Until the second quarter of the 20th century, medical treatment had little positive effect on when, or even whether, sick people recovered. With the discovery of ways to reverse the biochemical deficiencies that caused some diseases and the development of antibacterial drugs, it became possible to cure sick people. These early successes and the therapeutic optimism they engendered stimulated the biomedical research community to develop a host of more powerful agents to treat heart disease, cancer, neurological disorders, and other ailments. These increasing opportunities for productive intervention as well as a fundamental restructuring of the market away from nonprofit health care providers to for-profit entities and the expansion of the pharmaceutical, medical device, and insurance industries that saw opportunities to make money providing medical services, together with increasing expectations by the public, have led to spending an accelerating amount of money on medical services, reaching $2.6 trillion and nearly one-fifth of the United States’ entire gross domestic product in 2011 (Fig. 1-1).

This situation has led to continuous calls for reform from a wide spectrum of stakeholders, from business leaders who saw their costs skyrocketing, to labor leaders who saw health insurance costs putting downward pressure on wages, to advocates for the growing number of uninsured people who were simply priced out of the system, to political decision makers who saw out-of-control costs of providing medical care through government programs such as Medicare and Medicaid, jeopardizing other important government services.

Because of the fact that medical care touches everyone’s life in one way or another and because of the high stakes—financial and otherwise—for the individuals and organizations that provide these services, reforming the health care system has been a controversial and politically charged issue.

After over a year of increasingly partisan debate, in March 2010 the Democrats in Congress passed the Patient Protection and Affordable Care Act without a single Republican vote. On March 23, 2010, President Barack Obama signed the bill into law.

While this law has many provisions, including requiring people to have or purchase health insurance and imposing many regulations on the health insurance industry, it also recognizes that the current medical system is unsustainable financially and includes several provisions designed to get the costs of the medical system under control. (Indeed, one of the main facts driving the debate was the observation, from an ongoing research project at Dartmouth University, the Dartmouth Atlas of Health Care,* that 30% of the nation’s medical spending would be unnecessary if all regions of the United States the provided services at the level observed in low-spending regions that achieved that same equal quality.) The law established a Patient-Centered Outcomes Research Institute to conduct comparative effectiveness research on the “relative health outcomes, clinical effectiveness, and appropriateness” of different medical treatments. The law also created task forces on Preventive Services and Community Preventive Services to develop, update, and disseminate evidenced-based recommendations on the use of clinical and community prevention services.

These issues are, at their heart, statistical issues. Because of factors such as natural biological variability between individual patients and the placebo effect,** one usually cannot conclude that some therapy was beneficial on the basis of simple experience. Biostatistics provides the tools for turning clinical and laboratory experience into quantitative statements about whether and by how much a treatment or procedure affects a group of patients.

Hence, evidence collected and analyzed using biostatistical methods can potentially affect not only how clinicians choose to practice their profession but what choices are open to them. Intelligent participation in these decisions requires an understanding of biostatistical methods and models that will permit one to assess the quality of the evidence and the analysis of that evidence used to support one position or another.

Clinicians have not, by and large, participated in debates on these quantitative questions, probably because the issues appear too technical and seem to have little impact on their day-to-day activities. Clinicians need to be able to make more informed judgments about claims of medical efficacy so that they can participate more intelligently in the debate on how to allocate health care resources. These judgments will be based, in large part, on statistical reasoning.

* The research behind this statement, together with many other findings about geographical variations in medical services and health outcomes is available at www.dartmouthatlas.org.

** The placebo effect is a positive response to therapy per se as opposed to the therapy’s specific effects. For example, about one-third of people given placebos in place of painkillers report experiencing relief. We will discuss the placebo effect in detail later in this book.

Suppose researchers believe that administering some drug increases urine production in proportion to the dose and to study it they give different doses of the drug to five different people, plotting their urine production against the dose of drug. The resulting data, shown in Figure 1-2A, reveal a strong relationship between the drug dose and daily urine production in the five people who were studied. This result would probably lead the investigators to publish a paper stating that the drug was an effective diuretic.