Australasian guidelines for medical thromboprophylaxis are flawed
J Alasdair Millar, Southland Hospital
19 December 2011
In this issue of the Journal, Gibbs, Fletcher and other Australian colleagues report that compliance with Australasian guidelines (1) (ANZWPG) for medical thromboprophylaxis (of which Gibbs and Fletcher are co-authors) can be increased from 38% to 54% by implementation of a 'multifaceted nurse facilitated educational program'.(2) The finding that a degree of behavioral change can be detected after an 'educational' program including prompts and feedback is not surprising, though the 'improvement' is surprisingly modest. The real question raised by this paper is whether the thromboprophylaxis policy being promoted is well-founded, and the extent of clinical benefit.
Problems with guidelines for medical thromboprophylaxis
1. Assessment of thrombotic risk
In the ANZWPG, which are based on the corresponding ACCP guidelines,(3) patients are deemed 'high-risk' and eligible for thromboprophylaxis (as unfractionated or low molecular weight heparin) if they have any one of 7 risk factors ('ischaemic stroke, history of VTE, active cancer, decompensated heart failure, acute on chronic lung disease, acute inflammatory disease, age > 60 y'). This immediately raises several issues. The descriptors are operationally vague. Ischaemic stroke was an exclusion factor in recent clinical trials,(4-6) that is, no evidence of benefit from prophylaxis with low-molecular weight heparin exists for stroke patients. The risk factors have varying statistical weights but are in effect assigned the same weight because any one of them creates eligibility. 'Age > 60 years' is an arbitrary variation on the actual risk factor, correctly described in the ACCP Guidelines as 'increasing age'.(3) Since most general medical patients are > 60, this risk factor ensures that a high percentage of patients (in the Gibbs paper, 82%) are deemed to be at 'high risk'. The effect is to assign almost blanket eligibility for thromboprophylaxis in medical patients. This is incorrect in principle because anticoagulants have a significant major bleeding risk. Also, increasing age is a risk factor for bleeding with anticoagulants, as well as for thrombosis.(7-9) I have shown that the ANZWPG provide a benefit:hazard ratio less than unity, ie that they may do more harm than good, and that only eligibility based on risk factors of high weighting will produce net benefit.(10)
One might also ask about the significance of a finding that over 80% of a population is at 'high risk' of a disease. Theoretical considerations show that the maximum possible relative risk in a sub-population at high risk is the reciprocal of the proportion it represents.(11) Thus under the Australasian guidelines, the maximum relative risk in eligible patients is 1 ÷ 0.82, or 1.22, ie the added risk is 22%, on average, using the data from Gibbs et al.
2. Incidence of secondary thromboembolism in medical patients
The incidence of new thrombotic episodes in medical inpatients is given as 17%, 10-20% and 15% in the ANZWPG, the ACCP guidelines and the House of Commons Select Committee Report (12) (HCSCR) , respectively. The first two guidelines acknowledge that the figures include 'sub-clinical DVT' but neither state the proportion of events that are asymptomatic. The HCSCR talks of 'detectable' thrombosis but provides no further clarification, raising the possibility that MPs may have been misled on this point. Incidence data based on hospital morbidity coding statistics puts the incidence of symptomatic secondary thrombosis at 0.4-0.5%.(13,14) or less.(15) Thus over 95% of asymptomatic thrombotic events in medical patients do not progress to symptomatic disease. Several important questions arise from these facts:
1. Should prophylaxis aim to prevent asymptomatic disease? The ACCP guidelines state (reasonably) that the primary purpose of thromboprophylaxis is to prevent pulmonary embolus, but no guidelines discuss whether preventing asymptomatic disease is a meaningful aim. To this author it seems a non-sequitor.
2. Is prophylaxis of asymptomatic disease justified because it progresses to symptomatic disease in some patients? This beguiling argument may be valid if progression from asymptomatic to symptomatic disease were common, but for thromboprophylaxis the transition is a rarity. If the object is to prevent symptomatic disease, then the relevant determinants are the incidence of the symptomatic condition and the relative risk reduction of the proposed prophylaxis, which together determine the clinical gain, measured as the number needed to treat to prevent one event (NNT).
3. What is the source of the data? The use of data for asymptomatic disease derives from the randomized controlled thromboprophylaxis trials(4-6) in which detection of thrombosis was accomplished using techniques that detect asymptomatic thrombi. The reason for this was to ensure objective measurement and, because asymptomatic events are much more common, to decrease the size and hence the cost of the clinical trials. However, the result is that the trials measured the effect of thromboprophylaxis on proxy end-points.
4. What is the outcome? Using data based on asymptomatic events exaggerates the size of the clinical problem (by about 30-fold) and invalidates the outcomes of secondary analyses such as NNT and cost-effectiveness. For example, with a relative risk reduction from trials(4-6) of about 50%, a baseline incidence of 17% gives a NNT of 12 (to prevent each asymptomatic VTE event), or 400 (to prevent each symptomatic event), with proportionate differences in cost-effectiveness, respectively. A meta analysis supports the latter figure for pulmonary embolus.(16)
Anyone who works in the medical wards of any hospital knows that the given incidence figures when applied to clinical disease are simply nonsensical, and this has been confirmed in practice.(17,18)
Problems with the paper
In addition to problems with the Guidelines, the Gibbs et al paper contains various shortcomings:
' The authors claim (from autopsy studies) that 50-70% of symptomatic events, and also that 70-80% of fatal PE occur in non-surgical patients. The relevance of these data to the clinical problem is obscure and none of the 5 citations provides primary evidence in support. However, the paper correctly states that VTE complicates only 2-3 per 1000 (0.2-0.3%) of hospital medical admissions.
' The data on improved 'compliance' cannot be related to clinical outcomes because the authors did not take the opportunity to count the number of VTE or bleeding events before and after the nurse-led intervention. Thus the conclusion that the intervention resulted in 'significant improvements in patient care' is not warranted by the data.
' The increase from 38 to 54% is modest, and could be regarded as a failure of the intervention to produce the desired result, especially considering the resources expended.
' The paper reveals an undesirable level of commercial influence. The ANZWPG are sponsored by Sanofi-Aventis Australia Pty Ltd (this is not acknowledged). The 'Competing Interests' statement indicates that the study was carried out with financial support from Sanofi-aventis, and that one of the authors (DW) is an employee of that company. The reasons for including this person and his role in the work are not stated. The sponsorship probably included funding support for the nurse educator salaries. It is acknowledged that the two companies that provided 'editorial assistance' and undertook the statistical analysis respectively are also funded by Sanofi-aventis. The potential for prejudicial influence on drug use by commercial sponsorship (including the case of thromboprophylaxis(19)) is well recognised.(20,21) The conclusions of this paper, which have the effect of promoting medical thromboprophylaxis in spite of clear scientific objections, need to be seen in this context.
Conclusions
The ANZWPG are flawed. The eligibility criteria are expressed inadequately or are invalid. The Guidelines result in almost blanket use of prophylaxis in so-called 'high-risk' medical inpatients but the marginal risk is low. They are based on the implicit but dubious claim that prevention of asymptomatic disease is a good. The clinical gains are uncertain and do not outweigh the reported hazard of bleeding from anticoagulant drugs. In my view, the Gibbs et al paper promotes increased use of invalid guidelines. This is of obvious benefit to the paper's sponsor, but the clinical benefit is unclear.
References
1. The Australia and New Zealand Working Party on the Management and Prevention of Venous Thromboembolism. Prevention of Venous Thromboembolism 4th edition. Best Practice Guidelines for Australia and New Zealand. Health Education & Management Innovations Pty Ltd, 2007.
2. Gibbs H, Fletcher J, Blombury P, Collins R, Wheatley D. Venous thromboembolism prophylaxis guideline implementation is improved by nurse directed feedback and audit. Thrombosis J 2011 [to be added].
3. Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR, Colwell CW. Prevention of Venous Thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008; 133: 381S-453S.
4. Cohen AT, Davidson BL, Gallus AS, et al; ARTEMIS Investigators. Efficacy and safety of fondaparinux for the prevention of venous thromboembolism in older acute medical patients: randomized placebo controlled trial. BMJ 2006; 332: 325-327.
5. Samama MM, Cohen AT, Darmon JY, et al. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. Prophylaxis in Medical Patients with Enoxaparin study Group. N Engl J Med 1999; 341:793-800.
6. Leizorovicz A, Cohen AT, Turpie AG, et al. PREVENT Medical Thromboprophylaxis Study group. Randomised, placebo-controlled trial of dalteparin for the prevention of venous thromboembolism in acutely ill medical patients. Circulation 2004; 110: 874-879.
7. Hughes M, Lip GY. Risk factors for anticoagulation-related bleeding complications in patients with atrial fibrillation. Quart J Med 2007; 100:599.
8. Kadakiab MB, Desai NR, Alexander KP, Chen AY, Foody JM, Cannon CP, Wiviott SD, Scirica BM. Use of anticoagulant agents and risk of bleeding among patients admitted with myocardial infarction. J Am Coll Cardiol 2010; 3: 1166-77.
9. Mikkola KM, Patel SR, Parker JA, Goldhaber SZ. Increasing age is a major risk factor for haemorrhagic complications after pulmonary embolus thrombolysis. Am Heart J 1997; 134:69-72.
10. Millar JA. Selection of medical patients for prophylaxis of venous thromboembolism based on analysis of the benefit:hazard ratio. Internal Medicine Journal 39; 606-612.
11. Millar JA. Relative risk according to the proportion of a population deemed to be at high risk after risk factor analysis. NZ Med J 2011; 124: March 25. http://www.nzma.org.nz/journal/124-1331/4597/
12. House of Commons Select Committee on Health. Second Report, March 2005. http://www.publications.parliament.uk/pa/cm200405/cmselect/cmhealth/99/9906.htm#a4 (accessed 5 April 2011).
13. Millar JA, Lee GE, Ienco R. Prevalence of thromboembolism in medical inpatients (letter). Medical Journal of Australia 192; 724-726.
14. National Institute for Clinical Studies. The incidence and risk factors of venous thromboembolism in West Australian hospitals 1999 to 2001. Report commissioned from the School of Population Health, University of Western Australia. Canberra: NICS, 2005. http://www.nhmrc.gov.au/nics/material_resources/resources/incident_risk.htm (accessed 5 April 2011)
15. Gallagher M, Oliver K, Hurwitz M. Improving the use of venous thromboembolism prophylaxis in an Australian teaching hospital. Qual Saf Health Care 2009; 18: 408-12.
16. Dentali F, Douketis JD, Gianni M, Lim W, Crowther MA. Meta-analysis: anticoagulant prophylaxis to prevent symptomatic thromboembolism in hospitalised medical patients. Ann Intern Med 2007; 146:278-88.
17. Millar J A, Douglas JE, Bagley L, Densie, I. Improved criteria for medical thromboprophylaxis by application of risk factor weight: comparison with current Australasian guidelines. Unpublished.
18. Miras-Parra F, Navascues-Martinez E, Gomez-Outes A, et al. Utilisation and safety of bemiparin, a low molecularweight heparin, in medical patients: a prospective, uncontrolled cohort study. Clin Drug Invest 2005; 25:463'472.
19. Van Der Weyden MB. Doctors and the pharmaceutical industry: time for a national policy? Med J Aust 2009; 190: 407-8.
20. Sniderman AD, Furberg CD. Why guideline-making requires reform. J Am Med Soc 2009; 301: 429-31.
21. DeAngelis CD, Fontanarosa PB. Impugning the integrity of medical science. J Am Med Soc 2008; 299: 1833-35.
Australasian guidelines for medical thromboprophylaxis are flawed
19 December 2011
In this issue of the Journal, Gibbs, Fletcher and other Australian colleagues report that compliance with Australasian guidelines (1) (ANZWPG) for medical thromboprophylaxis (of which Gibbs and Fletcher are co-authors) can be increased from 38% to 54% by implementation of a 'multifaceted nurse facilitated educational program'.(2) The finding that a degree of behavioral change can be detected after an 'educational' program including prompts and feedback is not surprising, though the 'improvement' is surprisingly modest. The real question raised by this paper is whether the thromboprophylaxis policy being promoted is well-founded, and the extent of clinical benefit.
Problems with guidelines for medical thromboprophylaxis
1. Assessment of thrombotic risk
In the ANZWPG, which are based on the corresponding ACCP guidelines,(3) patients are deemed 'high-risk' and eligible for thromboprophylaxis (as unfractionated or low molecular weight heparin) if they have any one of 7 risk factors ('ischaemic stroke, history of VTE, active cancer, decompensated heart failure, acute on chronic lung disease, acute inflammatory disease, age > 60 y'). This immediately raises several issues. The descriptors are operationally vague. Ischaemic stroke was an exclusion factor in recent clinical trials,(4-6) that is, no evidence of benefit from prophylaxis with low-molecular weight heparin exists for stroke patients. The risk factors have varying statistical weights but are in effect assigned the same weight because any one of them creates eligibility. 'Age > 60 years' is an arbitrary variation on the actual risk factor, correctly described in the ACCP Guidelines as 'increasing age'.(3) Since most general medical patients are > 60, this risk factor ensures that a high percentage of patients (in the Gibbs paper, 82%) are deemed to be at 'high risk'. The effect is to assign almost blanket eligibility for thromboprophylaxis in medical patients. This is incorrect in principle because anticoagulants have a significant major bleeding risk. Also, increasing age is a risk factor for bleeding with anticoagulants, as well as for thrombosis.(7-9) I have shown that the ANZWPG provide a benefit:hazard ratio less than unity, ie that they may do more harm than good, and that only eligibility based on risk factors of high weighting will produce net benefit.(10)
One might also ask about the significance of a finding that over 80% of a population is at 'high risk' of a disease. Theoretical considerations show that the maximum possible relative risk in a sub-population at high risk is the reciprocal of the proportion it represents.(11) Thus under the Australasian guidelines, the maximum relative risk in eligible patients is 1 ÷ 0.82, or 1.22, ie the added risk is 22%, on average, using the data from Gibbs et al.
2. Incidence of secondary thromboembolism in medical patients
The incidence of new thrombotic episodes in medical inpatients is given as 17%, 10-20% and 15% in the ANZWPG, the ACCP guidelines and the House of Commons Select Committee Report (12) (HCSCR) , respectively. The first two guidelines acknowledge that the figures include 'sub-clinical DVT' but neither state the proportion of events that are asymptomatic. The HCSCR talks of 'detectable' thrombosis but provides no further clarification, raising the possibility that MPs may have been misled on this point. Incidence data based on hospital morbidity coding statistics puts the incidence of symptomatic secondary thrombosis at 0.4-0.5%.(13,14) or less.(15) Thus over 95% of asymptomatic thrombotic events in medical patients do not progress to symptomatic disease. Several important questions arise from these facts:
1. Should prophylaxis aim to prevent asymptomatic disease? The ACCP guidelines state (reasonably) that the primary purpose of thromboprophylaxis is to prevent pulmonary embolus, but no guidelines discuss whether preventing asymptomatic disease is a meaningful aim. To this author it seems a non-sequitor.
2. Is prophylaxis of asymptomatic disease justified because it progresses to symptomatic disease in some patients? This beguiling argument may be valid if progression from asymptomatic to symptomatic disease were common, but for thromboprophylaxis the transition is a rarity. If the object is to prevent symptomatic disease, then the relevant determinants are the incidence of the symptomatic condition and the relative risk reduction of the proposed prophylaxis, which together determine the clinical gain, measured as the number needed to treat to prevent one event (NNT).
3. What is the source of the data? The use of data for asymptomatic disease derives from the randomized controlled thromboprophylaxis trials(4-6) in which detection of thrombosis was accomplished using techniques that detect asymptomatic thrombi. The reason for this was to ensure objective measurement and, because asymptomatic events are much more common, to decrease the size and hence the cost of the clinical trials. However, the result is that the trials measured the effect of thromboprophylaxis on proxy end-points.
4. What is the outcome? Using data based on asymptomatic events exaggerates the size of the clinical problem (by about 30-fold) and invalidates the outcomes of secondary analyses such as NNT and cost-effectiveness. For example, with a relative risk reduction from trials(4-6) of about 50%, a baseline incidence of 17% gives a NNT of 12 (to prevent each asymptomatic VTE event), or 400 (to prevent each symptomatic event), with proportionate differences in cost-effectiveness, respectively. A meta analysis supports the latter figure for pulmonary embolus.(16)
Anyone who works in the medical wards of any hospital knows that the given incidence figures when applied to clinical disease are simply nonsensical, and this has been confirmed in practice.(17,18)
Problems with the paper
In addition to problems with the Guidelines, the Gibbs et al paper contains various shortcomings:
' The authors claim (from autopsy studies) that 50-70% of symptomatic events, and also that 70-80% of fatal PE occur in non-surgical patients. The relevance of these data to the clinical problem is obscure and none of the 5 citations provides primary evidence in support. However, the paper correctly states that VTE complicates only 2-3 per 1000 (0.2-0.3%) of hospital medical admissions.
' The data on improved 'compliance' cannot be related to clinical outcomes because the authors did not take the opportunity to count the number of VTE or bleeding events before and after the nurse-led intervention. Thus the conclusion that the intervention resulted in 'significant improvements in patient care' is not warranted by the data.
' The increase from 38 to 54% is modest, and could be regarded as a failure of the intervention to produce the desired result, especially considering the resources expended.
' The paper reveals an undesirable level of commercial influence. The ANZWPG are sponsored by Sanofi-Aventis Australia Pty Ltd (this is not acknowledged). The 'Competing Interests' statement indicates that the study was carried out with financial support from Sanofi-aventis, and that one of the authors (DW) is an employee of that company. The reasons for including this person and his role in the work are not stated. The sponsorship probably included funding support for the nurse educator salaries. It is acknowledged that the two companies that provided 'editorial assistance' and undertook the statistical analysis respectively are also funded by Sanofi-aventis. The potential for prejudicial influence on drug use by commercial sponsorship (including the case of thromboprophylaxis(19)) is well recognised.(20,21) The conclusions of this paper, which have the effect of promoting medical thromboprophylaxis in spite of clear scientific objections, need to be seen in this context.
Conclusions
The ANZWPG are flawed. The eligibility criteria are expressed inadequately or are invalid. The Guidelines result in almost blanket use of prophylaxis in so-called 'high-risk' medical inpatients but the marginal risk is low. They are based on the implicit but dubious claim that prevention of asymptomatic disease is a good. The clinical gains are uncertain and do not outweigh the reported hazard of bleeding from anticoagulant drugs. In my view, the Gibbs et al paper promotes increased use of invalid guidelines. This is of obvious benefit to the paper's sponsor, but the clinical benefit is unclear.
References
1. The Australia and New Zealand Working Party on the Management and Prevention of Venous Thromboembolism. Prevention of Venous Thromboembolism 4th edition. Best Practice Guidelines for Australia and New Zealand. Health Education & Management Innovations Pty Ltd, 2007.
2. Gibbs H, Fletcher J, Blombury P, Collins R, Wheatley D. Venous thromboembolism prophylaxis guideline implementation is improved by nurse directed feedback and audit. Thrombosis J 2011 [to be added].
3. Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR, Colwell CW. Prevention of Venous Thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008; 133: 381S-453S.
4. Cohen AT, Davidson BL, Gallus AS, et al; ARTEMIS Investigators. Efficacy and safety of fondaparinux for the prevention of venous thromboembolism in older acute medical patients: randomized placebo controlled trial. BMJ 2006; 332: 325-327.
5. Samama MM, Cohen AT, Darmon JY, et al. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. Prophylaxis in Medical Patients with Enoxaparin study Group. N Engl J Med 1999; 341:793-800.
6. Leizorovicz A, Cohen AT, Turpie AG, et al. PREVENT Medical Thromboprophylaxis Study group. Randomised, placebo-controlled trial of dalteparin for the prevention of venous thromboembolism in acutely ill medical patients. Circulation 2004; 110: 874-879.
7. Hughes M, Lip GY. Risk factors for anticoagulation-related bleeding complications in patients with atrial fibrillation. Quart J Med 2007; 100:599.
8. Kadakiab MB, Desai NR, Alexander KP, Chen AY, Foody JM, Cannon CP, Wiviott SD, Scirica BM. Use of anticoagulant agents and risk of bleeding among patients admitted with myocardial infarction. J Am Coll Cardiol 2010; 3: 1166-77.
9. Mikkola KM, Patel SR, Parker JA, Goldhaber SZ. Increasing age is a major risk factor for haemorrhagic complications after pulmonary embolus thrombolysis. Am Heart J 1997; 134:69-72.
10. Millar JA. Selection of medical patients for prophylaxis of venous thromboembolism based on analysis of the benefit:hazard ratio. Internal Medicine Journal 39; 606-612.
11. Millar JA. Relative risk according to the proportion of a population deemed to be at high risk after risk factor analysis. NZ Med J 2011; 124: March 25. http://www.nzma.org.nz/journal/124-1331/4597/
12. House of Commons Select Committee on Health. Second Report, March 2005. http://www.publications.parliament.uk/pa/cm200405/cmselect/cmhealth/99/9906.htm#a4 (accessed 5 April 2011).
13. Millar JA, Lee GE, Ienco R. Prevalence of thromboembolism in medical inpatients (letter). Medical Journal of Australia 192; 724-726.
14. National Institute for Clinical Studies. The incidence and risk factors of venous thromboembolism in West Australian hospitals 1999 to 2001. Report commissioned from the School of Population Health, University of Western Australia. Canberra: NICS, 2005. http://www.nhmrc.gov.au/nics/material_resources/resources/incident_risk.htm (accessed 5 April 2011)
15. Gallagher M, Oliver K, Hurwitz M. Improving the use of venous thromboembolism prophylaxis in an Australian teaching hospital. Qual Saf Health Care 2009; 18: 408-12.
16. Dentali F, Douketis JD, Gianni M, Lim W, Crowther MA. Meta-analysis: anticoagulant prophylaxis to prevent symptomatic thromboembolism in hospitalised medical patients. Ann Intern Med 2007; 146:278-88.
17. Millar J A, Douglas JE, Bagley L, Densie, I. Improved criteria for medical thromboprophylaxis by application of risk factor weight: comparison with current Australasian guidelines. Unpublished.
18. Miras-Parra F, Navascues-Martinez E, Gomez-Outes A, et al. Utilisation and safety of bemiparin, a low molecularweight heparin, in medical patients: a prospective, uncontrolled cohort study. Clin Drug Invest 2005; 25:463'472.
19. Van Der Weyden MB. Doctors and the pharmaceutical industry: time for a national policy? Med J Aust 2009; 190: 407-8.
20. Sniderman AD, Furberg CD. Why guideline-making requires reform. J Am Med Soc 2009; 301: 429-31.
21. DeAngelis CD, Fontanarosa PB. Impugning the integrity of medical science. J Am Med Soc 2008; 299: 1833-35.
Competing interests
None