Leukaemia (Hansard, 3 June 1991)

§ Mr. Dorrell

Yes. The Government are grateful to COMARE for its advice. We note that COMARE endorses the authors' caution in drawing conclusions from the data, and that the study findings do not warrant any specific recommendations for the Government to consider.

Following is the full text of COMARE's advice:

Comare statement on LRF study

Background

1. On 23 March 1991 the British Medical Journal published the findings of a study entitled "Parental Occupations of Children with Leukaemia in west Cumbria, north Humberside and Gateshead". This study was undertaken by the Leukaemia Research Fund Centre at the University of Leeds in conjunction with the Childrens' Cancer Unit at the University of Newcastle-upon-Tyne.
2. The authors' stated objective was to determine whether parental occupations and chemical and other specific exposures are risk factors for childhood leukaemia.
3. This study is independent of the programme of research recommended by the Committee on Medical Aspects of Radiation in the Environment (COMARE) which includes research aimed at elucidating any relationship between parental occupational exposure to ionizing radiation and development of childhood leukaemia and non-Hodgkin's lymphoma (NHL). Nevertheless, aspects of the study are clearly of relevance to the work of COMARE which has undertaken to update its advice to government as further research data concerning possible associations between ionising radiation and childhood leukaemia—NHL become available. In view of this, the Department of Health and the Health and Safety Executive have referred the results of this study to COMARE for consideration and advice.

The Study

4. 151 Children under 15, diagnosed with leukaemia/ NHL between 1974 and 1988 while resident in west Cumbria, north Humberside or Gateshead were ascertained from the Yorkshire Regional Children's Tumour Registry and the Northern Region Children's Malignant Disease Registry. Those who had been born in the same area as that in which they were resident at diagnosis met the criteria of eligibility for analysis. Interview data were obtained for 109 such cases.
5. Controls, obtained mainly from district health authority birth registers, were matched for sex, date of birth and health district of birth. The study design aimed to obtain two controls for each case. For those controls for whom interview data could not be obtained, replacement controls were recruited where possible but not all cases had two controls.
6. A complete history of employment and of exposure to specific substances and radiation was obtained for the biological parents from the time they started work to the end of the analysis period. Similar data were collected for any other adults living with the child for more than three months from before birth to the time of diagnosis. 9 industrial and occupational groupings, coded according to Office of Population, Censuses and Surveys (OPCS) classifications were tested as hypotheses. In addition, data on 15 specific exposures and a check list of 22 known chemical carcinogens were analysed. In all 480 comparisons were computed.
7. A fundamental part of the study design was a separate analysis of the data for evidence of risk through six pathways: each of the biological parents both prior to conception and during gestation (four pathways) and all adult male and female household contacts (includes biological parents) from birth to diagnosis (two pathways). In all cases where a 65 statistically significant result was found for exposure of bilogical fathers during gestation the fathers had also been exposed around the time of conception.
8. For the 25 fathers who reported radiation exposure prior to birth, a subclassification of their exposure to ionising radiation was produced after checks with the National Registry for Radiation Workers (NRRW) at the National Radiological Protection Board (NRPB) and British Nuclear Fuels PLC, Sellafield, Cumbria (BNFL). Exposures were classified as "certain" for those recorded on the NRRW or by BNFL and "possible" for other contract workers on nuclear sites and industrial radiographers. Those reporting exposures in occupational settings such as education and medicine were classified as "unlikely". The remainder reported non-ionising radiation exposure and included radar and radio operators.
9. Selected results of the occupational analyses were reported, including the results that were statistically significant in any one of the six pathways analysed. Adjustment was made for inter-dependence of some results and in some instances interpretation was aided by direct inspection of the data. In addition the authors undertook comparative analyses of the six exposure pathways.
10. The authors state that few risk factors were identified for mothers but note that the results show some associations for mothers working in food-related occupations or in catering, cleaning and hairdressing before conception. However they consider that the absence of any specific range of occupations accounting for the significantly raised risks suggests that the finding may be due to chance. The only other preconceptional risk for mothers that reached statistical significance was exposure to wood dust. Preconceptional exposure of biological fathers showed the greatest numbers of significant odds ratios. After adjustment for possible confounding factors independent contributions to risk in the preconceptional period were found for exposure to wood dust, radiation and benzene.
11. The authors urge that the results be interpreted cautiously because of the small numbers, overlap with another study and multiple exposure of some parents. They state further, that it is important to distinguish the periods of parental exposures since the risk factors identified were almost exclusively restricted to the time before the child's birth.

Comare's Advice to Government

General considerations

12. This study has been undertaken independently of COMARE's recommendations and the committee is grateful to the authors for making a draft available for discussion.
13. We agree with the authors that the results must be interpreted cautiously and we would stress that interpretation of this type of research is a complex matter. Demonstration of a statistical association does not necessarily imply a causative relationship. Proper interpretation requires that the study be considered in the context of other research, both epidemiological and biological. There is, in general, little consistency amongst the findings of epidemiological studies relating to possible links between parental occupation and childhood malignancies and the authors of this study draw attention to some of the reasons for this in their Discussion.
14. Although this statement contains some general observations which apply to all the findings, the remit of COMARE has necessarily led us to restrict detailed discussion to those findings which relate to radiation exposure.

Methodology

15. The use of retrospective questionnaires is inevitably open to some reporting bias.
People who agree to participate as controls in such studies may be intrinsically different from those who refuse. The findings for the controls may therefore be unrepresentative and replacement of non responders will not necessarily ameliorate this effect.
16. The original study design did not incorporate any means of validating the occupational and exposure histories. The interpretation of the results hinges on the validity of the data and the study's potential value would be enhanced considerably if validation could be achieved. The data relating to radiation exposures has been validated to some extent by 66 cross checking with the NRRW and BNFL records. However the validation is incomplete since it did not include those parents who did not report occupational exposure to radiation. Furthermore, validation was not attempted of the radiation exposures of contract workers and industrial radiographers.

Results

17. A large number (480) of comparisons were computed in the analysis and we share the authors' view that some significant associations might have been expected purely by chance. Indeed, we note that the number of significant associations found, falls within the limits of chance expectation. This observation reinforces both our and the authors' caveats against using the results to draw conclusions about causal links.
18. The authors report a threefold risk associated with radiation exposure during the preconceptional period which remained significant after adjusting for possible confounding factors. They state that the finding is not independent of that of Gardner et al. in West Cumbria because of the geographical overlap for cases and similarity of methods of selecting controls. However they also state that this risk is not confined to Cumbria and that exposed case fathers did not work exclusively in the nuclear industry.
19. The above statements refer to the results derived from the basic data prior to any attempt at validation by cross checking with the NRRW and BNFL records. However, for those fathers subsequently classified as having "certain" exposure to ionising radiation, the statistical significance of the result is dependent on the cases common to both this and the case control study of leukaemia and lymphoma among young people near Sellafield nuclear plant in West Cumbria undertaken by Gardner et al. Thus, while this study's findings offer some confirmation of those of Gardner they do not provide independent evidence of an association between objectively confirmed paternal exposure to ionising radiation prior to conception and development of leukaemia/NHL in children. The data do not provide firm evidence of a risk to children of fathers exposed to radiation outside the nuclear industry.

General Conclusions

20. This study has documented various parental occupational exposures as having a statistically significant association with cases of childhood leukaemia/NHL in the three areas studied. The authors have been cautious in drawing conclusions from the data and we feel that this is appropriate. We do not consider that the findings, of themselves, warrant our making any specific recommendations. However, we urge the authors to explore the possibility of extensive validation of the occupational and exposure histories and welcome the authors stated intention of considering further analyses.

Reference

1. Gardner M.J., Snee M.P. et al., Results of a Case Control Study of Leukaemia and Lymphoma among young people near Sellafield nuclear plant in west Cumbria, BMJ 1990, 300, 423–429.