Evidence based bloodletting

In the summer of 1996 I attended one of the early workshops on evidence based medicine, run by David Sackett in Oxford. One innovation was to name the small groups after eminent (but long dead) physicians. My group was called Pierre Louis.

A few days before the workshop I came across Pierre Louis on the introductory pages of a clinical epidemiology textbook. I photocopied his picture and took the old Parisian along with me to Oxford. Since then I've got to know him a little better.¹,²

Louis was born in 1787 in Ay, France, and studied medicine in Paris. After graduation he travelled around Russia and settled in Odessa. In 1820 a diphtheria epidemic revealed important gaps in his medical knowledge, and he returned to Paris to study under the eminent Francis Broussais, a sanguinary proponent of bloodletting. But Louis was unimpressed with his boss's didactic approach. At the age of 33, thinking he could learn more by just observing patients, he took an unpaid clinical post at La Charité Hospital in Paris.

Obsessed with systematically collecting empirical data, he carefully observed some 2000 patients over seven years. Using a standardised method (his so called “numerical method”), he counted and tabulated clinical events. The information he gathered allowed him to assess the merits of treatment for himself, rather than put his faith in the experts of the day.

Reviewing his records on 77 previously healthy patients admitted with pneumonia, Louis wondered heretically “whether bloodletting had any favourable influence on pneumonitis.” His comparison of early and late bloodletting produced a scandalous result. While 44% (18/41) of patients bled within the first four days died, only 25% (9/36) of those bled at a later date died. Louis proclaimed the benefits of bloodletting to be “much less than has been commonly believed.”

The medical establishment, heavy users of leeches, thought such averages unhelpful and likely to confuse doctors attending a patient. They refused to discard treatments “validated by both tradition and their own experience on account of somebody else's numbers.” The sanguine doctors of the time firmly believed that “medicine was about individual patients and not about groups.” Even advocates of the “numerical method” cautioned against the application of “mathematical reasoning to subjects which do not admit it.” Louis countered that “a leaf of a tree once well described may always be recognised.” But clinicians remained sceptical, and, despite Louis' evangelism, his “numerical method” disappeared from clinical practice.

At the workshop I stuck Louis' picture up in our seminar room. He observed our discussions all week. He was surprised how little the debate about applying information derived from groups of individuals to individual patients had changed. But he took some comfort from the abandonment of bloodletting.