Correspondence
4
5
Office for National Statistics. Population estimates by ethnic group. http://www.ons. gov.uk/ons/taxonomy/index.html?nscl= Population+Estimates+by+Ethnic+Group#tabdata-tables (accessed Dec 23, 2013). Ahmed SF, Franey C, McDevitt H, et al. Recent trends and clinical features of childhood vitamin D deficiency presenting to a children’s hospital in Glasgow. Arch Dis Child 2011; 96: 694–96.
Refining the American guidelines for prevention of cardiovascular disease We would like to propose a compromise in the debate on the clinical application of the recently proposed American guidelines for prevention of cardiovascular disease.1 Paul Ridker and Nancy Cook (Nov 30, p 1762)2 criticise the current guidelines’ assumption of a constant relative risk reduction and its subsequent focus on absolute (baseline) risk predictions. They plea for statins prescription based on treatment effects observed in specific trial populations. In our view, combining absolute risk predictions with individualised estimates of relative risk reduction is required to quantify the absolute treatment benefit of statins. Relative risk reduction across subgroups with different levels of baseline risk can be based on the results of the individual patient data meta-analysis of statin trials. 3 Ideally, the individualised relative risk reductions are estimated in a re-analysis of these trial data, by adding a statistical treatment interaction with the risk predictions according to 2013 guidelines.4 Ridker and Cook recommend recalibration of the guidelines’ new prediction model in additional external validation cohorts. Rather, we should use already available contemporary validation cohorts to adjust poorly calibrated risk predictions for time trends. To account for well recognised cardiovascular disease risk differences across the ethnic groups of Hispanics, Asians, and native-Americans, 598
recalibration might be based on available external data as well.5 In conclusion, we recommend building guidelines on adequate estimates of absolute treatment benefit, requiring a recalibrated absolute risk prediction model in conjunction with individualised estimates of the relative risk reduction. We declare that we have no conflicts of interest.
*David van Klaveren, Yvonne Vergouwe, Ewout W Steyerberg [email protected] Department of Public Health, Erasmus MC, 3015 Rotterdam, The Netherlands 1
2
3
4
5
Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation 2013; published online Nov 13. DOI:10.1161/01.cir.0000437738.63853.7a. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Kent DM, Rothwell PM, Ioannidis JP, Altman DG, Hayward RA. Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal. Trials 2010; 11: 85. D’Agostino RB Sr, Grundy S, Sullivan LM, Wilson P, Group CHDRP. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001; 286: 180–87.
The recent American Heart Association (AHA) and American College of Cardiology (ACC) guidelines1 recommend that more people are offered treatment to prevent heart attacks and strokes; they do this by lowering the risk cutoff from a 20% 10-year risk to 7·5% (the approximate risk of a person aged 60 years). Paul Ridker and Nancy Cook state that the risk calculator in the guidelines overestimates risk about two-fold.2 This variation, however, has little effect on discriminating between who will and will not have a heart attack or stroke (ie, on screening performance).3 Age has a much greater discriminatory effect than do risk factors such as
blood pressure and cholesterol despite their aetiological importance.3 Screening using age alone has advantages. It is simple and removes the need for risk estimation, which is only necessary when many screening factors are combined (eg, in antenatal screening for Down’s syndrome, where no single marker dominates over others). Therefore, screening using age alone avoids debate about the accuracy of risk estimation using different algorithms, none of which materially improve screening performance over age alone.4 Also, the focus should move from risk in the absence of intervention to the health benefit of intervention, which is what matters. If people took blood pressure and cholesterol lowering medicines from age 60 years without previous risk factor measurement, one third would benefit and they would, on average, gain 7 years of life without a heart attack or stroke. NW jointly holds European, Canadian, and US patents for a combinaton pill for the prevention of cardiovascular disease. JM declares that he has no conflicts of interest.
*Nicholas Wald, Joan Morris [email protected] Wolfson Institute of Preventive Medicine, London, EC1M 6BQ, UK 1
2
3
4
Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; published online Nov 13. DOI:10.1016/j. jacc.2013.11.002. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Wald NJ, Simmonds M, Morris JK. Screening for future cardiovascular disease using age alone compared with multiple risk factors and age. PLoS One 2011; 6: e18742. Simmonds MC, Wald NJ. Risk estimation versus screening performance: a comparison of six risk algorithms for cardiovascular disease. J Med Screen 2012; 19: 201–05.
Paul Ridker and Nancy Cook1 express concern that the new American Heart Association (AHA) and American College of Cardiology (ACC) risk calculator “systematically overestimates” observed risks.2 This www.thelancet.com Vol 383 February 15, 2014
Correspondence
Non-diabetic patients (n=64)
Diabetic patients (n=64)
ACC/AHA calculator vs other calculators3
64·4% (65·6)
65·2% (60·2)
Overall agreement among 25 calculators (18 for diabetics) excluding the ACC/AHA calculator2
64·3% (64)
72·6% (73)
Reynold’s Risk Score
82·8%
N/A
UK Prospective Diabetes Study
71·9%
81·3%
QRISK2-2011
78·1%
87·5%
Calculators with >75% categorisation agreement in either non-diabetics or diabetics with the ACC/AHA calculator
Australian Absolute CVD Risk
79·7%
76·6%
New Zealand CV Risk Charts
81·3%
78·1%
Australian Risk Charts
87·5%
73·4%
New Zealand Know Your Numbers
79·7%
57·8%
PROCAM (Health Check)
65·6%
79·7%
Data are average (median) or average. ACC/AHA=American College of Cardiology/American Heart Association.
Table: Agreement of the ACC/AHA risk calculator with a sample of other commonly used risk calculators
might well be, but how does this new calculator compare with other similar calculators? We decided to compare how this new calculator categorises patients into risk levels compared with other commonly used cardiovascular disease risk calculators. We used risk categorisation data from our previous study3 of 25 calculators in which 128 standardised patients over a broad range of risk (each of seven risk factors variably dichotomised to high and low values) were categorised and we compared these data with the risk categorisation obtained if one used the ACC/AHA calculator. To allow for easy comparison, we used the standard 10-year risk categories of less than 10%, between 10% and 20%, and more than 20%. The calculator agreement results are shown in the table. For non-diabetic patients, the agreement of the ACC/AHA calculator was very similar to the risk categorisation agreement for all calculators. For patients with diabetes, the average agreement was 7% lower and the median agreement 13% lower, with a tendency to assign lower risk slightly more often than would most other calculators. It seems that the ACC/AHA risk calculator categorises patients in a similar way to many of the other available calculators. www.thelancet.com Vol 383 February 15, 2014
Importantly, calculations of risk are estimates and categorisations are relatively arbitrary. Any risk estimate should only be used as a tool to facilitate the initial discussion of benefits and harms of treatments as part of shared decision making. We declare that we have no conflicts of interest.
*James McCormack, Hoan Linh Banh, G Micheal Allan [email protected] University of British Columbia, Faculty of Pharmaceutical Sciences, Vancouver V6T1Z3, BC Canada (JMcC); and Department of Family Medicine, University of Alberta, Edmonton, AB, Canada (HLB, GMA). 1
2
3
Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Goff DC, Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; published online Nov 13. DOI:10.1016/j.jacc.2013.11.005. Allan GM, Nouri F, Korownyk C, Kolber MR, Vandermeer B, McCormack J. Agreement among Cardiovascular Risk Calculators. Circulation 2013; 127: 1948–56.
The Comment by Paul Ridker and Nancy Cook1 concerning the new American Heart Association (AHA) and American College of Cardiology (ACC) guidelines deserves comment, primarily for the issues not discussed. The main problem with the new AHA/ ACC guidelines is that they are based on randomised controlled trials (RCTs),
all of which are affected by selection bias. This selection bias is a necessity of the RCT, which has to involve highrisk individuals, who are highly likely to sustain an atherothrombotic disease event within the trial timeframe. As a result, the findings of any RCT cannot be generalised to the population as a whole. This adds confusion to the primary prevention of atherothrombotic disease because the observational studies, which the AHA/ACC guidelines have disregarded, involve all comers and thus are more likely to involve a cohort that is much more representative of the entire population and to involve primary rather than secondary prevention scenarios. The abandonment of the low-density lipoprotein (LDL)-cholesterol goals of therapy seems foolish. The AHA/ACC guidelines seem to involve giving an arbitrary dose of statin and hoping for the best. As a result, some patients will be overtreated whereas others will be undertreated. If the patient has had his or her dyslipidaemia long enough, he or she will have subclinical atherothrombotic disease and will need dyslipidaemia therapy sufficient to attain an LDL-cholesterol level of 2·0 mmol/L if the precipitation method of high-density lipoprotein (HDL)cholesterol determination is used, but 1·8 mmol/L if the enzymatic method of HDL-cholesterol measurement is used.2 If the patient has only had his or her dyslipidaemia a relatively short time, he or she could probably be treated to lesser goals.3 Because this question is not answerable in most cases, the former goal would seem to be the safest. I declare that I have no conflicts of interest.
William E Feeman Jr [email protected] The Bowling Green Study, Bowling Green, OH43402, USA 1
2
3
Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Feeman WE. LDL lowering: evidence for a plaque non-progression threshold. Int J Cardiovasc Res 2013; 2: 5. Feeman WE Jr. Prediction of the population at risk of atherothrombotic disease. Exp Clin Cardiol 2004; 9: 235–41.
599
4
5
Office for National Statistics. Population estimates by ethnic group. http://www.ons. gov.uk/ons/taxonomy/index.html?nscl= Population+Estimates+by+Ethnic+Group#tabdata-tables (accessed Dec 23, 2013). Ahmed SF, Franey C, McDevitt H, et al. Recent trends and clinical features of childhood vitamin D deficiency presenting to a children’s hospital in Glasgow. Arch Dis Child 2011; 96: 694–96.
Refining the American guidelines for prevention of cardiovascular disease We would like to propose a compromise in the debate on the clinical application of the recently proposed American guidelines for prevention of cardiovascular disease.1 Paul Ridker and Nancy Cook (Nov 30, p 1762)2 criticise the current guidelines’ assumption of a constant relative risk reduction and its subsequent focus on absolute (baseline) risk predictions. They plea for statins prescription based on treatment effects observed in specific trial populations. In our view, combining absolute risk predictions with individualised estimates of relative risk reduction is required to quantify the absolute treatment benefit of statins. Relative risk reduction across subgroups with different levels of baseline risk can be based on the results of the individual patient data meta-analysis of statin trials. 3 Ideally, the individualised relative risk reductions are estimated in a re-analysis of these trial data, by adding a statistical treatment interaction with the risk predictions according to 2013 guidelines.4 Ridker and Cook recommend recalibration of the guidelines’ new prediction model in additional external validation cohorts. Rather, we should use already available contemporary validation cohorts to adjust poorly calibrated risk predictions for time trends. To account for well recognised cardiovascular disease risk differences across the ethnic groups of Hispanics, Asians, and native-Americans, 598
recalibration might be based on available external data as well.5 In conclusion, we recommend building guidelines on adequate estimates of absolute treatment benefit, requiring a recalibrated absolute risk prediction model in conjunction with individualised estimates of the relative risk reduction. We declare that we have no conflicts of interest.
*David van Klaveren, Yvonne Vergouwe, Ewout W Steyerberg [email protected] Department of Public Health, Erasmus MC, 3015 Rotterdam, The Netherlands 1
2
3
4
5
Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation 2013; published online Nov 13. DOI:10.1161/01.cir.0000437738.63853.7a. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Kent DM, Rothwell PM, Ioannidis JP, Altman DG, Hayward RA. Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal. Trials 2010; 11: 85. D’Agostino RB Sr, Grundy S, Sullivan LM, Wilson P, Group CHDRP. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA 2001; 286: 180–87.
The recent American Heart Association (AHA) and American College of Cardiology (ACC) guidelines1 recommend that more people are offered treatment to prevent heart attacks and strokes; they do this by lowering the risk cutoff from a 20% 10-year risk to 7·5% (the approximate risk of a person aged 60 years). Paul Ridker and Nancy Cook state that the risk calculator in the guidelines overestimates risk about two-fold.2 This variation, however, has little effect on discriminating between who will and will not have a heart attack or stroke (ie, on screening performance).3 Age has a much greater discriminatory effect than do risk factors such as
blood pressure and cholesterol despite their aetiological importance.3 Screening using age alone has advantages. It is simple and removes the need for risk estimation, which is only necessary when many screening factors are combined (eg, in antenatal screening for Down’s syndrome, where no single marker dominates over others). Therefore, screening using age alone avoids debate about the accuracy of risk estimation using different algorithms, none of which materially improve screening performance over age alone.4 Also, the focus should move from risk in the absence of intervention to the health benefit of intervention, which is what matters. If people took blood pressure and cholesterol lowering medicines from age 60 years without previous risk factor measurement, one third would benefit and they would, on average, gain 7 years of life without a heart attack or stroke. NW jointly holds European, Canadian, and US patents for a combinaton pill for the prevention of cardiovascular disease. JM declares that he has no conflicts of interest.
*Nicholas Wald, Joan Morris [email protected] Wolfson Institute of Preventive Medicine, London, EC1M 6BQ, UK 1
2
3
4
Stone NJ, Robinson J, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; published online Nov 13. DOI:10.1016/j. jacc.2013.11.002. Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Wald NJ, Simmonds M, Morris JK. Screening for future cardiovascular disease using age alone compared with multiple risk factors and age. PLoS One 2011; 6: e18742. Simmonds MC, Wald NJ. Risk estimation versus screening performance: a comparison of six risk algorithms for cardiovascular disease. J Med Screen 2012; 19: 201–05.
Paul Ridker and Nancy Cook1 express concern that the new American Heart Association (AHA) and American College of Cardiology (ACC) risk calculator “systematically overestimates” observed risks.2 This www.thelancet.com Vol 383 February 15, 2014
Correspondence
Non-diabetic patients (n=64)
Diabetic patients (n=64)
ACC/AHA calculator vs other calculators3
64·4% (65·6)
65·2% (60·2)
Overall agreement among 25 calculators (18 for diabetics) excluding the ACC/AHA calculator2
64·3% (64)
72·6% (73)
Reynold’s Risk Score
82·8%
N/A
UK Prospective Diabetes Study
71·9%
81·3%
QRISK2-2011
78·1%
87·5%
Calculators with >75% categorisation agreement in either non-diabetics or diabetics with the ACC/AHA calculator
Australian Absolute CVD Risk
79·7%
76·6%
New Zealand CV Risk Charts
81·3%
78·1%
Australian Risk Charts
87·5%
73·4%
New Zealand Know Your Numbers
79·7%
57·8%
PROCAM (Health Check)
65·6%
79·7%
Data are average (median) or average. ACC/AHA=American College of Cardiology/American Heart Association.
Table: Agreement of the ACC/AHA risk calculator with a sample of other commonly used risk calculators
might well be, but how does this new calculator compare with other similar calculators? We decided to compare how this new calculator categorises patients into risk levels compared with other commonly used cardiovascular disease risk calculators. We used risk categorisation data from our previous study3 of 25 calculators in which 128 standardised patients over a broad range of risk (each of seven risk factors variably dichotomised to high and low values) were categorised and we compared these data with the risk categorisation obtained if one used the ACC/AHA calculator. To allow for easy comparison, we used the standard 10-year risk categories of less than 10%, between 10% and 20%, and more than 20%. The calculator agreement results are shown in the table. For non-diabetic patients, the agreement of the ACC/AHA calculator was very similar to the risk categorisation agreement for all calculators. For patients with diabetes, the average agreement was 7% lower and the median agreement 13% lower, with a tendency to assign lower risk slightly more often than would most other calculators. It seems that the ACC/AHA risk calculator categorises patients in a similar way to many of the other available calculators. www.thelancet.com Vol 383 February 15, 2014
Importantly, calculations of risk are estimates and categorisations are relatively arbitrary. Any risk estimate should only be used as a tool to facilitate the initial discussion of benefits and harms of treatments as part of shared decision making. We declare that we have no conflicts of interest.
*James McCormack, Hoan Linh Banh, G Micheal Allan [email protected] University of British Columbia, Faculty of Pharmaceutical Sciences, Vancouver V6T1Z3, BC Canada (JMcC); and Department of Family Medicine, University of Alberta, Edmonton, AB, Canada (HLB, GMA). 1
2
3
Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Goff DC, Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2013; published online Nov 13. DOI:10.1016/j.jacc.2013.11.005. Allan GM, Nouri F, Korownyk C, Kolber MR, Vandermeer B, McCormack J. Agreement among Cardiovascular Risk Calculators. Circulation 2013; 127: 1948–56.
The Comment by Paul Ridker and Nancy Cook1 concerning the new American Heart Association (AHA) and American College of Cardiology (ACC) guidelines deserves comment, primarily for the issues not discussed. The main problem with the new AHA/ ACC guidelines is that they are based on randomised controlled trials (RCTs),
all of which are affected by selection bias. This selection bias is a necessity of the RCT, which has to involve highrisk individuals, who are highly likely to sustain an atherothrombotic disease event within the trial timeframe. As a result, the findings of any RCT cannot be generalised to the population as a whole. This adds confusion to the primary prevention of atherothrombotic disease because the observational studies, which the AHA/ACC guidelines have disregarded, involve all comers and thus are more likely to involve a cohort that is much more representative of the entire population and to involve primary rather than secondary prevention scenarios. The abandonment of the low-density lipoprotein (LDL)-cholesterol goals of therapy seems foolish. The AHA/ACC guidelines seem to involve giving an arbitrary dose of statin and hoping for the best. As a result, some patients will be overtreated whereas others will be undertreated. If the patient has had his or her dyslipidaemia long enough, he or she will have subclinical atherothrombotic disease and will need dyslipidaemia therapy sufficient to attain an LDL-cholesterol level of 2·0 mmol/L if the precipitation method of high-density lipoprotein (HDL)cholesterol determination is used, but 1·8 mmol/L if the enzymatic method of HDL-cholesterol measurement is used.2 If the patient has only had his or her dyslipidaemia a relatively short time, he or she could probably be treated to lesser goals.3 Because this question is not answerable in most cases, the former goal would seem to be the safest. I declare that I have no conflicts of interest.
William E Feeman Jr [email protected] The Bowling Green Study, Bowling Green, OH43402, USA 1
2
3
Ridker PM, Cook NR. Statins: new American guidelines for prevention of cardiovascular disease. Lancet 2013; 382: 1762–65. Feeman WE. LDL lowering: evidence for a plaque non-progression threshold. Int J Cardiovasc Res 2013; 2: 5. Feeman WE Jr. Prediction of the population at risk of atherothrombotic disease. Exp Clin Cardiol 2004; 9: 235–41.
599
