1 Background

Increasing levels of allergic disease amongst children in Westernised countries has lead to proposal of the 'Hygiene Hypothesis' [1, 2]. This suggests that a lack of exposure to common infection in early life leads to an increased risk of atopic sensitisation and allergic disease development. Investigation of the relationship between demographic factors (such as number of siblings, breastfeeding, hygiene levels, day care attendance) and infectious symptoms will lead to greater understanding of the factors that affect the level and type of infectious symptoms experienced in childhood.

Information on the incidence of overall common infection in infants is sparse, most data are collected relating to serious infections resulting in hospitalisation and serious morbidity or mortality. Others concentrate on one very specific, clinically diagnosed infectious disease, for example measles. Data from medical records will underestimate incidence of infections, as they do not include those that don't require a visit to the doctor. The ALSPAC study has collected data on the parental-reported information on various common medical conditions and occurrences that may lead to a visit to the GP including symptoms of infectious disease, accident, fits or wheezing [3]. Questionnaires were sent to parents when the child was 6, 18, 30, 42 and 57 months old with questions on 14 items. Previous research using the ALSPAC data-set has shown that levels of one symptom, cough, did not vary by maternal education but the percentage of those children taken to the GP or given medication decreased with increasing maternal education. The same paper also found children with a larger number of older siblings were more likely to have been reported to have a cough in the first 6 months of life [4]. Further description of the relationship between common infectious symptoms will help to investigate which children are most likely to have such symptoms and their relationship with attendance at the doctor.

2 Study aim and hypotheses

The aim of this research will be to use data collected within the Avon Longitudinal Study of Parents and Children to investigate relationships between reported levels of common infection requiring and not requiring a visit to the GP and demographic, social and environmental factors that may influence infection levels including day care attendance, hygiene practices, siblings and population mixing.

The hypothesis to be studied will be:

Do the number and type of infections reported in early life vary due to demographic factors such as number of siblings, day-care attendance and population mixing?

Within the United Kingdom there are few data available on the prevalence of common infectious symptoms experienced within early life in the general population and therefore little is known as to how well other factors relate them. The work carried out within this project will allow quantitative assessment of the most common infectious symptoms in young children and their relationship with proxy measures of infection used in epidemiology such as day care attendance. It will also allow an investigation into underlying classifications of infectious disease and symptom presentation.

3 Study design

3.1 Data

The work carried out will be using data already collected and available within the ALSPAC cohort. The ALSPAC study has collected data at 6 and 18 months on several symptoms and consultations in children including:

* Diarrhoea

* Vomiting

* Cough

* High temperature

* Snuffles/cold

* Ear ache

* Ear discharge

* Colic

* Rash

Parents then ticked one of the following choices for each symptom: 'yes and saw a doctor', 'yes but did not see a doctor' or 'no did not have'. Those listed above represent symptoms which could be linked to infections.

Information was also collected on the age and sex of other children living in the house when the child was aged 6 months and 18 months. At 15 months a questionnaire was administered including a section on childcare asking who regularly looks after the infant apart from the person completing the questionnaire. The choices were: partner, baby's grandparent, other relative, friend/neighbour, paid person outside the home, paid person in the baby's home, day nursery, other. Also collected were hours per week this occurred and age of the baby in months when it started. Also for each months of the child's life (up to 15 months) a chart is completed giving the number of hours of outside childcare, the person carrying this out and at what place. Residential postcode will be used to link to the 2001 census area in which the child resides and from this information about the area can be collected including population density, deprivation level and population mixing. Other general variables as used in other analyses (maternal age, time of questionnaire delivery, sex of the child etc.) will also be required.

3.2 Statistical analysis

Statistical analysis will be carried out including latent class analysis to investigate whether specific patterns of infectious symptoms are associated with any of the demographic variables recorded. The methods will allow the identification of the relationships between specific symptoms being present in early life and their relationship to different demographic factors e.g. Some demographic factors may be related to having a large number of mild heterogeneous symptoms whereas another may be related to having fewer, more severe infections with a specific symptom present such as a fever. The relationship between demographic factors and the likelihood of a parent taking a child to the doctors or giving the child medication can also be investigated. Previous analyses of symptom check-lists have been carried out using similar methods [5] and such analyses allow the identification of groups of individuals with similar behaviour or patterns of disease [6, 7] - in this case it will be those with similar infections and their relationship to demographic factors.

4 Experience of the research team

We are internationally recognised in the field of childhood cancer epidemiology and specifically current research into population mixing and other infectious measures and their association with childhood leukaemia (PMK/GL). The team have a large amount of experience in statistical analysis of epidemiological data including complex mobile phone record data in case-control studies of adult cancer (SH). Computer systems are already in place to allow the secure storage of data including a secure storage area on the Faculty of Medicine and Health server for keeping electronic data.

Within the Centre for Epidemiology and Biostatistics there are several internationally recognised statisticians specialising in structural equation modelling and the development of latent class analysis methods (Professor Mark Gilthorpe, Dr Yu-Kang Tu, Dr Robert West) and their expertise will be available for consultation on the best methods and use of statistical software.

5 References

1. Bloomfield SF, Stanwell-Smith R, Crevel RW, Pickup J (2006) Too clean, or not too clean: the hygiene hypothesis and home hygiene. Clinical and Experimental Allergy. 36: 402-25.

2. Strachan DP (2000) Family size, infection and atopy: the first decade of the "hygiene hypothesis". Thorax. 55 Suppl 1: S2-10.

3. Hay AD, Heron J, Ness A (2005) The prevalence of symptoms and consultations in pre-school children in the Avon Longitudinal Study of Parents and Children (ALSPAC): a prospective cohort study. Family Practice. 22: 367-74.

4. Dewey CR, Hawkins NS (1998) The relationship between the treatment of cough during early infancy and maternal education level, age and number of other children in the household. ALSPAC Study Team. Avon Longitudinal Study of Pregnancy and Childhood. Child Care Health Dev. 24: 217-27.

5. Sacker A, Wiggins RD, Clarke P, Bartley M (2003) Making sense of symptom checklists: a latent class approach to the first 9 years of the British Household Panel Survey. Journal of Public Health Medicine. 25: 215-22.

6. Steffen AD, Glanz K, Wilkens LR (2007) Identifying latent classes of adults at risk for skin cancer based on constitutional risk and sun protection behavior. Cancer Epidemiology, Biomarkers and Prevention. 16: 1422-7.

7. Spycher BD, Silverman M, Brooke AM, Minder CE, Kuehni CE (2008) Distinguishing phenotypes of childhood wheeze and cough using latent class analysis. European Respiratory Journal. 31: 974-81.