Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
ABSOLUTE AND RELATIVE RANKINGS IN COST-EFFECTIVENESS COMPARISONS
- Andrea Messori, Sabrina Trippoli, Monica Vaiani, Franca Vacca, Filippo Pelagotti, and Benedetta Santarlasci (2 December 2003)
-
Incremental NNT also needed
- John Sharvill (4 December 2003)
-
Should the individual patient follow the author's recommendations?
- Jeffrey Mann (4 December 2003)
-
Author's reply
- Rebecca N Warburton (16 January 2004)
|
|
|||
|
Andrea Messori, Coordinator Laboratorio SIFO di Farmacoeconomia, Careggi University Hospital, 50134 Firenze, Italy, Sabrina Trippoli, Monica Vaiani, Franca Vacca, Filippo Pelagotti, and Benedetta Santarlasci
Send response to journal:
|
To stimulate the debate on cost-effectiveness raised by the editorial of Warburton (1), we have picked up some "pharmacoeconomic" statements that have appeared in different contributions published recently in the BMJ or in the electronic BMJ. These different statements are the following: a) "The United Kingdom adopted guidelines on the use of statins for primary and secondary prevention of coronary heart disease that would have required either a 20% increase in drug costs or termination of more beneficial interventions"(1) b) "Most of the benefits of prevention can be achieved with aspirin and anti-hypertensive treatment at a fraction of the cost of simvastatin"(2) c) "The price of simvastatin would have to fall by 70%.... to be of similar cost effectiveness to antihypertensive treatment"(2) d) "Spending on alendronate seems to be quite inappropriate for [the Italian] national health system at least at the current price level for this agent. This amount of health of about 2,000 years of life has in fact been bought at a three-fold price than the average. In our previous studies, this index was favourable for statins and highly unfavourable for coxibs"(3) e) "Comparing the expected health gain of 65,400 years of life with the real health gain of more than 100,000 years of life shows that spending on statins is likely to be appropriate for [the Italian] national health system: this amount of health has in fact been bought at a price lower than the average"(4) f) "Comparing the expected health gain of 24,100 years of life with the real health gain of either 2,601 years or 900 years shows that spending on coxibs seems to be highly inappropriate for [the Italian] national health system at least at the current price level for these agents"(5) g) "A large number of studies indicate that statins are effective and cost-effective for secondary prevention of cardiovascular events. However, their role in primary prevention is more controversial"(6) h) "A joint interpretation of [these] arguments ....shows that spending by prescribing statins to both men and women can be acceptable on a cost-effectiveness basis mainly because there seems to be a selective substantial benefit in men. Can this cost-effectiveness ratio be further improved by prescribing statins to men only? This question is difficult to answer, but its economic implications are macroscopic."(7) In the first place, these statements indicate that statins have a better pharmacoeconomic profile for secondary prevention than for primary prevention and in men than in women. In addition, these statements show that, with regard to cost-effectiveness ranking, statins are worse than each of the following: a) aspirin alone; b) the combination of aspirin with a beta-blocker plus a diuretic; c) the combination of aspirin with a beta-blocker plus a diuretic plus enalapril. On the other hand, statins are better than each of the following: a) alendronate for preventing fractures in osteoporosis; b) coxibs for treating osteoarthritis or rheumatoid arthritis. The conclusion is the following. If the horizon of these analyses is restricted to a few agents indicated for a single disease condition (e.g. preventing cardiovascular events using aspirin alone vs. the combination of aspirin with a beta-blocker plus a diuretic vs. the combination of aspirin with a beta-blocker plus a diuretic plus enalapril) and if a disease-specific index (i.e. the cost to avoid a cardiovascular event) is used to construct the ranking, little information is provided by these comparisons. No doubt that one understands that statins are worse than aspirin and/or anti-hypertensive agents in relative terms. But no information is generated in absolute terms: in fact, even though statins are found to be worse than the two or three treatments mentioned above, they might well be better or much better than hundreds of other drugs reimbursed by national health systems. Or vice-versa. One solution to this problem is to avoid the use of disease-specific pharmacoeconomic indexes (e.g. the cost per event averted) and to give preference to "absolute" pharmacoeconomic indexes that allow comparisons between different drug classes (such as the cost per life year gained or the cost per quality-adjusted life year gained). In addition, this methodological preference for absolute indexes should be accompanied by a widespeard application of these analyses to as many treatments as possible throughout the entire drug formulary. This would allow us to make much more informative comparisons of (relative) cost-effectiveness across a large number of different drug treatments and would contribute to improving the allocation (or redistribution) of economic resources within (national or local) drug formularies. REFERENCES 1) Warburton RN. What do we gain from the sixth coronary heart disease drug? BMJ 2003;327:1237. 2) Marshall T. Coronary heart disease prevention: insights from modelling incremental cost effectiveness. BMJ 2003;327:1264 3) Messori A et al. Economic appropriateness of the expenditure for alendronate: cost-effectiveness analysis of national prescription data in Italy. http://bmj.com/cgi/eletters/327/7406/89#40333 4) Messori A et al. Spending on statins. http://bmj.com/cgi/eletters/327/7420/933-b#38400 5) Messori A et al. Economic appropriateness of the expenditure for coxibs: cost-effectiveness analysis of national prescription data in Italy. http://bmj.com/cgi/eletters/327/7420/933-b#39700 6) Messori A, Santarlasci B, Trippoli S, Vaiani M. "Polypill" to fight cardiovascular disease: Cost effectiveness of statins for primary prevention of cardiovascular events is questionable. BMJ 2003; 327: 808-9. 7) Vacca F et al. Cost effectiveness of statins in men vs women. http://bmj.com/cgi/eletters/327/7420/933-b#39209 Competing interests: None declared |
|||
|
|
|||
|
John Sharvill, Family Physician Deal England
Send response to journal:
|
Dose anybody publish data so that I can advise my patients the extra marginal benefit for adding a sixth drug to their regimes? With eveidence of compliance appearing to show on average of 50% at 2 years it would be good to see clear un-sponsored data to pass on to patients that adding a further pill is very likely or only marginally likely to benefit them so they can make an informed choice. Competing interests: None declared |
|||
|
|
|||
|
Jeffrey Mann, Retired physician Salt Lake City, UT 84103
Send response to journal:
|
The author states:- "Marshall's work shows that £100 000 can either prevent one coronary event or many more. It will prevent only 1.2 events if used (in accordance with current UK guidelines10) to prescribe simvastatin (after aspirin and antihypertensive agents) in patients at a five year risk of a coronary event of 15%. If used for aspirin the same £100 000 could prevent either 12.7 events (in patients at 5% risk) or 28.6 events (in patients at 10% risk)." It is true that aspirin may be more cost-effective from a societal perspective and that it could prevent 12.7 events in patients at a 5% risk of a coronary event over 5 years. However, why should the individual patient participate in this project if his likely individual benefit is so small? An individual who is in good health, with a 5% coronary risk over 5 years, can readily discern that his "average" yearly risk over the next 5 years of not having a coronary event is 99% and that daily aspirin therapy would change that figure to 99.3%. Why should a rational person subject himself to the inconvenience of taking daily aspirin, and the continuous ongoing risk of suffering a GIT bleed or henmorrhagic stroke, if the absolute risk of a coronary event is so small and the absolute degree of benefit (0.3% per year) so insubstantial? Jeff Mann. Competing interests: None declared |
|||
|
|
|||
|
Rebecca N Warburton, Assistant Professor, Michael Smith Foundation for Health Research Scholar School of Public Administration, University of Victoria
Send response to journal:
|
Rapid responses to my editorial have raised several interesting issues, without undermining my key conclusions. The same appears to be true of the reader responses to the paper by Tom Marshall (which was the subject of my editorial). Dr. Marshall, in his response to reader comments, noted that the purpose of economic evaluation studies is to improve decision-making. I agree, and emphasize that there is a tradeoff, in economic evaluation studies, between complexity (which may allow more exact and more sophisticated results) and clarity for readers (which is enhanced by making reasonable simplifying assumptions and focusing on essentials). There is also a tradeoff between complexity and costs; the extra cost of more comprehensive studies is only warranted if likely to result in better patient care. <P> Messori and Trippoli criticize Marshall's analysis because it is disease-specific, and note (correctly) that statins, although less cost- effective than other coronary disease preventives, might well be more cost -effective than hundreds of other drugs reimbursed by national health systems. They recommend analyses of most or all drugs in the national formulary, using quality-adjusted life years to measure effects (rather than events prevented as in the Marshall paper). While such studies would certainly allow broader and more complete comparisons, it seems unlikely that the costs of such a research approach would return sufficient benefits in terms of better decisions for patients. It is reasonable to focus analysis on costly, new, drug therapies for common conditions, whose use is recommended in national guidelines, ahead of examination of all drugs in the national formulary. Marshall's analysis is quite sufficient for its purpose of improving the efficiency of current preventive cardiology services. Sharvill asks whether there are published tables showing the added benefit (absolute risk reduction) of the sixth drug for specific patients. Indeed, this is information that clinicians and patients probably want, although not the focus of Marshall’s analysis. I do wish that Dr. Sharvill had been part of the editorial process and able to suggest the addition of this information to the paper, as I am unaware of published tables of this kind! It is possible (though not straightforward) to calculate this risk reduction using the information in Marshall’s Table 2. The calculation uses Column A (incremental relative risk reduction for each drug, which is the same for all patients) plus the adverse event rate and baseline risk. Essentially we must calculate the net (taking adverse events into account) cumulative reduction in baseline risk for each drug, then apply this reduction to a specific patient’s baseline coronary event risk (from risk tables). It is most informative to calculate the combined risk of either a coronary event or a serious adverse event; though in this case (based on available evidence) serious adverse events are modeled only for aspirin. Consistent with the Marshall paper, serious adverse events have been equated with a coronary event; essentially this makes the not unreasonable assumption that patients don’t thank their physicians for preventing a heart attack if instead they have a serious gastric bleed! Aspirin reduces coronary event risk to 72% of its baseline (multiplies by .72) but carries a 0.3% absolute risk of serious adverse events (adds 0.3% adverse event risk); Bendrofluazide and atenolol further reduce risk by 17% (multiply by .83), so that aspirin combined with initial antihypertensive treatment reduces risk to 62% of baseline (multiply by .72 x .83 = .60) and so on; Enalapril (added to previous treatments) reduces risk to 48% of baseline; Simvastatin to 33% of baseline; and Clopidogrel to 29% of baseline. The bottom line? For a patient with baseline 5-year risk of 10%, adding the sixth drug (Clopidogrel ) reduces 5-year risk from 3.6% to 3.2%. The 5-year number-needed-to-treat is 250! For a patient with baseline 5-year risk of 30%, Clopidogrel reduces 5-year risk from 10.3% to 9.1%: 5-year NNT = 83. Given the risks and inconvenience inherent in taking any drug, even if patients have no concerns about drug costs, these gains seem small. This leads nicely to the comment by Mann, who took my calculation of the number of coronary events that could be prevented by £100,000 to be a recommendation that these treatments be adopted; though actually no such recommendation was intended. Mann notes that an individual patient at 5% 5-year risk might well decide not to take aspirin, given the small absolute benefits, inconvenience, and risk of complications, and I agree. This patient would see 5-year coronary event risk reduced from 5% to 3.6%, and would also experience 0.3% risk of a serious adverse event; net 5-year risk would drop from 5% to 3.9%; 5-year NNT = 91. By comparison, the patient (treated in accordance with UK guidelines) at 15% 5-year risk, receiving Simvastatin as the fourth drug, would see 5-year risk drop from 7.6% to 5.3%, NNT = 44. The problem, in the latter case, is that the cost of achieving this reduction is too high, relative to other potential uses of health care funds. Both the Marshall paper and my editorial focused on the system-wide perspective, not on treatment recommendations for specific patients; but obviously both kinds of information are useful in coronary disease preventive programmes. Thanks to BMJ readers for their comments! Competing interests: None declared |
|||