early clinical statistics denis hadjiliadis md. background medicine up to 19 th century is...

29
Early Clinical Statistics Denis Hadjiliadis MD

Post on 20-Dec-2015

216 views

Category:

Documents


2 download

TRANSCRIPT

Early Clinical Statistics

Denis Hadjiliadis MD

Background

Medicine up to 19th century is interested in individual patients. Little effort for quantification and description of groups of diseases is given.

Symptoms are only part of the way disease expresses itself in a specific patient. Most theories express disease as an imbalance in one specific system (e.g. GI tract).

Historical background

The French Revolution in 1789, brought in new ideas, took away past privileges and beliefs.

Medical school curricula were changed and instruction stopped being a comment, in Latin, on Hippocrates and Galen. Emphasis was placed on clinical observation and hospital based, patient oriented teaching.

Historical background (cont.)

Great progress in all aspects of knowledge. Atomic theory in chemistry (Lavoisier and Dalton), use of electricity to separate elements in physics (Davey and Faraday).

Philosophy: Voltaire and Rousseau, talk about freedom of speech and thought.

Encyclopedia is written by Didero and Dalabert.

Laennec discovers stethoscope (1760s).

Early epidemiology

John Graunt (member of the Royal Society) published in 1662 his Natural and Political Observations Mentioned in a following Index and Made Upon the Bills of Mortality.

In 1747 Lind finds that citrus fruits prevent scurvy.

In 1760 Bernoulli concludes that inoculation against smallpox confers life-long immunity.

Early statistics

Pierre Charles Alexandre Louis (1787-1872) is considered the father of modern medical statistics.

Introduced numerical method in order to be able “to count the facts” and therefore reach unbiased and correct conclusions.

Pierre Charles Alexandre Louis

Born in 1787 in Ai (Marne river). Began studying law, but continued in

medicine in Rheims and later in Paris. Went to Russia in 1814 and practiced in

Odessa. Found he could not help his patients during a diphtheria epidemic in 1820 and returned to Paris for further study.

P.C.A. Louis (cont.)

Disappointed on the knowledge there, too, joined staff of l’Hopital de la Charite and l’Hopital de la Pitie for seven years.

Wrote medical books in 1825 on phthisis (TB) and on typhoid fever (1829). In the latter book he was able to first demonstrate clinical and autopsy differences among typhoid fever and other kinds of febrile illnesses.

Bloodletting controversy

Decision to tackle the ancient practice of bleeding for pretty much all diseases.

Leading medical authority in Paris was Francois Joseph Broussais. He believed that inflammation was the cause of all diseases.

Bloodletting and antiphlogistic diet could be used to treat all diseases.

In 1833 France imported 42 million leeches.

Study design

Louis performed a case-control study (in modern terminology) to evaluate the effect of bloodletting.

His study was published in Annales de Medecine Generales in 1828 and as a book in 1835. Translated in English in 1836.

Study design (cont.)

Design: chart review of 77 patients admitted to the hospital with pneumonia, who had no other medical problems prior to their illness.

Data recorded and extracted in a standardized fashion.

Study design (cont.)

Comparison, after controlling for other confounding factors such as age, gender and severity of illness of the effect of early (days 1-4) vs. late bloodletting (days 5 to recovery or death) on the duration and the outcome of pneumonia.

Results

Forty-one patients were bled early and thirty-six were bled late. Twenty-seven died; eighteen on the early bleeding group and nine on the late bleeding group.

The average number of bleedings were 2.8 and 2.3 per patient, respectively in the groups.

Results-Table

Survival data

Results (cont.)

Duration of disease was 17.7 days in the early bleeding group and 20.3 days in the late bleeding group.

In the original work there were some errors in the means calculations, which were corrected in the American translation by Jackson. The results were not significantly different.

Discussion

Louis did not condemn bloodletting as a treatment from this study. Results were “startling and apparently absurd” and the effect of bloodletting was “much less than has been commonly believed”.

His work had, however, an impressive effect and leech imports fell to a few thousand over the next 15 years.

Discussion (cont.)

The results were not completely accepted by all authorities and heated debates were common in the French Academy of Sciences and Academy of Medicine.

M. Double suggested that Louis “was attempting to substitute mathematical for logical analysis- to make arithmetic take the place of induction- and calculation that of reason”.

Discussion (cont.)

Paper would not stand modern peer review; it had selection bias, post-hoc analysis and other flaws if seen in the context of modern clinical trials.

Importance of confounding variables, sample size (“for among so large a collection, similarities of conditions will be met with, and all things being equal, the conclusions will be rigorous”) and standardized data collection is mentioned.

P.C.A. Louis

Louis recognized the importance of unbiased data collection and he collected data without being the primary physician of the patients.

Two of his mottos were: “ars medica tota in observatorus” and “I know the truth is in the things and not in my mind which judges them. The less I put of my own into these judgments, the surer I am to approach the truth” (Rousseau).

P.C.A. Louis (cont.)

He remained a hospital physician rest of his professional life in Hotel-Dieux hospital.

He was the president for the Society of Medical Observation (1832). They met weekly, to analyze and critique each other’s papers.

He retired in 1854, one year after his son died of tuberculosis.

P.C.A. Louis (cont.)

Many American visiting students studied under Louis.

Some notable names: Oliver Wendell Holmes (founding member of the AMA), Henry Ingersoll Bowditch (Clinical Conference of the Harvard Medical School), Alfred Stille and William Pepper Sr (Philadelphia), James Jackson Jr (his father translated and duplicated Louis’ work).

P.C.A. Louis (cont.)

Louis did not believe that microscopic examination and laboratory evaluation were useful in medicine.

In the latter part of the 19th century with the rapid progress in microbiology there was search for total and true answers. Louis methods did not give them. His work was not followed up for at least one more century.

Statistics after Louis

William Farr in England developed modern vital statistics by using Louis’ methode numerique.

Public health is born (Liverpool sanitation project, London cholera epidemic).

Statistician R.A. Fisher (of Fisher’s exact test) is said to have had the first ideas about random allocation and its importance. His first experiments were agricultural.

Statistics after Louis (cont.)

Statistics after Louis (cont.)

Gold in 1937 (JAMA) published the first randomized, double-blind, placebo control trial on the effect of xanthines on the treatment of cardiac pain. The trial was not originally double-blind.

The British Medical Research Council was the first to sponsor large scale double-blind, placebo control trials in the 1940s to evaluate TB treatment (streptomycin) and rheumatic fever prevention (penicillin).