Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Comparison of methods to identify high risk individuals
- Ronan M Conroy, Davida De La Harpe (27 June 2003)
-
Optimum strategies for patient selection in prevention of CHD
- Tom P Marshall (30 June 2003)
-
Absolute Risk assessment: which table to use?
- Davide Giavarina, Elena Barzon, Massimo Cigolini, Gabriella Mezzena, and Giuliano Soffiati (5 July 2003)
-
Integrated updated risk profiling
- John M Waddell, Caron Neal (7 July 2003)
-
Conclusion is an Oversimplification
- Gerd Assmann, Paul Cullen, Helmut Schulte (7 July 2003)
-
CAD risk assessment must start before the age of 50 years
- Benoit J BOLAND, Régis De Muylder, Geert Goderis, Jan Degryse (18 July 2003)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Ronan M Conroy, Lecturer in Biostatistics Royal College of Surgeons in Ireland, Dublin 2, Ireland, Davida De La Harpe
Send response to journal:
|
Sir, Wilson and colleagues report an interesting comparison of methods used to detect persons at significant risk of coronary heart disease.[1] They used a coronary risk function published by the Framingham group as a gold standard to measure coronary risk.[2] They reported that the Sheffield tables[3] performed well, identifying 99.9% of those at more than 15% risk. This is hardly a surprising result, as the Sheffield tables are calculated from the self same function that they themselves used to identify those at high risk. The gold standard, then, performed well against the gold standard. There are some other considerations which the paper raises. The use of the Framingham function to identify individuals at high risk of coronary heart disease means relying on a definition of coronary disease which differs in important respects from the definitions used in therapeutic trials, notably by including the onset of angina and ‘coronary insufficiency’ as endpoints. In women in the Framingham study, new onset angina accounted for 41% of all events in men and 56% in women[4], giving the term ‘coronary disease risk’ a different meaning in the two sexes. This makes it hard to relate the evidence base for risk factor modification to the persons identified by the Framingham function. The most important question, however, concerns the continuing focus on diseases – coronary heart disease, stroke – as separate entities, rather than acknowleding that the same people who are at high risk of coronary disease are at high risk of other cardiovascular diseases too. This is proposed by the SCORE project, which argues for the identification of persons at risk of any cardiovascular event.[5] Ronán M Conroy
References 1. Wilson S, Johnston A, Robson J, Poulter N, Collier D, Feder G, et al. Comparison of methods to identify individuals at increased risk of coronary disease from the general population. BMJ 2003;326(7404):1436-0. 2. Anderson KM, Wilson PW, Odell PM, Kannel WB. An updated coronary risk profile. A statement for health professionals. Circulation 1991 Jan;83(1):356-62. 3. Haq IU, Jackson PR, Yeo WW, Ramsay LE. Sheffield risk and treatment table for cholesterol lowering for primary prevention of coronary heart disease. Lancet 1995;346(8988):1467-71. 4. D'Agostino RB, Russell MW, Huse DM, Ellison RC, Silbershatz H, Wilson PW, et al. Primary and subsequent coronary risk appraisal: new results from the Framingham study. American Heart Journal 2000;139(2 Pt 1):272-81. 5. Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J 2003;24(11):987-1003. Competing interests: Ronan Conroy is the principal investigator of the SCORE project. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Tom P Marshall, Lecturer in Public Health Birmingham University B15 2TT
Send response to journal:
|
Dear Sir, Wilson et al’s paper on selection of patients for prevention of coronary heart disease addresses an important question.[1] They conclude that cholesterol testing on all patients over 50 is the optimum method evaluated. However this conclusion may be misleading. Firstly the strategies compared are not equal from a resource perspective. The strategy recommended by the authors requires full clinical assessment of 2920 patients; the strategy using a risk estimate assesses 1121 patients. Full clinical assessment requires three cholesterol tests.[2] If taken by a practice nurse, these strategies cost a total of £82,490 and £31,668 respectively. The strategy using a risk estimate can be adapted to identify a larger number of patients for assessment by setting the threshold for assessment at 14%, 13% or 10%. I repeated the authors’ method using the same data source.[3] A practice assessing the 2220 patients with CHD risk estimates >10% identifies 96% of patients truly at >15% ten-year CHD risk. This identifies more true positives, is less costly (£62,715) and has fewer false positives than the authors’ preferred strategy. Secondly, the strategy using a prior estimate of CVD risk described in the paper requires data on age, sex, blood pressure, diabetic status and smoking history. Some of these data - age, sex and diabetic status - are routinely held on practice databases. Others are recorded with varying frequency - blood pressure and smoking status. The authors do not make it clear where these data are to be obtained by a primary care team. However a previously described methodology for calculating prior estimates of CHD risk offers an alternative.[4] This uses only data on age, sex and diabetic status. All other data are generated from defaults based on the patient’s age, sex and diabetic status. I applied this methodology to the same group of patients identified by the authors from the Health Survey for England I generated the same group of patients. By assessing 2527 patients with CHD risk estimates >7% (using the Rouse-Marshall method) a practice identifies 93% of patients truly at >15% ten-year CHD risk. This is also less costly (£71,388), more sensitive and more specific than the authors’ preferred strategy and has the merit of not requiring access to data that may not be available on the practice database. Finally it is possible to apply a simplified version of the Rouse- Marshall strategy. This ranks the entire population by an age-sex ranking number: for men, their age; for women their age minus 13 (to reflect their lower average risk). Assessing the highest ranked 2650 patients under this approach is also less costly (£74,864), more sensitive (93%) and more specific than the authors’ preferred strategy and is equally simple. REFERENCES: 1. Wilson S, Johnston A, Robson J, Poulter N, Collier D, Feder G, and Caulfield M J. Comparison of methods to identify individuals at increased risk of coronary disease from the general population BMJ 2003; 326: 1436. 2 Wood D. Durrington P. Poulter N. McInnes G. Rees A. and Wray R. on behalf of the Societies. Joint British recommendations on prevention of coronary heart disease in clinical practice. Heart 1998;80:Supplement 2. 3 Health Survey for England 1998. 2000 The Stationery Office, London. 4 Marshall T., Rouse A. Resource implications and health benefits of primary prevention strategies for cardiovascular disease in people aged 30 to 74: mathematical modelling study. British Medical Journal 2002;325:197- 199. Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Davide Giavarina, Clinical Chemistry and Haematology Laboratory San Bortolo Hospital, 36100 Vicenza, (Italy), Elena Barzon, Massimo Cigolini, Gabriella Mezzena, and Giuliano Soffiati
Send response to journal:
|
Absolute Risk assessment:
which table to use? Davide Giavarina, Elena Barzon, Massimo Cigolini,
Gabriella Mezzena, Giuliano Soffiati. Sir, Wilson and colleagues [1] compared four different strategies for identifying people at high risk of coronary disease in the general population. They demonstrated that Sheffield tables [2] correctly identified 99.9% of subjects at more than 15% risk of coronary hearth disease, by using the Framingham equation [3] as a criterion standard. We evaluated a new chart for the risk assessment that has been recently introduced in Italy.[4] To this aim, we employed the Italian chart to estimate how many subjects would have been treated with statins in a population including 536 healthy blood donors and 213 patients with non-insulin dependent diabetes mellitus. Results were compared with those calculated by using the Framingham equation as reference method and with alternative risk assessment screening methods, including the New Zealand chart [5], the new Sheffield tables [2], the chart of the Joint Task Force of European Societies on Coronary Prevention [6], and criteria of the Joint British Societies.[7] Results are summarized in Table 1. Table 1: Comparison of the Framingham equation with
alternative screening methods and estimated risk assessment.
Our results show poor concordance between the different methods of risk-assessment and the Framingham equation, which can be accounted only in part to the approximation and simplification introduced by the tables. However, as demonstrated also by Wilson et al. [1], among the different screening methods, Sheffield tables were the most efficient in identifying people at high risk of coronary disease, since all patients were correctly classified. As for the Italian risk assessment chart, our results demonstrate that this method has a low sensitivity, since more than 80% of patients at risk for coronary disease were misdiagnosed. The benefit of cholesterol lowering therapy varies
with the absolute risk of coronary hearth disease in the treated population,
and the indication for statin treatment focuses on cost-effectiveness [8]. It is known that all risk tables and charts produce
results slightly different from those obtained from the complete Framingham
equation.[9,10] Calculation of risk using the full equation is preferred,
however when it is not possible, the Joint British Guidelines chart or the New
Sheffield risk tables are the best methods to select people at high risk from
the general population. 1. S Wilson, A Johnston, J Robson, N Poulter, D
Collier, G Feder, and M J Caulfield. Comparison of methods to identify
individuals at increased risk of coronary disease from the general population.
BMJ 2003; 326: 1436-41. 2. Wallis EJ, Ramsay LE, Ul Haq I, Ghahramani P,
Jackson PR, Rowland-Yeo K, Yeo WW. Coronary and cardiovascular risk estimation
for primary prevention: validation of a new Sheffield table in the 1995
Scottish health survey population. BMJ. 2000;320:671-6. 3. Anderson KM, Odell PM, Wilson WF, Kannel WB. Cardiovascular
disease risk profile. Am Hearth J 1990;121:293-8. 4. Menotti A, Lanti M, Puddu PE, Carratelli L, Mancini
M, Motolese M, Prati P, Zanchetti A. An Italian chart for cardiovascular risk
prediction. Its scientific basis. Ann Ital Med Int 2001;16:240-51. 5. Jackson R. Updated New Zealand cardiovascular
disease risk-benefit prediction guide. BMJ. 2000 Mar 11;320(7236):709-10. 6. Wood D, De Backer G, Faergeman O, Graham I, Mancia
G, Pyorala K. Prevention of coronary heart disease in clinical practice: recommendations
of the Second Joint Task Force of European and other Societies on Coronary
Prevention. Atherosclerosis. 1998;140:199-270 7. British Cardiac Society, British Hyperlipidaemia
Association, British Hypertension Society, and British Diabetic Association.
Joint British recommendations on prevention of coronary heart disease in
clinical practice: summary. BMJ 2000; 320: 705-708. 8. Durrington PN, Prais H, Bhatnagar D, France M,
Crowley V, Khan J, Morgan J. Indications for cholesterol-lowering medication:
comparison of risk-assessment methods. Lancet. 1999 23;353:278-8 9. Jones AF, Walker J, Jewkes C, Game FL, Bartlett WA,
Marshall T, Bayly GR. Comparative accuracy of cardiovascular risk prediction
methods in primary care patients. Heart 2001;85:37-43. 10. Bayly G. HDL-cholesterol and cardiac disease: which table to use? Ann
Clin Biochem 2002; 39:12-21 Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
John M Waddell, GP Principal Station Medical Grooup Blyth Northumberland NE24 6QN, Caron Neal
Send response to journal:
|
Over the last two years we have developed an audit routine to provide regularly, an update profile of Framingham risk of the whole population using EMIS ...the system used for over 50% of the English population. The parameters from within the pateints` computerised record for the Framingham are used in a system calculation to allow the score to be entered into the patients` computerised record (Dr Robson) For 30-74yr non-established- cardiovasular-disease group we continue a profile of the latest Framingham score enters..`Full` level, where all parameters present or `Estimated` where not. The scores are updated regularly as parameters change (perhaps controversial but enabling the NSF 4 ...to demonstrate reduction of number in the higher risk bands to lower potential risk bands) The routine corrects those allocated the wrong level (Full instead of Estimated ), gives a score to those newly eligible so that a clean profile is available and the names of those in any risk band in any target group (e.g.All , all except or including diabetes/hypertension, hypertensives with /without diabetes,diabetes..even family history (not a Framingham parameter per se)) are immediately available. Thus those with a merely Estimated risk over 15% within any group can be prioritised to see and complete all the parameters for Full score and instigate interventions EMIS have exported the `suite ` to the other practices in our PCG and the adjacent one ..we are in an area with among the highest CVD mortality ratios We have designed a training manual and have a worker to induce practice nursing and admin staff into the routine Thus this is implementing the fourth option of your article and provides a direct way of bringing in the Rouse-Marshall groups eg for whole group 30-74 (no CVD)..station June 2003
No= 4231 No systolic (no score)=311
risk in 10 yr
0-15 15-30 30plus total
Full 272 134 11 417
Estimated 3054 434 15 3503...
I note in the Wallis et al BMJ 320 671 (2000) article that the Sheffielld tables were used to indicate who to screen for cholesterol:HDL assuming they could have a TC:HDL up to 10. Thus many younger people with much lower ratios would be tested to find the 33/35 they found with the ratio >8. Setting the ratio up to 10 for the younger people therefore being inefficient and finding many with low actual risk The method we use employs a novel and only moderately elaborate use of the existing clinical system audit making this exercise feasible. It provides equitable and efficient prioritisation providing the Gold standard of direct notwithstanding imperfections such as (impractical) ideal to use several cholesterol results to counteract variance. Wallis EJ et al Coronary and Cardiovascular risk estimation for primary prevention: validation of a new Sheffield table in the 1995 Scottish health survey population Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Gerd Assmann, Professor of Laboratory Medicine Institute of Clinical Chemistry, University of Muenster, 48149 Muenster, Germany, Paul Cullen, Helmut Schulte
Send response to journal:
|
In their paper, Wilson et al. conclude that “measuring the cholesterol concentration of everyone aged 50 years and over is a simple and efficient method of identifying people at high risk of coronary disease in the general population”. In our opinion, this conclusion is an oversimplification resulting from a study with several important defects. First, the Sheffield table and the underlying Framingham algorithm were used as a the gold standard for determining absolute risk of coronary heart disease. Recently, we examined how well the Framingham algorithm predicted coronary events in Germany based on an analysis of the cohorts included in the Prospective Cardiovascular Münster (PROCAM) and the Augsburg cohort of the World Health Organisation Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA) studies.1 In both populations, the Framingham algorithm overestimated coronary risk by a factor of about two. The incidence of coronary heart disease in Britain is substantially higher than in Germany.2 Nevertheless, it would have been more appropriate for Wilson and colleagues to calculate absolute coronary heart disease risk in their population using either a British algorithm or a corrected Framingham formula. Second, men and women were considered together in the estimation of coronary risk, despite the fact that the estimated risk of coronary heart disease in non-diabetic women is between 2 and 4 times less than men of the same age up to the age of about 60 years. This difference is reflected in the differing prevalences of 10-year Framingham risk of >15% among men and women shown in Table 1 of the paper. The joint consideration of men and women dilutes the estimated risk of coronary heart disease, which can be substantial in men even below the age of 50 years, as explained below. Third, Wilson and colleagues suggest that cholesterol need not be measured in persons below the age of 50 years. However, in PROCAM, about 8% of men below this age had a 10-year risk of coronary heart disease of >15% as calculated using Framingham. It would therefore be helpful if Wilson and colleagues could provide the data shown in Table 2 for men and women separately. In addition, it would be useful to know the proportion of men below the age of 50 years in the English health survey population with a 10-year coronary risk of >15%. A particular aim of primary prevention of coronary heart disease is to prevent events occurring at an early age, as these produce an especially high burden of morbidity and cost. Moreover, since coronary heart disease risk increases sharply with increasing age, patients at borderine risk below the age of 50 also require attention if they are not to proceed inexorably into the high risk group. Thus, we conclude that restricting cholesterol measurements on the basis of age as suggested severely hampers the medical and economic benefits of primary prevention and will lead to an increased burden of disease and increased costs in the long term. 1. Hense HW, Schulte H, Löwel H, Assmann G, Keil U. Framingham risk function overestimates risk of coronary heart disease in men and women from Germany. Results from the MONICA Augsburg cohort and the PROCAM cohort. Eur Heart J 2003; 24: 937-945 2. Conroy RM, Pyörälä K, Fitzgerald AP et al. Estimation of 10-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J 2003; 24:987-1003 Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Benoit J BOLAND, general internist UCLouvain, B-1200 Brussels, Belgium, Régis De Muylder, Geert Goderis, Jan Degryse
Send response to journal:
|
EDITOR - The article by Wilson and colleagues compared four strategies to identify individuals at high risk of coronary disease in the general population.1 The authors concluded that the best method is the measurement of serum cholesterol in all individuals age 50 years and over. We agree that cholesterol measurement is not usefull in all adults but should be a second step for risk assessment in the subgroup of subjects at potential high risk. However, we disagree that age of 50 years or more is the right strategy to trigger risk assessment. Firstly, this age cut-off ignores many individuals who suffer a first major coronary events (MCE). Indeed, population-based registers indicate that up to 25% of all first MCE in men occur before the age of 55 years.2 Morover, coronary risk factors (RF) - e.g. smoking, elevated blood pressure and coronary family history - are frequent before the age of 50 years and should lead to formal coronary risk assessment and sometimes to specific life-style treatments.3 Secondly, the strategy suggested by Wilson and colleagues does not identify the individuals at intermediate coronary risk, who suffer one third of all first MCE and deserve preventive life-style interventions.4 Therefore, we recommend another strategy suited to the daily primary care practice, based on a systematic and quick (~ one minute) screening of clinical risk factors in all patients aged 30 years or more, namely smoking, history of dyslipidaemia, high blood pressure, diabetes, familial coronary disease, and age > 45 years.4 We showed that the absence of these widely available risk factors indicates low coronary risk, and no need of cholesterol measurement. Cholesterol testing and Framingham-based risk assessment are performed in all adults with any risk factor. Our strategy offers multiple advantages : the medical decision is quickly made without using any risk table; the clinical risk factors - which are often lacking in the medical records 5 - are systematically screened; cholesterol is only measured in patients with risk factor(s); and coronary risk is determined in all adult patients. This clinical strategy is able to identify all individuals at elevated risk, and was validated over ten years in a population-based cohort. It was tested among general practitioners who found it interesting, easy, applicable and useful.6 Competing interests : none References 1. Wilson S, Johnston A, Robson J, Poulter N, Collier D, Feder G, Caulfield MJ. Comparison of methods to identify indivduals at increased risk of coronary disease from the general population. BMJ 2003;326:1436- 40. (28 June.) 2. De Henauw S, De Backer G. Coronaire lijden in België (coronary heart disease in Belgium). Tijdschrift voor geneeskunde 1995;51:1607-16. 3. Brohet C, Janssens D, Beck D, Hannut R, Kulbertus H, Lavenne F et al. Cardiovascular risk factors in a sample of a rural Belgian population: the Bellux MONICA study. Acta Med Scand. 1998; Suppl 728:129-136. 4. Boland B, De Muylder R, Goderis G, Degryse J, Gueuning Y, Paulus D, Jeanjean M. An algorithm for cardiovascular risk prevention in general practice: development and validation. BJGP (submitted, June 2003). 5. McManus RJ, Mant J, Meulendijks CFM, et al. Comparison of estimates and calculations of risk of coronary heart disease by doctors and nurses using different calculation tools in general practice: a cross-sectional study. BMJ 2002; 324:459-464 6. De Muyder R, Goderis G, Paulus D, Degryse J, Jeanjean M, Boland B. A clinical algorithm for diagnosis of high cardiovascular risk. Acta Cardiologica 2003 ;58:59. Authors : B Boland 1-2, R De Muylder 1, G Goderis 1, J Degryse 3,
Competing interests: None declared |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||