Use of decision aids to support informed choices about screening
http://www.100md.com
《英国医生杂志》
1 Screening and Test Evaluation Program, School of Public Health, University of Sydney, Sydney, NSW 2006, Australia
Correspondence to: A Barratt alexb@health.usyd.edu.au
Decisions about screening can be more complex than those about treatment. Well designed decision aids could help patients understand the risks as well as the benefits
Introduction
Detecting inconsequential disease
People making treatment decisions have a disease for which they seek treatment. Screening, however, may uncover inconsequential disease (pseudodisease) that will never produce symptoms.10 11 Detection of inconsequential disease results in unnecessary diagnosis and treatment.6 10 12 This problem is most clearly recognised in PSA screening for prostate cancer13 but applies generally. For example, screening mammography results in overdetection of both invasive breast cancer14 and ductal carcinoma in situ,15 so although screening may reduce the risk of dying from breast cancer, it increases the chance of a breast cancer diagnosis.
The problem of inconsequential disease extends to screening for risk factors. For example, people identified as having high cholesterol concentrations or high blood pressure suffer immediate "labelling" effects16 17 and may view these conditions as diseases. In reality, however, many people with these risk factors will not develop symptomatic cardiovascular disease for many years, if ever.18
Inconsequential disease is considered by some to be the biggest drawback of screening.6 12 Therefore it is of concern that the general public are poorly aware of it.6 Decision aids must contain clear, numerical information on the chances of having inconsequential disease found and treated.
Providing information about follow up tests and treatments
Decision aids for treatments present information about several possible options and help people weigh these up. Screening, on the other hand, involves a longer, more complex process, including the initial test plus the follow up investigations and treatment. Informed decision making requires a consumer to understand the whole process before making a decision, yet few people realise the consequences of a positive screening result until they face that abnormal result. For example, few women who participate in cervical screening know much about the follow up tests and examinations for an abnormal smear test result, and knowledge levels are low, even in those who undergo colposcopy and biopsy.19 20 Thus developers of screening decision aids must include information about the whole screening process (including follow up tests and treatments).
Considerations in developing decision aids for screening interventions—and suggestions to developers
Screening leads to over-detection and over-treatment—Present the chances of having pseudodisease as well as clinically important disease detected by screening
Screening may include invasive follow up investigations and treatments—Give information about the whole of the early detection and treatment process
Benefits of screening are delayed, whereas harms are immediate—Present balanced information about the cumulative chance of benefits and harms over equivalent time frames
Few people experience benefits from screening compared with the number who would be expected to benefit from most treatments—Present very small numbers by using large and consistent denominators, for example, outcomes per 1000 or per 10 000 people screened
Individual values and preferences are critical to screening decision making—Screening decision aids need to accommodate flexibility in labelling the outcomes of screening as benefits or harms
Evidence base for screening decision aids is often limited—Explicitly declare where high quality evidence is lacking; use ranges or some other method to convey uncertainty in numerical estimates
Public attitude is that early detection and or prevention must be good—Explain that there is a choice and the reasons why people might decide to decline screening
Little regulation is in place to protect consumers from aggressive marketing, and there may be strong financial incentives to get people to participate in screening—Information about financial gains to the organisation offering the screening test may need to be included in decision aids
Presenting cumulative benefits and harms over an appropriate time
In the treatment context, people are sick and treatment may result in improvement in the near future. Thus an immediate treatment benefit may need to be traded off against immediate (or perhaps delayed) side effects. In screening, the benefits are usually to be experienced in the future, possibly quite far into the future, but the harms can be immediate. Thus immediate harms must be traded against an uncertain future benefit. For example, screening can reduce the risk of death from bowel cancer by about 23% if people attend screening regularly every two years, for 10 years.21 This possible benefit over the next 10 years has to be traded against the risk of a perforated bowel as a result of a colonoscopy, which could occur after the first or any subsequent screening test.21 To present balanced information about benefits and harms, therefore, decision aids need to present information about the cumulative risk of benefits and harms (such as the cumulative risk of a false positive result, or requiring a biopsy22) over an appropriate time frame (most likely 10 years or more).
Communicating outcomes with a low probability
The numbers needed to treat associated with most treatments for symptomatic conditions are relatively small.23 In contrast, the benefits of screening are relatively rare, and the numbers needed to screen are large. For example, the number of men who need to be screened for raised cholesterol concentration to prevent one death from heart disease within five years range from 21 067 for men aged 25-34 to 231 for men aged 55-64, and the numbers are greater for women.18 For cervical screening, an estimated 1000 women need to be screened for 35 years to prevent one death from cervical cancer.24
Conveying very low event rates is technically difficult. The cumulative deaths over 10 years from bowel cancer for Australian men in their 60s are about 98/10 000 without screening and about 75/10 000 with screening.21 Thus the diagrams used to aid screening decisions may need to use 1000 faces or even 10 000 faces to show event rates as whole figures instead of the 100 faces used in diagrams for treatments.25 Furthermore, it is important to provide a context within which to consider the impact of screening.26 For example, the decision aid might include the risk of death from other causes or all causes over a similar time.
Valuing outcomes
Research indicates that people vary in the way they value screening outcomes such as false positive results.6 27 28 This is consistent with qualitative reports we have received from participants in bowel, breast, and cervical screening programmes: some people consider a false positive a good result because they feel reassured that they have been thoroughly checked and found not to have the target disorder, while others regard a false positive as inconvenient, anxiety provoking, unpleasant, or dangerous. Thus the information presented by and values clarification exercises used in screening decision aids need to be flexible about classifying outcomes as pros and cons. The aid might allow people to assign positive (pro) or negative (con) status to different outcomes themselves, rather than use a fixed scheme for everyone.
Quality of evidence
For treatments, decision aids use evidence from randomised trials about outcomes of different options. Such high quality data are often not available for screening, partly because trials of screening need to be very large and may require follow up over many years, making them expensive and time consuming. Worse, by the time the trial reports, the screening test may have been substantially updated, so the applicability of the results of the trial is unclear. Thus developers may have to rely on observational data with all its inherent biases (including lead time bias, length time bias, and the healthy volunteer effect10 11), which should be explicitly stated.
Even when trials are available, there are usually only a few, and so estimates are subject to substantial uncertainty. For example, meta-analysis of trials of screening for bowel cancer using faecal occult blood tests shows a 23% relative risk reduction for death from bowel cancer for people who screen regularly, with a 95% confidence interval of 11% to 43%.21 There may be even greater uncertainty about estimates from randomised trials in other contexts—for example, mammography screening.29 It is challenging to adequately convey the uncertainty in the data about screening. Ranges may be better than point estimates, but their use may compromise the simplicity and clarity of communication.
Informed choice in the context of screening
Consumers are reasonably familiar with having different treatment options to choose between, as there have been different drugs or management options available for many conditions for a long time. The concept of having options with regard to screening is less familiar.6 Developers of decision aids for screening should be prepared for users to respond with "But why wouldn't I be screened for X? That's the responsible thing to do." We have found it necessary to explain the down sides of screening and reasons why you might choose not to be screened at the start of decision aids.
Furthermore, some healthcare providers may feel uncomfortable with the information provided in evidence based decision aids about screening and may disagree with consumers about the amount and balance of information. When pilot testing a mammography decision aid we found that healthcare providers thought it contained too much information and was slanted against screening, while women said the information was about the right length and slanted towards screening (D Stacey et al. Reasons for Hope, Ottawa, 25-27 Oct 2003). Our experience is consistent with previous research showing ultrasonographers felt they couldn't give women information contained in an evidence based leaflet about antenatal screening.30 Developers should probably expect some discomfort among healthcare providers, who may feel that the decision aid is unbalanced, that it runs counter to their view of screening, or that information about harms will cause anxiety among patients.
Summary points
The effects of decision aids about screening on decision quality and participation in screening are largely unknown
Public (and professional) understanding of the drawbacks of screening is limited
As well as information on benefits, decision aids need to include information on detection of inconsequential disease and about the risks of the full range of investigations and treatments which may be entailed if results are abnormal
Information on benefits and risks should be given over similar time frames
People may vary in viewing outcomes of screening as pros or as cons
Incentives to participate in screening
Financial incentives or performance targets are rarely related to acceptance of treatment. In the screening context, companies directly market "off the shelf" screening kits such as faecal occult blood test and genetic screening kits. Unlike treatments, screening tests are not well regulated by government, and the public is not well protected against the commercial promotion of screening tests.6 Even within authorised and accredited government funded screening programmes, there is a perceived need to meet participation targets in order to maximise population benefits.2 Healthcare funders and those delivering services may be reluctant to support the use of screening decision aids by consumers if they think it will reduce participation. The impact of screening decision aids on participation rates is unclear,3 4 but is likely to vary between different screening interventions. Developers may need to work closely with providers of the screening service to develop and evaluate screening decision aids. Any vested interest should be declared.
Conclusions
O'Connor AM, Stacey D, Entwistle V, Llewellyn-Thomas H, Rovner D, Holmes-Rovner M, et al. Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst Rev 2003;(2): CD001431.
Raffle AE. Information about screening—is it to achieve high uptake or to ensure informed choice? Health Expectations 2001;4: 92-8.
Jepson R, Forbes C, Sowden A, Lewis R. Increasing informed uptake and non-uptake of screening: evidence from a systematic review. Health Expectations 2001;4: 116-30.
Edwards A, Unigwe S, Elwyn G, Hood K. Effects of communicating individual risks in screening programmes: Cochrane systematic review. BMJ 2003;327: 703-9.
Yamey G, Wilkes M. The PSA storm. BMJ 2002;324: 431.
Schwartz LM, Woloshin S, Fowler FJ, Welch HG. Enthusiasm for cancer screening in the United States. JAMA 2004;291: 71-8.
Agency for Healthcare Research and Quality. US Preventive Services Task Force. www.ahrq.gov/clinic/uspstfix.htm (accessed 15 Jul 2004).
Montgomery AA, Fahey T. A systematic review of the use of computers in the management of hypertension. J Epidemiol Community Health 1998;52: 520-5.
American Cancer Society. Prevention and early detection. www.cancer.org/docroot/PED/content/PED_2_3X_Early_Detection.asp?sitearea=PED (accessed 27 Jul 2004).
Barratt A, Irwig L, Glasziou P, Cumming RG, Raffle A, Hicks N, Muir Gray JA, Guyatt GH. Users' guides to the medical literature. XVII. How to use guidelines and recommendations about screening. JAMA 1999;281: 2029-2034
Morrison AS. Screening in chronic disease. 2nd ed. New York: Oxford University Press, 1992: 23.
Black W, Welch H. Advances in diagnostic imaging and over-estimations of disease prevalence and the benefits of therapy. N Engl J Med 1993;328: 1237-43.
Etzioni R, Penson DR, Legler JM, di Tommaso D, Boer R, Gann PH, et al. Overdiagnosis due to prostate-specific antigen screening: lessons from US prostate cancer incidence trends. J Natl Cancer Inst 2002;94: 981-90.
Zahl PH, Strand BH, Maehlen J. Incidence of breast cancer in Norway and Sweden during introduction of nationwide screening: prospective cohort study. BMJ 2004;328: 921-4.
Ernster VL, Ballard-Barbash R, Barlow WE, et al. Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 2002:94: 1546-54.
Alderman MH, Lamport B. Labelling of hypertensives: a review of the data. J Clin Epidemiol 1990;43: 195-200.
Lefebvre RC, Hursey KG, Carleton RA. Labeling of participants in high blood pressure screening programs. Implications for blood cholesterol screening. Arch Intern Med 1988;148: 1993-7.
Khaw KT, Rose G. Cholesterol screening programmes: how much potential benefit? BMJ 1989;299: 606-7.
Slater DN. Are women sufficiently well informed to provide valid consent for the cervical smear test? Cytopathology 2000;11: 166-70.
Kavannagh AM, Broom DH. Women's understanding of abnormal cervical smear test results: a qualitative study. BMJ 1997;314: 1388-91.
Towler B, Irwig L, Glasziou P, Kewenter J, Weller D, Silagy C. A systematic review of the effects of screening for colorectal cancer using the faecal occult blood test, Hemoccult. BMJ 1998;317: 559-65.
Elmore JG, Barton MB, Moceri VM, Polk S, Arena PJ, Fletcher SW. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338: 1089-96.
American Medical Association. Users' guides to the medical literature: a manual for evidence based clinical practice. Chicago: AMA, 2002: 387-9.
Raffle A, Alden B, Quinn M, Babb PJ, Brett MT. Outcomes of screening to prevent cancer: analysis of cumulative incidence of cervical abnormality and modelling of cases and deaths prevented. BMJ 2003;326: 901.
Man-Song-Hing M, Laupacis A, O'Connor AM, Biggs J, Drake E, Yetisir E, et al. A patient decision aid regarding antithrombotic therapy for stroke prevention in atrial fibrillation. JAMA 1999;282: 737-43.
Woloshin S, Schwartz LM, Welch HG. Risk charts: putting cancer in context. J Natl Cancer Inst 2002;94: 799-804.
Scaf-Klomp W, Sanderman R, van de Wiel HBM, Otter R, van den Heuvel WJA. Distressed or relieved? Psychological side effects of breast cancer screening in the Netherlands. J Epidemiol Community Health 1997;51: 705-10.
Pernet AL, Wardle J, Bourne TH, Whitehead MI, Campbell S, Collins WP. A qualitative evaluation of the experience of surgery after false positive results in screening for familial ovarian cancer. Psycho-oncology 1992;1: 217-33.
Olsen O, Gotszche PC. Cochrane review on screening for breast cancer with mammography. Lancet 2001;358: 1340-2.
Oliver S, Rajan L, Turner H, Oakley A, Entwistle V, Watt I, et al. Informed choice for users of health services: views on ultrasonography leaflets of women in early pregnancy, midwives, and ultrasonographers. BMJ 1996;313: 1251-3.(Alexandra Barratt, senior)
Correspondence to: A Barratt alexb@health.usyd.edu.au
Decisions about screening can be more complex than those about treatment. Well designed decision aids could help patients understand the risks as well as the benefits
Introduction
Detecting inconsequential disease
People making treatment decisions have a disease for which they seek treatment. Screening, however, may uncover inconsequential disease (pseudodisease) that will never produce symptoms.10 11 Detection of inconsequential disease results in unnecessary diagnosis and treatment.6 10 12 This problem is most clearly recognised in PSA screening for prostate cancer13 but applies generally. For example, screening mammography results in overdetection of both invasive breast cancer14 and ductal carcinoma in situ,15 so although screening may reduce the risk of dying from breast cancer, it increases the chance of a breast cancer diagnosis.
The problem of inconsequential disease extends to screening for risk factors. For example, people identified as having high cholesterol concentrations or high blood pressure suffer immediate "labelling" effects16 17 and may view these conditions as diseases. In reality, however, many people with these risk factors will not develop symptomatic cardiovascular disease for many years, if ever.18
Inconsequential disease is considered by some to be the biggest drawback of screening.6 12 Therefore it is of concern that the general public are poorly aware of it.6 Decision aids must contain clear, numerical information on the chances of having inconsequential disease found and treated.
Providing information about follow up tests and treatments
Decision aids for treatments present information about several possible options and help people weigh these up. Screening, on the other hand, involves a longer, more complex process, including the initial test plus the follow up investigations and treatment. Informed decision making requires a consumer to understand the whole process before making a decision, yet few people realise the consequences of a positive screening result until they face that abnormal result. For example, few women who participate in cervical screening know much about the follow up tests and examinations for an abnormal smear test result, and knowledge levels are low, even in those who undergo colposcopy and biopsy.19 20 Thus developers of screening decision aids must include information about the whole screening process (including follow up tests and treatments).
Considerations in developing decision aids for screening interventions—and suggestions to developers
Screening leads to over-detection and over-treatment—Present the chances of having pseudodisease as well as clinically important disease detected by screening
Screening may include invasive follow up investigations and treatments—Give information about the whole of the early detection and treatment process
Benefits of screening are delayed, whereas harms are immediate—Present balanced information about the cumulative chance of benefits and harms over equivalent time frames
Few people experience benefits from screening compared with the number who would be expected to benefit from most treatments—Present very small numbers by using large and consistent denominators, for example, outcomes per 1000 or per 10 000 people screened
Individual values and preferences are critical to screening decision making—Screening decision aids need to accommodate flexibility in labelling the outcomes of screening as benefits or harms
Evidence base for screening decision aids is often limited—Explicitly declare where high quality evidence is lacking; use ranges or some other method to convey uncertainty in numerical estimates
Public attitude is that early detection and or prevention must be good—Explain that there is a choice and the reasons why people might decide to decline screening
Little regulation is in place to protect consumers from aggressive marketing, and there may be strong financial incentives to get people to participate in screening—Information about financial gains to the organisation offering the screening test may need to be included in decision aids
Presenting cumulative benefits and harms over an appropriate time
In the treatment context, people are sick and treatment may result in improvement in the near future. Thus an immediate treatment benefit may need to be traded off against immediate (or perhaps delayed) side effects. In screening, the benefits are usually to be experienced in the future, possibly quite far into the future, but the harms can be immediate. Thus immediate harms must be traded against an uncertain future benefit. For example, screening can reduce the risk of death from bowel cancer by about 23% if people attend screening regularly every two years, for 10 years.21 This possible benefit over the next 10 years has to be traded against the risk of a perforated bowel as a result of a colonoscopy, which could occur after the first or any subsequent screening test.21 To present balanced information about benefits and harms, therefore, decision aids need to present information about the cumulative risk of benefits and harms (such as the cumulative risk of a false positive result, or requiring a biopsy22) over an appropriate time frame (most likely 10 years or more).
Communicating outcomes with a low probability
The numbers needed to treat associated with most treatments for symptomatic conditions are relatively small.23 In contrast, the benefits of screening are relatively rare, and the numbers needed to screen are large. For example, the number of men who need to be screened for raised cholesterol concentration to prevent one death from heart disease within five years range from 21 067 for men aged 25-34 to 231 for men aged 55-64, and the numbers are greater for women.18 For cervical screening, an estimated 1000 women need to be screened for 35 years to prevent one death from cervical cancer.24
Conveying very low event rates is technically difficult. The cumulative deaths over 10 years from bowel cancer for Australian men in their 60s are about 98/10 000 without screening and about 75/10 000 with screening.21 Thus the diagrams used to aid screening decisions may need to use 1000 faces or even 10 000 faces to show event rates as whole figures instead of the 100 faces used in diagrams for treatments.25 Furthermore, it is important to provide a context within which to consider the impact of screening.26 For example, the decision aid might include the risk of death from other causes or all causes over a similar time.
Valuing outcomes
Research indicates that people vary in the way they value screening outcomes such as false positive results.6 27 28 This is consistent with qualitative reports we have received from participants in bowel, breast, and cervical screening programmes: some people consider a false positive a good result because they feel reassured that they have been thoroughly checked and found not to have the target disorder, while others regard a false positive as inconvenient, anxiety provoking, unpleasant, or dangerous. Thus the information presented by and values clarification exercises used in screening decision aids need to be flexible about classifying outcomes as pros and cons. The aid might allow people to assign positive (pro) or negative (con) status to different outcomes themselves, rather than use a fixed scheme for everyone.
Quality of evidence
For treatments, decision aids use evidence from randomised trials about outcomes of different options. Such high quality data are often not available for screening, partly because trials of screening need to be very large and may require follow up over many years, making them expensive and time consuming. Worse, by the time the trial reports, the screening test may have been substantially updated, so the applicability of the results of the trial is unclear. Thus developers may have to rely on observational data with all its inherent biases (including lead time bias, length time bias, and the healthy volunteer effect10 11), which should be explicitly stated.
Even when trials are available, there are usually only a few, and so estimates are subject to substantial uncertainty. For example, meta-analysis of trials of screening for bowel cancer using faecal occult blood tests shows a 23% relative risk reduction for death from bowel cancer for people who screen regularly, with a 95% confidence interval of 11% to 43%.21 There may be even greater uncertainty about estimates from randomised trials in other contexts—for example, mammography screening.29 It is challenging to adequately convey the uncertainty in the data about screening. Ranges may be better than point estimates, but their use may compromise the simplicity and clarity of communication.
Informed choice in the context of screening
Consumers are reasonably familiar with having different treatment options to choose between, as there have been different drugs or management options available for many conditions for a long time. The concept of having options with regard to screening is less familiar.6 Developers of decision aids for screening should be prepared for users to respond with "But why wouldn't I be screened for X? That's the responsible thing to do." We have found it necessary to explain the down sides of screening and reasons why you might choose not to be screened at the start of decision aids.
Furthermore, some healthcare providers may feel uncomfortable with the information provided in evidence based decision aids about screening and may disagree with consumers about the amount and balance of information. When pilot testing a mammography decision aid we found that healthcare providers thought it contained too much information and was slanted against screening, while women said the information was about the right length and slanted towards screening (D Stacey et al. Reasons for Hope, Ottawa, 25-27 Oct 2003). Our experience is consistent with previous research showing ultrasonographers felt they couldn't give women information contained in an evidence based leaflet about antenatal screening.30 Developers should probably expect some discomfort among healthcare providers, who may feel that the decision aid is unbalanced, that it runs counter to their view of screening, or that information about harms will cause anxiety among patients.
Summary points
The effects of decision aids about screening on decision quality and participation in screening are largely unknown
Public (and professional) understanding of the drawbacks of screening is limited
As well as information on benefits, decision aids need to include information on detection of inconsequential disease and about the risks of the full range of investigations and treatments which may be entailed if results are abnormal
Information on benefits and risks should be given over similar time frames
People may vary in viewing outcomes of screening as pros or as cons
Incentives to participate in screening
Financial incentives or performance targets are rarely related to acceptance of treatment. In the screening context, companies directly market "off the shelf" screening kits such as faecal occult blood test and genetic screening kits. Unlike treatments, screening tests are not well regulated by government, and the public is not well protected against the commercial promotion of screening tests.6 Even within authorised and accredited government funded screening programmes, there is a perceived need to meet participation targets in order to maximise population benefits.2 Healthcare funders and those delivering services may be reluctant to support the use of screening decision aids by consumers if they think it will reduce participation. The impact of screening decision aids on participation rates is unclear,3 4 but is likely to vary between different screening interventions. Developers may need to work closely with providers of the screening service to develop and evaluate screening decision aids. Any vested interest should be declared.
Conclusions
O'Connor AM, Stacey D, Entwistle V, Llewellyn-Thomas H, Rovner D, Holmes-Rovner M, et al. Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst Rev 2003;(2): CD001431.
Raffle AE. Information about screening—is it to achieve high uptake or to ensure informed choice? Health Expectations 2001;4: 92-8.
Jepson R, Forbes C, Sowden A, Lewis R. Increasing informed uptake and non-uptake of screening: evidence from a systematic review. Health Expectations 2001;4: 116-30.
Edwards A, Unigwe S, Elwyn G, Hood K. Effects of communicating individual risks in screening programmes: Cochrane systematic review. BMJ 2003;327: 703-9.
Yamey G, Wilkes M. The PSA storm. BMJ 2002;324: 431.
Schwartz LM, Woloshin S, Fowler FJ, Welch HG. Enthusiasm for cancer screening in the United States. JAMA 2004;291: 71-8.
Agency for Healthcare Research and Quality. US Preventive Services Task Force. www.ahrq.gov/clinic/uspstfix.htm (accessed 15 Jul 2004).
Montgomery AA, Fahey T. A systematic review of the use of computers in the management of hypertension. J Epidemiol Community Health 1998;52: 520-5.
American Cancer Society. Prevention and early detection. www.cancer.org/docroot/PED/content/PED_2_3X_Early_Detection.asp?sitearea=PED (accessed 27 Jul 2004).
Barratt A, Irwig L, Glasziou P, Cumming RG, Raffle A, Hicks N, Muir Gray JA, Guyatt GH. Users' guides to the medical literature. XVII. How to use guidelines and recommendations about screening. JAMA 1999;281: 2029-2034
Morrison AS. Screening in chronic disease. 2nd ed. New York: Oxford University Press, 1992: 23.
Black W, Welch H. Advances in diagnostic imaging and over-estimations of disease prevalence and the benefits of therapy. N Engl J Med 1993;328: 1237-43.
Etzioni R, Penson DR, Legler JM, di Tommaso D, Boer R, Gann PH, et al. Overdiagnosis due to prostate-specific antigen screening: lessons from US prostate cancer incidence trends. J Natl Cancer Inst 2002;94: 981-90.
Zahl PH, Strand BH, Maehlen J. Incidence of breast cancer in Norway and Sweden during introduction of nationwide screening: prospective cohort study. BMJ 2004;328: 921-4.
Ernster VL, Ballard-Barbash R, Barlow WE, et al. Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 2002:94: 1546-54.
Alderman MH, Lamport B. Labelling of hypertensives: a review of the data. J Clin Epidemiol 1990;43: 195-200.
Lefebvre RC, Hursey KG, Carleton RA. Labeling of participants in high blood pressure screening programs. Implications for blood cholesterol screening. Arch Intern Med 1988;148: 1993-7.
Khaw KT, Rose G. Cholesterol screening programmes: how much potential benefit? BMJ 1989;299: 606-7.
Slater DN. Are women sufficiently well informed to provide valid consent for the cervical smear test? Cytopathology 2000;11: 166-70.
Kavannagh AM, Broom DH. Women's understanding of abnormal cervical smear test results: a qualitative study. BMJ 1997;314: 1388-91.
Towler B, Irwig L, Glasziou P, Kewenter J, Weller D, Silagy C. A systematic review of the effects of screening for colorectal cancer using the faecal occult blood test, Hemoccult. BMJ 1998;317: 559-65.
Elmore JG, Barton MB, Moceri VM, Polk S, Arena PJ, Fletcher SW. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338: 1089-96.
American Medical Association. Users' guides to the medical literature: a manual for evidence based clinical practice. Chicago: AMA, 2002: 387-9.
Raffle A, Alden B, Quinn M, Babb PJ, Brett MT. Outcomes of screening to prevent cancer: analysis of cumulative incidence of cervical abnormality and modelling of cases and deaths prevented. BMJ 2003;326: 901.
Man-Song-Hing M, Laupacis A, O'Connor AM, Biggs J, Drake E, Yetisir E, et al. A patient decision aid regarding antithrombotic therapy for stroke prevention in atrial fibrillation. JAMA 1999;282: 737-43.
Woloshin S, Schwartz LM, Welch HG. Risk charts: putting cancer in context. J Natl Cancer Inst 2002;94: 799-804.
Scaf-Klomp W, Sanderman R, van de Wiel HBM, Otter R, van den Heuvel WJA. Distressed or relieved? Psychological side effects of breast cancer screening in the Netherlands. J Epidemiol Community Health 1997;51: 705-10.
Pernet AL, Wardle J, Bourne TH, Whitehead MI, Campbell S, Collins WP. A qualitative evaluation of the experience of surgery after false positive results in screening for familial ovarian cancer. Psycho-oncology 1992;1: 217-33.
Olsen O, Gotszche PC. Cochrane review on screening for breast cancer with mammography. Lancet 2001;358: 1340-2.
Oliver S, Rajan L, Turner H, Oakley A, Entwistle V, Watt I, et al. Informed choice for users of health services: views on ultrasonography leaflets of women in early pregnancy, midwives, and ultrasonographers. BMJ 1996;313: 1251-3.(Alexandra Barratt, senior)