National Academies Press: OpenBook

Modern Methods of Clinical Investigation (1990)

Chapter: 6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance

« Previous: 5. Strengths and Weaknesses of Health Insurance Data Systems for Assessing Outcomes
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 68
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 69
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 70
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 71
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 72
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 73
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 74
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 75
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 76
Suggested Citation:"6. Perscription-Event Monitoring: An Example of Total Population Post-Marketing Drug Serveillance." Institute of Medicine. 1990. Modern Methods of Clinical Investigation. Washington, DC: The National Academies Press. doi: 10.17226/1550.
×
Page 77

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.

6 Prescription-Event Monitoring: An Example of Total Population Post-Marketing Drug Surveillance WILLIAM H. W. INMAN In the early 1970s in the United Kingdom, many patients who had been treat- ed for heart disease with the beta-adrenergic blocking drug, practolol, suffered severe ocular and skin reactions and occasionally deafness or intestinal obstruc- tion. Although these events were to be found in the patients' medical records, most of their physicians had failed to consider the possibility of a causal link with practolol and had no reason to report the cases to the Committee on Safety of Medicines (CSM). The need to supplement the yellow card scheme for vol- untary reporting of suspected adverse reactions to drugs with a second scheme for recording events, irrespective of any recognized link with the use of new drugs, led to the establishment in 1980 of the United Kingdom's second national scheme, Prescription-Event Monitoring (PEM). This scheme was developed by the Drug Safety Research Unit (DSRU) in association with the University of Southampton. "THE NUMBERS GAME" Before describing PEM, it is worth reflecting on what can be described as "the numbers game." In Figure 6.1, I have attempted to indicate a number of variables which must be considered. The arrow at the bottom of the diagram represents the severity of an illness or symptom, ranging from trivial on the left to serious or life-threatening on the right. Above this is an arrow suggesting that, as severity increases, the acceptable level of risk of treatment may also increase from very low levels for a trivial complaint to much higher levels in the treatment of a serious or fatal illness. Paradoxically, as the acceptable risk level rises the number of patients 68

PRESCRIPTION-EVENT MONITORING Low Enthusiasm High Small a) ._ a) . _ o a) Q A Large ~- ywN;')~/ Low FIGURE 6.1 Variables In "die numbers game." Severity of Disease 1t High 69 Low Cost High required to detect or measure it and the cost of the study diminish. Very large risks (e.g., 1 in 10) can be measured in comparatively small numbers of patients (e.g., 100) at relatively low cost. On the other hand, to measure the acceptable level of risk of mortality for symptomatic treatment of a headache, which may be less than 1 in 1 million, would require huge numbers of patients and infinite resources. Unfortunately, nobody has ever been able to tell us precisely what is an acceptable level of risk for a particular treatment. If the authorities were able to define a risk level which would lead them to remove a drug from the market, people like myself would be better able to design appropriate studies and esti- mate their cost. I have added another variable at the top of the diagram to suggest a factor which people frequently forget. This is enthusiasm, which ranges from little or none for studies of very minute risks to considerable enthusiasm for studying large risks associated with the treatment of very serious diseases such as cancer. I could have added yet another dimension to the diagram, which is time. It is

70 WILLIAM H. W. INMAN worth remembering that some adverse effects of drugs might not become mani- fest for many years (e.g., carcinogenesis). Time adds to the costs of studies and also diminishes the enthusiasm of people to undertake them. Few people of my age, for example, would be keen to commence a carcinogenicity study, the results of which might be submitted to the New England Journal of Medicine by my grandson. Clinical trials in 2,000-3,000 patients are likely to measure only the compar- atively large risks associated with treatment of the more serious illnesses. They have the advantages of comparatively low cost and high enthusiasm of those involved. At the other extreme, voluntary reporting is the only practical and affordable way to detect very rare events. There is no way that any nation can afford to set up post-marketing studies capable of measuring risks in the region of 1 in 1 million or less. Even if every detail of patient care were recorded elec- tronically for the whole population, the numbers and time involved in collecting sufficient data to be sure that a headache treatment was safe would be pro- hibitive. Somewhere between these two extremes, between the clinical trial on the one hand and voluntary reporting on the other, are a number of schemes aimed at measuring middle-level risks, those in the region of 1 in 100 to 1 in 10,000. VOLUNTARY REPORTING SYSTEMS From 1964 to 1980, I was responsible for managing CSM's yellow card reporting scheme in the United Kingdom, which is very similar to the voluntary reporting scheme operated by the Food and Drug Administration in the United States. I believe that, while voluntary reporting is the only practicable way to identify very rare or unusual events, we must be careful about using data derived from voluntary reports as the sole basis for decision making, particular- ly when the events are fairly common. Gastrointestinal hemorrhage associated with the use of non-steroidal anti-inflammatory drugs (NSAIDs) is a good example of a fairly common event. Each year, it occurs in between 1 in 1,000 and 1 in 200 people of an age commonly associated with arthritis or rheuma- tism, even in the absence of treatment with an NSAID. For this kind of event, it is always necessary to seek measurements of incidences from other types of study. In 1976 I described the "seven deadly sins" of reporting doctors (1~. First is the complacent belief that only safe drugs are marketed. Next is fear of litiga- tion. Third, feelings of guilt about having done something for the benefit of the patient which has gone wrong. Fourth, and perhaps worst of the deadly sins, is an ambition to collect and publish a personal series of cases. It may be nice to see one's name in print, but it is not good to keep quiet about early observations and thereby delay general recognition of an important hazard. Fifth is igno- rance of the need for reporting or of the reporting mechanism. Sixth is difh~- dence about deciding whether or not an event is an adverse drug reaction or

PRESCRIPTION-EVENT MONITORING 7 something which would occur spontaneously, tinged perhaps with concern about appearing ignorant to those to whom the event would be reported. Finally, the seventh sin is old-fashioned lethargy. All seven deadly sins may have worked in producing the practolol disaster in my country, which led to blindness and a number of deaths. In that disaster it was estimated that 100,000 patients had been treated for a year or more before a single yellow card report of conjunctivitis reached the CSM (2~. PO ST- MARKETING SURVEILLANCE In 1976 I proposed a scheme known as "recorded release," which would use our unique ability in the United Kingdom to assemble all prescriptions written under the National Health Service (1~. This scheme and a number of later vari- ants, such as "retrospective assessment of drug safety," were all turned down on the grounds of impracticality by those who advised the ministers of the day. After four years of negotiations with the various bodies concerned with the ethics of what has become known as pharmacoepidemiology, and particularly with issues of confidentiality and patient consent, I set up the DSRU within the faculty of medicine at the University of Southampton in June 1980. Our first PEM study commenced in 1981. Since that time our staff has increased from three to nearly fifty. In 1986, for a number of administrative and financial rea- sons, the management of the DSRU was transferred from the university to a charitable trust known as the Drug Safety Research Trust. Because no one can predict which drug will be the next practolol, the only sensible policy is to study all new chemical entities marketed on a wide scale in general practice. We depend to a considerable extent on the drug industry for support, but we like to do the work first and hope that we will be reimbursed realistically after we have completed the study. The DSRU does not undertake contract work. Our staff are not allowed to receive regular retainers and may not hold shares in drug companies. We do not offer payment to doctors prior to their writing a drug prescription, and thus we do not in any way influence their selection of patients for treatment. PRESCRIPTION-EVENT MONITORING Prescription-Event Monitoring is based on the unique facility provided by the Prescription Pricing Authority (PPA) in the United Kingdom. All National Health Service prescriptions issued by general practitioners find their way to this central agency so that pharmacists can be remunerated. This facility has been available for more than 40 years, but only in 1980 was it possible to get agreement to use prescriptions, which are highly confidential documents, to identify very large cohorts of patients for epidemiological study (3~. PEM is conducted only in England, in a study population approaching 50 million patients because, in Wales particularly but also in Scotland and Northern

72 WILLIAM H. W. INMAN Ireland, large numbers of patients share the same name. An individual is often extremely difficult to identify from the limited details shown on the prescrip- tion. In England some 350 million prescriptions are written each year, and the PPA employs approximately 2,000 clerical staff who each process about 100 prescriptions per hour. Patients are not identified in the PPA data base. To identify them for research purposes, the PPA has to prepare a computer-generat- ed "picking list" for each new chemical identity on our PEM list. The prescrip- tions are then pulled out of the files by hand and photocopied for transmission to the DSRU; up to 1 million prescriptions are handled each year for the pur- pose of PEM. Let us briefly consider the word "event." Event monitoring was the idea of professor David Finney, who is currently a trustee of the DRSU and a founding member of the Adverse Reactions Subcommittee of the CSM. Finney pointed out that if you ask doctors to report "events," without worrying whether they are drug related, you may get a great deal more information. An "event" includes any new diagnosis, any reason for referring a patient to a hospital, any unex- pected improvement in a patient's condition, any change of treatment, any sus- pected adverse reaction or indeed any significant word which a doctor has thought important enough to record in a patient's notes. Each of the question- naires that we use to obtain clinical information from the general practitioner carries a very simple example: a broken leg is undoubtedly an "event." If we were comparing three drugs used for the same disease, and found that patients treated with drug B suffered four or five times as many fractures as those treated with A or C, we might suspect that we had a problem with hypotension, dizzi- ness, or even softening of the bones. This simple concept of an event has been communicated to more than 20,000 general practitioners who participate enthu- siastically. An important aspect of event monitoring is that, because the report- ing of an event does not require a medical opinion as to its cause, it may well carry considerably less medico-legal risk than reporting an adverse reaction. An adverse reaction, after all, is an admission that something has gone wrong as a direct consequence of the physicians's decision to advise use of the drug. Routinely, as soon as a new chemical entity is marketed for use in general practice we inform the PPA. In due course, increasingly large numbers of pho- tocopied prescriptions are sent to the DSRU. We process them on our computer and after an interval that depends on the type of drug being studied, we post our questionnaires (green forms) to the doctors. We expect a response of about 70 percent. The DSRU has a comprehensive system for following up individual case reports. All deaths are followed up. With the doctor's permission we contact the Family Practitioner Committee, to whom the notes usually have been returned after the patient's death. We are thus able to study the lifetime medical records of each patient who has died. Similarly, a non-fatal but serious event will be followed up by reference to a physician or surgeon at the hospital. We

PRESCRIPTION-EVENT MONITORING 73 also obtain copies of all death certificates from the Office of Population Censuses and Surveys (OPCS), and have developed but not yet used a flagging technique to identify cohorts of patients through the OPCS so that we would be notified of any patient deaths. This process could be used for long-term moni- toring of efficacy and serious problems such as carcinogenicity. Occasionally we have "banked" large numbers of prescriptions without taking any further action, so that, if a problem should arise 10 or 15 years later, a population is available for retrospective research. During the last year or so we have experimented with a "red alert scheme." In collaboration with the CSM we issued a special variant of the yellow card on receipt of each prescription identifying a new patient. On the whole the scheme proved unsatisfactory. Many doctors were confused about what should be reported on the new yellow card. Although they were instructed to keep the card in the patient's notes and to use it only on the rare occasions when a seri- ous or life-threatening adverse reaction occurred, large numbers of cards came back reporting that nothing had gone wrong or describing trivial events. We have now substituted a two-track system. A conventional green form is sent to the doctor within about three months of writing the first prescription. Then, where necessary, we return the same form one or two years later for an update. The expanding data base offers all sorts of opportunities for outside research workers. For example, we are running an exercise with the help of the Merck Foundation which checks the results of our own follow-up of deaths with what has been written on the death certificate. Some very interesting differences are obvious. For example, if one believes what is written on death certificates, one might think there is almost no risk of dying from open heart surgery. Deaths tend to be due to less exciting causes, such as bronchopneumonia, which in some cases appears to develop before the patient has left the operating room! We attempt to study all new chemical entities in addition to any older drug which may have caused problems. There are some exclusions. We do not rou- tinely look at parenteral preparations because they are not generally used on a sufficiently wide scale by general practitioners. At the moment we do not study vaccines or topical preparations. So many new chemical entities are being released for marketing that the resources of the PPA and the capacity of our computer are somewhat stretched. For each drug we attempt to select about 20,000 patients, with the objective of an absolute minimum of 10,000 well-doc- umented cases at the end of each study. To date, we have completed 11 studies. We abandoned 6 others because the drugs did not sell and we could not build up a worthwhile cohort over a period of three or four years. A drug which is slow to penetrate the market can cause a great deal of additional work because the ratio of repeat prescriptions to new patient identifications is large. Currently we have more than 20 studies in progress and 5 or 6 others in the pipeline. PEM has occasionally been used to test hypotheses. For example, we looked into erythromycin estolate because yellow card reporting suggested the possibil-

74 Wlrl:IAA! H. W. INMAN ity that the estolate was relatively more likely to cause jaundice than other forms of erythromycin. A comparative PEM study found no difference in the incidence of jaundice (4~. Although this does not exclude the estolate as an occasional cause of jaundice, there was certainly no evidence that it is a rela- tively more common cause. In another study, we looked at about 16,000 patients treated with emepronium bromide, which was associated with occa- sional reports of esophagitis. Approximately 450 doctors reported that patients had indeed experienced some swallowing difficulty after taking the tablets; these positive replies were so numerous that we were obligated to investigate the matter further using a four-page questionnaire. The secondary enquiry found that only a handful of patients had severe esophagitis and only one life had been threatened. COMPARATIVE STUDIES OF NSAIDS The value of PEM has been well illustrated by our studies of seven NSAIDs. Benoxaprofen was associated with a large number of reports of skin rashes, almost all due to photosensitivity. The possibility that photosensitivity might persist long after the drug had been withdrawn prompted us to conduct a special experiment three or four years later. We were also interested in determining the completeness of reporting to the CSM. We discovered, on following up more than 900 reports received by the DSRU, that at least one-third of these cases had also been reported to the CSM. We were able to show that reporting to the CSM had not been influenced significantly by the "Oraflex"jaundice publicity. This suggested that we might need to modify our view that adverse reactions are grossly under-reported, by the voluntary methods. It is quite likely that many minor adverse reactions are under-reported but when the side effect is serious the reports to the CSM may be more complete than has been thought. We analyzed indomethacin administered in the sophisticated delivery system known as "Osmosin," which releases the drug progressively as the capsule pass- es through the gut and thereby minimizes gastric intolerance. We discovered to our surprise that the rate for reports of gastritis or dyspepsia with this product was considerably greater than with four other products studied earlier (benoxaprofen, fenbufen, zomepirac, and piroxicam). However, when the drug was removed from the market, patients who had received it still experienced a much higher rate of dyspepsia and gastritis. When we ranked the drugs accord- ing to the frequency with which patients experienced this side effect, we found that the rate during treatment was directly proportional to the rate following treatment. The rates for indomethacin during and after treatment were the high- est and those for piroxicam were the lowest. Our tentative conclusion was that, whenever a company promotes an NSAID as being less liable to produce cer- tain side effects, doctors will tend to prescribe it for those patients who are most likely to develop those same side effects. One important consequence is that

PRESCRIPTION-EVENT MONITORING 75 doctors will then report relatively more events with the seemingly less toxic drug. This has led us to the general conclusion that if you make something safer, people will take greater risks with it and thus cancel out the advantage (5~. When we came to look at the more serious side effects of NSAIDs, notably the complications of peptic ulceration, we failed to find any difference among the seven drugs in the frequency of gastrointestinal hemorrhage or perforation. Nor did we find any important differences between the rates for these complica- tions during and after treatment. We have to accept, of course, that many patients who stopped treatment with one NSAID would have been switched to another. POST-MARKETING SURVEILLANCE AND POST-MARKETING CLINICAL TRIALS An important distinction needs to be made between post-marketing surveil- lance and post-marketing clinical trials. Post-marketing surveillance is conduct- ed under "real life" conditions in which events are ascertained in patients for whom a therapeutic decision has already been made. The surveillance proce- dure must not influence the choice of treatment. In post-marketing clinical Dials, on the other hand, a patient's treatment is deliberately changed and is followed up prospectively. In these circumstances a patient must be a fully informed volun- teer. In the United Kingdom, Intercontinental Medical Statistics Limited (IMS) recently established a Post-marketing Surveillance Unit which is engaged in what seem to be post-marketing clinical trials rather than PMS. Their procedure is quite different from ours. They write to physicians encouraging them to take part in a "PMS" study for which IMS offers financial remuneration. If the study were restricted to patients who had already commenced treatment one could argue that it was truly post-marketing surveillance. In practice, however, their approach encourages physicians to change patients' treatments and the study is therefore technically a promotional post-marketing clinical trial. This distinc- tion is extremely important because the results of post-marketing clinical trials of this kind differ very significantly from those of real-life post-marketing surveillance. In the former, there may be a considerable element of selection. For example, doctors may avoid use of the new drugs in high-risk groups such as elderly patients, pregnant women, children, and so on. We recently encountered an important example of this difference. We com- pared the results of a large post-marketing clinical trial of enalapril in approxi- mately 11,700 patients, and a PEM study of more than 13,000 patients conduct- ed by the DSRU. In the first, study patients were observed for only six weeks, and there were eight deaths. In the PEM study, which covered a year of obser- vation, there were 1,098 deaths. Adjusting for the difference in duration of the studies, this rate was about 80 times greater than in the clinical trial. There

76 WIll;IAM H. W. INMAN were 152 reports of renal failure, of which 75 were fatal. No cases of renal fail- ure had been reported in the clinical trial. Fortunately for enalapril, the CSM agreed to wait until we fully investigated the reports of renal failure and in par- ticular the 75 deaths. After intensive study involving colleagues from the Post- Graduate Medical School in London, Sir Colin Dollery and Christopher Speirs, we were able to show that all but perhaps 10 of the fatal cases of renal failure could be accounted for by pre-existing renal disease. Even in the 10 deaths in which enalapril might have played a part, other factors such as excessive use of diuretics or hyperkalemia could well have precipitated the renal failure (6,7~. The large differences between these two studies can be accounted for almost entirely by selection. The company study was conducted in low-risk patients with middle to moderate hypertension. The PEM study reflected, very precise- ly, a widely varying range of patients receiving enalapril. Many were suffering from advanced congestive heart failure and a proportion had pre-existing renal disease. If the company made a mistake it was calling their study post-market- ing surveillance when it was plainly a post-marketing clinical trial. Another problem, which is causing us considerable concern, is the distortion of early prescribing practice by some company studies in which a small number of doctors are encouraged to prescribe for large numbers of patients. Recently, for example, we encountered a drug which was the subject of a company study where 5 percent of the doctors who prescribed it accounted for more than half the total U.K. market. In one extreme case a doctor had prescribed the drug for 235 patients during the six months following its introduction. This drug was licensed for use only for rheumatoid arthritis. Sooner or later another unexpected incident like practolol is almost inevitable. Our best hope lies in its early detection and containment. The speed with which we identify a hazard depends upon the speed with which we can gather sufficient information about the largest possible number of patients. Competition for patients early in the market life of a new drug will inevitably fragment the available data base and lead to a failure to identify an unexpected hazard. We have already seen several examples where promotional studies con- ducted by drug companies or market research organizations have seriously delayed the progress of PEM. Guidelines for post-marketing surveillance in the United Kingdom insist that it should not be promotional. If this were true there would be no need for such studies to compete with the two national systems, PEM and the yellow card scheme. After these systems establish that a new drug has an acceptable level of safety, the companies should be able to continue their sales drive with greater confidence.

PRESCRIPTION-EVENT MONITORING 77 CONCLUSION Despite a close working relationship with the government and the pharma- ceutical industry, the DRSU's greatest asset is its independence from bow. This has given it credibility and considerable influence, particularly in regulatory cir- cles. Using the United Kingdom's unique ability to identify all patients who receive a particular drug, we can study its performance rapidly in large numbers of patients. The cost is quite modest; a one-year study of 10,000 patients, including follow-up of any who die or develop serious adverse reactions, costs on average $400,000. We have all been looking for inexpensive monitoring of health outcomes in the real world of clinical practice. I would like to think that, in Prescr~ption-Event Monitoring, we have gone some way toward achieving Mat goal. REFERENCES Inman WHW. Detection and investigation of drug safety problems. In Gent M, Shigamatsu I teds). Epidemiological Issues in Reported Drug-Induced Illnesses. Hamilton, Ontario: McMaster University Library Press, 1976. Inman WHW, Weber JCP. In Inman WHW (ed). Monitoring for Drug Safety. Second edition. Lancaster, England: MTP Press, 1986:37. Inman WHW, Rawson NSB, Wilton LV. Prescription-Event Monitoring. In Inman WHW (ed). Monitoring for Drug Safety. Second edition. Lancaster, England: MTP Press, 1986:213-235. 4. Inman WHW, Rawson NSB. Erythromycin estolate and jaundice. British Medical Journal 1983;286:1954-1955. Inman WHW. Risks in medical intervention. In Cooper M. (ed). Risk: Man-Made Hazards to Man. Oxford: Oxford University Press, 1985. 6. Inman WHW, Rawson NSB, Wilton LV, Pearce GL, Speirs CL. Post-marketing surveillance of enalapril. I: Results of prescription-event monitoring. British Medical Journal 1988;297:826-829. 7. Speirs CJ, Dollery CT, Inman WHW, Rawson NSB, Wilton LV. Post-marketing surveillance of enalapril. II: Investigation of the potential role of enalapril in deaths with renal failure. British Medical Journal 1988;297:830-832.

Next: 7. The Role of Decision Analysis in the Translation of Research Findings into Clinical Practice »
Modern Methods of Clinical Investigation Get This Book
×
Buy Hardback | $49.95
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The very rapid pace of advances in biomedical research promises us a wide range of new drugs, medical devices, and clinical procedures. The extent to which these discoveries will benefit the public, however, depends in large part on the methods we choose for developing and testing them.

Modern Methods of Clinical Investigation focuses on strategies for clinical evaluation and their role in uncovering the actual benefits and risks of medical innovation.

Essays explore differences in our current systems for evaluating drugs, medical devices, and clinical procedures; health insurance databases as a tool for assessing treatment outcomes; the role of the medical profession, the Food and Drug Administration, and industry in stimulating the use of evaluative methods; and more.

This book will be of special interest to policymakers, regulators, executives in the medical industry, clinical researchers, and physicians.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!