Prevention and Cost

Health Affairs, Jan-Feb 2009, has several articles on preventing chronic illness, examining the value and cost-effectiveness of various prevention strategies. In “Preventing chronic disease: an important investment but don’t count on cost savings,”[1] Louise Russell uses the techniques of cost-effectiveness analysis (CEA) to look at a number of prevention and screening strategies and looks at the cost per QALY (quality-adjusted life year) compared to treating the disease. What becomes clear is that the more that the prevention strategy is targeted to a high-risk group, the more cost-effective it is, that is, the less the cost per additional QALY.

“For low-risk men ages 45-54 with LDL cholesterol between 160 and 189 mg/dl, the additional cost of statins [moderate cholesterol-lowering agents] in 1997 was $270,000 per healthy year, or about $400,000 today. For smokers ages 45-54 with LDL above 190 mg/dl, high blood pressure, and poor HDL [good] cholesterol, cost per healthy year is much less: $57,000 in 1997, or $85,000 today. For men with established heart disease, statins are highly cost-effective: under $10,000 per healthy year in 1997, or $15,000 today.”

The primary reason for the difference, of course, is the large number of people who need to be treated to prevent one person from dying; the higher risk the population considered, the lower that ratio. The other variable, obviously, is the frequency of screening. Doing a test every year costs twice as much as doing it every two years; unless the condition occurs and progresses so quickly that twice as much disease is uncovered and prevented (or treated successfully) the CEA will be worse. A good example is Pap smear tests for cervical cancer, arguably the best (most effective in preventing morbidity and mortality) screening test we have. “…screening every 3 to 5 years is cost-effective compared with no screening, but costs rise rapidly with more frequent screening. Compared to screening every 2 years, annual screening cost $1 million per healthy year in 1995, or more than $3 million today.” There are certainly women who are at higher risk of cervical cancer, and should be screened annually, but this illustrates that, even in the case of the best screening test, going from the benefit to some to screening for all is not always – or even usually – a good thing.

There are other issues besides cost when people who have a low likelihood of a disease are screened. They may be “false positives” – have a positive test but not have the disease – and undergo more tests that could be unpleasant, expensive, risky, or all three – before they find out they are really disease free. This is a difficult concept, even for medical students, but the reason is that if a disease is very uncommon – as most are – then even a good test may find more false than true positives.[2]

In the same issue, Ron Goetzl talks about different conceptions of prevention. These vary greatly, with the public (and legislators) often meaning very different things than academics or health professionals; moreover, health professionals are often afflicted with biases toward prevention of diseases in their “area”; a specialist will see a much higher percent of people with a very advanced or serious form of a disease than a generalist (because they have often been referred) and thus may have a perceptual overestimate of the commonness (prevalence) of that disease. Goetzl describes a variety of types of prevention: screening (discussed above), immunizations, and various public policies that may limit risk: “…raising taxes on cigarettes, limiting employees’ exposure to toxic substances, mandating seat-belt use, screening for cancer, restricting alcohol scales to minors, building bicycle paths, and eliminating sales of sugary beverages in schools”.[3] He notes that while these policy interventions, and immunizations, may be prevention, screenings are actually identification of disease in its early stages when treatment can prevent progression. He then talks about health promotion, a kind of primary prevention that rarely or ever requires medical intervention, “…managing one’s weight, being physically active, eating a healthy diet, not smoking, drinking moderate amounts of alcohol[4], getting enough rest, surrounding oneself with family and friends, driving safely, managing stress, and, in general, living what most would agree is a healthy lifestyle.” Well, anytime one says “most would agree” one has abandoned the world of evidence for belief systems, but you get the point. Pretty much all these things are cost-effective and work very well. Goetzl also addresses secondary prevention – treatment of early disease to prevent worse disease, such as treating high cholesterol, high blood pressure, or stopping smoking. He notes that “Although these interventions rarely save money, certain ones offer high value in terms of adding QALYs at a relatively low cost”.

If cost-effectiveness analysis can help us to understand what the cost per benefit (usually $ per QALY) is, it is also important to understand that I may be willing to spend a lot more on ME per QALY than on you, or particularly someone I don’t know;. Thus the importance of having standards, not individuals choosing enormously expensive screening interventions for themselves (which can often, as noted above, backfire by finding “disease” that is not present, not treatable or less risky than the diagnostic and treatment interventions.

Goetzl’s most important point is to counter the argument that prevention is not worthwhile because …”people have to die of something – all prevention does is postpone the time of eventual death and introduce new and more costly diseases that are the consequences of aging. As Woolf points out, the aim of prevention is not to replace one disease with another, but to compress the time one is sick or disabled before one’s ultimate demise.”

Much disease can be prevented, much suffering can be averted. Some of the preventions are behavioral changes such diet and exercise and not smoking, and real progress has been made in this arena. Others require social policies – seatbelt laws, safer cars, even safer guns. Yet others use social policies to facilitate individual behavior change – eliminating sweets from schools, making public places smoke free, prominently displaying calorie counts on fast food items. And, yes, some are medical: immunizations and some screening tests and some treatments for early disease. But people cannot expect that everything can be prevented by medicine, or found early, or successfully treated even when treatment is offered – and even when it is “successful” (that is, diseases sometimes get better or do not progress regardless of treatment, but the benefit is attributed to the treatment if it has been instituted (prostate cancer, for example).

Just because something makes sense doesn’t make it true; it is a research question, not an answer. If something doesn’t make sense, you probably won’t study it, but if it does you need to do the research to find out if it is true. People – their biology and their behavior – is too complex for something to be true just because it makes sense in one dimension. And certainly not because you just “want” it to be.



[1] Russell, LB, “Preventing chronic disease: an important investment but don’t count on cost savings”, Health Affairs, Jan/Feb 2009;28(1):42-45.
[2] Consider a disease which is very common; say 1 in 1000 people have it. And let’s say the test for it correctly identifies 9 of 10 people who have it and misses 1 (this is known as 90% sensitivity), and is negative in 9 of 10 people who DON’T have it (90% specificity). If we test the 1000 people, the one person who has it will probably have a positive test (true positive). But 10% of the 999 who don’t have it – 99.9 people – will have a false-positive test. So there are 99.9 false positives for every true positive; a person with a positive test is 100 times as likely to NOT have the disease as to have it (1% positive predictive value). However, if we take a population where the risk is 10 times as great, 1 in 100, then of the 1000 people there will be 9 (90% of 10) true positives and 90 (90% of 900) false positives, so the positive predictive value is up to 10% (1 in 10 positives is a true positive).
[3] Goetzl, RZ, “Do prevention or treatment services save money? The wrong debate”, Health Affairs, Jan/Feb 2009;28(1):37-41
[4] “ drinking moderate amounts of alcohol…” – his phrase, not mine. To suggest “not smoking” but “drinking moderate amounts of alcohol” is the most healthful demonstrates a clear bias and is not evidence based. From a health perspective, “not drinking alcohol” is the best choice.