Big Data in the 1800s in surgical science: A social history of early large data set development in urologic surgery in Paris and Glasgow

Abstract

“Big Data” in health and medicine in the 21st century differs from “Big Data” used in health and medicine in the 1700s and 1800s. However, the old data sets share one key component: large numbers. The term “Big Data” is not synonymous with large numbers. Large numbers are a key component of Big Data in health and medicine, both for understanding the full range of how a disease presents in a human for diagnosis, and for understanding if one treatment of a disease is better than another treatment or better than just leaving the patient on his or her own without therapy. In this paper, we examine the first considerations of Big Data in medicine in Paris in the early 1800s when urologic surgeon Jean Civiale collected the first large numbers. Civiale collected the large numbers to defend the efficacy of his urologic instrument, the lithotrite, and the surgical procedure he developed, lithotrity, for the removal of bladder stones compared with earlier, more invasive surgical approaches. We examine how large numbers were adjudicated in social decision-making in the Académie des sciences, Paris, when a dispute arose among French urologic surgeons about the importance of large numbers in surgical science. After Civiale’s successful defense of his instrument and procedure in Paris, we examine how his approach to Big Data (large numbers) impacted data collection by George Buchanan in his use of the procedure at the Royal Hospital Infirmary in Glasgow.

Introduction

In this paper, we will examine Big Data of the 1800s in Paris and Glasgow as it was adjudicated in surgical science in a dispute among French urological surgeons. This scientific dispute focused on a new urological surgical instrument and procedure whose use had spread across Europe. The data adjudicator and interpreter in this case was a commission selected by the Académie des sciences, Paris.

The term “Big Data” today is typically defined in terms of algorithmic processes carried out by computers. For example, the Oxford Dictionary (2014) defines “Big Data” in the area of computing specifically as “data sets that are too large and complex to manipulate or interrogate with standard methods or tools.” However, these types of large data sets have existed in many eras prior to the invention of computers to carry out algorithmic processes. Our focus in this paper will be on large numbers—as viewed by urologic surgeons and statisticians in the 1800s—as a component of Big Data.

Before computing by machine, when one computed only by hand or in one’s head, the term of probabilistic and statistical importance was “large numbers.” The recognition for the need for large numbers first came about in science, but not in the medical or surgical sciences, rather in arguments in astronomy. In 1755, Thomas Simpson commented on the need for large numbers in astronomical observations:

It is well known that the method practiced by astronomers, to diminish the errors arising from the imperfections of instruments, and of the organs of sense, by taking the mean of several observations, has not been so generally received, but that some persons of note have publicly maintained, that one single observation, taken with due care, was as much to be relied on, as the mean of a great number.

Simpson (1755) further argued that “the more observations of experiments there are made, the less will the conclusion be liable to err, provided they admit of being repeated under the same circumstances.”

In 1837, Simeon Denis Poisson, also an astronomer but foremost a statistician interested in applying probabilistic models to data outside of astronomy, developed his law of large numbers and encouraged surgeons of his day to study large numbers of patients. This was necessary in probability and statistics, as Simpson (1755) notes, “… to diminish the error arising from the imperfections of instruments, and of the organs of sense.” We are interested in the examination of the first large data sets used in surgery and of how the Académie des sciences used “Big Data” in France in the 1800s.

Surgeons have a long history of accumulation of survival versus mortality data in the use of new surgical instruments and procedures for operations. This is particularly true in the military, as many new instruments and procedures are invented and used for the first time in wartime (Vekerdy, 2005), for example, involving ballistic trauma to the abdomen (Mahoney et al., 2005) and temporary vascular shunts in damage control in wartime vascular injury (Britt et al., 2002).

Jean Civiale (1792–1867), a urologic surgeon in Paris, developed and maintained one of the largest data sets on a particular instrument (the lithotrite) and procedure (lithotrity) during peacetime. Early instruments for the removal of small stones from the bladder through the urethra included the urethra-forceps of Sir Astley Cooper and Sir Benjamin Brodie (Hawkins, 1864). Civiale studied under Dupuytren and won prizes from the Académie des sciences in 1827 (6000 Francs) and 1829 (1000 Francs) for his lithotrity procedure (Haydn, 1841). Lithotrity involved “bruising,” “crushing,” or otherwise fragmenting a bladder stone for removal through the urethra without cutting through the bladder wall. The uniqueness of the lithotrity procedure was its less invasive nature in removing stones from the bladder compared with lithotomy (cutting through the bladder wall) (Herr, 2009; Kiefer, 1968).

Civiale’s data set

As a surgeon with a new surgical instrument and a new procedure, Civiale kept track of his urologic cases that distinguished his instrument and procedure from his predecessors and contemporaries. This approach was a necessity in Paris in the early 1800s, as the Académie des sciences had a strong role as an adjudicator of disputes among physicians and surgeons regarding the assessment of newly developed surgical instruments and procedures aimed at replacing older instruments and procedures for the same or similar surgical problem.

In its role as an adjudicator of disputes among physicians and surgeons, the Académie des sciences served an early regulatory function in French society by bringing the typical “physician versus physician” and “surgeon versus surgeon” disputes before experts commissioned by the academy. In the case of the Civiale instrument, the lithotrite, and his procedure, lithotrity, Civiale had what were then considered to be large numbers at hand to defend his procedure. The Académie des sciences selected experts to evaluate Civiale’s procedure, including those from the field involved in the dispute and Poisson, the main statistical expert of the time.

Poisson, although first interested in the development, refinement, and application of mathematical models to the prediction of the orbits of the planets, also recognized that these mathematical models could be used in other realms of government in a variety of ways. Yet, Civiale’s numbers in terms of his own personal data, supported by data from urologists across Europe, demonstrated that his less invasive procedure, lithotrity, had a much lower relative risk of the patient dying than lithotomy in all its forms, including the Marian operation, the Allarton operation, and others (Gant, 1878).

Jacobson’s and Heurteloup’s instruments to break down the stone were replacing Civiale’s method of perforating (drilling) of the stone in the bladder. Also, the operations and the instruments themselves were becoming more simplified and lithotrity was replacing the earlier complicated apparatuses. At the time, problem patients were men with prostatic hypertrophy that would either have retained fragments of the stone in the bladder or have fragments lodged in the prostatic portion of the urethra.

Parisian urological surgeons disparaged Civiale’s instrument and procedure and took their case to the Académie des sciences for adjudication and evaluation (Crosland, 2002). Writing in 1837, Lee et al. note that lithotrity (with an instrument inserted into the bladder) was replacing lithotomy (the cutting of the bladder for stone removal in a surgical operation) throughout Europe, although at the time, the authors did not know if the lithotrite and its procedure, lithotrity, admitted to such a universal application as its advocates pronounced. Civiale’s data set provided the data the l’Académie des Sciences needed to vindicate his lithotrite and lithotrity as a less morbid treatment for bladder stones.

Urologist discussions of survival

In 1862, the British Medical Journal published a report from a hospital gazette showing that Civiale’s data from 1860 to 1861 focused on both survival/mortality and quality of life. In terms of survival, the following was noted about Civiale’s cases:

Taking all his cases in 1860 and 1981 together, M. Civiale has had 120 calculous patients: 115 males and 5 females. Of these, lithotrity has been performed on 88; of whom, 3 have died, 79 have recovered; and in 6 functional disturbances remain—independent, however, both of the stone and of the operation. Lithotomy has been performed on 17 patients; of whom 8 are cured, 2 have fistulae remaining, and 7 have died. The remaining 15 patients have not been operated on; 6 have died, and 9 are still alive.

However, not only were Civiale’s own patients included as data points to support his procedure, but his lithotrity procedure had also gained widespread support from urologists over Europe who were collecting data on their patients. Civiale added these sets to his own data in his report to the Academie des sciences. Belinaye (1837) notes that with lithotomy, 1024 of 5443 patients died, and with lithotrity, 5 of 245 patients died.

Glasgow urologist’s commentary on Civiale’s data set and Buchanan’s data set in Glasgow

George Buchanan (1868), a surgeon and lecturer on Clinical Surgery at Glasgow Royal Infirmary, gives an insight into how data from one country become incorporated into surgical practice in the 1800s. Buchanan notes that the lithotrity procedure was not used extensively in Glasgow for the following reasons: cost of the instrument; and once a lithotrite had been donated, the surgeon’s lack of experience and a lack of understanding of which patients would most benefit from lithotrity over lithotomy. Buchanan then notes that once an expert surgeon demonstrated the use of the lithotrite and the procedure, its use increased. For Buchanan, the additional reason why data are not accumulated in large data sets is that no qualified person was willing to take the time to put the data together in a data set for analysis and interpretation.

In his 1868 paper, Buchanan also described the problems in patient selection for lithotrity in terms of characteristics of stone (size, difficulty in crushing, and the shape of the resultant fragments to be removed from the bladder through the urethra with the lithotrite); the characteristics of the patient and his anatomy (degree of debility, presence of stricture, presence of an enlarged prostate, atony of the bladder, and chronic cystitis); and the general health and age of the patient. Given the importance of patient characteristics in determining patient risk, Buchanan further notes the following about the relative mortality of the two surgical procedures (surgery versus lithotrity): “It is almost impossible to come to anything like an accurate result [of the chance of the patient’s survival] by taking numbers only; still, by taking a very extensive survey, an approximation to truth may be made.” In a pamphlet published in 1880, Buchanan notes that mortality with the performance of lithotrity in his hands was “1 death in 7½ cases for all ages, and 1 in 16½ in children under 15 years.” But Buchanan did not list the total number of procedures he had personally performed in patients with bladder stones.

Civiale’s procedure and his supporting numerical data as reviewed by the Académie des sciences

In terms of Civiale’s patient numbers, 5443 patients underwent surgery for removal of the bladder stone(s), and 245 patients underwent Civiale’s lithotrity procedure. The Académie des sciences commissioned Poisson et al. (1835) to evaluate Civiale’s data. Commentators (Black, 2001; Matthews, 2001; Tröhler, 2001; Vandenbroucke, 2001) note that this type of presentation of data by a surgeon to an august body of experts—Poisson, an expert statistician; Dulong, a physician-chemist; Larrey, a great military surgeon who was Napoleon’s surgeon; and Double, a representative of the medical establishment—was a unique form of peer review for Civiale’s era. These evaluators noted that these numbers were “all supplied by the practice of the greatest surgeons alive.” Physicians in Civiale’s era considered the set of 5443+ patient cases a large number. Even though the reported numbers were large, derived from experts across Europe, and considered even today as “hard data” (survival and mortality data (Boivin, 2014)), the Académie des sciences subjected Civiale’s data to rigorous review.

Poisson et al. (1835) noted that these comparative calculations were not made on very accurate bases (retrospective review of reported cases); therefore, they could not take the place of science. In addition, the expert reports lacked details of how the expert urologists carried out the assessments. Thus, for the Académie des sciences, these reports fell short of the goal of determining the numerical proportions of mortality after incision with any degree of accuracy. Yet, the commission commended Civiale for recognizing both the importance of large numbers and many of the problems interpreting large numbers, especially when those large numbers involved a compilation of data sets from different urologic surgeons. The Académie des sciences concluded that Civiale’s work comparing surgical approaches (surgery versus lithotrity) was moving in the right direction. In the commission’s report on Civiale’s data, Civiale was encouraged “to pursue his statistical research to increase the volume of data, and to provide more circumstantial detail to make it more conclusive…”

Different views on importance of Civiale’s data: Surgeon versus historian versus statistician

As noted by Buchanan, surgeons of Civiale’s day were already interested in both mortality data and other data, such as the demographics of the patient that made a procedure like lithotrity more or less suitable to be undertaken. From a historian’s perspective, Tröhler (2001) argues that Civiale’s report on his lithotrite and lithotrity procedure presented to the Académie des sciences in 1835 was part of a formal evaluation of Civiale’s numbers. This evaluation by urologic surgeons and Poisson, as a participant evaluator, needs to be viewed in “the context of a contest in contemporary French medical literature about the applicability of ‘statistics’ to ‘medicine’” (Tröhler, 2001). However, Civiale’s numbers on the relative mortality of his lithotrity procedure versus lithotomy (cystotomy) were so good from a mortality reduction standpoint that expert urologists throughout Europe adopted the lithotrity procedure in their own practice and continued with the procedure based on their own clinical experience with both lithotomy and lithotrity. Civiale’s data set on bladder stone removal was the most extensive of its day, and his less invasive procedure of lithotrity and his instrument (which other urologists in France and around Europe used quickly and improved upon readily) continued to develop over time. Even Civiale started using Heurteloup’s instruments in his lithotrity practice.

Summary and conclusions

Civiale had the large numbers necessary for statistical analysis, yet the success of his lithotrity procedure was its reduced invasiveness over lithotomy as recognizable without the need for a statistic. Buchanan, a professor of surgery, was interested in clarifying how deaths could be misattributed in such large data sets. Urologic surgeons showed strong understanding of how numbers could be used in comparing surgical procedures on key parameters like age and death, well ahead of physicians comparing treatments in the medical domain.

The impetus for consideration of large numbers in health and medical decision-making in the 1800s rested on surgeons’ collections of survival and mortality data on the new instruments they developed and the new procedures they invented. Internal medicine physicians and psychiatrists lagged behind in this era of large numbers but caught up, as France’s adoption of hospital-based care allowed not only postmortem examination but also the in-hospital recording of patients’ histories, symptoms, signs, and physical examination findings before they died and went to postmortem examination.

Robert Koch did not discover the tuberculosis bacillus until the end of the 1800s, and it was a while before physicians recognized this bacillus as one cause of many cases of phthisis that had ravaged Europe and the rest of the world through preceding generations. In health and medicine, for physicians other than surgeons, the importance of large numbers only started to be appreciated in the 1800s, even though mathematicians saw their need in society much earlier. But even the statisticians of the day, like Poisson, saw the application of statistics, not so much in health and medicine, but more so in prediction in the courts and the law. Thus, the appreciation of large numbers in internal medicine and psychiatry would be further delayed. Physicians need large numbers as a key component of Big Data for two purposes: to understand the full range of the ways a disease may present in a human and to compare one treatment of a disease against another treatment in patients.

Today, the term “lithotripsy” (synonymous with lithotrity in Civiale’s day) dominates the urologic research literature and urologists are crushing concrements in the urinary tract through lithotripsy (rather than surgery). Instead of a lithotrite, urologists studying various crushing procedures for stones in the urinary system are comparing instruments like the holmium laser versus the ultrasound probe. For example, Jakóbczyk et al. (2011) reported data on 164 patients who underwent endoscopic lithotripsy. Of these, 98 were treated with a holmium laser and 66 with an ultrasound probe. Before surgery, all patients underwent urological ultrasound and radiological imaging to localize the stone, and the effectiveness was measured using radiological and renal ultrasound imaging. An effective state was reached if the stone was invisible by radiograph and ultrasound imaging after the procedure.

Laser lithotripsy was 100% effective in the bladder (25 of 25 patients), and 89% effective in stones localized in the ureters. Using an ultrasound probe, the procedure was 79% effective (52 out of 66 procedures). With laser lithotripsy, two of the patients were found to have inflammatory changes of the ureter which necessitated undergoing open surgery. A third patient sustained a perforation of the ureter with laser lithotripsy. The authors concluded the following: “The results prove that laser lithotripsy is a method of high effectiveness with a low risk of complication. It might surpass sonotrode and become its alternative. Both methods have both advantages and disadvantages.”

Today, regional studies are being conducted to identify and explain inconsistencies to formulate best practice patterns for use in clinical urology across regions of a country or across countries. Brown et al. (2014) conducted one such study in the United States to evaluate such regional differences in shockwave lithotripsy. The researchers examined a prospectively maintained database retrospectively and evaluated regional differences across four distinct American geographic regions involving 2240 procedures. The authors found regional differences in all parameters. In their conclusions, the researchers state: “Inconsistencies in regional adaptation of best practices may identify opportunities for further education.”

The most notable differences in the reporting of procedures of the urologic sciences today compared with the urologic sciences of the 1800s are the safety of patients and study participants and the care with which data is analyzed statistically. Two factors mediate the need for large numbers in research studies in urology today: the primary factor is the need to protect study participants by not exposing them to unnecessary risks (Mazur, 2007), and the second factor is the need to study a large enough group of patients to yield statistically significant results.