Fleur Fisher and colleagues "The risks of making assumptions about
consent, rapid response 14 May 2002" (1) contest the need for identifiable
data in cancer research and in public health.
We documented the likely minimum loss of data from patients' denial
of disease (2) if informed consent were to be required for cancer
registration. In fact, further and unquantifiable data loss would be
caused by administrative difficulties in obtaining and documenting
consent. This loss would almost certainly be even higher than the 11%
indicated for the Icelandic Health Sector Database (3): a small Cerebral
Palsy Registry in Northern Ireland has reported data losses of 32-40% as a
result of a requirement for informed consent, despite active follow-up by
a dedicated research worker and despite the patients having frequent
contact with service providers (4). In the USA, although 3.2% of a
hospital-based sample declined authorization for their medical records to
be included in a study, a further 17.5% did not respond to the request for
authorization, and taken together, these subjects differed systematically
in a number of ways from the subjects who had provided active, informed
consent. (5) The introduction of a requirement for informed consent in
parts of Germany, (6) and Hungary (7) led to serious ascertainment
problems in their cancer registries in the 1990s.
Dr Fisher and colleagues focused only on research. Cancer registries
have many other roles. They carry out surveillance of cancer trends;
monitor the impact of government policy on cancer incidence, treatment and
survival; help protect patient safety by auditing the quality of cancer
screening services; help patients by collaborating - with the patient's
informed consent - with clinical genetic services, and investigate cancer
clusters. Clusters are often based on small numbers of cases, and are
highly susceptible to erroneous conclusions if data are missing.
Registries also facilitate research into the causes and prevention of
cancer. The use of cancer registry data by researchers is subject to
ethical approval and, if patients are to be contacted, their informed
consent.
Cancer registries operate strict policies to protect patient
confidentiality and they anonymise data as soon as is possible before
analysis. There has never been a breach of that confidentiality.
But cancer registries cannot deliver these benefits to society if
their data are incomplete. Cancer registration is statutory in many other
nations, including Australia, New Zealand, Canada, USA and the Nordic
countries, so the UK would hardly be alone in recognising the importance
of this crucial public health surveillance tool by legislating to ensure
it can function effectively. In countries where cancer is a statutorily
notifiable disease, we are not aware of any evidence that patients are
reluctant to disclose information to health care professionals, nor of any
campaigns to repeal the underlying legislation.
After rigorous scrutiny by the Patient Information Advisory Group,
and approval following full debate in both the Commons and the Lords,
Parliament last month approved new regulations for England and Wales that
exempt cancer registration from the common law on confidentiality under
tightly specified conditions. These will be reviewed annually.
Dr Fisher and colleagues refer to “a growing literature on the
technical aspects of managing de-identified data and sampling bias, and on
successful experiences overseas”. We see no evidence of research output
with a major impact on public health from such approaches and there are
conflicting views (8) about their practicality. We remain to be convinced
that it is possible for society to have both total personal autonomy as
outlined by Fisher et al., and high-quality, useful data capable of making
a genuine impact on public health.
References
1. Fisher F, Anderson R, Chester M, Cowley C, Hassey A, Thornton.
The risks of making assumptions about consent. BMJ Rapid Response (14 May
2002)
2. Gavin AT, Fitzpatrick D, Middleton RJ, Coleman MP. Patients’ denial of
disease may pose difficulty for achieving informed consent. BMJ 2002,
324:974
Rapid Response:
Response to Fleur Fisher and colleagues
Fleur Fisher and colleagues "The risks of making assumptions about
consent, rapid response 14 May 2002" (1) contest the need for identifiable
data in cancer research and in public health.
We documented the likely minimum loss of data from patients' denial
of disease (2) if informed consent were to be required for cancer
registration. In fact, further and unquantifiable data loss would be
caused by administrative difficulties in obtaining and documenting
consent. This loss would almost certainly be even higher than the 11%
indicated for the Icelandic Health Sector Database (3): a small Cerebral
Palsy Registry in Northern Ireland has reported data losses of 32-40% as a
result of a requirement for informed consent, despite active follow-up by
a dedicated research worker and despite the patients having frequent
contact with service providers (4). In the USA, although 3.2% of a
hospital-based sample declined authorization for their medical records to
be included in a study, a further 17.5% did not respond to the request for
authorization, and taken together, these subjects differed systematically
in a number of ways from the subjects who had provided active, informed
consent. (5) The introduction of a requirement for informed consent in
parts of Germany, (6) and Hungary (7) led to serious ascertainment
problems in their cancer registries in the 1990s.
Dr Fisher and colleagues focused only on research. Cancer registries
have many other roles. They carry out surveillance of cancer trends;
monitor the impact of government policy on cancer incidence, treatment and
survival; help protect patient safety by auditing the quality of cancer
screening services; help patients by collaborating - with the patient's
informed consent - with clinical genetic services, and investigate cancer
clusters. Clusters are often based on small numbers of cases, and are
highly susceptible to erroneous conclusions if data are missing.
Registries also facilitate research into the causes and prevention of
cancer. The use of cancer registry data by researchers is subject to
ethical approval and, if patients are to be contacted, their informed
consent.
Cancer registries operate strict policies to protect patient
confidentiality and they anonymise data as soon as is possible before
analysis. There has never been a breach of that confidentiality.
But cancer registries cannot deliver these benefits to society if
their data are incomplete. Cancer registration is statutory in many other
nations, including Australia, New Zealand, Canada, USA and the Nordic
countries, so the UK would hardly be alone in recognising the importance
of this crucial public health surveillance tool by legislating to ensure
it can function effectively. In countries where cancer is a statutorily
notifiable disease, we are not aware of any evidence that patients are
reluctant to disclose information to health care professionals, nor of any
campaigns to repeal the underlying legislation.
After rigorous scrutiny by the Patient Information Advisory Group,
and approval following full debate in both the Commons and the Lords,
Parliament last month approved new regulations for England and Wales that
exempt cancer registration from the common law on confidentiality under
tightly specified conditions. These will be reviewed annually.
Dr Fisher and colleagues refer to “a growing literature on the
technical aspects of managing de-identified data and sampling bias, and on
successful experiences overseas”. We see no evidence of research output
with a major impact on public health from such approaches and there are
conflicting views (8) about their practicality. We remain to be convinced
that it is possible for society to have both total personal autonomy as
outlined by Fisher et al., and high-quality, useful data capable of making
a genuine impact on public health.
References
1. Fisher F, Anderson R, Chester M, Cowley C, Hassey A, Thornton.
The risks of making assumptions about consent. BMJ Rapid Response (14 May
2002)
2. Gavin AT, Fitzpatrick D, Middleton RJ, Coleman MP. Patients’ denial of
disease may pose difficulty for achieving informed consent. BMJ 2002,
324:974
3. Association of Iceanders for Ethics in Science and Medicine.
www.manvernd.is/english
4. Parkes J, Dolk H, Hill N, Patterson S. Cerebral Palsy in Northern
Ireland: 1981-93. Paediatric and Prenatal Epidemiology 2001, 15:278-286
5. Jacobson S~J, Xia Z, Campion ME et al. Potential Effect of
Authorization Bias on Medical Record Research. Mayo Clin Proc 1999; 74:
330-338
6. Becker N. Cancer Epidemiology and Privacy Laws: Recent Trends in
Germany. Eur J Cancer 1993; 29A: 661-663
7. Whelan SL (editor). IACR Newsletter. December 2000. No 31. Pp 16-17.
http://www.iacr.com.fr/dec2000htm
8. Dudeck J. Informed consent for cancer registration. Lancet Oncology
2001; 2: 8-9
Competing interests: No competing interests