The Statistics of Health Decision Making- Causes and Consequences of Illiteracy

statimagesGerd Gigerenzer has produced much work that is relevant and easy to understand for the laymen.  Gigerenzer has defended the effectiveness and our abilities to use heuristics.  He has also tirelessly promoted statistics education and transparent presentation of statistical information.  This post tries to summarize some unique and interesting material with respect to the causes and consequences of our statistical illiteracy.

Gerd Gigerenzer along with Wolfgang Gaissmaier, Elke Kurz-Milcke, Lisa M. Schwartz, and Steven Woloshin suggest that the long opposition to health statistics can be traced back to the struggle between three 19th-century visions of the physician: artist, determinist, or statistician.  They argue that these professional ideals go hand in hand with patients’ corresponding ideals, which even today determine the mixture of feelings about health statistics: The artist embodies paternalism and requests blind trust from the patient, the determinist strives for perfect knowledge of causes and invites the illusion of certainty in patients, and the statistician relies on facts rather than medical charisma, paving the way for shared decision making. The first two ideals effectively deter interest in health statistics.

This cartoon was included as a part of the 2008 monograph. Considering its age, it has held up well.

There are various players in public health with goals that can conflict with transparent risk communication—goals such as pushing a political agenda, attracting media attention, selling a new drug, increasing compliance with screening, or trying to impress physicians. Conflicts of interest lead to omission of relevant information and the use of nontransparent framing.

In medical journals, papers often report the benefits and harms of treatments  with different measures, benefits in big numbers (relative risk reduction), but harms in small numbers (absolute risk increases). Doctors and patients continue to use those numbers. In some cases, those numbers mask the fact that the harms are greater than the benefits. And although medical journals have been characterized as creatures of the pharmaceutical industry–relying on drug money to survive-there are other players in nontransparent information.  Government and other public health organizations have agendas and are concerned with appearing to provide inconsistent information to the public.  They are prone to going along.  If woman stop getting mammograms and later improved cancer treatment could have saved their lives, who is going to get blamed?

Gigerenzer promotes transparent forms of providing statistical information that  include absolute risks, natural frequencies, mortality rates, and, in general, statements about frequencies or depictions of frequencies in pictures. Nontransparent forms include relative risks, conditional probabilities such as sensitivities and specificities, survival rates, and statements about single events that do not specify the reference class.

The two consequences of statistical illiteracy coupled with misleading advertising are emotional manipulation and impediments to informed consent and shared decision making.