Mar 2006

Consumer-Directed Health Plans

Joseph P. Newhouse, PhD
Virtual Mentor. 2006;8(3):174-176. doi: 10.1001/virtualmentor.2006.8.3.oped2-0603.


Consumer-directed health plans are certainly attracting attention, but what effects will they actually have on health and health spending if they catch on? Forecasts are often wrong, but in this case we have strong evidence about the probable effect of a key component of such plans—namely the large deductible that is partially or perhaps even wholly funded by an employer.

This strong evidence comes from the RAND Health Insurance Experiment, for which I served as principal investigator, and, as I will explain, it accounts for the provision in many such plans whereby any unused dollars in the account can be preserved over time, accumulate tax free, and perhaps even be owned by the individual [1]. This experiment, conducted 30 years ago, randomly assigned families to plans with varying cost-sharing arrangements. Some families received all their medical care for free; others were enrolled in a plan that required a large deductible. In today's dollars the deductible in the RAND Experiment was larger than that in most contemporary consumer-directed plans. The experiment's deductible was reduced for lower-income families, whereas this is not a feature in most of today's consumer-directed plans. The study subjects were all under 65 years of age and participated in the experiment for either 3 or 5 years.

In addition to the large deductible, the experiment had another feature that mimicked employer-funded accounts. We wanted to prevent sick individuals from declining to participate if offered the deductible plan but accepting if offered the plan with free care, which would skew the results to show an excessively large effect of the deductible. To avoid this bias, the participants with deductible plans received a "hold-harmless" side payment, independent of any medical spending, so that they would never have less to spend on nonmedical goods and services than if they remained on their prior insurance. Indeed, they would almost always be better off by enrolling in the experiment. Instead of coming as a deposit in an employer-funded account, however, the hold-harmless payments came as monthly checks to the families. These payments were found to have negligible effect on use of medical services.

The results of the experiment showed that families with the large deductible spent about 30 percent less on medical care in each year than families with free care. Each person in the deductible group made about 2 fewer physician visits annually and had fewer hospitalizations. For the great majority of participants there were no measurable adverse health consequences from the reduction in use, but hypertension was less well controlled in those with high blood pressure. This poorer control of hypertension was estimated to raise mortality about 10 percent in the affected group.

Based on these results, it is reasonable to expect the deductible feature of consumer-directed plans to reduce spending substantially. The reduction, however, is likely to be a one-time phenomenon; although there is no evidence of this in the experiment, it is not likely in my view that the steady rate of increase in annual medical spending will be much affected. Even so, the one-off reduction in spending of the magnitude seen in the experiment is notable.

The advocates of consumer-directed plans believe that the large deductibles will induce individuals to take better care of themselves, but there was no evidence of this in the experiment. Smoking rates and weight, for example, were unaffected by the large deductible.

What about the negative health effects? Disease management, which was not part of the experiment, should be part of such plans; better management of chronic disease could mitigate such negative effects as poorly controlled hypertension. We know from studies of savings behavior that unaided consumers do not always make wise decisions, especially when those decisions entail present sacrifices for future gain, as, for example, complying with a prescribed antihypertensive regimen that causes side effects. The ability to improve compliance affords an opportunity for disease management.

Advocates of consumer-directed plans also claim that prices for medical services will become more competitive, with corresponding benefits for patients. But in the experiment those on the large deductible plan did not choose physicians whose services cost less per relative value unit than did those in the fully subsidized care arm of the experiment. Some consumer-directed plans are making an effort to inform patients about prices or at least to tier physicians based on price. Whether the experiment's results would have differed had patients had better information about cost differences among physicians is unknowable.

A question looming over all the prior results is whether the findings of the RAND experiment would hold if it were repeated today, some 30 years later. Doubtless there would be differences—if for no other reason than that medical technology has advanced in the past 3 decades—but would these differences be material? Damon Runyon once said that, in a fight between a big guy and a little guy, the big guy does not always win, but that's the way to bet. In a similar vein, I would be willing to bet that if the experiment were rerun today, the large deductible would continue to cause a major reduction in use with minimal adverse health consequences among much of the employed population. And those consequences could be mitigated by smarter cost-sharing, such as exempting chronic maintenance drugs from the deductible.


  1. Newhouse JP, Insurance Experiment Group. Free for All? Lessons from the RAND Health Insurance Experiment. Cambridge, Mass: Harvard University Press; 1993.


Virtual Mentor. 2006;8(3):174-176.



The viewpoints expressed in this article are those of the author(s) and do not necessarily reflect the views and policies of the AMA.