The Costs of Making Practice More Cost-effective

To the Editor: Dr Mason and colleagues¹ derived an economic model, based on England's National Healthcare System (NHS), for determining when it might be cost-effective to try to change physician behavior. Although health care is not nationalized in the United States, some health care services for eligible elderly and disabled patients are federally funded through Medicare. Three differences between the NHS and Medicare must be recognized before the model of Mason et al¹ can be applied to Medicare policy.

First, under Medicare, physicians incur costs when changing their practices. In the NHS, it may be reasonable to assume no costs associated with practice change; physicians are salaried and apparently have an allocation of time dedicated to such pursuits. For US physicians who are obtaining Medicare reimbursement for services, time spent away from patient care represents lost income. Although the low Medicare reimbursement rates may mitigate the effect somewhat, this financial disincentive may decrease the intervention's effectiveness. The combination of higher costs and lower effectiveness increases the loading costs substantially.

Second, medications are not liabilities in the Medicare system. In the model of Mason et al,¹ the primary direct costs to the NHS are medication costs; downstream reductions in health care service utilization associated with their use are excluded.² Medicare does not pay for medications and, therefore, incurs no direct pharmacy costs. Patients may rationally choose to purchase more expensive medications if they are also more effective or have fewer adverse effects. To the extent that patients' use of health care services decrease because of privately funded medication use—fewer visits because of fewer adverse effects or fewer hospitalizations because of more effective treatment—Medicare obtains a windfall.

Finally, in the model of Mason et al, benefits that accrue to NHS would be liabilities for Medicare. Additional life-years gained represent an extension of Medicare's liability for health care services funding for patients and US medical costs are increasing more rapidly than the 5% discount rate used in the model.² Therefore, in the United States, additional life-years do not represent benefits, but more, and more costly, health care services payments.

These considerations do not diminish the model's utility. On the contrary, they offer insight into why cost-effectiveness analyses may not be very informative in the United States. Within the US health care system, physicians' self-interest is frequently financially, not socially, driven. Only by addressing those incentives can leaders of health care systems motivate physicians to change.