Comparative Effectiveness Research (CER), as it name suggests, compares how well different medicines treat a given disease. Politicians claim that using CER findings can help improve quality and decrease cost. If one treatment produces better health outcomes on average than another and also costs less, we should always make people use that treatment, right?
Not according to a working paper by Basu and Philipson (2010). Let us assume that health outcomes for Drug A are better than the health outcomes for Drug B. If the results from this research were released, the public’s demand for Drug A would increase and the demand for Drug B would decrease. However, this may not save money. The demand for Drug B will decrease since fewer people want to buy it; this will reduce expenditures. As the demand for Drug A increases, however, the price and quantity purchased will increase. Thus, the net effect on spending is indeterminate.
In addition, if insurance companies or the government decided to subsidize or cover the entire cost of treatment using Drug A, the demand will increase even more.
Basu and Philipson, however, assume that the marginal cost to produce a drug increases as the quantity rises (i.e., the supply curve is upward sloping). However, because there are economies of scale in the production of pharmaceuticals, the price of Drug A could actually decrease (increasing returns to scale) or stay the same (constant returns to scale) as demand increased.
The key assumption in the above analysis is that it assumes that all people with a given disease respond in a homogeneous way to Drug A. If two-thirds of people have better health outcomes when treated with Drug A and one-third have better health outcomes when treated with Drug B, then it may be suboptimal to cover Drug A, but not Drug B.
To prove this, Basu and Philipson look at the Clinical Antipsychotic Trials of Intervention Effectiveness project (CATIE). They find that “if Medicaid would have eliminated coverage for the least cost-effective treatments of the CATIE trial then under homogeneous effects, it would save about 90% of the $1.3B Medicaid class sales annually in non-elderly adult patient with schizophrenia. However, taking into account the observed heterogeneity in treatment effects, it would incur a loss of health valued annually at about 98% of class spending and thus a net loss of about 8% of annual class spending.”
However, one of their key assumptions is that not covering the less effective medicines means that no patients will take the uncovered drug. This may be a fair assumption for the Medicaid population, but not for the population at large. There is a big distinction that needs to be made between not covering a drug (but allowing for the purchase to purchase it out of of their own pocket) and prohibiting the drug entirely. Basu and Philipson are looking at the most extreme case where not covering the drug means, de facto, that the drug will not be taken, but this need not be the case.
What is important to take from this research, however, is that the drug that is most effective on average may not be the best drug for everyone. One must take into account heterogeneous treatment effects when designing any insurance benefit plan.
- Anirban Basu and Tomas J Philipson (2010) “The Impact of Comparative Effectiveness Research on Health and Health Care Spending,” NBER WP #15633.