Emergence of quantification in clinical investigation and the quest for certainty in therapeutics: The road from Hammurabi to Kefauver

Origins 

Although the priestly medicine that dominated early antiquity was set aside for the more pragmatic and rational medicine that evolved in the Golden Age of Greece, Hippocratic medicine continued to be centered on individual observation and decision by an experienced physician. To quote Hippocrates (460–370 B.C.) on therapy: “One must attend in medical practice not primarily to plausible [ie, probable] theories, but to experience [ie, lessons learned from trial and error] combined with reason [ie, deductive conjecture].”3

Clearly, this method is not experimental medicine as we understand it now, yet it was this approach, which was based on careful observation and documentation of trials and errors and deliberate decision based on experience that continued to evolve in Greece, thrived in Ptolemaic Alexandria and culminated 6 centuries later in the works of Galen (131–200 A.D.). Galen has been appropriately credited with emphasis on repeated observation (“to become expert in all matters…by personally inspecting them not once or twice, but often”), including experimentation on animals, which he used in his own studies. His demonstration of the kidneys as the source of urine is a classic experimental study by any standard.4 In his therapeutics, Galen also relied on experimentation but has been faulted for his statement that “All who drink this remedy recover in a short time, except those whom it does not help, who all die. Therefore, it is obvious that it fails only in incurable cases.” Although anyone would be derided for such a statement, when read without its extravagant claim, which was the norm for that era, how different is it really from what we now tell the family members of patients who succumb to sepsis, despite the use of appropriate antibiotics?

What the above quote rather illustrates is the authority and thoroughness with which Galen summarized, classified, and presented the medical knowledge that had accrued theretofore. The languishing of the sciences in general, and of medicine in particular, during the period of decline and fall of the Roman Empire that followed and the ensuing Dark Ages of the Early Medieval period, accounts for the continued dominance of Galenic medicine, which because of its firm basis on divine creation, was espoused and disseminated by the increasingly dominant Church, which made science subservient to religion.

Arabic medicine, which thrived during the eight to the twelfth centuries, remained deferential to Galenic and Hippocratic medicine in its methodology but made significant and novel contributions to therapeutics. In fact, Avicenna (980–1037) in his encyclopedic Canon sets clear rules for testing the increasing number of new drugs that were being introduced. He specifically states that: “experimentation must be done with the human body, for testing a drug on a lion or a horse might not prove anything about its effect on man.”5, 6

The challenge to Galenism that began in the closing years of the Middle Ages flourished during the Renaissance, when direct observation of nature and the study of man began to be emphasized. One of the first to do so is an intriguing but vilified figure of medicine, Paracelsus (1493–1541). Shortly after his appointment to the medical faculty in Basle, Paracelsus posted the list of his lectures in the vernacular German, rather than the traditional Latin, and once his courses began, he threw the classical medical texts, including those of Galen, into a bonfire. For his contribution to therapeutics, Paracelsus has been credited for founding the pharmaceutical sciences.7, 8

Actually, the revolt against Galenism is best epitomized in the work of Vesalius (1514–1564), who in the course of his dissections demonstrated several of the inaccuracies of human anatomy described by Galen, whom he directly criticized in his De Humanis Corporis Fabrica (1543). Erosion of Galenism reached its apogee in the description of the circulation by William Harvey (1578–1657). Although best remembered for his description of the circulation, Harvey’s systematic methodological approach to presenting his conclusions after 26 years of careful observation involving humans and animals was what actually revolutionized the clinical sciences.9 Another eminent figure who exemplifies the revolutionary spirit of the seventeenth century is Santorio Sanctorius (1561–1636). The balance studies he conducted on himself are the first reported instances of self-experimentation and of metabolic studies that bear on quantification and scientific instrumentation that were to change the clinical sciences in the years that followed.7, 10

The foundations of clinical research laid by Paracelsus, Vesalius, Harvey, and Sanctorius did not reach their full fruition until the nineteenth century. By then, the basis of the scientific method had been laid by the new physics of Galileo Galelei (1564–1642) and the new chemistry of Robert Boyle (1627–1691). Although these scientists were recognized and applauded by clinicians, their discoveries were not applied to medicine and therapeutics. Instead, reliance continued to be placed on the age-old approach of Hippocrates and Galen—personal experience and deductive reasoning. Apart from the impact of the burgeoning sciences basic to medicine (chemistry, physiology, and pathology), what ultimately changed things was the equally important gradual introduction of quantitative methods into clinical investigation. This development was the result of changes that began during the century between 1750 and 1850, as an increasing number of mathematicians became aware of the possibilities inherent in the calculus of probabilities. What began as a theoretical exercise in calculus soon evolved into statistical analysis that was to revolutionize clinical investigation.7, 11, 12, 13

Foundations: quantitation and certainty 

Initially, it had been the signs and symptoms with which patients presented that were more or less reliable that constituted the evidence of an illness, formed the basis of a diagnosis, and determined its therapy. By the end of the Renaissance, as postmortem examinations became possible and observations began to be compiled, their correlation with signs and symptoms led to the transformation of the evidence of an illness to evidence of a disease that affected an organ.7, 13, 14, 15 This transformation, which provided the basis for a new classification begun by G. B. Morgagni (1682–1771), reached its apogee as morbid anatomy flourished in the nineteenth century, when the increase in the number of hospitals and their use in teaching and investigation allowed for correlating bedside observations, physical findings, chemical analysis, and subsequent findings at postmortem to define and classify new diseases. An innovative approach, but one that remained descriptive in nature, which made its greatest initial strides in the Parisian schools of medicine, where some of the giants of medicine such as Cabanis (1757–1808), Pinel (1745–1826), Bichat (1771–1802), and Dupuytren (1777–1835) made their greatest contributions.15, 16 This new approach is perhaps best exemplified in the reports of Richard Bright (1789–1858) from Guy’s Hospital, especially after a ward was assigned to him for the study of patients with kidney disease.17

Among the ablest successors of Bichat was F. Broussais (1772–1838), who continued to emphasize objective research and localization of disease but remained dogmatic in manner and didactic in therapy—notably in his fanatic support of bleeding and leeching. In this exciting arena, several events converged between 1815 and 1850 to bring about the introduction of mathematical methods into clinical investigation. These methods changed the variability of the conjectural discipline of the art of medicine that had existed theretofore into the relatively clearer and exact science of modern medicine.5, 7, 11, 12, 13, 14, 15, 17 In establishing the genealogy of this revolutionary paradigm shift, medical statistics can be considered to have begun with the work of Pierre Charles-Alexandre Louis (1787–1872), a student of Broussais and a contemporary of Laennec (1781–1826), with whom, for a brief period, he shared the wards of La Charité.18, 19, 20

Tentative attempts to use statistics had already been made, but these attempts were inexact and sporadic, and no established methodical approach existed for rigorous data collection and analysis. The perseverant, disciplined, and systematic approach of Louis makes him stand out in history for introducing the precision of his “numerical method” and to be credited for being the first to make statistical analysis the basis of clinical investigation.7, 11, 15, 17, 18, 19, 20 Although Louis followed the prevailing method of correlating clinical with postmortem findings and made his initial contributions in defining the diagnostic features, natural history, and pathologic findings of tuberculosis and typhoid fever, his concern for cures soon directed his attention to therapeutics. In a classic comparative study of patients with pneumonia, he clearly demonstrated the relatively poorer outcome of those treated with bleeding (Recherches sur les Effets de la Saignée, 1835). His report soon led to the abandonment of leeching in Paris and antagonized his teacher Broussais.18, 19, 20

Objections soon were made to the numerical method of Louis, which was debated in the Academy of Sciences in Paris and in the Royal Society in London. Some argued that statistical averages eliminated individual differences and one could not use this method in the care of a single patient because individuals varied greatly; and that a real danger existed that the concept of an average case (l’homme moyen) in clinical medicine would lead to the use of routine treatments while ignoring individual differences.5, 6, 7 Such an objection seems strange considering that at that time, therapeutic options consisted mainly of bleeding, leeching, and clysters, in addition to herbals that dated back to the days of Galen. Actually, the size and content of the pharmacopoeias that were being compiled notwithstanding, little advances in available medications had been made since the first century A.D., when Dioscorides compiled his De Materia Medica. Relevant in this regard is the expressed contempt of Paracelsus for those who used a single panacea for all stomach aches, when as he put it, “the physician must know that there are a hundred, indeed more than a thousand kinds of stomach” diseases.7, 8

Whereas most authorities and practitioners of the period maintained that medicine “cannot be reduced to calculus,” the numerical method of Louis continued to gain ground.17, 18, 19, 20 A principal limitation of the numerical method was its use of simple arithmetic, which made no allowance for “probable error” and, in retrospect, might have promulgated small statistical differences. However, shortly after its introduction, Jules Gavarret (1809–1890), a former military engineer turned physician, resolved the issue in 1840 by introducing the concepts of variance and confidence intervals; he used Louis’ data to illustrate his inferences. The importance of enumeration in collecting data that was then interpreted in terms of possible error and probabilities was now on sound footing.7, 11, 21, 22, 23

Apart from his use of clinical statistics and the emphasis of their importance in medical investigation, Louis trained and stimulated many of the subsequent generation of luminaries who went on to establish medical statistics and a quantitative approach to epidemiology in Europe and the United States.17, 18, 19, 20, 21, 22, 23 Two of his students, William Farr (1807–1893) and William A. Guy (1800–1885), were responsible for founding in 1834 the Statistical Society of London, predecessor of the present Royal Statistical Society.21, 22, 23

New beginnings 

Where does the kidney stand in all this? 

The numerical method of Louis found one of its first applications in the multicenter study of urolithiasis.29 One of his American students, Francis Delafield (1841–1915), went on to apply it to the study of the causes of Bright’s disease.30 Doubtless other examples of their application to kidney disease exist. Their recognition awaits further research in the origins of nephrology.

In a visionary statement in the introductory comments made at the foundation of the Epidemiological Society in London in 1850, Mr. Babington, its president states, “The existence of urea in animal fluids of those labouring under albuminuria is another fruit of modern discovery, and from these, and many similar instances of chemical research we are led to hope much from its application, even to the objects which come within the scope of this society.”26 As previously noted, the dominance of the bacteriological era in epidemiological studies delayed the application of the numerical method to medicine in general and, in due time, to the study of chronic kidney disease in particular. In the 1950s, when nephrology was still in its germination phase, pyelonephritis was considered the main cause of kidney failure, and, “inapparent pyelonephritis” was promulgated as the “missing link” whose eradication was “essential if the termination in hypertension and uremia is to be prevented.”31 Two decades passed before this “great crusade” was finally put to rest, a story briefly and humorously recounted in a 1977 JAMA editorial subtitled, “Will The Real Pyelonephritis Please Stand Up.”32

The full application of the sophisticated mathematical methods to diseases of the kidney had to await the emergence of nephrology as a discipline. Now, studies initiated after the introduction of RCTs are providing evidence for the emerging epidemic of chronic kidney disease, based on the retrospective analysis of data gathered in the 1970s in clinical trials such as the Hypertension Detection and Follow-Up Program and the Multiple Risk Factor Intervention Trial (MRFIT).

In the relatively brief period since the discipline of nephrology was established, our understanding of chronic kidney disease and its epidemiology has expanded exponentially. Application of the mathematical methods of precision that took millennia to develop found an immediate and fertile ground in the nascent discipline. As the articles in the present issue of Advances Chronic Kidney Disease illustrate, much has been accomplished, but more remains to be done.