The discovery of anesthesia mandated research to address a simple question’how to conduct the patient safely through the operative procedure, and how to satisfy surgical requirements without injuring or killing the patient? We needed a clinically relevant assessment of what we came to call depth of anesthesia, we needed a controllable way to deliver anesthesia to achieve the “correct” depth, and we needed it right away. John Snow gave it to us. Golly, he was good. He invented the notion of depth of anesthesia, and he expressed that as five “degrees”:

“I shall divide the effects of ether into five stages or degrees; premising, however, that the division is, in some measure, arbitrary…In the first degree of etherization I shall include the various changes of feeling that a person may experience, whilst he still retains a correct consciousness of where he is, and what is occurring around him, and a capacity to direct his voluntary movements. In…the second degree, mental functions may be exercised, and voluntary actions performed, but in a disordered manner. In the third degree, there is no evidence of any mental function being exercised, and consequently no voluntary motions occur; but muscular contractions, in addition to those concerned in respiration, may sometimes take place as the effect of the ether, or of external impressions. In the fourth degree, no movements are seen except those of respiration, and they are incapable of being influenced by external impressions. In the fifth degree…the respiratory movements are more or less paralysed (sic), and become difficult, feeble, or irregular.” (pp 1–2) [10]

Snow described how to achieve these degrees in his two books on delivery of the anesthetics of the day, diethyl ether [10] and chloroform [11]. We spent a century tinkering with the basic notions, Arthur Guedel being the greatest tinkerer. In World War I, he gave us not five degrees, but four stages, the third (surgical anesthesia; roughly equivalent to Snow’s fourth degree) containing four planes (Fig. 39.2) [12]. Along the way, in 1895, Cushing (a neurosurgeon) and Codman added measurement of blood pressure and heart rate, and the anesthetic record [13].

What Snow, Cushing, and Guedel taught us required attention to the patient’s response to the anesthetic. What we saw in the patient, things such as the character of breathing, or the size and responsiveness of the pupils to light, or changes in blood pressure, guided administration of more or less anesthetic or dictated cessation of anesthetic administration. Our observations had an enormous subjective element, meaning that delivery of anesthesia was as much an art as a science. This changed with our increased knowledge of anesthetic phenomenology in the 1950s, and the ability to deliver a precisely known amount of anesthetic that, additionally, we could directly measure.

In the mid-1900s we knew little about what we came to call anesthetic pharmacokinetics. We were ignorant about factors governing movement of anesthetics within the body (and thus we were hindered in our understanding of how to control and make use of such limits). Snow knew it was important, but could do little about it. In 1924, Howard Haggard explored the absorption, distribution and elimination of di-ethyl ether in dogs, dividing the distribution into that which went to the brain, and that which went elsewhere [14,15]. In 1950, Seymour Kety (1915–2000) provided the first kinetic models, unintelligible for most mortals, and wrong in details (it failed, for example, to recognize the importance of differential distribution of the delivery of blood to different tissues) [16].

In 1954, John Severinghaus (b. 1922) started us properly down the pharmacokinetic road with his seminal study of nitrous oxide, the first investigation of uptake of an inhaled anesthetic in humans [17]. His investigation led to a more general, but empirical, description of uptake known as the square root of time rule, a rule stating that anesthetic uptake decreases as a function of the square root of time. It applied to anesthetics other than nitrous oxide [18]. Hal Price improved on the concepts of tissue distribution of intravenous anesthetics, noting that tissue bulk and blood flow had to be accounted for in our thinking of the redistribution of an anesthetic like thiopental (Pentothal®) [19]. His concepts led the way with thiopental and injected anesthetics, and prompted a broader view, one applicable to the inhaled anesthetics. In parallel with Price, Eger (b. 1930) studied inhaled anesthetic kinetics, doing so with a vengeance, measuring solubilities and uptake of all clinically-used and many experimental anesthetics [20]. In the late 1950s, he developed an iterative mathematical model explaining how inhaled anesthetics move into and throughout the body [21], a model that with a few modifications applies to all inhaled anesthetics used today. Other investigators used analog models to accomplish a similar explanation [22,23], but these could not describe the effect of the inspired anesthetic concentration on pharmacokinetics.

In 1963, we got a measure of anesthetic potency, a measure that would provide a standard that allowed a comparison among anesthetic effects. The measure of potency was MAC, the minimum alveolar concentration of an inhaled anesthetic producing immobility in 50% of subjects given a stimulus such as a surgical incision [24,25]. Three decades later, we got the equivalent of MAC for injected drugs such as propofol [26]. MAC allowed the comparative measurement of effects of inhaled anesthetics, everything from cardiorespiratory effects to what anesthetics did to brain and muscle and kidney and liver and pregnant ladies and infants and obese people and old people and… [27]. These phenomenology studies (“what does X do to Y?”) poured out and dominated research through the 1970s and 1980s. Such studies continue to the present.

For the most part, important outcome studies appeared in the second half of the 20th century. But we must acknowledge earlier work, for example, Snow’s careful dissection of the causes of death associated with chloroform anesthesia [11]. He repeatedly refers to a “sudden paralysis of the heart”, possibly ventricular fibrillation or possibly severe myocardial depression. A less successful outcome study with the same theme was the work of the Hyderabad commission [28]. In 1902, echoing Snow’s observations, Edward Embley provided support for the view that chloroform might cause sudden death by cardiac depression, believing that this resulted from a direct depressant effect of the vapor or by vagal irritability [29−31].

In 1954, at about the same time as we began to understand inhaled anesthetic pharmacokinetics, we got the first of the great outcomes studies, the work of Henry Beecher (1904–1976) and Donald Todd [32]. These investigators directed a consortium of 10 University hospitals in an investigation of the factors associated with death after anesthesia. Despite an absence of computers (after all, it was the early 1950s) they collected data on 599,548 patients coming to the astonishing conclusion that if a patient undergoing major surgery received curare, that patient was approximately 6 times more likely to die. The finding raised a ruckus, especially since it was published in a surgical journal and surgeons subsequently ordered anesthetists to stop using curare. No one likes being told what to do. Of course, the main problem was that apparently American anesthetists didn’t understand that curare was apt to leave a residual paralysis that required reversal of the neuromuscular blockade [33]; the anesthetist did not recognize that although the patient might be breathing and appear grossly normal, he was still at risk from a remaining weakness, at risk of inadequate ventilation or an inability to react properly to vomiting’with vomit going into the lungs. You’d think we anesthetists would learn, but half a century later, another outcomes study again found that inadequate reversal of the effects of muscle relaxants grossly increased mortality [34].

Introduction of the inhaled anesthetic halothane, in the 1950s, was a smashing success. It quickly replaced all other potent inhaled anesthetics, including chloroform, cyclopropane and ether (halothane didn’t explode but ether and cyclopropane did, and halothane had the added advantage of being less pungent than ether). Halothane also displaced the now infrequently used chloroform because chloroform injured the liver and halothane didn’t seem to do so. Then in the late 1950s and early 1960s, sporadic cases of severe liver injury appeared after anesthesia with halothane, and we anesthetists feared that we might lose this new friend. The result was the National Halothane Study [35], a study of more than 800,000 patients given various anesthetics. The results suggested a rare, but believable connection between halothane and severe liver injury and death, a connection with signs and symptoms indicating that injury might be a consequence of an allergic response of the liver to halothane (or its metabolites).

The number of studies of outcomes, consequent to the activities of anesthetists, has increased dramatically. Outcome studies differ from phenomenology studies in that the outcomes intend to connect an action (of an anesthetic or a manipulation of some sort such as body temperature [36], oxygen concentration [37], or administration of beta blockers [38]) to either an adverse or beneficial outcome. In contrast, phenomenology studies connect an activity to a change where the change is neither inherently good nor bad. Sometimes outcome studies require data collection from large numbers of patients, as in the Beecher-Todd study or in the National Halothane Study, but more recent outcome studies might include just tens, hundreds, or a few thousand patients. Such small numbers were needed to demonstrate in the 1990s, that beta adrenergic blockage might provide long-term protection against postoperative myocardial injury [38].

One dramatic outcome study in the 1950s, was inadvertently conducted on patients suffering from tetanus [39]. These patients were given nitrous oxide to breathe, in order to decrease the pain of the muscle spasms produced by tetanus. The pain decreased, but after a few days, the number of white blood cells decreased and the patients began to die. The connection to nitrous oxide was clear, although the reason was not. The use of nitrous oxide continued in ordinary surgical patients because it appeared that the problem was only from prolonged breathing of the nitrous oxide. Anesthetists debated whether a more subtle noxiousness of this most ancient of anesthetics resulted from shorter exposures, with a consensus that the case was not proven, and the anesthetic had a long record of benefit and safety [40]. In 2007, an outcomes study of 2000 patients, given or not given nitrous oxide, suggested that nitrous oxide increased morbidity [41], possibly doing so for reasons related to the unique capacity of nitrous oxide to slowly inactivate a vital enzyme, methionine synthase [42]. However, not all recent outcome studies, even of 1000 patients, reveal an adverse effect of nitrous oxide [43]. The noxious effect of nitrous oxide continues to be debated, but the concerns over such noxiousness, including increased postoperative nausea and vomiting, have decreased the use of nitrous oxide in clinical practice.

The numbers of outcome studies are already large and will increase further. They will increase because of the need to supply better patient care, and to provide “evidence-based medicine.” Their application may not only increase patient well-being, but will likely decrease the cost of patient care by causing us to discard ineffective remedies that often have the additional disadvantage of being expensive. This latter issue also concerns research on the comparative effectiveness of medical treatments and whether the incremental benefit of more effective treatments justifies the added costs.

As indicated above, the major anesthesia research in the 2–3 decades following World War II was phenomenological, plus an increasing, somewhat later interest in outcomes. The 1970s and 1980s saw the start of a third major focus: how did anesthetics and anesthetic adjuvants work? A few such inquiries had been made earlier. In 1846, Claude Bernard (1813–1878) demonstrated that curare acted by interfering with the transmission of nervous impulses to muscle, at what is called the motor end-plate, the juncture of nerve and muscle [3]. Around 1900, Meyer [44] and Overton [45] connected anesthetic potency to an affinity for fatty substances, implying that anesthetics acted by doing something to the lipid parts of neurons, perhaps something to the lipid bi-layers that encircle all cells, including the cells of the brain.

But only sporadic discoveries were made until the 1980s, when we began to make use of the great advances in our understanding of receptors, and in genetic engineering. Nick Franks and Bill Lieb deserve credit for shifting our attention away from thinking that anesthetics, especially inhaled anesthetics, must work by some action on lipids in membrane bilayers. As noted above, Meyer and Overton had focused on the correlation of anesthetic potency with affinity to lipid. In 1984, Franks and Lieb demonstrated that an equally good correlation could be made with an effect on a protein [46]. Then the floodgates opened, with many investigators showing that anesthetics affected numerous and specific receptors (proteins) in ways that plausibly explained the actions of anesthetics and anesthetic adjuvants, such as opioids and muscle relaxants.