with 29·2% admitted aged <1 year, 18·3% aged 1–4 years, 27·2% aged 5–9 years, 12·8% aged 10–14 years and 12·5% aged ⩾15 years. Of those with meningococcal meningitis 53·5% were males, as were 58·1% with Haemophilus meningitis and 62·2% with enteroviral meningitis.
Number of children admitted for meningococcal, Haemophilus or enteroviral meningitis showing age at admission and sex
Meningococcal meningitis
There were 41 patients with meningococcal meningitis born during 1970–1974 (0·06% of all births in the period, N = 68 939), 28 born during 1975–1979 (0·05% of births, N = 56 492), 23 during 1980–1984 (0·04%, N = 60 394), and 35 during 1985–1989 (0·06%, N = 62 834). There was no consistent trend in the incidence of meningococcal meningitis across birth cohorts (χ 1 2 linear association = 0·37, P = 0·54). In univariate analysis, meningococcal meningitis was significantly associated with low maternal social class, with mothers who smoked in pregnancy, and with increasing numbers of siblings (Table 2). Only 17·5% of mothers of children who subsequently developed meningococcal meningitis were in social classes 1 or 2 (the most affluent) compared to 35·8% in the whole study population (Table 2). Unmarried motherhood and lack of breastfeeding were significantly, but modestly, associated with meningococcal meningitis (Table 2). Several of the perinatal factors were themselves inter-related (e.g. social class, maternal smoking, parity, birth weight). In the multivariate model, the only factor that remained significantly associated with hospital admission for meningococcal meningitis after taking account of all other factors in the model (which included social class), was maternal smoking [odds ratio (OR) 2·1, 95% confidence interval (CI) 1·2–3·7]. Within each band of social class, children who developed meningococcal meningitis were more likely than other children to have mothers who smoked (Table 3). For example, considering mothers in social classes 1 or 2 (the most affluent classes), 36·4% of mothers whose children developed meningococcal meningitis smoked compared to 12·5% of mothers whose children did not develop meningitis. Considering mothers in social classes 4 and 5 (least affluent), the respective percentages were 44·4% and 32·0%.
Maternal and perinatal characteristics of babies who, in later life, were admitted to hospital with meningococcal, Haemophilus or enteroviral meningitis
The numbers for each characteristic do not always add to the total number of children in the study as some records did not contain the corresponding information (notably, because the data items for the characteristic were only collected in certain calendar years).
† All cases were enteroviral.
* P value, for heterogeneity within the risk factor group, between 0·02 and 0·05; ** P value between 0·001 and 0·02; *** P < 0·001.
Comparisons within social class groups of whether the child had meningococcal meningitis and whether the mother smoked during pregnancy
Haemophilus meningitis
Considering those who developed Haemophilus meningitis, 23 were born during 1970–1974 (0·03% of all births in the period), 31 (0·05%) during 1975–1979, 43 (0·07%) during 1980–1984 and 63 (0·1%) during 1985–1989 respectively. An increasing linear trend in incidence was found across birth cohorts (χ 1 2 linear association = 23·8, P < 0·001). Parity was significantly associated with hospital admission for Haemophilus meningitis in children: Haemophilus meningitis was uncommon in first-born children. Only 18% of children with Haemophilus meningitis were first-born, compared to 42% of all children in the study. Low social class was significantly associated with admission for Haemophilus meningitis, as was mode of delivery (Table 2). In the multivariate model, parity was the only factor that remained significant. Compared to first-born (as the reference group), the odds of Haemophilus meningitis in second-, third-, fourth- and fifth-born children were, respectively, 3·31 (95% CI 1·97–5·57), 3·25 (95% CI 1·77–5·96), 4·68 (95% CI 2·23–9·86) and 3·12 (95% CI 1·06–9·21).
Enteroviral meningitis
There were 108 people born during 1970–1974 diagnosed with enteroviral meningitis (0·16% of all births), 91 (0·16%) during 1975–1979, 71 (0·12%) during 1980–1984, and 42 (0·07%) during 1985–1989. A decreasing linear trend was found across birth cohorts, (χ 1 2 linear association = 24·5, P < 0·001). Enteroviral meningitis in the child was significantly associated with maternal parity, maternal smoking during pregnancy, babies’ low birth weight, non-vertex presentation at birth, and a maternal history of enteroviral meningitis (Table 2). It was also considerably more common in male than female offspring (Table 2). The majority of children with enteroviral meningitis were first or second born. In the multivariate model, male sex (OR 1·7, 95% CI 1·2–2·3), low birth weight (OR 2·2, 95% CI 1·3–3·6; <2500 g vs. ⩾2500 g) and presentation other than vertex (OR 1·9, 95% CI 1·1–3·3) remained the only significant risk factors for enteroviral meningitis. There were six families in which the mother and her child both had hospital admissions for enteroviral meningitis. In all six families, the mother's admission occurred before that of the child. The time intervals between admission of mother and child were 4 days, 61 days, 1½ years, 2½ years, 5 years, and 20 years.
DISCUSSION
A strength of this analysis is that it was undertaken in a large, defined population that covered over 30 years of data collection and almost 250 000 births. Another strength is that information about perinatal risk factors and the main outcome measures – meningococcal, Haemophilus and enteroviral meningitis in the child – were originally recorded independently of each another. They were subsequently brought together by record linkage. This means that data about each risk factor could not have been influenced by the presence or absence of the outcome measure, as could be the case in interview-based studies of people with meningitis. The study is therefore not subject to potential biases, such as interviewer and recall bias, that can affect interview-based case-control studies.
We acknowledge the limitations of our study design. We were not able to identify data from resident children who may have been admitted to hospital outside the ORLS area or about children who were admitted with meningitis after moving outside the ORLS area. We do not have information on the diagnostic criteria or methods used in making a clinical diagnosis or laboratory confirmation of the diagnosis. The sensitivity of the diagnostic method used for enteroviral meningitis during most of the study period, most likely viral culture of cerebrospinal fluid (CSF), is less accurate than more recent CSF polymerase chain reaction (PCR) methods [Reference Sawyer11, Reference Jeffery12]. In addition, we do not have follow-up information on breastfeeding after discharge and as a result have no information on the duration or exclusivity of breastfeeding. We have no way of testing the validity of the data on breastfeeding.
To test the likely validity of the data on maternal smoking, we undertook a separate analysis of the smoking status of mothers in the dataset with inflammatory bowel disease. We did this because it is well recognized that smoking is associated with an increased risk of Crohn's disease and that it protects against ulcerative colitis [Reference Mahid13]. In the whole ORLS perinatal dataset from 1975 to 1989 (during which times hospitals were asked to record maternal smoking) there was a smoking history for 146 811 pregnancies. In these, 23·7% (34 728) of mothers smoked (34 728/146 811). Considering pregnancies of mothers with Crohn's disease, 37·3% (62/166) of the mothers smoked; considering pregnancies of mothers with ulcerative colitis, only 7% (9/129) smoked. We conclude that the smoking data in our dataset are likely to be valid and that the ORLS data on maternal smoking in pregnancy are good proxies for longer-term smoking by the woman.
The data are, of course, archival and old. Nonetheless they include data items that are not available in routinely collected hospital maternity statistics even now in England – notably, the mother's smoking history, whether the baby was breastfed, the family's occupational social class, as well as such items as ABO blood group and rhesus. The principal findings – smoking and meningococcal disease, family size and Haemophilus, familial occurrence of enteroviral meningitis – are unlikely to be affected by whether the data are recent or old. It is possible, however, that recent introduction of immunization against some meningococcal strains and against Haemophilus may have modified the impact of such factors in the modern era. The programme of MenC vaccination was initiated in 1999, and its effects were therefore outside the scope of our analysis. Widespread use of immunization against Haemophilus influenzae was largely introduced in 1992 and largely post-dates our study period [Reference Ladhani14].
Meningococcal meningitis
Low social class and maternal smoking were the strongest predictors of meningococcal meningitis in children. Previous studies have shown socioeconomic factors to be associated with meningococcal risk [Reference Heyderman10, Reference Kriz, Bobak and Kriz15–Reference Stuart17]. A study by Yusuf et al. [Reference Yusuf6] found an association between factors such as unmarried motherhood and low maternal education (<12 years schooling) with meningococcal disease [Reference Yusuf6]. In our study, a parity of three or more and unmarried motherhood were also risk factors for meningococcal meningitis, but, when social class was taken into account, they were not independent predictors of risk. We add further evidence that maternal smoking is an independent predictor of meningococcal meningitis in children [Reference Sorensen3, Reference Yusuf6, Reference Kriz, Bobak and Kriz15, Reference Fischer18]. Our adjusted odds ratio for this, at 2·1, is very similar to that found in a Danish record-linkage study (OR 1·8) [Reference Sorensen3]. The Danish study also found low birth weight and premature birth to be risk factors for meningococcal disease [Reference Sorensen3].
Haemophilus meningitis
An important determinant of admission of a child to hospital for Haemophilus meningitis was having older children in the family. First-born children were significantly less likely to have Haemophilus meningitis than children with older siblings. It has been previously reported that the occurrence of Haemophilus meningitis in single-child families is very low, suggesting that older siblings introduce it into the home [Reference Goldacre19]. Siblings [Reference Takala20] and household overcrowding [Reference Takala and Clements7, Reference Pereiro8, Reference Cochi21, Reference Arnold, Makintube and Istre22] have been attributed as risk factors for Haemophilus infection, with a dose-response effect reported for increasing risk with increasing numbers of children sharing a bedroom [Reference Arnold, Makintube and Istre22]. We found an increasing linear trend in H. influenzae over the time period studied. An increase over time was also reported in Sweden prior to the introduction of Hib vaccine [Reference Silfverdal, Bodin and Olcen23, Reference Salwen, Vikerfors and Olcen24]. While research in the USA found that an increased risk of contracting H. influenzae was associated with the use of daycare centers [Reference Redmond and Pichichero25], the Swedish research suggested that the increase in the incidence of H. influenzae could not be explained by the use of child daycare services themselves [Reference Silfverdal, Bodin and Olcen23]. The Swedish researchers suggested changes in bacterial virulence or host susceptibility might have contributed to the observed increase [Reference Salwen, Vikerfors and Olcen24], or it may be due to a low breastfeeding rate in the population as the incidence was correlated with breastfeeding rates over time in a negative way [Reference Silfverdal, Bodin and Olcen23]. Although our results do not show a significant relationship between breastfeeding and Haemophilus meningitis, previous studies have found breastfeeding to be a strong protective factor against H. influenzae [Reference Takala and Clements7, Reference Cochi21, Reference Silfverdal, Bodin and Olcen23, Reference Salwen, Vikerfors and Olcen24, Reference Silfverdal26].
Enteroviral meningitis
Data on perinatal risk factors for enteroviral meningitis are sparse. We confirm the findings of a record-linkage study performed in Denmark that reported that male sex and low birth weight were associated with non-polio enteroviral meningitis [Reference Hviid and Melbye9]. Other studies have noted a strong association with male sex [Reference Lee27, Reference Lin28]. The Danish study also reported several other risk factors for enteroviral meningitis including a large number of younger children in the household, gestational age, Apgar score, Caesarean section, season, age, calendar period, number of adults in the household, and urbanization [Reference Hviid and Melbye9]. We cannot offer an explanation for our finding of an association with non-vertex presentation. It is possible that, although significant, it is nonetheless a chance finding. A study in the USA found that meningitis in enterovirus-infected infants was more frequent in those with a certain enteroviral serotype (echoviruses 30, 11, and type B coxsackie viruses) [Reference Dagan, Jenista and Menegus29]. Transmission of enterovirus may occur nosocomially or through vertical transmission from mother to child (either transplacentally or at the time of delivery) [Reference Hawkes and Vaudry30]. Most mother-and-child pairs in our study were remote in time suggesting either familial susceptibility or longstanding carriage, or both.
CONCLUSIONS
In summary, this study used record-linkage analysis to examine maternal and perinatal risk factors for subsequent hospitalization for meningococcal, Haemophilus, and enteroviral meningitis in Oxfordshire over a 30-year period. We confirm the findings of previous studies that low social class and maternal smoking are risk factors for meningococcal meningitis. This study also shows that Haemophilus meningitis tends to be associated with having older siblings in the household. Finally, hospitalization for enteroviral meningitis was associated with a number of factors including enteroviral meningitis in the mother.
ACKNOWLEDGEMENTS
Over many years, the linked data files were built by Leicester Gill, Matt Davidson and Myfanwy Griffith, Unit of Health-Care Epidemiology, University of Oxford. This work was supported by the English National Institute for Health Research (grant number, Department of Health ref. RNC/035/02). This was an independent study, the funding source had no involvement in the study design, data collection, data analysis and interpretation, writing of the report, or the decision to submit the article for publication. The views expressed are not necessarily those of the funding body.
DECLARATION OF INTEREST
None.
