1
Risk and medical innovation
A historical perspective
Thomas Schlich1
Medical innovations do not always turn out the way they are supposed to. Any therapeutic, diagnostic, or preventive measure can have unexpected and undesired effects, making uncertainty inherent in medicine. As the sociologist Renée C.Fox states, “clinical advances change the content of medical uncertainty and alter its contours, but they do not drive it away.”2 Advances even create new uncertainties. To the extent that modes of diagnosing and treating disease and illness have become more powerful, they have also grown more dangerous, exposing patients to more potential harm through anticipated and unanticipated negative consequences. Doctors, patients, state authorities, and others have developed strategies for dealing with this problem. Most importantly, they have increasingly framed the problem in terms of risk. Using the notion of “risk” implies a specific strategy for dealing with uncertainty, based on the calculation of probabilities.3 Probability-based logic has been employed “to approach the uncertainties of diagnosis, therapy, and prognosis, and the clinical judgment that lies at their heart” since the eighteenth century.4
The term “risk” is derived from the French word “risque” and first appeared in its anglicized form in England in the early nineteenth century. It was originally used in a neutral way in order to refer to a wager made by individuals after taking the probabilities of losses and gains into account. In more recent times it has come to refer exclusively to negative outcomes; to the likelihood of some adverse effect of a hazard.5 Social scientists acknowledge the fuzziness of the term with its range of synonyms like “gamble, hazard, danger, probability, uncertainty, and odds ratio”.6 Technically, one can speak of risk when the probability estimates of an event are known, or at least knowable, while “uncertainty”, by contrast, implies that these probabilities are inestimable or unknown. In the real world, however, the distinction is not so sharp, and risk and uncertainty are better conceived of as poles on a continuum, with “risk” being used as shorthand for forms of technology and social organization that have the potential to harm people.7 Despite its fuzziness, the notion of risk is typically linked to the particular purpose of reducing, modifying, or anticipating the extent or nature of uncertainty in decision-making processes.8 Thus, using the term “risk” implies human agency. According to Niklas Luhmann one speaks of “dangers” if the potential damage is attributed to causes outside one's own control, as is the case, for example, with natural disasters. “Risk” implies that damage is perceived as being the consequence of one's own decision, and could be prevented by using, for example, technologies of risk control.9
In contemporary medicine the issue of risk has become so common that social scientists speak of a veritable “risk epidemic” in the medical literature since the 1960s.10 The concept of risk has also been employed in different ways in this later period, but is most often used to refer to the likelihood of falling ill. Again, the notion of risk carries with it the connotation of human agency in the origin of illness, as opposed to fate or nature, and this has made it easier to apply the risk concept to medical intervention. Since the mid-twentieth century new technologies, procedures, and drugs have increasingly been evaluated in terms of risk.11 Because of the fear of the risks involved, medical innovation has often been perceived as a mixed blessing.12 This general disillusionment with the advances of science and technology has led sociologists like Ulrich Beck to proclaim a new phase of “reflexive” modernity in which “progress” is increasingly problematized by the production of risks that pervade and transform modern societies.13
The contributions to this volume explore the historical dimension of the issue of risk in medical innovation. They look at how, in the nineteenth and twentieth centuries, historical actors have dealt with the uncertainty associated with medical innovations in different contexts. Medical innovation is understood as the process of introducing a new medical technique or drug, a process during which both the context and the technology itself may change. Therefore, sharp distinctions between invention and discovery, innovation and diffusion threaten to oversimplify the complex process of the technology's acceptance.14 Risk plays a key role in this process, since whether a medical novelty gets accepted or not is, in part, the result of a process of negotiating its potential benefits and dangers.15 Considerations of risk belong to the “real, messy, contested and complex debates by which, over time, some procedures were accepted in preference to others”16 and thus changed the face of modern medicine.
As will be evident from the case studies presented in this volume, different kinds of innovations resulted in diverse strategies for increasing certainty. Highly visible surgical procedures required a different logic from other therapeutic modalities such as new drugs whose effects cannot be immediately identified. Preventive measures like vaccination, lifestyle drugs such as the contraceptive pill, life-saving measures for cancer, medicines for pain management, or diagnostic techniques for ascertaining brain death have all been associated with distinctive rationales of risk management. What all the examples have in common is the attempt by developers, users or regulators to make the consequences of innovation more predictable and controllable using documentation, calculation, and regulation. It is also obvious that the mathematical control of uncertainty by using techniques such as probability functioned primarily as an ideal, serving as a model or as a rhetorical device, but only rarely put into practice in a comprehensive manner. The approach was apparently taken from the context of theories about disease causation and then applied to the problem of medical innovation. This seems to have happened around the middle of the twentieth century, when the notion of risk became more widespread in various areas of society, but this is still an open research question. However, elements of the fully developed risk concept such as quantification, calculation, and probability-logic were adopted much earlier from fields such as life insurance and population studies and used for the purpose of regulating medical innovations. In the following passages I aim to provide a short historical survey of the origins of these elements.
Creating certainty
From a historical perspective, strategies for dealing with uncertainties in medical knowledge and practice that involve applying mathematical methods can be traced back to the period of the Enlightenment.17 The idea of risk first emerged in an economic context of new, extended and more internationalized markets. Risk referred to the danger of losing money in business, especially on loans, in gambling and in insurance, and was related to attempts to hedge one's bets.18 The earliest application of risk to health issues took place in another business context, namely when life insurance companies needed a basis to decide on the acceptance of applicants for a policy. This was again a quantified and calculable precaution for predicting and, if possible, avoiding financial loss. As early as 1762, the first life insurance company, The Equitable in England, charged extra premiums for applicants with gout, a hernia, and no history of smallpox.19 The companies were able to use mortality tables as a basis for determining their policy. The technique of drawing up mortality tables goes back to vital statistics, a field of enquiry that produced knowledge about the statistical regularity of health conditions within populations. As raw material for these statistics, investigators used records of births, deaths, and often marriages in a locality or country. The first examples are the bills of mortality published in 1562 by the City of London to keep track of plague deaths.20
Vital statistics became more sophisticated when the mathematics of business practices were applied to them. In the seventeenth century, the London merchant John Gaunt started applying commercial accounting practices to those mortality lists, and the resulting tables were published in 1662 under the title “Natural and Political Observations Made Upon the Bills of Mortality”. They enabled a systematic comparison between parishes and city neighborhoods, causes of death and gender specificities, as well as the past and present states of affairs. British physicians subsequently started to use this type of procedure for dealing with other medical and public health questions as well.21
Taking up the vital statistics approach and broadening their perspective beyond the traditional individualistic approach of medicine, doctors and administrators could now focus on the well-being of specific groups within the population—such as the laboring poor, soldiers, women, and children, rather than considering only the individual patient. Thus, doctors assumed a role in Enlightenment population politics, an increasingly important area for strengthening and stabilizing the nation state. Starting in the 1830s, statistics also contributed to public health measures. In his 1829 textbook on vital statistics, Elements of Medical Statistics, Francis Bisset Hawkins predicted that the application of statistics to medicine would permit not only the determination of the effectiveness of treatments, and provisions of a basis for reliable diagnoses, but also the evaluation of the impact of living conditions on life, health, and labor.22
In this way medicine became one of many areas in which counting and accounting constituted an important way of thinking. The population-based mathematical approach was broadened to cover more medical fields of application, including accounting for the efficiency of new medical institutions. The British hospital and dispensary movement, for example, induced doctors and lay sponsors alike to establish recovery and death rates within these institutions and so to calculate the success rates of specific cures. The evaluation of clinical experience was another new field of application for numerical calculations. An analysis of British medical journals from 1733 to 1829 shows the gradual reduction in dependence on single case reports and a growth in the publication of larger series, some of which were even analysed by what can be called proto-statistical methods. Especially in eighteenth-century Britain, doctors perceived the need for the empirical evaluation of remedies by comparative trials with results expressed in numbers. These physicians wanted to base clinical medicine on elementary numerical analysis of compilations of cases and observations made on distinct groups of patients, not on individuals. For them, arithmetic calculation was a way out of the maze of contradictory observations. As a result, by the first decades of the nineteenth century a more or less tacit acknowledgment of the utility—and even necessity—of numerical observations in clinical medicine emerged in the British medical literature.23
The strategy of making the potential dangers of medical innovation calculable and controllable by the quantification and calculation of probabilities can be observed early on in the disputes over smallpox inoculation in the eighteenth century. Smallpox inoculation had its own hazards: patients could die from it, and they could also spread the disease by infecting others with inoculated smallpox. This made contemporaries wonder whether the benefit outweighed the risk. The discussions on that topic, for example those between Daniel Bernoulli and Jean le Rond d'Alembert in the 1760s, are considered classics in the history of probabilistic thinking.24 In England, the physician Thomas Nettleton (1683–1742) introduced a particular approach to this problem, using what he called “Merchants Logick”. Merchant's logic argued that physicians should calculate the utility of particular practices by summing up the costs and benefits among a population of patients. Nettleton's was only one of various forms of merchant's logic that cropped up in eig...