Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
CHAPTER II-1. APPROACHES TO THE STUDY: A RANDOMIZED TRIAL OR A STUDY OF PAST DATA William H. Forrest, Jr. Stanford University School of Medicine, Palo Alto, California Byron W. Brown, Jr. University of Minnesota School of Public Health, Minneapolis, Minnesota J. Weldon Bellville Stanford University School of Medicine, Palo Alto, California The choice of an experimental design for the National Halothane Study lay between a random- ized clinical trial and a population study of past data. This chapter discusses the merits and shortcomings of each, relative to availability and quality of data, simplicity of data collection, cooperation of collaborators, time, manpower costs, and simplicity of procedure. RANDOMIZED CLINICAL TRIAL The principle of randomization automatically tends to balance the treatment groups on all factors, such as age of patient and operative risk, whether the factors are known or unknown. This ensures a valid experiment and provides a sound probability basis for scientific inference. Ran- domization is thus a powerful technique and should be used when it is feasible. But infer- ences can be correct without randomization; i.e., studies of past data can provide a basis for in- ference. Inferences from nonrandomized studies are common, for example, in geology, and indeed in many areas of medicine (as in studies of smoking and health). However, in the absence of randomization, corrections must be made for the influence of extraneous factors; sometimes this is adequate to yield inferences that provide a basis for fruitful research, and the replace- ment of fancy with fact (even tenuous fact) has advanced many an investigation. A randomized clinical trial initially ap- peared to be more suitable than a population study for the National Halothane Study. But the randomized trial is approximated only with con- siderable art, effort, and expense. In a clinical trial of halothane versus other anesthetic agents, which would be aimed at distinguishing between extremely small risks, the numbers of institu- tions and patients required would be enormous, and the problems of coordination, cooperation, staffing, and training to ensure adequate ran- domization of all important factors, sobering. There is also the question of whether undetected discrepancies in randomization would be smaller than the undetected biases remaining in a study of past data after adjustment for relevant factors. And, finally, there is some doubt that the con- siderable medical resources required would be profitably invested in the detection of what might well be a very small lethal effect. In addition to such considerations of the feasibility of a randomized clinical trial, it is necessary to weigh the propriety of continuing to use a drug already under suspicion. If the sus- picion were strong enough to preclude, on ethical grounds, the use of one of the agents, a study of past data would be the only alternative. In the usual prospective epidemiologic study, treatments are not assigned to subjects ran- domly. For example, persons cannot be directed to maintain an assigned serum cholesterol level. When treatments can be and are randomly as- signed by the investigator, the investigation be- comes an experiment, and the length of the chain of inferences from treatment to outcome is re- duced. The inferences themselves are usually more absolute. A nonrandomized prospective study of halothane (i.e., without randomization of halothane tO'Some patients and other anesthetic agents to other patients) was never considered, because it seemed feasible to allocate anesthetic agents randomly to the large proportion of pa- tients for whom there would be no absolute agent of choice. Such a controlled prospective study of halothane would be appropriately called a "ran- domized clinical trial." The proposed National Halothane Pilot Study originally was designed as a randomized clinical trial in which patients scheduled for surgery under general anesthesia were to be selected at random for exposure to one of several anesthetic agents. Pediatric, obstetric, and emergency patients were to be excluded. The surgeon or anesthesiologist, however, had the option to ex- clude such controversial or critical cases as cholecystectomy and open-heart surgery, but only before the designation of an anesthetic agent. Each hospital was to submit each day's surgery schedule, explaining fully which patients were and were not accepted, and the reasons for their selection. The anesthesiologist was to have no knowledge of the agent to be used until he was to prepare for the induction of anesthesia. Pre- operative, intraoperative, and postoperative data were to be recorded. While attempts were being made to improve the protocol and a few institutions were collecting pilot data, halothane became widely publicized as a possible cause of postoperative massive hepatic necrosis. At that time, at least one participating 47
institution had decided to use halothane only in a small proportion of special cases, but most col- laborators were willing to cooperate in a ran- domized trial and continued to use halothane after the cancellation of the randomized study and during the course of the retrospective study. Thus, the ethical issue might well not have been an overriding factor in the choice of approach if the clinical trial had been the only way of obtain- ing data on which to base an inference. POPULATION STUDY OF PAST DATA A study of past data takes less time and manpower to' set up and conduct than a clinical trial. It was believed that the National Halothane Study could, in 1 year, collect information on a sample of all the general anesthetics admin- istered over a period of 4 or 5 years in 30 to 40 institutions. Moreover, the cost was estimated at about one-third that of a randomized clinical trial. In designing a cooperative study, it is diffi- cult at times to keep the size within the limits of feasibility. However, the temptation to develop an elaborate, comprehensive protocol is much greater in a clinical trial than in a study of past data. For example, participants in preliminary discussions about the protocol of a prospective study suggested that efforts be made to measure hepatic function in all patients, with a 6-week follow-up of each. Despite the laboratory and logistic problems, there would have been great reluctance to let the opportunities afforded by such a trial go by without extra laboratory study and clinical examination of each patient, which would inevitably have led to more missing data, more unfinished cases, less enthusiasm among the participants, and more difficulties in inter- pretation of the data - which could have led to failure of the study. We decided that a study of past data should be carried out primarily because it would permit a smaller workload and a shorter completion schedule, and because imbalances in the data would probably be no more serious than unde- tectable discrepancies in random sampling. This decision was justified further by the belief that a study of past data would provide information that would be useful in planning a randomized clinical trial if the latter proved essential or desirable. It was also argued that the only indica- tions of the toxicity of halothane were those men- tioned in scattered cases in the literature or re- ported to Subcommittee members in private communications. If a systematic study of past data disclosed similar incidents following surgery with other agents, or if the incidence was so low that it was not considered a public health prob- lem, the previous "evidence" might justifiably be dismissed or relegated to its proper perspec- tive and the matter dropped. In the study of past data, there are two com- mon methods of investigation: the outcome-retro- spective method and the historical population study. In the outcome-retrospective method, groups of cases are selected for study on the basis of their outcome. In studying massive hepatic necrosis after halothane, records of surgical patients who died with hepatic damage within 6 weeks of general anesthesia might be obtained and compared with records of patients who died without hepatic damage or with records of a sample of all postanesthesia patients, in- cluding survivors. In this approach, the com- parison group may be chosen by matching. The selected groups would be examined for differences with respect to the occurrence of the imputed causal factor. In a historical population study, cases are selected to represent the general population. In this instance, patients would be selected from among those who underwent surgery with general anesthesia. Then the anes- thetic groups would be compared with respect to outcome, e.g., death or incidence of hepatic damage. The outcome-retrospective approach was rejected for several reasons: (1) Incidence rates cannot be computed from data gathered retrospectively unless the magnitude of the source populations that have experienced several outcomes is known. (2) The sources of bias pointed out by Berkson (1) can be investigated and taken into consideration only with great uncertainty. (3) The outcome criterion plays a control role; an adequate choice could not have been made in this Study because the nature of the hepatic damage was not (and still is not) well defined. A study of past experience with halothane and other agents would ordinarily call for selec- tion of surgical patients who received halothane and other patients who received other agents, with a follow-up of the records to determine the number of deaths and the incidence of massive hepatic necrosis in each group. This method seemed feasible and informative. In each hospital, the patients who received halothane and those who received other agents would be counted. Furthermore, all patients who died in the hospital within 6 weeks of general anesthesia could be listed and separated according to anesthetic agent; the presence or absence of massive hepatic damage could then be determined on the basis of necropsy reports. The hospital chart for each patient who survived surgery would not need to be examined, but a sample of all surgical charts would furnish estimates of the numbers of pa- tients to whom halothane and other agents had been administered, according to age, sex, type of operations, etc. This would constitute a stand- ard population study of past data, except that the numbers of patients at risk in various subgroups would be estimated from samples, rather than directly enumerated. Such sampling has no sig- nificant effect in principle, but its impact on statistical techniques of analysis is considerable. In view of the over-all advantages, however, this 48
approach was selected for the National Halothane sign, and procedures adopted for the population Study, and new methods of analysis were de- study of past data are discussed in Chapter veloped to handle the statistical problems. II-2. The protocol for the pilot prospective study was still being developed when the retrospective REFERENCE approach was selected. It is appended to this chapter for historical purposes, in the hope that u Berkson, J. Limitations of the application of it could be helpful to those considering a clinical fourfold table analysis to hospital data, trial. Details of the protocol, experimental de- Biometrics Bull. 2:47-53, 1946. 40
APPENDIX TO CHAPTER II-l PROTOCOL FOR PILOT PROSPECTIVE STUDY* The great advances of modern surgery have in considerable measure depended on simulta- neous advances in anesthesia. Perhaps the most urgent need in the recent past has been for a potent, safe, and nonexplosive anesthetic. The introduction of halothane (Fluothane) in 1956 seemed to meet that need. This agent was intro- duced after studies of unprecedented thoroughness in animals and humans. Halothane has now been administered to an estimated 10 million patients and appears to have been associated with an ex- traordinary record of safety. However, in the first few months of 1963 several deaths caused by hepatic necrosis were reported in patients who had received halothane, and the serious possibility has been raised that the anesthetic may have been responsible for those deaths. It is not possible to determine whether any of the individual deaths reported were related to the anesthetic. Death may have been caused by coincidental infectious hepatitis or by other new drugs, some of which are also suspected of caus- ing hepatic damage and were received by some of the patients. The number of deaths, and hence the incidence, is small, particularly in compari- son with deaths from other causes in surgical patients. However, it appears that additional deaths from hepatic necrosis have occurred re- cently, and have not been reported in the litera- ture, but only by word of mouth. Other anesthet- ics, as well as halothane, had been received by some of these patients. The recent experience points to an urgent need to determine whether halothane damages the liver. That it might has long been suspected because of its chemical similarity to chloroform, which is known to produce fatal hepatic damage. The urgency to examine the safety of halothane is especially great in view of the remarkably favorable qualities associated with its use and in view of its current popularity (it is estimated that half the operations in the United States are performed with halothane). OBJECTIVES The proposed research plan is designed to answer the following questions: (1) Does halothane damage the liver? More pre- cisely, is there an association between the use of halothane and the occasional appear- ance of postoperative hepatic damage whether fatal or nonfatal? (2) If so, what is the incidence of postoperative hepatic damage after halothane, and is it greater than the incidence after other com- monly used anesthetic agents? â¢Adapted from working protocol prepared in August 1963; implemented in only one institution. (3) Under what circumstances does halothane damage the liver? Are some patients more susceptible to hepatic damage than others? For example, is the patient undergoing chole- cystectomy, with the possibility of some degree of pre-existing hepatic disease, more susceptible to the toxic effects (if any) of halothane? (This question and others like it are of less importance than those in para- graphs 1 and 2, and are less likely to be answered.) (4) If fatal hepatic necrosis occurs in some pa- tients after exposure to halothane, does halothane possess sufficient advantages to justify accepting the risk to the liver? For example, the fragmentary evidence available from retrospective studies in progress sug- gests that the incidence of fatal hepatic necrosis after halothane may be around one in 5000 or 10,000. An over-all mortality of approximately one in 1000 has been attributed to general anesthesia by Beecher and Todd, who found, in the same study, a death rate of approximately one in 75 from all causes in surgical patients. It is not unreasonable to expect that the favorable qualities of halo- thane, if they are as great as has been claimed, would be reflected in a decrease in mortality associated with general anesthesia or in total surgical mortality. (5) What is the over-all incidence of postoperative hepatic damage? It should be possible by comparison with the control group to separate at least the effect of halothane, and to obtain some information concerning the effects of other factors on postoperative hepatic diffi- culties. THE CHOICE OF AN EXPERIMENTAL APPROACH There are a number of serious problems in designing a clinical trial to answer the foregoing questions. The most difficult is the apparent infrequency of the complication in question. Although retrospective information is scanty, it appears most unlikely that fatal hepatic necrosis occurs more frequently than once in every 5000 cases. Information on the occurrence of nonfatal hepatic failure is, if anything, less satisfactory, but its incidence may be five or 10 times that of fatal hepatic necrosis. A second serious difficulty, and one that has hampered the collection of dependable case re- ports from past experience, is the apparently late appearance of the clinical manifestations of hepatic damage. Many patients have been dis- charged on the 5th or 6th postoperative day, to be readmitted as late as the 12th or 14th postopera- tive day with jaundice, usually thought to be a manifestation of infectious hepatitis. 50
It would be highly desirable to be able to recognize such patients before their discharge, i.e., in the early postoperative period, and it has been suggested that serum glutamic oxaloacetic transamiua.se (SGOT) might be measured on ap- proximately the 5th postoperative day, in an attempt to determine hepatic damage that is not yet clinically manifest. Unfortunately, data are not available that allow us to assume that meas- urement of SGOT will help in this matter. If a drug has a direct toxic effect on the liver, it is highly likely that this would become apparent by the 5th postoperative day, both clinically and by tests of hepatic function. But it is suspected that, if halothane does damage the liver, the damage is not caused in the same manner as that caused by chloroform and carbon tetrachloride (i.e., by a direct toxic effect), but rather results from some other mechanism, such as a sensitivity reaction. If that is so, and if hepatic damage can occur as late as 2 weeks postoperatively, there is no reason to assume that any test on the 5th post- operative day could predict it. Another major difficulty is the effect of bias on the interpretation of individual case material. It is agreed that the end points should be fatal and nonfatal clinical disease. The diagnosis of hepatic damage and, in fact, the simple observa- tion of lesser degrees of jaundice, is open to considerable subjective interpretation. A patient with a low-grade fever or a moderate increase in white-cell count, observed under indirect lighting in such a way that the color of the sclerae may not be easily determined, is much more likely to be subjected to a battery of hepatic tests and intensive study if he has re- ceived halothane than if he has received a drug that is not under suspicion. Even the interpreta- tion of fatal cases is subject to serious bias. The possibility of answering the questions listed above from an examination of past experi- ence seems unlikely. Examination of published reports alone is of no help, because the finger of suspicion has been raised; hepatic necrosis after halothane can be expected to be reported, where- as hepatic necrosis after agents not under sus- picion almost certainly is not being reported. An examination of the total experience of a single institution over the last 5 years, which has already been attempted, showed that anes- thetic agents are used in a highly selective way and one may expect to find that as many as 90 percent of patients who have undergone a par- ticular operation have received halothane. Another difficulty that has been encountered in attempts to derive information from card in- dexes in hospitals is that hepatic-related deaths are not likely to be adequately coded; e.g., in one large institution, during a period in which 160,000 anesthetics had been given, not a single instance of hepatic necrosis was cross-indexed in the system. The approach most likely to accomplish the objectives would be a prospective study comparing halothane with nonhalogenated agents, the two treatment groups being carefully randomized. It is anticipated that a large volume of patients will be needed, and, accordingly, a large number of institutions. It is therefore essential that the experimental approach be as simple as possible to permit coordination. In view of the apparently late appearance of hepatic damage in the suspected cases, careful follow-up will be essential. The recognition of hepatic damage will depend on careful clinical observation in the postoperative period, and in the 2-week follow-up. Furthermore, patients still in the hospital on the 5th postoperative day will undergo an SGOT test and, if SGOT is ele- vated, detailed studies of hepatic function. The usefulness of the SGOT procedure for this pur- pose is by no means clear and its evaluation will be an important objective of the pilot study. A urine sample will be taken from all patients to determine the presence or absence of bile on or about the 14th day. If bile is present, blood will be drawn for complete hepatic-function studies and the patient's physician will be informed of the possibility of hepatic damage. Many of the difficulties of experimental design in a prospective study of the kind con- templated can be related to the fact that (as originally planned) the choice of anesthetic agent would be known to the investigative team and to the physicians caring for the patient. At first glance it would appear impossible to conceal the choice of agent, and not necessary if the evalua- tion of results could be kept entirely objective. If the screening of patients could be safely en- trusted to the SGOT test, as we hoped would be possible, and if the evaluation of hepatic damage could be precisely related to disturbances in hepatic function, the need for a "blind" study might appear minimal. Even if this idealized approach were possi- ble, as now seems unlikely, one can anticipate many opportunities for serious error in an "open" study. The recognition of the need for a pro- longed follow-up and the emphasis on clinical evidence of hepatic damage make it imperative to use the blind approach. The reasons are sug- gested in the paragraphs that follow. It is planned that a nurse technician will follow the clinical course of each patient in the hospital and record specific information on the Study charts. Such information will be derived largely from the patient's record, but specific clinical observations will also be made, e.g., she will examine the sclerae for icterus. If she be- lieves the sclerae to be yellow, she will draw blood for appropriate hepatic-function tests, in- cluding measurement of bilirubin, and inform the internist member of the Study team of the possibility of hepatic damage. Inasmuch as highly competent physicians, in our experience, have been grossly inaccurate in the detection of lesser degrees of scleral icterus (as documented by serum bilirubin determinations done later in the 51
same day), it is too much to expect that a nurse can do better. It goes without saying that, if she has any preconception that halothane may damage the liver, her suspicion will be greater if she knows that halothane has been used. If it were possible (which is doubtful) to define the functions of the nurse technician in such a way that her observations could be un- questionably objective, a more serious difficulty immediately intrudes itself, namely, that the actual care of each patient will be in the hands of physicians entirely separate from the Study team. These physicians will make observations and order laboratory tests as they believe indi- cated. The patient given halothane would be likely to undergo tests of hepatic function earlier than other patients, and there is no question but that disturbances made evident by hepatic- function tests will influence the clinical evalua- tion of the postoperative state of the liver. In a retrospective study of the hepatic effect of anesthesia in patients who had undergone surgery for portal hypertension, we encountered precisely this problem: patients who had received one anesthetic agent had multiple hepatic-function tests, whereas those who had received a different agent had few tests. It was clearly shown that this contrast seriously biased the final clinical evaluation. It is proposed that the follow-up visit, pre- sumably at the patient's home, will include, and depend very heavily on, a urine test for bile. This would appear to be wholly objective. However, 100 percent follow-up will be impossible and the amount of effort made to obtain the urine speci- men will most likely vary with the degree of suspicion, i.e., with the anesthetic agent used. Furthermore, subjective bias becomes an even greater problem if there is any effort to evalu- ate the patient's general condition - fatigue, jaundice, tenderness over the liver, etc. In the evaluation of deaths, one would at last seem to be on firm ground. The microscopic sections of the liver will be evaluated by a panel of pathologists and specialists in medical aspects of hepatic disease who will know nothing about the clinical course and whose interpretation can be depended on to be completely objective. How- ever, it is clear that clinical events will play a very important role in the evaluation of each death. In any final analysis of mortality, it will be desirable to eliminate patients who died from other obvious causes; the physicians most capa- ble of such evaluations are those directly con- cerned with the care of the patients. Again, the bias that inevitably accompanies knowledge of the anesthetic technique can negate any efforts to make objective judgments. A very valuable kind of information that will be particularly important in any attempt to evaluate effects other than those on the liver is the separation of deaths into "patient's disease" "surgical deaths," and "anesthetic deaths," as is common in many teaching departments. If the anesthetic death rate is significantly affected in either group, that can markedly alter our final interpretation of the significance of any anes- thetic effects on the liver that may be detected. The separation of deaths into these categories, however, is highly subjective, with a high prob- ability of variation in interpretation. These evaluations must be made by the physicians on the scene, as emphasized by Beecher and Todd, and would clearly be very sensitive to bias. One might question whether the choice of anesthetic agent can ethically be concealed and whether such concealment might prejudice the care of the patient. This need not be the case if all patients are treated as though they had re- ceived halothane, which is not unreasonable, inasmuch as it is agreed that each patient is suitable to receive halothane before he is ad- mitted for the Study. One might also wonder whether it is even possible to conceal the choice of anesthetic agent. In actual practice today, the surgeon rarely knows which agent is being used, and is usually grateful to be able to turn over everything in this area to the anesthesiologist. To be sure, the recent halothane scare has forced him to be more conscious of which agent is used. If the surgical team is to remain blind to the choice of agent, the surgeon must be willing to accept this condition. His very willingness to participate in the proposed Study implies his agreement as to the importance of the project and its objectives; thus, he is presumably willing to accept any reasonable and necessary provisions. It should be clear, of course, that whenever it becomes essential to identify the anesthetic agent, the code may be broken; but that is not expected to be common. If an explosive anesthetic agent is to be used in the control group, it will be difficult to conceal the fact that halothane is not being used. To correct for this in institutions where explosive agents will be used in the control group, one might simply suggest that all patients, whether halothane or control, be treated as though an explosive agent were to be used. On the same subject, there is something to be said for avoiding explosive agents entirely in the control group, in that, particularly with the increased use of electrocoagulation for control of bleeding, the choice is usually between one nonexplosive technique and another, and a com- parison between halothane and a technique in- volving intravenous drugs, particularly in terms of their effects other than those related to the liver, may be more meaningful. It is clear that, if the identity of the anes- thetic agent is to remain unknown, conduction anesthesia cannot be accepted as control tech- nique. SAMPLING PLAN It is to be hoped that the results of the Study will be applicable to as wide a group of patients 52
as possible; by this token, it is preferable to make as few exclusions as possible. If, for ex- ample, the Study patients were limited to good- risk young men 20 to 30 years old, and no dif- ferences in hepatic effects were found, it would be hard to claim that this result could be ex- pected to apply to all surgical patients. Special attention has been focused on the possible relationship between biliary tract opera- tions and hepatic necrosis after halothane anes- thesia. Thus it might appear to be particularly desirable to know whether this group is espec- ially sensitive to any possible effects of halothane on the liver, but one might also be reluctant to subject such patients to halothane if it is expected that there is an increased risk. One might omit such patients initially, and include cholecystec- tomies later if no differences were found in other patients. Inclusion from the outset, however, has a very special advantage in that if the incidence of hepatic difficulties after cholecystectomy is greater, we are likely to have an answer earlier than if these patients are excluded initially. There is, in fact, good justification for their in- clusion: preliminary conclusions from on-going retrospective studies suggest that hepatic necrosis appears no more frequently after cholecystectomy under halothane than after cholecystectomy under other agents. It is pos- sible that more information on this group of patients can be obtained soon and can help to form the basis for their inclusion in the Study. It would appear to be appropriate to include cardiac surgery, unless an individual institution considers halothane essential to safe anesthesia for these patients (as is the case in many medi- cal centers). Children of all ages may be in- cluded, but blood samples for SGOT tests will be omitted below the age of 10 (possibly 8) years. Evaluation of such children will necessarily be clinical during the early postoperative period, but the 14-day urine test will be included. Because the emphasis on the SGOT test for screening will be less than originally planned (or, in any event, because its inclusion now is still on a pilot basis for later reconsideration), there appears to be no reason to exclude patients who may be discharged before the 4th or 5th day; of course, such patients should be seen on the 14th day and a urine sample should be taken then. Obstetric patients and the newborn, as originally planned, will be excluded from the Study. Emergency procedures are excluded, simply because of the technical problems involved in the collection of preoperative data, the drawing of blood samples, etc. SPECIFIC PROCEDURES Each institution will select its subjects from the operating schedule on the following basis: (1) Emergency patients, obstetric patients, and neonates will be rejected automatically. (2) The following categories will be subject to local interpretation: (a) noncooperating surgeon, (b) noncooperating anesthesiologist, (c) noncooperating patient, (d) jaundiced patient, (e) patient undergoing surgery for gall bladder disease, and (f) patient who will not tolerate halothane or other inhalation agents. Every attempt should be made to use patients who will have a hospital stay of 5 days or longer. Therefore, before final selection, patients should be rated as follows: if the stay is to be 5 days or longer, rate as group 1; and if the stay is to be snorter than 5 days or the patient is under 10 years old, rate as group 2. Include group 1 patients in pref- erence to group 2 patients. If there are more in group 1 than one could normally handle, select them alphabetically according to anes- thesiologist; if there are too few in group 1, include patients from group 2. When the anesthesiologist designates a Study patient according to the definitions and restric- tions in the protocol, he must code all possible open boxes of the National Halothane Study Form (NHSF) (Fig. 1) before the random selection of the anesthetic agent. Presumably, this would be done during preoperative rounds. Just before the induction of anesthesia, the agent is identified. A decision to exclude a patient must be made before identification of the anesthetic agent. The procedure for selecting anesthetic agents is as follows. Random numbers will be taken from the Statistical Tables for Biological, Agricultural and Medical Research;* these tables allow for 15,000 serially selected random num- bers for assignment of agent. Serially numbered envelopes with cards will be prepared from random numbers designating halothane or non- halothane at a central office and distributed to participants. When the envelopes have been dis- tributed to the participating institutions, the small envelopes should be assembled and a signoff sheet placed at a location convenient for the attending anesthesiologist. He should pick them up just before administration of the anes- thetic. It should be stressed that all open boxes on the NHSF are to be coded before the anesthetic selection is made. The number on the selection envelope should agree with the number on the assignment sheet and with the serial number of the NHSF. Forms will have to be stored in a place convenient to the area of preoperative rounds. A register should be signed, indicating for each serial number the patient's name, the time the card was picked up, and the anesthesi- ologist's name. â¢Fisher, R. A., and F. Yates. Stat.istir.aJ Tables for Biological, Agricultural and Medical Research, p. 126 (5th ed.) London: Oliver & Boyd, Ltd., 1957. 334-553 O-69â5 53
If, after one has selected the agent card, the case is canceled or the choice is not carried through, a note of explanation should accompany the appropriate NHSF. If a patient is excluded from the Study after the selection of the anes- thetic, an after-the-fact deletion form must be completed, including the serial number of the NHSF and assignment packet, the patient's name, the anesthesiologist's name, the institution code, and statements of whether the case was canceled (and, if so, whether the cancellation was related to the anesthetic choice) and whether a new indi- cation became available after assignment. A copy of each day's operating schedule must be prepared and submitted with the Form One, Part One (see Fig. 1). Each case should be marked "Study," "Excess," or "Excluded." "Ex- cess" refers to patients found suitable for in- clusion, but in excess of needs or capacity for that day and therefore omitted. "Excluded" cases should be so indicated and the explanation for each case given. The operating schedule and the completed Part One should be mailed to the central office at the end of each day, with the signoff sheet for that day and other data if neces- sary (e.g., deletion forms). After preanesthetic medication and before induction of anesthesia, a blood sample is to be obtained for the determination of SGOT level. This value serves as the preanesthetic control, SCOTi. The induction of anesthesia is left to the discretion of the anesthesiologist within the definition of the protocol (i.e., halothane as ordinarily used, or other agents as ordinarily used). Among the anesthetic techniques excluded from the Study are: other halogenated agents, regional anesthesia (regional anesthesia in combination with general anesthesia is accept- able), local anesthesia, and hypnosis. At the end of the anesthesia, the remaining open boxes of the form should be completed. This satisfies the anesthesiologist's obligation for completion of the form, because it is planned that a nurse technician, who will not know the identity of the anesthetic, will record postoperative data on a separate form. The original copy of the anesthetic record should be removed from the chart, because that would identify the anesthetic technique. The carbon copy should not show the anesthetic tech- nique and may remain on the chart. Thus, we maintain a blind approach. The original of the anesthetic record should not reappear on the chart until discharge of the patient from the Study (not necessarily from the hospital). A blood sample is to be obtained on the 5th* postoperative day for SCOT determination (SCOT2), as well as white-blood-cell and differ- ential counts. If the postoperative SCOT level is â¢May be drawn on the 4th day if the patient is likely to be dis- charged before the 5th day, or on the 6th day to miss a week- end, or if for some other reason the 5th day is missed. higher than 75 units, detailed hepatic-function studies are to be carried out, including serum bilirubin, alkaline phosphatase, thymol turbidity, cephalin flocculation, and punch biopsy of liver if possible (and perhaps prothrombin time). Follow-up study is to be done as indicated. For patients who have multiple operations, the same random selection is to be used for sub- sequent procedures. Patients exhibiting hepatic damage after anesthesia will be designated as reactors. In such a patient, the choice of anes- thesia for subsequent operations is left to the discretion of the attending anesthesiologist. However, if he thinks that the patient can be re- entered into the Study within the bounds of the protocol, this should be done. If one does not include the patient as a randomly selected sub- ject, then the special instruction should be fol- lowed. Hepatic function will be studied after each subsequent anesthesia, irrespective of the agent used. (Any patient readmitted to Study for a second or later procedure should undergo pre- operative and 5th-day postoperative batteries of hepatic-function tests, as listed in the previous paragraph.) RECORDING OF DATA Data available before, during, and after operation: The following data are to be recorded by the anesthesiologist for subsequent statistical analysis: (1) Preanesthetic: Name, unit number, date of birth, sex, physical status, number of anes- thetic exposures within past year, preopera- tive diagnosis, significant preoperative com- plications, drug therapy and duration, alcoholic intake, history of recent blood transfusion, preanesthetic medication, and SCOTi level; specific statement of jaundice, recent or past; dietary habits, intravenous or oral intake, including specific statement of caloric intake during 36 hr immediately pre- operatively; history, including biliary tract disease, pancreatitis, malnutrition, weight loss (specify), obstetric history (specify complications), allergy, drug reaction, dental work, and recent congestive heart failure. (2) Physical findings: Jaundice, edema, ascites, spider nevi, liver size, splenomegaly, cyano- sis, clubbing, pulmonary congestion, venous congestion, fever, and hypertension (specify). (3) Anesthetic: Primary and secondary anes- thetics, other medications, and duration of anesthesia and operation; respiratory, circu- latory, and other complications; fluid therapy and blood transfusion; controlled, assisted, or spontaneous respirations; concentration of halothane (if used) and type of vaporizer; halothane freshly opened?; lowest blood pressure (specify); and use of vasopressors. 54
A specially trained nurse will record the following on Form One, Part Two: (4) Postanesthetic: Highest temperature, fever, chills, rash, nausea, vomiting, upper abdominal pain, anorexia, hypotension (specify lowest systolic pressure), duration of hypotension, use of vasopressors, and jaundice (if present, specify whether ac- companied by pruritus, dark urine, light stools, ascites, edema, stupor, or hemor- rhage). The patient will be instructed on discharge to send to the hospital via special delivery mail a urine sample on the morning of the 14th post- operative day, and will be given a special con- tainer and mailing envelope for this purpose. The nurse technician will telephone the patient on the 14th day to remind him to send the sample, and in addition will question the patient regarding those signs and symptoms listed above, as postoperative follow-up. If the urine has not arrived by the following day, a second telephone call will be made to try to obtain it. If this is not successful, a house visit may be nec- essary. The internist member of the team will fol- low patients if: (1) the SCOT2 is above 75 units; (2) hepatic damage is suspected by the nurse technician or by ward surgeons; or (3) urinary bilirubin is elevated on the 14th day. It is im- perative that both the nurse technician and the internist remain blind to the anesthetic pro- cedure at all times. Necropsy data: If a necropsy is obtained, the pathologist should remain blind to the anesthetic agent. A copy of his pathology report will be sent by the nurse technician, a secretary, or another locally designated person to the Collection Center. In addition, necropsy material will be requested for submission to the Pathology Review Board. It is urged that in all cases, whether or not permis- sion for necropsy is anticipated, a postmortem needle biopsy be obtained immediately after death (the time between death and biopsy should be recorded). A report on the findings of the needle biopsy should be sent to the Collection Center; again, it is hoped, the pathologist will be blind to the anesthetic agent. Laboratory data: The clerk will record pertinent laboratory data. All patients will have a preoperative hemoglobin or hematocrit measurement, white- cell and differential counts, and SCOTi test. All patients will be studied for SGOT and urinary bilirubin on the 5th day, if still in the hospital. All patients will be studied for urinary bilirubin on the 14th day. If the SCOT2 is elevated or if either of the urine bilirubin readings is elevated, a liver profile is in order. If the clinical course of the patient is such that the internist is called to see him, then the results of the laboratory tests should be recorded. Data on patients undergoing multiple operations: Any patient reoperated on since the begin- ning of the Study should be suspect. If the history of previous inclusion can be obtained, it should be noted on Form One, Part One that the patient is a repeat under previous inclusion. He can be categorized as reactor or nonreactor. "Reactor" (as distinct from "hepatic dam- age") will have to be defined most specifically, and should probably include elevated SCOT2, clinical jaundice confirmed by serum bilirubin above 2.0 mg percent, and bile in 14-day urine. A repeat should have a liver profile done on day 1, as well as on day 5, whether he was previ- ously a reactor or not. For the purpose of this Study, "hepatic damage" will be defined on the basis of elevated SGOT 2, clinical jaundice shown by a serum bilirubin above 2.0 mg percent or disturbances in other tests, and bile in the 14-day urine. HANDLING AND GENERAL ANALYSIS AND INTERPRETATION OF DATA The statistical evaluation of the data will be based on the correlation between the laboratory findings and the total clinical picture. It is hoped that a cutoff point can be defined after pilot data are collected, so that patients may be divided into specific categories. In the presence of hepatic damage, other possible etiologic factors, such as history of blood transfusion and therapy with drugs of hepatotoxic potential, may be implicated. In a patient whose SCOTi is elevated, the significance of hepatic damage, if present, may be most diffi- cult to assess. Such cases will be treated sep- arately in the final analysis, as will patients who are reactors and who nave multiple anesthetics, including those who enter the Study more than once. Handling of forms and data processing (1) 1st postoperative day - Send via airmail to Collecting Center: (a) Form One, Part One, completed, (b) complete surgical schedule, with descrip- tion of exclusions, and (c) sign-off sheet and after-the-fact deletion form. (2) Nonreactor 15th postoperative day - Send Form One, Part Two to Collecting Center via airmail, and retain Part Three. (3) Reactor, nonfatal - Send Form One, Part Two when discharged from Study by internist. (4) Reactor, fatal - Send Form One, Parts Two and Three with necropsy report to Collecting Center; send necropsy material to consult- ants for pathology review (blind). 55
C enter handling of data (1) Secretary receives daily report from each institution, checks for errors of omission, and sends immediate request for correction on separate sheet. (2) Registration and logging of reports: (a) number from each institution, (b) number of errors from each institution, and (c) weekly reports to Dr. Forrest. (3) Entry of diagnosis by international code.* â¢international Classification of Diseases. 1955 Revision. Vol- umes 1 and 2. Geneva: World Health Organization, 1957. (4) Card coding and printout to check for errors. (5) Filing of data cards. (6) Monthly report from each institution to Dr. Seeley, Dr. Bunker, and Dr. For- rest: (a) total number of entries, (b) number completed and coded, (c) number of errors, and (d) suggestions. Coordinator will make frequent visits to participating institutions for indoctrination, lab- oratory check, adherence to protocol, adherence to double-blind method, troubleshooting, and discussion of reports. 56
NATIONAL FLUOTHANE STUDY DATA FORM FORM ONE - PART ONE PATIENTS NAME. ANFSTHFSIOI OniST. TECHNICIAN CLERK- Mi 1 1 M M M 1 J U 1 1 U MOBIYAl MY MOUTH Y«. HOSI1YAI CHAIT KUMMt AM MX INTAKE (EC COM 14* IF X> ALCOHOUC HOI ENT Ml Of UT POOD OtY WITHIN AMSTHinC MUG THMAfT UfOSUUS MtUIO 0 Ph*notn,«iin.i 00 l^ntie«n>Jt«nli 1.1 RuBiM.-.. 2-Men.mine O»- 1-2 F!,Mt-.n. rfDi. lnh,b.to,. 2-1 »r mora 2-Anlibiotici I-NOM IliHn 1 1-toW u u u u u :~ l;i! M M.ll.d... 'Hkl 0-Nt'r'm.*' VMIII 2 i .,, l,,.q I/Mb 2-FW/R**. J An.lq.«,r 4-At«r«ctic 5> plwll i-l+J + l MWAYrm J.UNWCI HOOD lUMMSION WITHY 2 M.d.r.1. 4 V rUNtmiON lIACnOH HAHTS ' """ , > No Writ>. 4 MM. ⢠- NMM 1 - Good bM»i»lm»l 7 - V 0 - l Mo.'k. ON. 1 . F«bHI« IF.., 4 -il.qkt ⢠. V i . 4 MO. Ik, I r., 2 - OHio IFo,, 1 - uoj., .1. t . V 4-1 «â¢Â»-"* Sl.r.oJi'6 7 2 pi.. > V.,hi 4 Poj, l,,»g HlMtkcM S Onl An , d.*bitl, 1-9 ion. t-Anti-Tuborcirlcr ⢠! 1 4 t'-'ii., VMb fwDftY vMk. H» o... 100 'Mb t-IV 1.>1 100 4-1 or mor. ⢠Ar,, M.,.bcl.U. i 9 M«llltl 4 - 'j «,», . f . IJM.Mk, 7 P0« IV â¢â¢â¢r 100 K,"oth.n. 7-2 Comb.n.. â¢ti«ri t4». Tk., On Ep.i.d, IPO I IV kntfcn NO 1 c-...- f A-.itk.hc. Mil 1 1 I 1 U U U U rnvioui MUMMY ANBTHmC AMNT nwu«cnoN HttOKRATTVI DMCHOSIS PHI-OrOATIVl OUftHOW HfTSJGCkTiVt 535*555 MCUniON SM C"d* lor 69. /I 0 No; T Y.I 1 u"' I.YM! LLU U U U U M L j U U U MHMM DWAtlON CONCMY1A. MCTHOO Of HOYHAM SYtTM R.UOTH MMH4> Of TWHmtAlY YMtHATION VArOUO* YYfl II Wl AMinaiU AHOTHOU A«NT AM PRIMARY SECONDARY TMTIAHY AMNYS OLAXANY ⢠-None 00 9k,, U.J., 5 , Spool. ..ea, - Okk, . «l 1 Op.. 1 Not J it 1 - HW«MMM d 2 - CYJapo «-.d * ' Eitlor ..di 1 - otk., H A,, -i 5J1.â¢.. "''- ⢠MB 1 *-j'S»>ii . >9kr, -2.11o2.1 . Otk,, Oik., N». H»>.,l . A.IH.II. . Mhn. .-iiol . H,p.l...... - Hk,, .11 l.11 I! I - SHnU - 7.«bn. -ll.ltalt [ | [â j u u ' â¢-"»- -"-'-« UV«IK)«Y 0»UYm HT. lUNIIVHON C..r 9 -,, .O*»r24.l ⢠- No ⢠- H«.d Hnd N.cl | . AAdl.ll. *l t .'r»il.ntcn..l ,Pl.ni. I.TH 1 -TlwMic (l«tr« 4-Uw«rUr "4> rY Tr.et. P*rin.otn D a JE.Ir.ikor.cie 7.t.tr,m,l 3 Upp.r AMWfllMl 1 lnl.«p.r ) f . IntrHrM 4 - U.> r AMMnlMl 1 litr*p«r ) f . OtiMr to irf M 1 1 1 1 1 1 II M I n n ED «*°'i WOT, MY OT FOUOW.UC ITVMI D,,.,i l-d.r.ci PMtm â¢M C « >W> HOi Câ¬fk * WK I4KiT. | . »..i,.l J. H-gk SSOT, 1 . Oik., MJtt â¢ta n Ii i rtxrft tm D D D D C2«Htorf ranifaAI MSMMITIOK t .. l-lfc I-Y« 1 -»..«, r,Ulrf 1. li>o dilMl. |.I «»«k 0 i l o , man 2 . M,, p., , ,,i â«..â,. . . . d, r , . « r " D | 4. A,...^rt»l ,«.,.,,.â M"ll! ...,).,... 1 ^,id..l d..w 7-7..^. t-t -..1 , Figure 1.âNational Halothane Study Form One. This form consists of three parts, which are virtually Identical (the pa- tient's name appears only on Part One). Part Two is designed for ease of punching to Hollerith cards. Pan Three Is stiffer, for use as a permanent record of the Study. 57
NATIONAL FLUOTHANE STUDY DATA FORM FORM ONE - PART TWO ANESTHESIOLOGIST: TECHNICIAN CLERK:. LLJ I I I I II I I I I I I I I U U U U U U LLLU U 70 21 22 2) 14 U U U U U 24 17 2* 10 u u u u u II J2 )) 14 )5 16 17 II JV 40 41 42 41 44 45 4* 47 41 49 50 51 52 51 54 55 St 57 51 UUUUUUUUU U M 6i 67 *J 64 AS U u U U 7; 71 7« n D rm rm n 57 100 101 101 10! CM IDS n a n n n n »7 'I D D 107 101 104 MO IH Hi 119 120 121 122 123 124 125 12* 127 121 124 110 11l 112 111 114 US '1* 117 HI I If I 141 141 Ml 144 145 I4* 147 141 144 ISO 151 151 151 154 1SS 156 ''- "'* IS9 l Figure 1 (Cont'd) 58
NATIONAL FLUOTHANE STUDY FORM ONE - PART THREE DATA FORM ANESTHESIOLOGIST: TECHNICIAN CLERK:. II 1 1 1 1 J LI II U U U I I HOSPITAL DAY MONTH Y«. HOSfITAL CHAIT NUMIH AM SIX INTAKE RECENT PtE-OF AHCSTHtTtC DftUG THfftAFY fHl-MKI COM U4 2f BC'J, POOD fXfOSUUS 0 Non9 ALCOHOLIC HliTOBY WITHIN 0 PK,r-oth,8I,n.. l14 Hn.) 00 l-AMieen>ukant. 1 Si-l ' U U U U U 0- Sight Non9 1-1 FhMtli«iM 2-Monemine O»i- WMk s !-,,,-,. 2-2 Fiuattiana dm* Inhibitor 2 B.I »dor). ChrwOc WMki 0 Norm.! 2-1 or mo.. J Anr.b,oriei W,.ki 2- F ,ir If r.g Huatkana 4- ( AMboKc ) 4 At...clie K-l pl.ii2 1 - Slight W«*b 1-PoOr/Rtf. 4-1 non- St.r.odi * >2^1 W»«ki 4 Poor , l.r.fl. Huoth«M 1-Or.tl Antidi.b*tie M P1"> 1 W*«ki K-2 â¢Â«Â«- 4*Anti-Tub*rcuUr 1234 MIPATim JAUNDICE HOOD TtANSnmOM TftANSFUSION DUTAIY HACTION HAHTS 2 - Modvrat* N.» V.-.h.-, 4 M., ⢠-Non e 1 - Good l«>nmrrt«nt WMkt PflrDwr Fluo,h«ri, 7.Su.toMm.oVi f -Other 0 - J M*- .«.. Q N« 1 - ' .- . , . 1 . F.ir 4. Sl.qM W».ki 5 IV ,Dt, 100 4-1 Or mor. â¢-AMi-M.toboJit* 1 ).»M«. >thi 1 - Ya« 1 - Ot.«r 1 . foor I . Mod.. . t . W».k. 4-tV 1... tOO "0". f-ThiMid*. 7 I . - i IV f luoth*n. i - 9 Mai thi 4- S9>«r* â¢*or (00 7-2 Comb.na. ? - 12 Mi- .|4. ⢠PO I IV .,.»n Mor* Th.n OIM Epii«d* 100 ' »t.on «-Otn»r f An.ith.tis Unknown U U U U U rimous PIIMAIY ANfSTHinc Aoon DISFUNCTION uvn INCLUSION Soo CoJo lor M-71 Ftl.O««ATl>f DIAGNOSIS PttOraUTTVI OIA&NOSIS Mt-OKItAttVE MA6N" 3? 0-No: 1-YM I §J»o: 1-YM LLU u u U U u u 'Ti u u U MUGS DURATION COHCIN7RA METHOD Of PtOTHAM SYSTW or TIOH HIM ART v FLUOTHANf PRIMARY SECONDARY TERTIARY â¢ILAXANT MIMNG ANESTHESIA AMNT ENTHATiON VAfORLUD TYFI DtfOSUU AGIHTS AMSTHfSIA -NoM .1hn. -Li»d«r.S 1. Sp«>it*neoui - Ohio - 98 - Fluot.c 1 -Op-M. 1 - Not Juit 1 - Huo'^nr | . EihYl4jn9 d I., i-u, .ââ,,â, V .H«t l.tkr». -.*»⢠1 - o Of«r.-d 2 . CYclopropana 4 - l»rb,turat* SucclnYlckolin. - AUwti c 1 - 2.tbn. . I.I to I.S Anlii.d Cop. K«HI* Op*n Drop 2 - S«MM-Clo,*d ^ -Ju,tOp.«d I-«*â¢' 7-AMl9..k 4 - NHraui Oiid* 1 - Oth«r H A,I*nt So'lom*!*""'"" . !.,! .>. .1. . Mb*. -l^i.2 I' ControlUd Sc-.in V*rnitrvl ..nloqu,non,um 4.f kn. -2.1 to2.S MB » . Oth*r Non-H Afl*nt Oik., . tiorold * '. K, -|.I to Il 7.9 kn. - 12.1 to II u u U U 1.4 kn. . ll.l to 24 m HOfSY - ' i - - - f krt. - O>or 24. 1 U Ofl«ATI», srra TtANSFUsraN SniAL NUMill D -ThorMJa l-ii » I - Abdom,n.l Exroporilooool/HMi. 1 - 4 - Lowor UriMrY TrMt, PorlMum 2 - ? . I .i-..-.,iâ, Upp., Akdom., . - I ..,..â., I t . lr,hKr«I.l IIMr.port f.OHMr D D a DAT OF «VOI, D i m n an FOLLOW-Uf STUOKS D GINIiAL DISfOSITION -Yat 1 D*«th r*Ut*d to I..- dii**i« 2 - D*«th unr*lat*d to li>*r di.*«i 1 - Ali>* portial r«ce>orlr. a>idant 4 - Ali>e, pBtial -.,o..- r, no D t-IOotn 1-1 -r.k 1 - ...« 4-4 woohi I.l »io.i ⢠V .Mil, Figure 1 (Cont'd) 59
CHAPTER II-2. THE PROTOCOL FOR THE NATIONAL HALOTHANE STUDY (Including a Discussion of the Problems of Data Collection) William H. Forrest, Jr. Stanford University School of Medicine, Palo Alto, California J. Weldon Bellville Stanford University School of Medicine, Palo Alto, California Byron W. Brown, Jr. University of Minnesota School of Public Health, Minneapolis, Minnesota John P. Bunker Stanford University School of Medicine, Palo Alto, California When attention turned from a prospective study to a study of past data, an entirely new protocol had to be written and new forms had to be designed. In developing the protocol, it was necessary to determine: (1) the kinds of data to be collected, (2) the number of cases to be re- viewed, and (3) the period to be covered. To ex- pedite the collection of a large mass of infor- mation, forms for transfer of data to Hollerith cards had to be designed. And finally, to maxi- mize the possibility of the successful use of the protocol and forms, questionnaires were pre- pared to evaluate the pilot collection of data. DEVELOPMENT OF PROTOCOL Selection of Variables The most difficult problem in the develop- ment of the protocol was the selection of the var- iables on which data were to be collected. To en- sure that large-scale data collection would be kept simple and manageable, all pertinent vari- ables were listed and then rated according to: (1) need--the value of each variable in an- swering the null hypotheses, (2) probabilityâthe probability that the in- formation was in the chart (missing data would create difficult problems in sub- sequent statistical analysis), and (3) ease--the ease with which the variable could be found in the chart (in a review of 50,000 to 60,000 charts, simplicity is vital to ensure accuracy and complete- ness). From these ratings, the variables of highest aggregate score were selected. Except for phys- ical status and duration of anesthesia, both of which rated high in need (1) and ease (3) but low in probability (2), the plan for selection of vari- ables was rigidly followed and those which rated high in all three categories were chosen. The variables selected were age, sex, preoperative physical status, duration of anesthesia, anesthetic agent, operation, previous operation within 4 years, previous administration of halo thane within 1 year, necropsy, and presence of massive hepatic necrosis. Determination of Sample Size Considering the possible low incidence of massive hepatic necrosis (MHN), it was esti- mated that approximately 1 million cases of gen- eral anesthesia should be studied. Even with so large a number, differences between anesthetic agents might not be detectable if the projected incidence of MHN, one in 10,000 anesthetic pro- cedures, prevailed. (This estimate was based on the over-all use of halothane and the number of cases reported either in the literature or to the Committee.) Period of the Study Halothane was being used extensively as early as January 1959, so that was chosen as the starting month for our investigation. December 1962 was considered an appropriate cutoff month to avoid collecting data from the "biased era" of MHN reports. The Pilot Study A general format, including forms and a questionnaire, was prepared to abstract the rec- ords of all deaths and randomly selected cases and to screen necropsy records for MHN. The forms were designed for ease of completion and decoding to Hollerith cards; the questionnaire was designed to evaluate (1) the sources of in- formation, (2) the problems of data collection, (3) the clarity of the forms and protocol, (4) the time expended to collect the data, (5) the types of employees needed, and (6) the anticipated cost. Following this format, 38 volunteer institutions each submitted pilot data for 1 month (December 1962) to the Data Collection Center. The informa- tion gathered was used to make recommendations 60
to the National Research Council for contractual arrangements, to improve the forms and protocol, and to delineate problems of protocol interpre- tation that could be discussed during site visits. It was estimated from a time analysis that each institution could analyze the data for 1 month in 1 day, and each principal investigator was to adjust his contract and request personnel to handle the workload at that rate. A few insti- tutions could not adhere to the protocol and with- drew; others required special considerations, but were allowed no major deviations from basic protocol. Thirty-five institutions were granted contracts and used the following protocol and forms to collect data. PROTOCOL Introduction A retrospective study is one based on re- corded experience. Its main advantage is its economy of effort when the event of interest is infrequent. In the present study, the event of interest is death following surgery conducted under general anesthesia. Consider a hospital with a 2 percent death rate in the 6-week post- surgical period. In a forward or prospective study, it would be necessary to analyze records for 12,000 cases to find 240 deaths. In a retro- spective study, the records of 240 deaths can be studied; even if coupled with an equal number of controls, the total number of records to be ana- lyzed would be only 480, not 12,000. The retrospective method does, however, have some disadvantages. First, some of the desired information may not have been recorded. Second, the two groups under study, test and control (in this case, halothane and nonhalothane), may not be comparable with regard to all im- portant variables other than the anesthetic agent. This might vitiate conclusions based on com- parison of the halothane-nonhalothane ratio among the deaths with the over-all ratio in sur- gical cases, and whether the two groups are comparable cannot be learned by comparing only the deaths. Suppose, for example, that half the halothane deaths were males and half the non- halothane deaths were males. This would tell us nothing about the risk of a halothane death or whether the risk varied by sex. If 30 percent of all halothane surgeries were males and 50 per- cent of all nonhalothane surgeries were males, then a 50 percent sex ratio among the deaths for both groups would imply that halothane caused an adverse reaction in males. In short, the comparison of the anesthetics among the deaths can be misleading unless supplemented by the study of anesthetics among representative cases that did not result in death. The two disadvantages are being dealt with as follows: 1. The pilot study for the period December 1962 showed that the information that is required is almost always in existence and is generally accessible. 2. To identify any tendencies to use halothane in more severe cases, less severe cases, particular operations, etc., a random sample of all surgical cases is required; such a sample can not only disclose these tendencies, if present, but can also allow adjustment for them; the pilot study dem- onstrated that the sample drawn for the test resembled the random sample closely, and for this reason the number of ran- domly selected charts to be abstracted per month for the 4-year analysis has been reduced from 50 to 25. Objectives This study is concerned with cases in which general anesthesia was used during the period 1 January 1959 to 31 December 1962, and has two objectives: 1. To compare the incidence of massive hepatic necrosis within 6 weeks of gen- eral anesthesia with and without halothane. 2. To compare the over-all incidence of death within 6 weeks of general anesthesia with and without halothane. Experimental Design: General Information There are a number of serious problems in designing a clinical trial to answer the questions we are asking. The most difficult is the apparent infrequency of the complication of interest; on the basis of only scanty retrospective informa- tion, it seems most unlikely that fatal hepatic necrosis occurs more frequently than once in every 5000 cases. Thus, a very large number of surgical cases must be committed to the study in order to get even a small number of cases of fatal hepatic necrosis. A multi-institutional pool- ing of data from 1959 through 1962 should pro- vide an adequate number of cases. Two facets of the investigation are impor- tant. The protocol requires that all surgical deaths, defined for the purpose of this study as deaths that occur within 6 weeks of surgery, be recorded and the charts abstracted. It also re- quires that charts be abstracted of a random sample of all surgical procedures. By abstracting all surgical deaths we can obtain information on the biases that affect the distribution of necrop- sied deaths. For example, the acutely ill patient after halothane anesthesia who develops liver complications may be more likely to have a ne- cropsy than the patient who did not receive halo- thane. Inclusion of all surgical deaths constitutes a built-in check on this possibility. Random sam- pling of all procedures will furnish baseline data on the use of halothane and nonhalothane anesthe- sia in poor risks compared with good risks, in one type of surgery compared with another, etc.; only then can valid conclusions be drawn from data concerning deaths. 61
The procedures are designed to yield infor- mation on the following in each participating hos- pital: (1) total cases of general anesthesia with halothane and without halothane, (2) total surgical deaths, (3) total necropsies, (4) types of patients given halothane and not given halothane. (5) types of patients necropsied, and (6) massive hepatic necrosis with halothane and without halothane. Summary of Procedure In order to avoid the removal of too many charts from the record room at one time, as well as to allow us to look at data on a month-to- month basis, the steps described below should be followed month by month, starting with De- cember 1962 and working back to and including January 1959. The data should always be kept separated by month. STEP 1 Using the surgical log or duplicate an- esthesia records (or the charts them- selves), count the number of general- anesthetic cases done with and the number done without halothane during the month, and complete Part A of Form I, Check anesthetic agents against pa- tients' records later wthdrawn for ab- stracting (Steps 5 and 6). STEP 2 Using the surgical log or anesthetic rec- ord number, (see the detailed instruc- tions below) select a random sample of 25 general-anesthetic cases, and list their chart number and dates of operation on Part B of Form I, regardless of final disposition of patient. STEP 3 Obtain the charts that correspond to the cases listed on Part B of Form L STEP 4 Obtain the records of every hospital death for the month, and complete Form II, listing their chart numbers and dates of operation and death. It is important that all death charts for the month be reviewed so that the surgical deaths as defined in the preceding section can be set aside for abstraction. Since it will be necessary as a first step in this pro- cedure to list all the hospital deaths for the month, we must have this complete listing on Form II. STEP 5 Obtain the hospital chart for each death listed on Form I I that occurred from the time of induction of anesthesia to within 6 weeks of a general-anesthetic surgical procedure in the same hospital. STEP 6 Mix the hospital charts listed on Form I and the charts of those patients listed on Form I I who died within the 6-week period (Step 5) and abstract each chart on Form m (mixing will aid in the uni- form abstracting of deaths and survivals). STEP 7 For each record of death abstracted on Form m that was necropsied, refer to the necropsy report and complete Form IV. STEP 8 If massive hepatic necrosis is found in any necropsy recorded on Form IV, ob- tain a duplication of the relevant portions of the hospital chart and necropsy rec- ords for each case. Complete Form V and transmit it with the duplicate records to the Data Collection Center. Detailed Instruction for Recording Data Form I, Part A Count all operative procedures in surgery re- quiring services of anesthetists or anesthesiolo- gists in your institution during the month of the report. Exclude obstetrical experience. Count all general-anesthetic cases for the month and rec- ord whether they are halothane or nonhalothane and, in addition, record whether any other halo- genated hydrocarbon was used. A halothane an- esthetic is one in which halothane is used regard- less of time interval or concentration. A non- halothane anesthetic is one in which halothane is not used at all. If, because of the practice in your hospital or other circumstances, you use a com- bination of regional, or topical, or local with general anesthesia, these cases should be included in the total count done with general anesthesia and categorized appropriately as above. Form I, Part B Select randomly, month by month from the surgical log or other appropriate source, cases in which a general anesthetic has been used and list on Form I, Part B. A case is eligible for random selection if it qualifies for inclusion in Part A of this form. The procedure for random selection follows. If you have tallied 400 or more general an- esthetics for this month (Item 2, Part A, Form I), select from this month's surgical log every an- esthetic chart number with a 5 in the units column and any odd number in the ten column (i.e., every anesthetic chart number ending in 15, 35, 55, 75, or 95) and enter them and the other required in- formation on Form I, Part B. This will probably yield 20 or more charts; enter as many as pos- sible consecutively up to 25. If you have tallied 200 to 399 general anes- thetics for this month (Item 2, Part A, Form I), select from this month's surgical log every an- esthetic chart number ending in 5 and enter them and the other required information on Form I, Part B. This will probably yield 20 or more charts; enter as many as possible consecutively up to 25. If you have tallied fewer than 200 general an- esthetics for this month (Item 2, Part A, Form I). select from this month's surgical log every an- esthetic chart number ending in 3 or 8 and enter 63
them and the other required information on Form I, part B. Enter as many as possible consecutively up to 25. Be sure to indicate the chart number and date of operation. The information later abstracted from the chart must pertain to a specific pro- cedure and might become confused if a given pa- tient had more than one surgical procedure per admission. If a chart is abstracted as a death and comes up in the random sample, it should also be in- cluded here and not omitted for any reason. If a patient has multiple procedures and his unit num- ber is selected, he may appear more than once within the same month in the random or control sample. He should be included as many times as randomly selected or death selected and a Form III should be completed for each selection even if much of the information will be repetitious. Form II List all deaths for this month and complete Form II, including hospital chart number, date of death and date of operation. Obtain the records for the deaths and sort them for surgical deaths as defined under "Experimental Design: General Information." Put the charts aside for abstraction. It is imperative that all death charts be checked since the patient may be considered "medical" at the time of death even if he underwent surgery within 6 weeks during a prior admission. Form III Collect the randomly selected control charts (Form I, Part B) and surgical death charts for the month in question, mix them, and abstract each on Form HI. Assign deaths to the month in which they occurred regardless of when the anes- thetic was given. Form III is designed so Hollerith card punch- ing can be done directly from the form after proof- reading for errors or omissions. The boxes cor- respond to the 80 entry columns of the Hollerith card. Specific instructions for each box or set of boxes are set forth below. FORM m - INSTRUCTIONS INSTITUTION'S CODE NUMBER Each institution will be assigned a number, which should be entered as follows: Example: INSTITUTION'S CODE NUMBER 0 1 PATIENT'S UNIT OR CHART NUMBER Enter the number of the chart (hospital number). Both letters and numbers can be entered directly. If there are fewer than eight characters in the chart number, leave blank spaces at the left. Example: PATIENT'S UNIT OR CHART NUMBER B 1 5 5 6 7 2 NOTE: If you are using some other number to make the random selection of charts, do not put it in here, but enter on Form I, Part B, as directed in the protocol. We may have to refer to this entry to get more information from record room when necessary. 63
AGE Enter the code for the age of the patient. If the age is not recorded on the chart, enter "-" (minus or dash). Example: 6 (Patien 0 = rt 9 5 = 50-59 1« lC - 19 6 = 60-69 2 = 20- 29 7 = 70-79 3 = 30- 39 8 = 80-89 4 = 40- 49 9 = 90-99 - = Missing Information SEX Enter code for sex of patient . If the sex is not recorded on the chart, enter "-" (minus or dash). Example: t (Patient is female) 1= Male 2 = Female - = Missing Information DATE OF DISCHARGE OR DEATH Enter date of death or discharge of patient as month, day, and only the last digit of year. Example: (Patient discharged December 3, 1962) 1 2 0 3 2 Month Day Year Date of Discharge or Death NOTE: Use zeroes to fill otherwise empty boxes in this sequence and all data box sequences. This helps us to guard against inadvertently entering digits into wrong boxes. REASON FOR ABSTRACTION Enter the code designating the reason for abstracting this chart . Example: (This is a death abstraction) Reason for Abstraction 0 = Death 1 = Random 64
AUTOPSY Enter the code for necropsy or no necropsy but leave this box blank if the chart was selected randomly. Example: i i (Patient had necropsy death abstraction) Autopsy 0 = No 1 = Yes MASSIVE LIVER NECROSIS Enter the appropriate code for massive liver necrosis. If the diagnosis is by necropsy, record 0 or 1, whichever is appropriate. If there is a clinical diagnosis of massive liver necrosis as recorded in the death certificate, enter 2. This information should appear on the line marked "How death was signed out on death certificate." ("2" should be entered ONLY IF NO AUTOPSY.) Example: - â (Death autopsy showed massive liver Massive necrosis) Liver Necrosis 0 =No 1 = Yes Path. 2 = Yes Clin. CAUSE (S) OF DEATH-HOW DEATH WAS SIGNED OUT ON DEATH CERTIFICATE Write in this item. Leave blank if random abstraction. DIAGNOSES PERTAINING TO THIS ABSTRACTION Write in this item; do not code. NOTE: List only those diagnoses pertinent to this admission, i.e., the chart which you are abstracting. OPERATION FOR WHICH THIS CHART IS BEING ABSTRACTED Write in the operation that was performed under the anesthetic for which this chart is being ab- stracted. The operation may include one or many separate systems of the body but can comprise one and only one anesthetic experience. If the patient has had more than one anesthetic experience, include only that information pertinent to this abstraction. If this chart is being abstracted for death, code the last operation in the boxes marked "Operation for which this chart is being abstracted." Any previous operation should be coded in the boxes "four most recent operations and dates." 65
CODE OPERATION Enter code for above operation from code list of operations in the Appendix to these instructions. The list is not intended to be complete, but is adequate for the purpose of relating anesthetic experi- ence to operative procedures on certain body systems. Where multiple operative procedures are per- formed, select the major procedure from the list (with the exception related to operative code num- ber 42). Example: Patient had D & C 12/8/62 and abdominal hysterectomy and appendectomy 12/11/62. Both procedures might come up for random abstraction. If so, two Form m's should be completed. Find code for operation in the appendix and enter as follows: First Form III Second Form III (D & C) (Abdominal hysterectomy-appendectomy) NOTE: The D & C would appear on the second Form III as part of the history of recorded operations. This example does not imply that all patients with two operations need to have two abstraction form (III). THERE SHOULD BE ONE ABSTRACTION FOR EACH RANDOM SELECTION. DATE OF OPERATION Enter month, day, and year of surgery for which record abstraction is being done. For years, enter last digit only. From previous example: CODE OPERATION DATE OF OPERATION Month Day Year First Form m 1 2 Second Form III 1 2 1 1 2 66
SAMPLE DATE Enter month and year for which these data are abstracted. These should correspond to month and year of surgery for random abstraction, or month and year of death for death abstraction. Example: Random abstraction for surgery of 12-08-62 1 2 2 UP TO FIVE AGENTS (ANY ORDER) USED IN OPERATION ABOVE Enter the code number for the anesthetic agents used in the operation coded and written out above. The codes for anesthetic agents are listed on each Form III. Example: Patient had D & C under pentothal, nitrous oxide anesthesia. 0 5 0 8 (Up to five agents (any order) used in operation above) 01 Halo thane 02 Methoxyflurane 03 Cyclopropane 04 Etc. (as on Form ra) NOTE: If more than five agents were used, circle all agents used on Form m and enter those five thought to contribute most to anesthetic state. You will notice that "local," "topical," and "other" are listed in code. When these are used in conjunction with gen- eral anesthesia they are to be listed. No chart should be abstracted, either death or random, in which the agent used was regional, topical or local, to the exclusion of general anesthesia. ANESTHETIC LENGTH MINUTES Enter minutes directly. Example: 8 8 .Anesthetic Length Minutes (1 hour 28 min) 87
ANESTHETIC RISK (PHYSICAL STATUS) Enter code as listed, first ascertaining which coding system you are abstracting and then finding proper column representing that category. Enter code representing that column. Enter "-" (minus or dash) for missing information. In either event, the code is entered in the box below. Examples: Old System 1 2 3 567 Old 1 2 3 4 5 6 7 New 1 2 3 4 1E2E 3E4E 5 Code 1 2 3 4 5 6 7 Physical Status (Poor physical status, elective) Examples: New System 123 (Poor physical status, emergency) SERIAL NUMBER Old 1 2 3 4 5 6 7 New 1 2 3 4 1E2E 3E4E 5 Codel 2 3 4 5 6 7 Physical Status Death and random abstractions should be serially numbered for each month and category. FOUR MOST RECENT OPERATIONS AND DATES (LAST FOUR YEARS) (Limit to previous surgery in your hospital.) Enter operation code number and corresponding date, one for each anesthetic experience as described previously. Look up operation code in Appendix. Example: Patient had appendectomy 1958 Open reduction fractured humerus 1960 D & C May, 1961 Operation Month Year Code Date Operation Month Year Code Date 68
Four Most Recent Operations and Dates (Last four years) (Enter above and below) NOTE: In many instances month will not be available - just enter year. PREVIOUS HALOTHANE EXPOSURES Enter the number of times the patient has previously received halothane either by history or documented record. NOTE: If history is not recorded on the chart, we can count this as no history, realizing of course the limitation of our definition. Example: Patient had halothane on previous admission two years ago as well as halothane for this abstraction. # Previous halothane exposures DATES OF MOST RECENT EXPOSURES TO HALOTHANE Enter month and year of most recent exposures to halothane (space for three entries). Example: Halothane 1961 (month not known) Halothane November 1962 Halothane for this abstraction Month Year Month Year Month Year Dates of most recent exposures to halothane FORM IV Form IV should be completed under close supervision of the responsible investigator or one of his principal associates. Format here is quite flexible because necropsy reports vary considerably from institution to institution and within a given institution. However, complete reporting of final diagnosis is necessary. FORM V This form pertains to the transmittal of photostatic or otherwise duplicated copies of surgical charts of massive hepatic necrosis and to the forwarding of necropsy material. Every chart of a case diagnosed as massive hepatic necrosis at necropsy must be photostated. Include history, physical, all other progress notes, laboratory studies, and other important diagnostic studies. Delete permits, authorizations, nurse's notes, and other nonpertinent information. In addition, for all cases of mas- sive hepatic necrosis, forward: 1. Block of liver unstained with specification of fixing method. 2. If not available, send slide (s) of liver with note on how it was fixed and stained. If you desire that they be returned, so state. 334-553 O-69â6 69
The slides are to be reviewed by a panel of pathologists. Please label slides or block with insti- tution code number and name as well as month of death, chart number and necropsy number. If there were no cases of massive hepatic necrosis, send blank Form V so indicating. Precautions in Procedure 1. The sample of general anesthetic procedures must be drawn with the instruction and super- vision of someone familiar with sampling techniques. We are prepared to provide information and service at Stanford to those who require help on random sampling techniques. After the sample is selected, its value can still be compromised by failing to obtain and abstract all the case records designated; thus, a strong effort should be made to find all designated charts. 2. Care should be taken to ensure that halothane and nonhalothane cases are abstracted uni- formly. This will be ensured in part by the instruction to abstract as much history and pre- operative information as possible before the anesthetic agent is looked up and recorded. The same caution must be exercised in interpreting the necropsy information. 3. If an alternate method of random selection is used, a detailed resume of your technique should be given. Mailing Instructions Send each month's packet of data and other relevant material as soon as completed to: William H. Forrest, Jr., M.D. Sutter Hospitals Research Foundation 52nd and F Streets Sacramento, California Send tissue blocks or slides to: Paul R. Glunz, M.D. Armed Forces Institute of Pathology Walter Reed Army Medical Center Washington, D.C. 70
FORM I National Halothane Study - NAS-NRC B.O.B. #68-6375 November 1963 Exp. 30 June 1965 PART A: - ANESTHETIC CHOICE SUMMARY Period covered in this report: Month Year Institution's Code Number. 1. The total number of surgical procedures during this period was 2. The number in 1 done with general anesthesia was 3. The number in 2 known to have involved the use of halothane was 4. The number in 2 known not to have involved the use of halothane was. 5. How many in 3 also involved methoxyflurane or other fluorinated hydrocarbons. 6. How many in 4 involved methoxyflurane or other fluorinated hydrocarbons PART B: - LIST OF CHARTS FOR ABSTRACTION FOR SAME MONTH AND YEAR AS ABOVE Identifying Number - Chart or Other Number Used in Random Selection* Date of Operation 1 2 3 4 5 6 7 8 9 10 11 12 I3 14 15 16 17 18 19 20 21 22 23 24 25 â¢If other than chart numbers, specify the method used in selection of numbers. 71
FORM n National Halothane Study - NAS-NRC November 1963 B.O.B. #68-6375 Exp. 30 June 1965 LIST OF HOSPITAL DEATHS Data Collector's Name Date of Collection All hospital deaths for (Month) . Total Number of deaths: Non-Surgical. . (Year) Institution's Code Number. .Surgical Hospital Chart # Date Death Date Operation Hospital Chart # Date Death Date Operation 72
FORM m National Halothane Study - NAS-NRC November 1963 B.O.B. #68-6375 Exp. 30 June 1965 ABSTRACTOR'S NAME- INSTITUTION'S CODE NUMBER PATIENT'S UNIT OR CHART NUMBER AGE 0=0-9 5 = 50-59 1= 10-19 6 = 60-69 2 = 20-29 7 = 70-79 3 = 30-39 8 = 80-89 4 = 40-49 9 = 90-99 - = Missing Info. SEX 1 = Male 2 = Female - = Missing Info. Month Day Year Reason Necropsy Massive Date of Discharge or Death for Abstraction 0 = Death 1 = Random 0= No 1= Yes 0 1 2 Hepatic Necrosis 'No Yes Path. -- Yes Clin. CAUSE(S) OF DEATH-HOW DEATH WAS SIGNED OUT ON DEATH CERTIFICATE (Write above) DIAGNOSES PERTAINING TO THIS ABSTRACTION (Write below) 4. S. 6 OPERATION FOR WHICH THIS CHART IS BEING ABSTRACTED (Write below) CODE OPERATION DATE OF OPERATION Month Day Year SAMPLE DATE Month Year (Up to five agents (any order) used in operation above) 01 Halothane 02 Methoxyflu- rane 03 Cyclopropane 04 Ether 05 06 07 08 09 Nitrous Oxide Trilene Ethylene Pentothal Brevital 10 11 12 13 Surital Other Barb. Narcotic Antagonist 14 Relaxant 15 Spinal 16 Epidural 17 Local 18 Topical 19 Fluomar 20 Vinethene 21 Other 73
FORM m (Cont'd) Anesthetic Length Minutes Anesthetic Risk Old 1 2 3 4 5 6 7 New 1 2 3 4 1E2E 3E4E 5 Code 1 2 3 4 5 6 7 (Leave Blank) Operation Code Month Year Date Operation Code Month Year Date *FOUR MOST RECENT OPERATIONS AND DATES â- (Last four years) (Enter above and below) Operation Code #Previous halothane exposures Month Year Date Operation Code Month Year Date Month Year Month Year Month Year Dates of most recent exposures to halothane *LIST BELOW THOSE OPERATIONS NOT CODED IN APPENDIX AND DATES 74
FORM IV National Halothane Study - NAS-NRC B.O.B. #68-6375 November 1963 Exp. 30 June 1965 ABSTRACT OF NECROPSY REPORT Hospital Chart # Necropsy # Institution's Code* Abstractor's Name LIVER (Check one if applicable) Massive Hepatic Necrosis Focal Hepatic Necrosis Other Hepatic Disease(s) List final necropsy diagnosis 75
FORM V National Halothane Study - NAS-NRC B.O.B. #68-6375 November 1963 Exp. 30 June 1965 Necropsy Information for Cases of Massive Hepatic Necrosis: Period Covered in this Report: Month Year Institution's Code Number Number of cases of massive hepatic necrosis List of Pertinent Clinical and Necropsy Records Forwarded to Computation Center Chart f Necropsy # Date of Death No. of Pages of Records List of Necropsy Specimens Forwarded to Armed Forces Institute of Pathology Specimen (s) - specify block, slide(s) Chart # Necropsy # Date of Death or both 76
APPENDIX National Halothane Study - NAS-NRC November 1963 OPERATION CODES Head and Neck: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Mouth, dental, T&A, tongue, lymph node biopsy (head-neck), myringotomy Nose all, s.m.r., fracture Eye all Ear all, except myringotomy Radical neck Fracture, maxilla, mandible, excision or biopsy Exocrine biopsy or excision Excision embryonic remnants Neck, cervical sympathectomy, resection of 1st rib Thyroid, all Craniotomy Burr holes, only, ventriculogram Pneumoencephalogram Arteriogram, aorta or major vessel, cardiac catheterization Carotid artery endarterectomy, graft Endoscopy-trachea, esophagus, larynx Endoscopic excision biopsy, trachea, esoph- agus, larynx Tracheostomy Laryngectomy Thorax: 26 27 28 29 30 31 32 33 34 35 36 37 38 Lung biopsy Lung, all except biopsy Hiatus hernia, transthoracic Breast biopsy, abscess, simple mastectomy Mammoplasty Radical mastectomy Heart and great vessels, pump Heart and great vessels, no pump Mediastinum, all Abdomen, Gastro-Intestinal: 39 Hiatus hernia (abd. approach) 40 Cholecystectomy, alone 41 Cholecystectomy and bile ducts 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 Cholecystectomy with other major procedures* Expl. lap., biopsy, lysis of adhes. Gastric resection, pyloroplasty, vagotomy Close ruptured ulcer Small bowel, all Large bowel, all Appendectomy Abdominal perineal resection Closure evisceration Hemorrhoidectomy and sigmoidoscopy Herniorrhaphy, incisional, umbilical, in- guinal, femoral, ventral Large vessel surgery, graft endarterectomy Sympathectomy, bilateral adrenalectomy Splenectomy, partial hepatectomy, pancrea- tectomy, porto-caval shunt, spleno-renal shunt Female: 60 61 62 63 64 65 66 67 68 D&C, bartholin cystectomy, culpotomy, in- sertion radium, cervical biop. Caesarian section Ectopic pregnancy Hysterectomy, all including bilateral sal- pingo-oophorectomy, or other variation of hysterectomy, such as A&P repair Pelvic laparotomy, adhesion, suspension, oophorectomy, salpingectomy, cystectomy Radical pelvic exenteration A&P colporrhaphy Male: 70 71 72 Prostate, t.u.r., biopsy, suprapubic, retropubic Hydrocoele, orchiectomy, circumcision Urinary Tract: 73 74 75 76 77 78 79 Cystoscopy, retrograde, fulgeration, bladder tumor Kidney, all Ureter, all, ureterolithotomy Bladder, all except electric fulgeration â¢For combined procedures the major procedure should be selected except for those cases which have a Cholecystectomy as pan of the procedure, and for these use Code 42. 77
APPENDIX- - Continued OPERATION CODESâContinued Extremities: 80 81 82 83 84 85 86 87 88 89 Fractures, all closed reduction Muscle, fascia, tendon, nerve surgery, neu- rorrhaphy Joints, open all Open reduction, all bones except femur Open reduction, femur Amputations, fingers, toes Amputations, hand, upper extremity, foot, lower extremity Saph. vein lig. and strip. Skin and Miscellaneous: 90 Plastic surgery, skin only, debridement, biopsy, excision small lesions. 91 92 Minor procedures miscellaneous 93 Major procedures, miscellaenous 94 95 Pilonidal I&D or excision, I&D abscess superficial 96 97 98 Laminectomy, cervical 99 Laminectomy, thoracic, lumbar NOTE: IF YOU CANNOT FIND THE OPERATION IN THE APPENDIX WHICH YOU HAVE WRITTEN ON FORM III THEN LEAVE THE CODE AREA BLANK. UNDER "FOUR MOST RECENT OPERATIONS" WRITE IN AT THE BOTTOM OF FORM m THOSE YOU CANNOT FIND IN APPENDDL 78
Addenda to Protocol Addendum [ (June 1964): Implementation of Collection and Review of Pathology Data - National Halothane Study Personnel of the Data Collection Center are to review all Form IVs (Necropsy Summaries) to check for consistency of the coding of massive hepatic necrosis. The necropsy information is to be divided into the following categories: massive necrosis, possibly massive necrosis, other ne- crosis, fatty metamorphosis, hepatitis, other liver disease and no liver disease. All cases categorized as massive, possibly massive and hepatitis not already submitted will be requested. A critical review of the submitter's diagnoses compared to the categories of this sort will be made. Addendum II (August 1964): Implementation of Collection and Review of Pathology Data - National Halothane Study All necropsy summaries except those where no mention is made of the liver are to be re- viewed by a consultant pathologist for any sug- gestion that the liver may have played a role in the patient's death; in such case, a photostatic copy of the complete necropsy report will be re- quested and reviewed. The morphologic descrip- tion of the liver will be examined for evidence of necrosis where no specific diagnosis was made. Slides will be requested on all cases of hepatic necrosis of any degree. Another consultant pathologist will review the above requested slides and set aside for the Panel's review cases with submassive or massive necrosis of the liver. Cases with infarcts (single or multiple), necrosis in relation to tumors (pri- mary or secondary), and focal or local necrosis associated with liver infections, peritonitis or other diseases are to be excluded. Also, livers showing fatty metamorphosis, collapse and scar- ring in the absence of necrosis, and livers showing marked degrees of autolysis are to be excluded. Special attention is to be paid to autolyzed livers in which the diagnosis of submassive or massive necrosis could be made in spite of the postmortem degeneration. These cases should be included for Panel review even though a detailed study of the type of necrosis may not be possible because of the postmortem autolysis. The cases selected from the above review are to be submitted to the Pathology Panel for completion of Form VI only. Chart abstracts will be completed for these cases also. An abstract of the chart of all patients sus- pected of massive hepatic necrosis will be pre- pared for subsequent review by pathologists. Forms Vm and IX were prepared to make the clinical information easily retrievable. Addendum III (August 1964): Preparation and Distribution of Hepatic Necrosis Slides - National Halothane Study Massive hepatic necrosis blocks will be pre- pared into slides of three in a set and will be distributed one set to each of the pathologists. Massive hepatic necrosis slides (no blocks submitted) will be appropriately numbered and grouped concurrently to be sent as a "round robin" amongst the pathologists. Addendum IV (August 1964): To Facilitate Analysis of Results of the Review of Liver Slides by Pathologists - National Halothane Study In order to facilitate analysis of the results of the review of the liver slides by the Pathology Panel, Forms VI and VII are to be designed and the following protocol established. Each pathol- ogist will review slides suspected of massive hepatic necrosis and complete Form VI (mor- phologic description). Each pathologist is to sub- mit the Form VIs to the Data Collection Center in Sacramento. Upon receipt of the Form VI, the Data Collection Center will send to the pathologist a chart abstract that corresponds to the slide reviewed on the Form VL The pathologist will review the chart abstract and the slide and com- plete Form VTL All of the above review by the Pathology Panel will be done without knowledge of anesthetic agent. Addendum V (May 1965): To Facilitate Analysis of Results of the Review of Liver Slides by Pathologists - National Halothane Study A consultant pathologist will review all orig- inally submitted slides of massive hepatic necro- sis and exclude from consideration those cases not qualifying to criteria established in Addendum II. 79
NATIONAL HALOTHANE STUDY FORM VI STUDY CASE*. . L LOBULAR CHANGES 1. PRE-EXISTING LOBULAR ARCHITECTURE NORMAL ABNORMAL DEGREE1 2. AUTOLYSISi 3. NECROSIS LOCALIZATION CENTRAL _ TYPE COAGULATIVE. PARACENTRAL. MIDZONAL PERIPHERAL. FOCAL INDETERMINATE 4. CYTOPLASMIC VACUOLIZATION: FATTY LARGE. SMALL. OTHER: SPECIFY LOSS OF CELLS ACIDOPHILIC BODIES PSEUDOXANTHOMA CELLS. 5. PIGMENTATION HEPATIC CELLS: LIPOFUSCIN. IRON BILE KUPFFER CELLS: LIPOFUSCIN. IRON BILE FORMALIN: NOTE: 1 Grade 'DEGREE" from 0-4+ where applicable. 2 When "AUTOLYSIS" precludes evaluation the form need not be completed. 80
FORM VI (Cont'd) STUDY CASE # DEGREE 6. INFLAMMATION. NECROTIC AREAS DIFFUSE FOCAL NON NECROTIC AREAS DIFFUSE FOCAL CELL TYPE: POLYS LYMPHS. EOS HISTO. 7. BILE STASIS _ CENTRAL PERIPHERAL. DIFFUSE LAKES 8. REGENERATION 9. CENTRAL VEIN ENDOPHLEBITIS PORTAL TRIADS DEGREE 1. INFLAMMATION ALL PORTAL AREAS. SCATTERED PORTAL AREAS. CELL TYPE: POLYS LYMPHS EOS HISTO 2. BILE STASISâ DUCT DUCTULE. 3. SCARRING 4. DUCTULE PROLIFERATION 81
NATIONAL HALOTHANE STUDY FORM VII STUDY CASE # A. Taking into account the clinical features and necropsy findings in this case, the histologic changes in the liver can be attributed to: 1. A single specific etiologic factor 2. The combined effects of multiple etiologic factors 3. Any one of several etiologic factors or 4. Do not warrant any etiologic diagnosis B. The histologic changes in the liver are consistent with the effects of: 1. Halothane 2. Any hepatotoxin 3. Any drug-sensitization reaction 4. Hepatitis virus. 5. Shock, anoxemia and/or sepsis - 6. Any other etiologic factor specify: or 7. Do not warrant any specific etiologic diagnosis 8. Cannot be interpreted because of postmortem autolysis. PATHOLOGIST
NATIONAL HALOTHANE STUDY FORM VIII CHART ABSTRACT IL SPECIFIC SURVEY FOR HEPATIC DISORDERS: Known Liver Disease (prior to surgery) Diagnosis: Jaundice: Hepatomegaly: Hepatic Tenderness: Other Yes No Unknown Alcoholism Yes No Amount: Type: Duration: Other comments: Unknown Hepatotoxic exposure Type: Duration: Comment: Yes No Unknown Homologous serum hepatitis, possible exposure Dental work under local: Injections during preceding 6 months: Blood transfusions: (prior to hospitalization) Tattoos: Other: Yes No Unknown Right-sided cardiac failure: Etiology: Duration: Severity: Response to therapy: Time relation to surgery: Other: Yes No Unknown. Allergies: Drug or agent Yes. No .Unknown.. Type of reaction Last exposure 83
NATIONAL HALOTHANE STUDY FORM DC CHART ABSTRACT III. CLINICAL COURSE POSTOP: The chart has been reviewed for the below listed signs and symptoms. If present, the date of onset, duration of episode and other comments will be found on the following page. YES NO UNKNOWN Rash Fever Chills â¢- Ascites Hypotension Shock Peripheral cyanosis Cardiac failure Edema Hepatomegaly Jaundice Hepatic tenderness Hepatic encephalopathy Purpura Bleeding Hematuria Splenomegaly Anemia X-rays 84
PROBLEMS OF DATA COLLECTION Random Selection Each step of the final protocol (dated January 1964) was clearly outlined and forms were pro- vided to simplify data collection, as well as trans- scription to Hollerith cards. A review of the methods used in each institution, deviations from protocol, and evaluation of data collection and missing data are presented in the following para- graphs. Monthly Totals There were five sources of information for a total count of surgical cases done under general anesthesia. Of the 35 institutions, 26 used the surgical log, two used the anesthetic log, two used Hollerith records, one used McBee punchcards and records, and four used the original records of each anesthetic administration. To check the accuracy of the information transferred from anesthetic records to the log- book and thus the accuracy of total count, it was suggested in the protocol that each selected ran- dom (Form I, Part B) be compared with the in- formation source of the total count (logbook). In practice, this check was cumbersome and time- consuming, so the principal investigators were asked to compare the randoms for September of each year with the surgical or other log. This was done after the September data were sub- mitted, by returning a list of the random cases (Form I, Part B) for that month to the principal investigator and requiring him to record the an- esthetic agent that appeared in the log corre- sponding to the particular chart numbers. In some institutions, other methods of checking were used, and in others, more frequent checks were made. Most institutions found very few errors in the log; a few discovered as many as eight in 1 month. In view of (1) errors in the original transcription of the anesthetic agent from the anesthetic record to the log, which would naturally lead to some dis- crepancies between what was used and what was reported in Form I, Part A, and (2) more impor- tantly, the absence of total counts of anesthetic agents other than halothane, it was decided that the administration rates of all agents would be estimated from the random selection. After all 48 monthly counts (Form I, Part A) were recorded on Hollerith cards, they were listed for each institution and hand-checked for inconsistencies. Indications of possible errors were found in two hospitals, on the basis of ap- parent inconsistencies in the totals. A check on this revealed that one institution had no counts for January through April 1959, because it had not opened until May 1959, and another had moved to a smaller operating-room suite during the Study. Missing data were not a problem with regard to Form I, Part A. Depending on the number of general anes- thetics administered in a particular month, ran- doms were selected using the patients' chart numbers, anesthetic chart numbers, or surgical log numbers, by choosing an appropriate 10 or unit digit calculated to yield 25 randoms per month/ These earmarked charts were selected in chronologic order from the log until 25 cases with general anesthesia were collected or the list was exhausted. Of the 32 institutions that followed this method, 10 (institutions 2, 8, 9, 13, 14, 16, 25, 27, 29, and 34) used other than chart numbers; therefore, their random chart numbers did not show a pattern that would indicate the method of selection. Missing Charts The random selection listing (Form I, Part B) was checked to ensure that no eligible charts were overlooked. (The protocol required a listing of all eligible randoms, whether or not they could be found and submitted; adherence to this requirement was a subject for special attention during site visits.) When a chart was tallied as missing, the Data Collection Center asked for a search of records. If the chart was unavailable, every effort was made to secure information from duplicate anesthetic records, surgical logs, death certificates, and pathology reports. Several in- stitutions had a high incidence of missing charts in the initial selection; but, through excellent co- operation and persistent follow-up procedures, many of these charts were retrieved (Table 1), so that only a very few were missing in the final tally. Frequently, there were no missing random charts but the number of randoms did not reach 25 per month, primarily because it was difficult to estimate the number of rejects (regional and local anesthetics) that would be encountered among the designated cases. There were 19 in- stitutions with fewer than the maximum of 25 for TABLE 1. âMISSING RANDOM CHARTS I'!:t. ilii.'il.nl No. randoms Institution submitted No. 1 5 15 12 3 2 12 5 13 2 1 11 1,113 13 1,122 U 1,123 15 1,166 16 1,194 20 1,207 25 1,137 27 1,131 28 1,057 30 786 33 1,199 strations 0.09 26,255 0.45 6,892 1.34 27,085 1.03 30,273 0.25 16,097 0.17 47, 610 1.06 8,447 0.44 27,259 1.23 7,063 0.25 4,398 0.08 26,825 334-553 O-69â7 85
some months but with no missing randoms (insti- tutions 4, 5, 6, 7, 8, 9, 10, 12, 17, 18, 19, 21, 23, 24, 26, 29, 31, 32, and 35). Departures from Protocol for Random Selection Acceptable Deviations Institution 3 used the surgical log and chose every 20th patient. If a regional anesthestic was encountered, the preceding record was chosen. Institution 21 used a different system of digits each month as the basis for selection from De- cember 1962 to June 1962 (data were collected for the most recent month, December 1962, and then in reverse order for each preceding month); protocol procedures were then used for the re- mainder of the Study period. Institution 27 seri- ally numbered the surgical log and selected all numbers that ended in 5. The deviations from protocol permitted for institutions 3 and 27 were considered of minor importance. Institution 21 deviated in a manner that could conceivably introduce bias if the selec- tion of numbers was based on some preconceived notion about the data to be abstracted. Unacceptable Deviations The following deviations were considered unacceptable: (1) omission of an earmarked random chart for any reason (i. e., chart not available), (2) submission of more than 25 random charts, and (3) replacement of a missing chart with another. Selection of Deaths - Form II The hospital death list, usually kept in the record room, was the source of the death selec- tion. The protocol stipulated that all hospital deaths be recorded on Form I I and that all death charts be read to determine which met the pro- tocol requirement, i.e., death within 6 weeks of the administration of a general anesthetic. The major difficulty in adherence to the protocol in this phase was again the location of missing charts. The importance of including the entire death list on Form I I was emphasized in the pro- tocol, and presumably no hospital deaths that followed general anesthesia could go undetected. However, it was often difficult to locate missing charts and, although there was excellent follow-up and every reasonable attempt was made to find missing charts, many were not found. Attempts to categorize missing charts by searching other documents, such as death certificates and ne- cropsy protocols, were usually not fruitful. There were nine institutions with missing death charts (Table 2). In two (14 and 35), enough charts were missing to make the validity of any data on death or MHN derived from them ques- tionable. By examining Form II and establishing the ratio of surgical to nonsurgical deaths in each hospital over the entire 4-year period, we could estimate the number of missing surgical charts. Thus, for institution 14, it was estimated that four of the 35 missing death charts were probably surgical; and for institution 35, 75 of the 187 missing death charts were probably surgical. To avoid errors in estimating the death rates, an upper limit of 5 percent missing surgical death charts was established; therefore, the data for institution 35 were not included in the statistical analysis. Institution 11 13 14 15 16 20 25 30 35 TABLE 2 Missing death charts Estimated Over-all surgical Total Percent deaths death in charts study missing 9 1 35 7 1 1 10 2 187 2 0 4 1 0 0 1 0 75 555 246 141 586 519 727 578 223 1,348 0.4 0.0 2.8 0.2 0.0 0.0 0.2 0.0 5.6 Random and Death Abstractions - Form m All selected random and death charts were abstracted to complete Form III. Missing infor- mation was accounted for by appropriate coding to avoid confusion with blank spaces for data not yet entered. To ensure accurate over-all cate- gorization and identification of all Form IIIs, they were checked manually for correct coding of in- stitution, month, year, and operation. The data were then transferred to Hollerith cards, veri- fied, and checked for error by programs devel- oped to find inconsistencies, omissions, and errors more quickly and accurately. Frequent communication between the Data Collection Cen- ter and the principal investigator made it pos- sible to correct most of the errors. A code for operations was developed that was simple to use and would not require highly trained personnel or complicated codes, such as the International Classification of Diseases and Operations. Categorization based on regional site was most important in the development of this code. Unassigned codes were available in case additional classifications for operation were needed during the Study. All institutions were urged to leave columns 21-22 (operation code) blank if they were not sure of the code for a specific operation. By sorting all the uncoded Form IIIs and having one person code them at the Data Collection Center, it was possible to obtain consistency of coding for the less-common operations. If one type of operation frequently appeared uncoded, a code was assigned and all participants were notified by means of a flyer. 86
Evaluation of Missing Data and Errors - Form III Manual and programed computer checks re- vealed many errors and omissions, which the appropriate institutions were asked to correct. The general response to such requests was ex- cellent and most of the elusive data were re- trieved. Many institutions had difficulty in lo- cating charts for abstraction, and it was obviously much more of a problem in some institutions than in others. The number of missing charts may have reflected management problems in the rec- ord room, but it appeared more likely to have re- flected conditions in an active center where charts were in constant circulation. Selection of Massive Hepatic Necrosis Cases - Forms IV, V Necropsy protocols, summaries, or both were examined by the principal investigators or their designated assistants for evidence of MHN. In the pilot study it was often difficult or impos- sible to assess the degree of necrosis from the information on the necropsy protocols. For in- stance, in reviewing a necropsy protocol, it was difficult to state whether the recorded informa- tion, such as "marked centrilobular necrosis," actually represented massive hepatic necrosis. To help the principal investigators make judg- ments and to provide data for later evaluation of judgments, we advised them to include a sum- mary for each necropsy and to be liberal in their interpretations of MHN so that all borderline cases would be submitted for review by the Pa- thology Panel. All cases of MHN (as determined by the principal investigators) were reported on Forms III, IV, and V. Investigators were urged during site visits to select MHN cases without knowledge of anesthetic agent if possible. It was hoped that there would be minimal bias in MHN reporting, but actually the agent was known about half the time. A copy of the chart of every MHN case was submitted to the Data Collection Center, and a block of tissue and microscopic slides of the liver were submitted to the Armed Forces Institute of Pathology (AFIP) in Washington, D.C. DATA EVALUATION AND EXTENSION OF PATHOLOGY PROTOCOL Evaluation of Form IV Review of Form IV for consistency of chart numbers, necropsy numbers, and correlation with reports of degree of necrosis in Form m, column 19, was not difficult and was rewarding. It became apparent that there were occasional discrepancies in the extent of hepatic disease reported in the necropsy summary (body of Form IV), the classi- fication of hepatic disease at the top of Form IV, and the indication that necrosis occurred as re- corded in column 19 of Form III. This and similar inconsistencies led to a complete review of the Form IV information. All necropsy information provided on Form IV was reviewed at the Data Collection Center, without knowledge of the anesthetic, in an effort to detect overlooked MHN. Cases not previously submitted but possibly representing massive or submassive necrosis or hepatitis were sought from participating institutions. Pathology Review of Form IV and MHN Slides To investigate discrepancies in the reports on Form IV, as delineated by the review of a major portion of the Form IVs submitted, the Subcommittee amended the protocol (Addenda III and IV). It also commissioned two consultant pa- thologists (1) to review the information on Form IV from all the necropsy reports, except those which did not mention the liver, and sort out and request microscopic slides on all cases in which hepatic necrosis could have been present, and (2) to review these slides and sort out cases in which there were appreciable degrees of MHN. These, along with original slides, were then han- dled as described below. Handling and Circulation of MHN Slides All material sent to AFIP was prepared for circulation to the Pathology Panel. When enough material was submitted, sets of slides were pre- pared for each pathologist so that each set con- tained material stained with standard techniques and unstained sections for special staining. Un- fortunately, not all material sent to AFIP was in the form of blocks; in some cases, only one slide that had been prepared at the investigating in- stitution was available. Protocol for Review of Bona Fide Cases As the MHN charts for abstraction and the hepatic slides for review became available, it was obvious that addenda to the protocol were essential to standardize the Panel's review of the slides and chart abstracts and to provide forms for recording this information. In addition, four new forms (VI through DC) were designed to record the Pathology Panel's morphologic de- scription of the hepatic slides and etiologic fac- tors in the cause of the heptatic necrosis. The forms were also used to record, in tabular fash- ion, the data from the chart abstractions, and to outline the general format of the chart abstrac- tions. Microscopic sections from all cases were examined by the five members of the Pathology Panel. The extent of necrosis was rated by each pathologist independently on a scale of 0 to 4. These ratings were averaged and the cases were separated into three categories: (1) massive ne- crosis (average score, 2.6 or above), (2) inter- mediate necrosis (average score, 1.6 to 2.5), and (3) minimal necrosis (average score, 1.5 or be- low). Minimal necrosis was considered to be a commonplace and negligible occurrence in any 87
necropsy population and these cases were there- after disregarded. Assignment of necrosis and description of morphology were made without knowledge of anesthetic or clinical history. On receipt of Form VI at the Data Collection Center, an abstract of the corresponding clinical history, from which the identity of the anesthetic agent was deleted, was sent to each pathologist. The Pathology Panel then reviewed the slides with access to the chart abstracts that were pre- pared according to Forms Vm and IX and in- cluded all clinical factors of importance except the anesthetic agent. Only the office copy in- cluded the anesthetic agent. The purpose of the second review was to elicit opinions from each member of the Pathology Panel as to possible or probable etiology of the observed hepatic ne- crosis as required by Form VIL Review of Form VI and Protocol Addendum Slides from approximately 100 cases scored by all five pathologists (Form VI) were reviewed for consistency of scoring and evaluation of scor- ing technique. In all but 11 of the first 100 slides reviewed, consistency of scoring was good; there were only few instances of a spread of more than one degree of necrosis. In 11 cases, however, there was considerable inconsistency in scoring; scores ranged from 0 to 4 or, in some cases, slides were not interpretable owing to autolysis. The morphologic description and the necropsy summary reports on these 11 cases were ex- amined, and it was found that such factors as ne- crosis near malignant tissue or abscess and marked autolysis accounted for the discrepancies in the scoring. The pathologist who had already been commissioned to review the additional slides (solicited in Addendum IV) was designated to screen the original MHN cases to eliminate those in which there was marked autolysis or other in- terfering variables. Problems of Pathology Data Collection Except for the coding of necrosis, it was dif- ficult to obtain consistency in coding the pathol- ogy forms. Variables were frequently not coded, indicating either an absence of the variable or missing information. No means were found in the protocol to detect these differences. Missing Pathology Data - Charts and Slides Slides Submitted by Principal Investigator Six cases of necrosis were reported for which no slides were submitted; two involved cyclopropane, one ether, and three Other.* In three cases the wrong tissue was submitted and no substitute tissue was available; two involved cyclopropane and one thiopental-nitrous oxide. In two cases slides were lost and not replaced; both involved cyclopropane. Slides Requested by Pathologist Consultants Twenty-six slides that were requested on the basis of review of necropsy summary could not be obtained; two involved halothane, 11 thiopental- oxide, five cyclopropane, and eight Other. For four slides requested and obtained for review, corresponding clinical charts were never ob- tained; two were scored as massive hepatic ne- crosis (one involved thiopental-nitrous oxide and one ether), and two were scored as inter- mediate necrosis (both involved thiopental- nitrous oxide). *For comparisons, anesthetic agents were separated into the following anesthetic practices; (1) halothane, (2) nitrous oxide-barbiturate, (3) cyclopropane, (4) ether, and (5) "Other." These categories are described more fully in Appendix 1 to Chapter IV-2. 88
PART III. REPORT ON HEPATIC NECROSIS
![National Halothane Study: a Study of the Possible Association Between Halothane Anesthesia and Postoperative Hepatic Necrosis; Report. Edited by John P. Bunker [and Others]](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAADsQAAA7EAZUrDhsAAAANSURBVBhXYzh8+PB/AAffA0nNPuCLAAAAAElFTkSuQmCC)