Testing Medicine

Most people are surprised to learn that rigorous application of formal rules of evidence to evaluate medical research and decide on the best treatments is a recent phenomenon. They might be just as surprised to learn that health policy is not yet subject to the same approach.

So-called “evidence-based medicine” entails a hierarchy of empirical proof that ranks medical studies according to their quality. Physiological research on animals, or of human responses in laboratory settings, rank low on the hierarchy. Observational studies that compare outcomes for patients who receive particular treatments and control subject who do not rank higher, but may still be misleading.

Convincing studies of drugs and surgical procedures usually come only from randomized trials, in which patients receive treatment or don’t according to a process analogous to a coin flip. Well-conducted randomized trials incorporate additional safeguards against bias, including use of placebo medication that allows investigators to blind patients and caregivers to whether patients are, or are not, receiving active treatment.

Doctors have made many errors that they could have avoided had these principles been applied. For example, for a decade leading doctors advocated hormone replacement therapy to lower cardiovascular risk in post-menopausal women. These recommendations, based on physiological and observational studies, proved profoundly misleading. Randomized trials ultimately showed no benefit, and a clear increase in breast cancer and vascular blood clots.

Similarly, cardiologists once prescribed drugs to reduce sudden death in patients after heart attacks. The drugs suppressed arrhythmias – disturbances of the heartbeat associated with sudden death. The logic seemed clear: suppress the asymptomatic but nasty-looking arrhythmias, and you will reduce sudden death. Unfortunately, when investigators undertook randomized trials, they found that treated patients had a higher death rate. Clinicians stopped using the drugs, but not before causing a large number of unnecessary deaths.

But awareness that principles of evidence-based medicine should also guide health policy has been slower to take root. Here, although randomized trials are not feasible, it remains possible to follow evidence-based medicine’s primary rule: to base action on systematic summaries of the highest quality evidence available.

Consider the issue of whether countries should fund health care publicly, through taxes, or privately, through health insurance and user fees. Private health insurance has proved enormously wasteful, with large administrative expenditures on activities that include developing insurance packages, marketing those packages, and assessing claims. Public funding eliminates these activities, resulting in far lower administrative costs.

Likewise, user fees discourage the poor and elderly from accessing needed health services, whereas public funding largely eliminates inequities in the provision of care while producing health outcomes that are as good as, if not better than, mixed public-private funding models. Public funding also benefits a country’s economy, because large employers are spared the expense of providing health benefits to their workers.

Thus, the evidence suggests that public funding of all key aspects of medical care – physician and hospital services, drugs, and devices –- offers benefits of equity, efficiency, and industrial advantage.

The medical community has come to accept the need for systematic reviews to guide decisions regarding drugs and surgical therapies, but their use in health policy is only now taking hold. Systematic reviews have summarized the evidence regarding the provision of hospital services by large for-profit companies versus not-for-profit providers.

The primary studies forming the basis of these reviews come largely from the United States, where for-profit and not-for-profit providers work side by side in the same environment, and where large administrative databases allow accurate detection of death rates. But, while these observational studies are threatened by bias from patient selection – sicker patients mean higher death rates, for example ­– the available administrative databases document patient characteristics such as age, severity of illness, and co-morbidity, including diabetes and high blood pressure. This information allows statistical approaches that level the playing field, helping to protect against biases associated with patient selection.

These systematic reviews have shown that for-profit hospital care leads to higher death rates than not-for-profit care, despite higher charges to third-party payers. Systematic reviews have demonstrated the same phenomenon in outpatient dialysis facilities – lower death rates in not-for-profit facilities – and have also shown higher quality care in not-for-profit nursing homes.

The reason that not-for-profit providers can achieve higher quality care with better health outcomes at lower cost is evident. While for-profit providers have higher administrative costs and larger executive salaries, their main burden relative to not-for-profit provision is the need to generate returns for their shareholders. These additional costs mean that approximately 20% of the total income that not-for-profit providers can devote to patient care is unavailable. The result is cutting corners, deficiencies in care, and poorer outcomes for patients.

Principles of evidence-based medicine have transformed the way we look at clinical interventions and may prevent repetitions of public-health disasters such as the inappropriate promotion of hormone replacement therapy and anti-arrhythmic drugs. Application of these principles to health policy can lead to wiser decisions about how to run our health systems. The available evidence provides strong support for public funding of health services delivered by not-for-profit providers.