APPENDIX 12
Letter to the Clerk of the Committee from
Professor Julian Peto, Head of Section of Epidemiology, and Sir
Richard Doll, Emeritus Professor, Oxford University, the Institute
of Cancer Research
There is one simple action that would greatly
enhance UK cancer research (and medical research generally) while
reducing Government expenditure. This would however require government
intervention, and therefore deserves your Committee's particular
consideration.
Enforcement of the new Data Protection Act will
prevent a large proportion of the most useful cancer epidemiology.
Such research usually requires access to entire groups of individuals,
either to abstract records or for personal interview. Whether
the target group is workers at an asbestos factory, patients with
a particular cancer, or simply the general population to select
random controls, the requirement for prior consent by each individual
before the reseacher can be told of their existence leads to selective
exclusion of a large proportion and renders the results unreliable.
It also adds enormously to the costs of such studies. To illustrate
the problem, the Institute of Cancer Research has been commissioned
by the Health and Safety Executive to interview all mesothelioma
patients aged under 60 in Britain, together with randomly chosen
controls. We need to identify consultants with eligible patients
(not the patients' own names) from the Hospital Episode Statistics
file held by the DH, and random population controls from Health
Authority files. The DH has refused to notify consultants to us,
and several Health Authorities have refused to identify suitable
controls, solely because of the new Data Protection Act.
The old MRC guidelines, now superseded by the
Act, explicitly permitted access to medical and other records
for bona fide medical research purposes without the subjects'
knowledge. The universal requirement for stringent review of all
such studies by the local or regional Research Ethics Committee
ensures that such access will not be abused. During several decades
of epidemiological research during which we have contacted many
tens of thousands of patients and controls we have never known
of a subject in such studies who was seriously upset by being
contacted. We always write to the subject's GP to ask whether
there is any reason why we should not contact them, and in a small
proportion the GP feels that for medical or psychosocial reasons
it would be inappropriate to do so. Of the remainder, about ten
per cent of random controls and perhaps five per cent of patients
prefer not to participate when we contact them, but far less than
one per cent complain about being contacted. Moreover, when the
purpose of the research is discussed with this tiny minority of
complainants, who have usually failed to read the information
sheet carefully, they are always reassured and often agree to
participate. In contrast, if the Health Authority or GP has to
contact the subject to get a signed consent form returned to them
before they can notify us the response rate can drop to less than
50 per cent. It is thus absolutely vital that the principles of
the old MRC guidelines be re-established. Such a change in the
Act would greatly reduce research costs, which in many studies
are borne by the Government, and would also substantially reduce
administrative costs within the NHS.
Revision of the Data Protection Act is by far
the most important single issue in British epidemiological research,
but there is one further government action which would also contribute
enormously to many aspects of cancer research while reducing government
expenditure. Cancer registry data on incidence and survival rates
are the basis for the widely publicised evidence that cancer survival
in the UK varies between districts and is worse than in mainland
Europe or the U.S. Future trends in cancer registry data will
also be used to evaluate current initiatives to improve cancer
services, as they have been for the impact of the national breast
and cervical screening programmes. British cancer registry data
are, however, so unsatisfactory that even broad conclusions on
international or regional differences in survival are unreliable.
Most cancers are registered after a long delay and many are never
registered. In London, for example, 25 per cent of surviving cancer
patients have still not been registered five years after diagnosis
(Bullard et al British Journal of Cancer, 82 pp
1111-1116). Cancer registries are also a vital source of patients
for research, and many epidemiological studies are seriously compromised
by late and incomplete cancer registration.
The solution to this major problem is simple,
but requires Government intervention. A common PC-based data system
must be adopted in all pathology labs, or at least those under
direct NHS control. One of the existing systems should be chosen,
so the initial cost would be trivial, and there would be large
future savings in local NHS expenditure on both hardware and software.
Such a system would provide automatic monthly notification to
Cancer Registries, either online or by posting a floppy disc,
of all newly diagnosed cancers.
There are other changes that would improve cancer
registration, notably the incorporation of the cancer registries
within the Office for National Statistics with central funding
and uniform registration procedures, but these are less important
and more difficult to implement.
8 March 2000
|
