Childhood cancer research in Oxford I: the Oxford Survey of Childhood Cancers.

BACKGROUND: Significant research on the epidemiology and natural history of childhood cancer took place in the Universities of Oxford and Birmingham over sixty years. This is the first of three papers recording this work and describes the Oxford Survey of Childhood Cancers (OSCC), the largest case-control survey of childhood cancer ever undertaken. METHODS: The OSCC studied deaths in Britain from 1953 to 1981. Parents were interviewed and medical records from ante-natal clinics and treatment centres were followed up and abstracted. The survey left Oxford in 1975 and was run subsequently from Birmingham. The data are now being documented and archived to make them available for future study. RESULTS: Many papers have resulted from this survey, most notably those relating to the association first reported therein between childhood cancer and ante-natal X-raying. This paper is a historical review of the OSCC. CONCLUSIONS: In spite of many analyses of the study, this historic data set has continuing value because of the large number of examples of some very rare tumours and the detailed clinical and family history data that are available; and also because of the possibility of carrying out new analyses to investigate emerging research issues.

Leukaemia in young children in the vicinity of British nuclear power plants: a case-control study.

BACKGROUND: Concern about the risk of leukaemia in children living near nuclear power plants (NPPs) persists. Previous British analyses have been area based and consequently thought to be less effective than case-control studies. METHODS: Cases of childhood leukaemia and non-Hodgkin lymphoma (LNHL) born and diagnosed in Great Britain between 1962 and 2007, with matched cancer-free controls, were analysed by logistic regression to estimate the risk of residential proximity at birth and diagnosis to the nearest NPP, adjusting for relevant variables. RESULTS: For 9821 children with LNHL under the age of 5 years, the estimated extra risk associated with residential proximity to an NPP at birth was negative-interpolated Odds Ratio (OR) at 5 km was 0.86 (0.49-1.52). The comparison of 10 618 children with LNHL under five with 16 760 similarly aged children with other cancers also gave a negative estimate of the extra risk of residential proximity at diagnosis-interpolated OR at 5 km was 0.86 (0.62-1.18). CONCLUSION: Our results show little evidence of an increase in risk of LNHL to children aged under 5 years from living in the vicinity of an NPP. Risk estimates are incompatible with comparable ones published in a recent German case-control study.

Childhood leukaemia near British nuclear installations: methodological issues and recent results.

In 2008, the German Childhood Cancer Registry published the results of the Kinderkrebs in der Umgebung von Kernkraftwerken (KiKK) study of childhood cancer and leukaemia around German nuclear power stations. The positive findings appeared to conflict with the results of a recent British analysis carried out by the Committee on Medical Aspects of Radiation in the Environment (COMARE), published in 2005. The present paper first describes the COMARE study, which was based on data from the National Registry of Children's Tumours (NRCT); in particular, the methodology used in this study is described. Although the results of the COMARE study were negative for childhood leukaemia, this apparent discrepancy could be accounted for by a number of differences in approach, especially those relating to the distances from the power stations and the ages of the children studied. The present study was designed to match the KiKK study as far as possible. The incidence observed (18 cases within 5 km against 14.58 expected, p = 0.21) was not significantly raised. The risk estimate for proximity in the regression fitted was actually negative, though the confidence intervals involved are so wide that the difference from that reported in the KiKK study is only marginally statistically significant (p = 0.063).

Childhood leukaemia and non-Hodgkin's lymphoma in relation to proximity to railways.

We investigated whether living close to railway lines is a risk factor for childhood leukaemia and non-Hodgkin's lymphoma in electoral wards in England and Wales, 1966-1987. The national rail network, 1966-1987, was digitised and the numbers of cases in each ward were related to two measures of environmental exposure to railways: a proximity and a density function, contributions to these functions being weighted by the frequency of use and time in use of each stretch of railway. Poisson regression was used to derive rate ratios in relation to these measures of exposure to railways, both unadjusted and adjusted for population mixing. We found no association between risk of leukaemia and railway proximity (unadjusted rate ratio for trend from the lowest to the median value=1.006, 95% CI: 0.998 - 1.013, P=0.14) and a very small association with railway density, of marginal statistical significance (rate ratio for trend=1.001, 95% CI: 1.000 - 1.003, P=0.05). This effect depended on two deprived, urban wards with high railway density and high population mixing and became nonsignificant (P=0.09) after allowing for population mixing. The very weak association between railway density and risk of childhood leukaemia is likely to be a consequence of the association between population mixing and proximity to railways in very deprived, urban wards.

Population mixing and childhood leukaemia and non-Hodgkin's lymphoma in census wards in England and Wales, 1966-87.

We found an increased risk of childhood leukaemia with higher levels of inward migration, particularly from outside the region (rate ratio=1.9, 95%CI: 1.2-2.9, P<0.01). This significant effect was observed only in urban areas, although a marked but non-significant effect was seen in affluent, rural areas.

Childhood leukaemia clustering--fact or artefact?

There is a general belief that the clustering of childhood leukaemia is a widespread phenomenon and that it provides evidence for appreciable environmental influence on the incidence of the disease. We discuss this issue critically, identifying different kinds of clustering and their possible aetiological mechanisms and examining some analyses of British data. We argue that, in some cases, analyses have been used which lead to dubious conclusions, and that, allowing for multiple testing and anecdotal reporting, the total evidence for clustering is at best weak.

A classification of disease mapping methods.

This paper considers the underlying principles of depicting disease incidence on geographical maps and uses them to attempt a comparative classification of methods. After a discussion of the possibilities for incorporating time, we consider projection methods, some of which have been used to portray information in a manner supposed to be independent of population density. We then distinguish between non-parametric and model-based methods, including models for areal data using Bayesian ideas. Data in point form are also discussed and it is argued that the relative risk function provides a fundamental model useful for assessing different methods as a whole, some of which are known to be flawed and many of which are untested as regards their statistical properties.

Uranium-235 and childhood leukaemia around Greenham Common airfield.

There has been considerable publicity recently concerning the possible release of enriched uranium from the Greenham Common USAF base near Newbury in Berkshire. Evidence for the release relies on an internal report of the Atomic Weapons Research Establishment at Aldermaston, the authors of which postulated that it resulted from a fire in 1958 involving a B47 bomber standing on the runway. Their report contained a much publicised contour map of excess 235U levels estimated from the ratio of 235U to 238U in 26 evergreen leaf samples examined. The current concern of the inhabitants of Newbury centres mostly on the incidence of leukaemia, which was known beforehand to be slightly elevated in parts of West Berkshire, at least for young children. A number of cases have received considerable press publicity, with suggestions that their homes are located close to the base or the flight-path. The reports are, however, anecdotal and are not based on a complete register of cases. We have examined the evidence for this putative association by re-analysing the uranium data and determining the spatial relationship to the base of cases of childhood leukaemia diagnosed in the years 1966-87. We conclude that, although the excess uranium found has a non-random distribution, it does not support the pattern depicted by the contours and bears no relation to the incidence of childhood leukaemia for the period we examined. In any case, the increase in level of environmental radiation as a result of the putative release must be very small and is at variance with the reporting in some of the national press.

Controlling for socioeconomic confounding using regression methods.

STUDY OBJECTIVE: To describe the advantages of using Poisson regression methods as an alternative to standardisation when computing expected numbers of disease occurrences adjusted for possible confounding factors. The problem of assessing the adequacy of model fit when the expectations are small is addressed by analytical calculations and by simulation. The method is illustrated with data from the national register of childhood tumours. DESIGN: The tumour data are recorded in a national register. SETTING: England, Scotland, and Wales. SUBJECTS: The cases considered are all children registered with leukaemia or non-Hodgkin lymphoma under the age of 15 years between 1966-87. MAIN RESULTS: The methods show a significant variation of leukaemia incidence in relation to the Register General's standard region and a negative association with socioeconomic deprivation, as measured by the Townsend index. After allowing for these variables, the incidence seems to be reasonably homogeneous throughout the population, in the sense that the residual deviance does not seem to be much larger than would be expected by chance. CONCLUSIONS: The methods described have major advantages over standardisation in controlling for confounding, both in terms of flexibility of factor selection and assessment and also in the ability to determine whether there is residual variability of incidence after allowing for these factors.

The choice of test for detecting raised disease risk near a point source.

This paper considers the problem of testing for excess risk near a point source of risk S, as might be postulated to exist near a nuclear installation, for example. The data are assumed to be in the form of counts in small areas for which expected numbers of cases have also been calculated by reference to national rates. It is shown how to derive the most powerful test against any given alternative hypothesis; such 'linear risk score' tests are further considered, particularly that which scores each case according to the reciprocal of the rank of the distance from S. These tests are compared with two tests due to Stone for general ordered alternatives and the important distinction is drawn between conditional and unconditional versions of the tests. Their behaviour is illustrated by application to data on childhood leukaemia in relation to nuclear installations in Britain.

Apparent association between benzene and childhood leukaemia: methodological doubts concerning a report by Knox.

A recent study by Knox concludes that cases and "clusters" of two or more cases of childhood leukaemia and non-Hodgkin's lymphoma occur closer to many kinds of industrial installation than to supposedly comparable control locations. It is argued that these findings could be largely or entirely artefactual, the apparent differences arising out of the inappropriateness of the control data. Knox used randomly selected postcode units as controls, a procedure that leads to the comparison of individuals located in areas with typically quite different population densities from those for the cases. The resulting potential for bias is explored and the arguments are exemplified by analysing household data based on postcodes.

Distribution of childhood leukaemias and non-Hodgkin's lymphomas near nuclear installations in England and Wales.

OBJECTIVE: To examine the relation between the risk of childhood leukaemia and non-Hodgkin's lymphoma and proximity of residence to nuclear installations in England and Wales. DESIGN: Observed and expected numbers of cases were calculated and analysed by standard methods based on ratios of observed to expected counts and by a new statistical test, the linear risk score test, based on ranks and designed to be sensitive to excess incidence in close proximity to a putative source of risk. SETTING: Electoral wards within 25 km of 23 nuclear installations and six control sites that had been investigated for suitability for generating stations but never used. SUBJECTS: Children below age 15 in England and Wales, 1966-87. MAIN OUTCOME MEASURE: Registration of any leukaemia or non-Hodgkin's lymphoma. RESULTS: In none of the 25 km circles around the installations was the incidence ratio significantly greater than 1.0. The only significant results for the linear risk score test were for Sellafield (P = 0.00002) and Burghfield (P = 0.031). The circles for Aldermaston and Burghfield overlap; the incidence ratio was 1.10 in each. One of the control sites gave a significant linear risk score test result (P = 0.020). All the tests carried out were one sided with P values estimated by simulation. CONCLUSION: There is no evidence of a general increase of childhood leukaemia or non-Hodgkin's lymphoma around nuclear installations. Apart from Sellafield, the evidence for distance related risk is very weak.

Estimation of relative risk functions.

The risk associated with different levels of a quantitative factor X is often measured relative to the level corresponding to X = 0. There are situations, however, where there is no natural zero for X, for example where the risk factor is the age of an individual. In this case it is more natural to measure risk relative to an overall average for the study population. To use the whole population in this way also raises the possibility of regarding X as truly continuous, rather than as a grouped variable. This gives rise to the concept of a relative risk function. Methods for estimating such functions are discussed, concentrating for the most part on the discrete case. The extension to higher dimensions permits the investigation of joint effects of several factors, while the problem of controlling for confounding variables can be handled by fitting multiplicative risk models. Relating the latter to the log-linear model permits the estimation of adjusted relative risk functions. The method is illustrated using data on childhood cancer. The continuous case can in principle be handled in a similar way using density estimation techniques.

An application of density estimation to geographical epidemiology.

A relative risk function over a geographical region is defined and it is shown that it can be estimated effectively using kernel density estimation separately for the spatial distribution of disease cases and for a sample of controls. This procedure is demonstrated using data on childhood leukaemia in the vicinity of the Sellafield nuclear reprocessing plant in Cumbria, U.K. Various modifications to the method are proposed, including the use of an adaptive kernel. The final plot demonstrates a sharp peak at Sellafield and a reasonably smooth surface over the rest of the region, despite the small number of cases in the series.

On statistical methods for analysing the geographical distribution of cancer cases near nuclear installations.

There is great public concern, often based on anecdotal reports, about risks from ionising radiation. Recent interest has been directed at an excess of leukaemia cases in the locality of civil nuclear installations at Sellafield and Sizewell, and epidemiologists have a duty to pursue such information vigorously. This paper sets out to show that the epidemiological methods most commonly used can be improved upon. When analysing geographical data it is necessary to consider location. The most obvious quantification of location is ranked distance, though other measures which may be more meaningful in relation to aetiology may be substituted. A test based on distance ranks, the "Poisson maximum test", depends on the maximum of observed relative risk in regions of increasing size, but with significance level adjusted for selection. Applying this test to data from Sellafield and Sizewell shows that the excess of leukaemia incidence observed at Seascale, near Sellafield, is not an artefact due to data selection by region, and that the excess probably results from a genuine, if as yet unidentified cause (there being little evidence of any other locational association once the Seascale cases have been removed). So far as Sizewell is concerned, geographical proximity to the nuclear power station does not seem particularly important.

A new calculation of the carcinogenic risk of obstetric X-raying.

The association between obstetric X-raying and childhood cancer was first identified by the Oxford Survey of Childhood Cancers in 1956. The present re-analysis exploits the case-control matching of the study while incorporating the effects of important risk determinants, notably year of birth, trimester of exposure and number of films exposed. The decline in risk over time is closely mirrored by the estimated decline in dose per film and, by constraining these two relationships to be parallel, time-invariant estimates of the extra risk per mGy are obtained. For example, it is now estimated that irradiating 10(6) foetuses with 1 mGy of X-rays would, in the absence of other causes of death, yield 175 extra cases of cancer and leukaemia in the first 15 years of life.

Relationship between lower oesophageal contractility, clinical signs and halothane concentration during general anaesthesia and surgery in man.

The effects of a range of concentrations of halothane upon lower oesophageal contractility (LOC) and on defined clinical signs has been studied in patients undergoing surgery. Changes in clinical signs were assigned a numerical value by means of a scoring system. One hundred and eighty-one sets of measurements were made in 46 patients exposed to concentrations of halothane between 2.0 and 0.5 minimum alveolar concentration (MAC). The results were examined to identify relationships between (i) the clinical signs and alveolar halothane concentration, (ii) the clinical signs and LOC and (iii) the changes in LOC and alveolar concentration; significant correlations were found between these variables. Decreasing alveolar halothane concentration was associated with an increase in LOC and these increases in LOC were also associated with increases in the clinical score.

Pre-natal irradiation and childhood malignancy: a review of British data from the Oxford Survey.

This paper reviews data relating to obstetric radiography from the Oxford Survey of Childhood Cancers, i.e. for deaths in Britain from 1953 to 1967. Some 8513 cases were traced and used in the analyses, together with an equal number of matched controls. The relative risk estimate (1-47 overall) does not vary significantly between different tumour groups, for different ages at death, nor between sexes. Other epidemiological factors-sibship position, maternal age, social class, region of residence and maternal morbidity-are analysed and show varying degrees of association, but not sufficient to "explain" the observed risk in terms of a selection effect. The dependence of the risk on the number of films exposed is highly significant and adequately described by a linear relationship. The timing of and reason for the exposure are also examined. Analysis of the risk by year of birth shows a pattern of steadily declining risk for both solid and haematopoietic tumours; this may be partly attributable to lower radiation doses per film exposed but is also due to the smaller numbers of films used. A consequence may well be that the risk-always of small clinical significance-would become virtually undetectable in future investigations.

Association between malignant disease in children and maternal virus infections.

The possibility of an association between virus infections during pregnancy and subsequent malignant disease in the child has been investigated using retrospective data from the Oxford Survey of Childhood Cancers. Such an association appears to exist for influenza, chickenpox, and possibly rubella. For influenza and rubella the estimated risk is small; the data do not permit an estimate to be made directly in the case of chickenpox. It is suggested that there may be a specific association between maternal chickenpox and tumours of the nervous system.It is important to emphasize that, even if the relative risk associated with these viruses is fairly large, the number of cases of childhood cancer and leukaemia actually attributable to them is probably very small.