MRC Centre of Epidemiology for Child Health
- Home
- About Us
- People
- Research
- Cohort Studies
- Statistical Resources
- Impact & Achievements
- Education & Training
- Data Management
- Events
- News
How to find us
MRC Centre of Epidemiology for Child Health
UCL Institute of Child Health
30 Guilford Street
London, WC1N 1EH
Tel: +44 (0)20 7905 2362
Research quick links
- Congenital disorders
- Childhood origins of adult disease
- Electronic health records
- Genetic epidemiology
- Growth & development
- Health inequalities
- Infections
- Life course research
- Obesity, nutrition & physical activity
- Research for policy & practice
- Screening & surveillance
- Statistical methods
- Vision & eyes
Statistics courses
The centre runs a variety of introductory courses for health professionals who need to understand research methodology and statistical analyses. We also provide training in R statistical software.
Statistical methods
Statisticians in our centre have an international reputation for their work developing new statistical methods to analyse human growth and development and constructing age-related reference ranges in a variety of conditions and measures.
Our statisticians come from a range of backgrounds - including engineering, economics and genetics - and employ a diversity of methods and approaches. They collaborate both within the centre, across UCL and internationally.
The integration of epidemiology and biostatistics in the centre is of critical importance. Epidemiologists need quantitative skills, but close collaboration with biostatisticians enables a step change in approach and provides opportunities for developing relevant statistical methodology.
Working with longitudinal and complex data
One of our cross-cutting research themes funded by the MRC is developing statistical methods and tools that can be used to analyse longitudinal and other complex data. In epidemiology many of the data we use are longitudinal - collected at multiple points in time. We continue to employ new methods and develop tools and approaches to working with longitudinal and hierarchical data and the challenges they pose to researchers.
Current and recent research
Age Related Reference
Blah blah.
- Age-related references
- Genetics
- Growth curves
- Quantile methods
- Measuring obesity
- Analysing physical activity
- Analysing visual function
Age-related references
Our research with children involves changes due to growth and development that need taking into account when analysing data gathered over a period of time. For this age-related norms are necessary, and where they do not exist they must first be established.
Work in the centre on growth, visual function, lung function and obesity, as well as other areas relies on these age-related references.
Professor Tim Cole’s LMS method has been the basis for construction of many age-related references over the past two decades, including the World Health Organization 2006, British 1990 and UK-WHO growth charts, used to monitor the growth and development of children. We have also developed methods for age-related references of ordinal data, useful for assessing visual function.
Statisticians working in this area: Tim Cole, Phillippa Cumberland, Angie Wade
Genetics
A number of our researchers are involved in developing statistical methods for the analysis of data in genetic epidemiology studies. Dr Mario Cortina Borja is currently collaborating on trials investigating epistasis in Alzheimer's disease and he has developed methodology for the meta-analysis of statistics measuring gene-gene interactions in complex diseases.
Statisticians working in this area: Mario Cortina Borja, Xiayi Ke
Growth curves
Professor Tim Cole is currently developing the SITAR method of growth curve analysis. SITAR is a way of summarising multiple growth curves as a single summary curve, and estimating for each subject a set of three parameters to convert the average curve to fit their own growth curve. This is useful for analysing studies where growth is affected by some intervention, and also for life course studies where early growth may affect later outcome.
Statisticians working in this area: Tim Cole
Quantile Methods
A quantile is the value that corresponds to a specified proportion of a population. For example, the median is the value above or below which we find 50% of the ordered data. The median is what Sir Francis Galton once termed 'middle-most value', which expresses the vox populi.
Quantiles and probabilities of a random variable represent two sides of the same coin. Typically, introductory statistical courses focus on modelling the probability distribution of a phenomenon given the observed data. However, it is less known that analogous inferences can be drawn from modelling sample quantiles.
There are a number of advantages in using quantile methods, including robustness of the estimates, ease of modelling and interpretation of results. Dr Marco Geraci has developed linear quantile mixed models, which allow modelling conditional regression quantiles of clustered data that may exhibit complex distributional features such as skewness and heteroscedascity. The R package lqmm is freely available from the Comprehensive R Archive Network. Researchers at the Karolinska Institutet are collaborating on computational aspects of the LQMM project. Dr Geraci is also developing conditional quantile-specific risk curves to assess risk factors in mediation analysis, which plays an important role in causal inference, and quantile regression methods to analyse complex data.
Statisticians working in this area: Marco Geraci
Measuring obesity
Childhood obesity is a major health problem in the UK, linked to an increased risk of a number of chronic diseases in adulthood including type 2 diabetes and heart disease. In adults, body mass index (BMI) is used to identify those who are obese or overweight, but in children BMI varies with age. Professor Tim Cole helped to develop the International Obesity TaskForce cut-offs to provide internationally comparable rates of overweight and obesity in children.
Statisticians working in this area: Tim Cole, Leah Li
Analysing physical activity
Physical activity in childhood is linked to health later in life. Research into physical activity uses data collected by accelerometers - devices that measure movement. We are working on methods to improve the analysis of accelerometer data. The aim is to reduce and summarise the vast amounts of data that are collected, to make them more useful to researchers.
Statisticians working in this area: Mario Cortina Borja, Marco Geraci, Francesco Sera
Analysing visual function
As well as requiring age-related standards for visual function, the analysis of data relating to visual function requires methods that take into account hierarchical data (since there are data derived from both eyes), and longitudinal data collected over a period of time which must account for the development of the eye as well as the development of the individual as a whole.
This requires specialist methods, and Phillippa Cumberland is working to refine existing statistical approaches and develop new methods which will allow researchers to improve the analysis of ophthalmic data.
Phillippa is lead for paediatric ophthalmology in the UK Ophthalmic Statistics Group.
Statisticians working in this area: Phillippa Cumberland
Page last modified on 15 mar 13 08:38
