Anesthesiologists, like all other specialists, need to examine carefully their clinical practices so that excessive costs and waste can be reduced without compromising patient care or safety. While costs of drugs used for anesthesia constitute only a small fraction of total health care cost, they are highly visible costs which are easy for administrators to scrutinize. Although cost savings in an individual case may be small, the total savings may be impressive because of the large volume of cases performed. In a recent analysis of strategies to decrease PACU costs, Dexter and Tinker found that anesthesiologists have "little control over PACU economics via the choice of anesthetic drugs". Greater savings could be achieved by timing the arrival of patients into the PACU to reduce the peak requirement of nursing personnel. Hospital and operating room management would be better served by concentrating on these simple measures to improve efficiency rather than forcing anesthesiologists to base drug usage on acquisition costs. Even in countries that have nationalized health services, salaries make up the largest part of the costs, and the expenses in delaying an operation by 30 min exceeds the costs of a 2 h propofol infusion. It is becoming increasingly apparent that attempts at better scheduling of cases, more efficient processing of patients in the PACU to optimize admission rates, and reduced wastage of anesthetic and surgical supplies lead to greater savings than reducing anesthetic-related drug costs. Nevertheless, it is still important for anesthesiologists to participate in the ongoing effort to reduce medical costs without affecting the quality of patient care. Quality care and fiscally sound decision-making are not necessarily mutually exclusive. Simple, effective cost containment measures that all anesthesiologists can practice include using low fresh gas flow rates with inhalation agents and opening sterile packages and drug ampules only if the contents will be used. The choice of an anesthetic agent for routine use depends not only on its demonstrated efficacy and side effect profile, but also on economic factors. It is important to perform careful pharmacoeconomic evaluations of these newer drugs, including assessing all associated costs and benefits for subsets of patients undergoing different types of surgical procedures. These evaluations should also include input from patients regarding their personal preferences. Excessive emphasis on the acquisition costs of drugs may lead to blanket bans on the use of new drugs because of their higher costs rather than permitting physicians to individualize therapy according to their clinical experience and the perceived needs of a given patient. Institutional and individual variations in clinical practices, their associated costs and outcomes may alter conclusions about acceptability and economic evaluation of a particular drug or technique. The information in this review can be used to provide a rational basis for incorporating cost considerations into the decision-making process regarding the drugs, devices and techniques used in anesthesiology.
Concerns regarding soaring health care costs have led to debate of the best methods to contain these costs.  When health insurance agencies paid physicians and hospitals on an indemnity basis, the relationship between institutional costs and revenues discouraged attempts to control costs and encouraged development and widespread use of new drugs and technology. [1,2] Because direct payment for procedures performed encouraged more procedures, new methods of payment, such as diagnosis-related groups (DRG), managed care programs and capitation fees for health care services, were introduced to decrease health care costs.  In managed care, there is an integration of financing and delivery of health care, whereas capitation involves payment of a fixed amount for all medical services. These two programs have been successful in reducing expenditures, at least in the short term, because payments tend to remain constant regardless of the number of services provided.  In this new health care environment, clinical departments and services with high resource utilization such as the operating rooms will no longer be considered “profit centers” but rather “cost centers” for hospitals.  The current challenge is to maintain the same high quality of care while consuming fewer resources. [1,6–8] Clinical practice patterns including work-force modifications are now being examined to determine if they are associated with the best outcome at the most reasonable cost, [9–11] a concept termed value-based anesthesia care.  Increased value can be obtained by either achieving the same outcome at a lower cost or a better outcome at the same or a higher total cost. 
If anesthesiologists fail to participate in this process of determining the value obtained for the resources used, the result may be that nonmedical personnel will dictate changes in the organizational structure and in the utilization of pharmaceutical products, equipment, and supplies in our practice. [7,10–12] This article will review the terminology, methods, limitations, and some reported results of economic analyses of anesthetic techniques and devices. The ethics of cost-containment, the basis for pharmaceutical company charges for drugs, and issues of who should administer the anesthetic are beyond the scope of the article. Given institutional variations in costs and clinical practices (including surgical, anesthesia and recovery times) for the same procedure, these examples can be used as templates for constructing economic evaluations that reflect practices in the anesthesiologist's own institution.
The four most commonly used methods for economic analysis in health care are: [13–15]
1) Cost-minimization. This involves comparison of acquisition costs of various alternative drug regimens without regard to the outcome or associated side effects (e.g., emesis, delayed awakening and discharge). Unless there is equality of outcomes, other methods should be used. [16,17]
2) Cost-benefit analysis. This is a comparison of pertinent costs and the consequences or outcome (benefit) in monetary terms. [18–20]
3) Cost-effectiveness analysis. This expresses the costs of an intervention in units of success or effect (e.g., cost per mmHg reduction in blood pressure, costs per patient free from a postoperative complication). This analysis, rather than a cost-benefit analysis often is performed because of the difficulty in converting outcomes to monetary values. [14,21]
4) Cost-utility analysis. This analysis is similar to cost-effectiveness analysis, wherein the measure of effectiveness includes patient preferences and satisfaction about their quality of life by expressing outcome in terms of the quality of adjusted life years (QALY). 
Estimate of Total Costs
Assessment of costs is an essential part of any evaluation of the economic impact of a health care intervention and is similar for all four methods of analysis. The description and analysis of the costs of drug therapy to health care systems is termed pharmacoeconomics. 
Definitions and types of costs
Definitions of the cost accounting terms average, marginal, fixed, semi-fixed, and variable costs are listed in Table 1. The total costs associated with a medical intervention consist of direct and indirect costs. The direct costs of drug treatment are not limited to the acquisition cost of the amount of drug administered, but they include the costs of drug wasted, equipment to administer the medication (e.g., intravenous sets, syringes), pharmacy dispensing costs, and the costs of managing any drug-induced side effects. From the patient's perspective, total costs include nonmedical direct out-of-pocket costs (e.g., transportation to the site of therapy, family lodging, and home help care), along with indirect costs from the lost wages of the patient and the family member taking care of the patient. [15,24]
The term indirect costs is used differently by physicians, accountants, and health care economists. Although some anesthesiologists include costs of managing side effects and delayed recovery as indirect costs,* most health care economists would describe these as associated direct costs. [26–28] Accountants include all fixed costs (e.g., administration, engineering, housekeeping, utilities, upkeep, bond payments, depreciation) in their calculation of indirect costs, whereas economists usually refer to indirect costs as the costs related to lost productivity. [26,29]
Costs versus charges
Evaluations of costs in anesthetic practices sometimes refer to costs, charges, and health insurance company payments as if they are synonymous; however, these terms are not interchangeable.  Although payment for a patient charge may represent cost to the insurance company or payer, it does not necessarily represent the true costs of providing the service. The identification of the true costs of services provided by an institution is complex and involves use of techniques developed by cost accountants and industrial engineers.  Some studies used a conversion factor to convert hospital charges to presumed costs (cost-to-charge ratio). [31,32]
The rationale for using charges as a substitute for costs is based on the assumption that the cost-to-charge ratio is constant for all services. However, hospitals frequently mark-up the charges for services in one area to invest in new facilities and to pay for unreimbursed care provided elsewhere in the facility (cost-shifting). Macario et al.  determined a mean cost-to-charge ratio of 0.42 for all services to hospitalized patients undergoing four common operations (discectomy, appendectomy, prostatectomy, and laparoscopic cholecystectomy). However, the mean cost-to-charge ratios for the surgery admission unit, radiology department, blood bank, patient ward and laboratory (0.92, 0.63, 0.53, and 0.50, respectively) were significantly higher than for the anesthesia department (0.29) at this university hospital. Studies using charges as a substitute for costs would overestimate resource consumption by the anesthesia department compared with these other hospital departments.
Perspective of person assessing costs
Cost assessments differ depending on the perspective of the person making the assessment. Common perspective in health care cost analyses include 1) society as a whole, 2) health insurance companies or other payers, 3) health care providers, and 4) the patient. What is relevant from one perspective may be irrelevant from another. For example, cost of an illness to an insurance company is represented by payments for outpatient and hospital care, along with professional and drug fees. Costs from the patient's perspective include insurance premiums, out-of-pocket expenses (copayments, deductibles, drugs, uncovered services, cost of travel to the physician's office or hospital facility), and lost wages. For a hospital, the costs of resources used in the management of an illness include personnel costs, acquisition costs of drugs and equipment, and the prorated costs of the hospital's physical plant (e.g., rent, bond payments, utilities). Because labor costs constitute a large proportion of costs, [34,35] accurate determination of costs requires time and motion studies. The hospital administration may arbitrarily assign specific costs to individual departments. For example, cost of drugs used in the operating suites and post-anesthesia care units (PACU) is often assigned separately to the budgets of the anesthesia and nursing departments, respectively. Thus, it is important to identify the perspective from which the health care economic analysis is performed. Many health care economists recommend that wide-ranging policy decisions on drug use should be based on cost assessments that take the widest possible perspective (i.e., that of society as a whole). [17,24] Cost considerations from a narrower perspective may overlook shifting cost from one segment of society to another (e.g., early discharge shifts recovery costs from the ambulatory surgery facility to patients, their family, and to their primary care practitioner ). Direct and indirect costs should be included in costs of the medical or surgical management of a specific condition when the perspective taken is that of society as a whole.
Potential for cost reductions in the operating rooms
Macario et al.  noted that half of the intraoperative anesthesia costs could be influenced by the choice of agents and anesthetic techniques as they were direct, variable costs. Although savings per case would be small, the potential for major savings were large because the individual savings would be multiplied by the thousands of operations performed each year.  In a recent editorial, Orkin  noted that the influence of anesthesiologists on perioperative costs extended beyond the intraoperative phase. Enhanced pain therapy with epidural infusions and patient-controlled analgesic devices may result in fewer complications and a shorter length of stay in hospital. Traditionally, these increased costs of analgesic therapy (but not the benefits of earlier discharge or increased patient satisfaction) have been assigned to the department of anesthesia. A wider perspective should be taken to identify cost savings for perioperative anesthetic interventions.
Intraoperative anesthesia costs constituted only 5.6% of total hospital costs, whereas one third of total hospital costs for surgical inpatients were incurred in the operating room suites,  with personnel costs being a major component. Many opportunities to reduce costs in the operating room suites may be found if a lower-cost configuration of personnel can provide the same quality of services. These workforce modifications include cross-training of personnel, replacement of scrub nurses by operating room technicians, and a more flexible work schedule arrangement to match supply and demand for personnel.  Structural changes in facilities (e.g., constructing a separate phase II PACU facility wherein family members can aid in the care of the patient) may help to reduce costs by promoting a more efficient use of personnel. For pediatric patients in particular, the presence of parents in the recovery room may reduce the need for sedation and its consequent side effects of delayed discharge, airway obstruction, and an increased duration of oxygen therapy.
Time-and-motion studies have been used to identify personnel costs, assuming that there is a linear relationship between labor costs and the time spent in providing a service. [37,38] It is important to remember that from the perspective of the institution, these costs are semi-fixed and not variable.  Even if the patient spends an extra 15–30 min in the PACU, institutional costs may not be affected, unless there is a consequent bottleneck in the flow of patients through the operating room or procedure suites  or it happens with sufficient frequency that additional staff are hired. Similarly, if all other aspects of care remain unchanged, more rapid awakening after anesthesia may not be associated with decreased institutional costs unless an additional case can be completed with the same staff. However, if use of short-acting drugs permits the safe transfer of patients directly from the operating room to the less labor-intensive phase II recovery area, it may be possible to provide appropriate patient care with fewer personnel or overtime payments, leading to cost savings. In an institution working at near capacity, the more rapid movement of patients through the post-anesthetic recovery phase may obviate the need to hire more personnel. Marketing claims of a reduction in personnel or in other associated direct costs with the use of a drug or device should be subjected to the same rigorous scrutiny as claims of scientific merit. 
Practical Applications of Economic Evaluations
Definition and methods of analysis. Cost-minimization involves a determination of the least amount of money required to provide a service without regard to patient outcome. Although physicians are responsible for health care expenditures, they often remain unaware of either the specific costs or charges associated with their interventions. [41,42] For example, only 21% of a group of 19 anesthesiologists in a teaching institution in the United States were aware of the charges for equipment they used on a regular basis (e.g., epidural and radial artery catheters).  Johnstone and Jozefczyk  demonstrated that a simple education program informing anesthesiologists of the costs of anesthetic drugs at their tertiary care center in the United States was associated with a 23% reduction in monthly drug expenditures without affecting quality of care. However, such behavioral changes were short-lived, and within 2 months of completing the education program, use of the more expensive drugs in the hospital returned to baseline. Because the study failed to differentiate between drug usage by anesthesiologists in the operating rooms and by other physicians in the intensive care unit (ICU), increased use of high-cost drugs such as vecuronium and midazolam in the ICU rather than in the operating rooms may have been responsible for the results. These authors suggested that extraordinary efforts by drug company representatives were a factor in the return to high-cost drug usage.  An alternative explanation is that it is difficult to maintain changes in physician practice patterns without repeated reinforcement.  It also has been suggested that a perceived direct reward for cost-minimization activities is required because there will be strong opposition to any rigid system that places arbitrary restrictions on drug or equipment use.  With a change in health care payments from a per service to a per capita basis, there may be a direct financial incentive for anesthesiologists to reduce drug and equipment expenditures in all aspects of clinical practice. Anesthesiologists working in more cost-constrained environments, such as the British National Health Service, are aware of drug and equipment costs and the need to limit expenses. [35,45]
Examples of cost-minimization. Single versus multiple dose vials. An example of such a cost-minimization practice is provided by comparing costs of single-dose and multidose drug vials. The acquisition price of a unit dose withdrawn from a multidose preparation may be lower than for single-dose preparations.  For example, the average wholesale price quoted for 500 mg thiopental from a 5 gm preparation is lower than for a single-dose preparation (1994 prices, 3.48 dollars for the multidose vs. 8.44 dollars for a single- dose preparation ). Similarly, the average wholesale price of ketorolac 15 mg, 30 mg, and 60 mg was quoted as 7.05 dollars, 7.52 dollars, and 7.75 dollars, respectively in 1994.  Thus, there is a major potential cost savings to a hospital pharmacy with fractionating (unit) doses of expensive medications, such as antibiotics, midazolam, ketorolac, and ondansetron, particularly for pediatric patients,  provided the additional work will not require hiring more personnel. The disadvantages of the use of multidose vials include the extra costs of personnel and equipment for withdrawing a unit dose and the associated potential for drug wastage and contamination with blood and bacteria during the withdrawal. However, in one study, drug expenditures were claimed to be reduced by having a pharmacist prepare solutions for intravenous use under a laminar flow hood compared with having nurses mix the drugs on the hospital ward before administration.  Although this practice was associated with increased pharmacy labor costs, there were overall cost savings to the institution because of reduced drug wastage, decreased use of supplies and lower nursing labor costs.  In anesthetic practice, cost savings from the use of multidose vials depend on adherence to proper technique: the practice of using drugs from the same syringe on multiple patients is an example of false economy because any gains from reduced drug use may be negated by the potential for cross infection. 
Fresh gas flow rates.
Another simple example of a cost-minimization practice is the preferential use of low fresh gas flow rates. In one study, reducing fresh gas flow rates from 8 l/min to 4 l/min was associated with a 55% decrease in the cost of isoflurane without altering quality of care.  Although it is possible to use much lower flow rates (< 1 l/min) during isoflurane anesthesia, the outflow concentrations of vapor required during the early, rapid-uptake phase may exceed the limits of standard isoflurane vaporizers at this flow rate and may be achieved only by direct injection of liquid isoflurane into the circuit. Many clinicians may find this impractical, but the practice of reducing flow rates to 1–3 l/min after an initial high flow rate for 5–10 min when uptake is rapid is a simple, effective cost savings measure. With less soluble gases, it is practical to use low flow rates much earlier, even during the rapid uptake phase, to achieve greater cost savings. The acquisition costs of potent anesthetic agents for 1 MAC-h of anesthesia at a flow rate of less than 1 l/min are lower for desflurane than isoflurane, but isoflurane usage has a lower cost if flow rates exceed 1 l/min, even after allowing for several initial minutes of high gas flow rates to provide a sufficient quantity of isoflurane vapor.  Direct cost comparisons of sevoflurane and desflurane are not available in the peer-reviewed literature, but the methods used to estimate the costs of agent required to achieve 1 MAC-h of anesthesia have been described.  At the minimum flow rate recommended by the Food and Drug Administration (FDA) for sevoflurane (2 l/min), the costs of 1 MAC-h of sevoflurane and desflurane are similar (6.77 dollars vs. 6.75 dollars). Because desflurane may be safely administered at even lower flow rates, a comparison with greater clinical relevance would involve the costs of desflurane at flow rates of 1 l/min and sevoflurane at 2 l/min. At these flow rates, sevoflurane would be significantly more expensive than desflurane (6.77 dollars per MAC-h for sevoflurane vs. 4.62 dollars for desflurane).  These cost-minimization analyses do not adjust for costs of different delivery systems (vaporizers) or differences with respect to induction times, adjunctive drugs, or early recovery (e.g., earlier awakening, tracheal extubation, and decreased time between the end of surgery in one patient and the incision in the followed case).
Older versus newer drugs-Neuromuscular blocking drugs. Recent suggestions for cost-minimization (or cost-containment) in anesthetic practice include preferential use of the less expensive induction agents (e.g., thiopental, methohexital), inhalation anesthetics (e.g., halothane, enflurane), opioids (e.g., morphine, meperidine, fentanyl), and neuromuscular blocking agents (e.g., pancuronium) without consideration of any benefits associated with the use of the newer, more expensive agents, or adverse events related to older drugs.  Some older drugs (e.g., metocurine) may be as expensive as newer drugs. A large number of patients undergo anesthesia wherein the pharmacokinetic properties of drugs such as alfentanil, propofol, vecuronium, and midazolam may offer no particular advantages over less expensive, longer-lasting drugs.  For example, one institution claimed to have achieved annual savings of 100,000–150,000 dollars by substituting fentanyl for the more expensive opioid sufentanil.  Even greater savings in their drug budget could have been achieved by using morphine. [25,55] However, the cost savings may have been offset by prolonged awakening and extubation times. It remains to be seen if the faster recovery from the new, ultra-short-acting opioid remifentanil will be associated with any true cost savings as a result of earlier discharge from either the more intense observation areas such as the PACU or from the ambulatory surgery center. The incidence of pain and PONV after use of remifentanil may be more important factors in times to discharge than the times to eye opening and response to commands.  It is important that these cost factors be considered in the design of clinical trials before the approval of new drugs by regulatory agencies. [18,57] Emerging market forces in the health care industry may be more successful in convincing pharmaceutical manufacturers to provide such data to practitioners and the regulatory agencies.
Studies involving use of neuromuscular blocking drugs also suggest that there are marked differences in costs between different muscle relaxants with the newer agents, usually, but not always, costing more than the older “generic” drugs. [58,59] As mentioned previously, metocurine is an exception. Substituting pancuronium for vecuronium in long cases (> 90 min) with no contraindication to the use of either drug can result in annual savings of more than 100,000 dollars in the drug budget of the anesthesia department with an annual caseload of 16,000 patients.  However, at a major university medical center in the United States, instituting explicit practice guidelines that limited neuromuscular blocking drug use to pancuronium for cases longer than 90 min increased the average time from end of surgery to PACU arrival (11 min vs. 14 min; SD not provided; P < 0.05).  In this preliminary report, the clinical importance of this additional time was unclear because the authors stated no case was canceled or additional costs incurred from overtime payments to OR and PACU personnel as a result of this change.
The number of patients having clinically significant residual paralysis on arrival in the PACU is decreased with the use of atracurium and vecuronium compared with pancuronium.  In the report mentioned previously, the incidence of mechanical ventilation in the PACU increased from 2.7% to 3.2%, but this did not reach statistical significance.  The authors did not detect any major adverse outcome that could be related to the use of the practice guidelines but acknowledged that 36,000 patients would have to be studied to obtain a power of 0.8 for this conclusion.  The experience of spending more time in patient observation and airway manipulation in the immediate postoperative period when older, longer-acting agents are used may influence anesthesiologists' practice patterns, especially in an atmosphere of increased pressure not to slow the processing of patients through the operating room suites. [54,62]
In the study by Johnstone and Jozefczyk,  a resurgence in the use of shorter-acting drugs occurred after a brief decrease during the education program, despite a failure to detect changes in outcome. Johnstone and Jozefczyk claimed sales efforts of pharmaceutical company representatives motivated anesthesiologists to keep using more expensive drugs.  However, these sales efforts have failed to popularize pipecuronium and doxacurium in clinical practice. We suggest that preference for using repeated doses of an intermediate-acting drug rather than a single dose of a longer-acting drug is based on the perception that the former practice permits improved titration of the anesthetic drug when duration of surgery is unpredictable.  Marketing attempts to encourage use of flumazenil to antagonize residual sedation also have not been successful, perhaps because of the potential for resedation because the effect of flumazenil is shorter than the effect of the benzodiazepine agonist drug.
Pharmaceutical companies are aware of the increased interest in acquisition costs of drugs and are taking steps to price their drugs to maintain market share after the loss of patent protection or the introduction of a competing drug in the same class.  It is of interest that the acquisition price of an intubating dose of the newest neuromuscular agents, rocuronium and cis-atracurium, are similar to vecuronium. However, once patent protection of vecuronium expires, generic versions should cost less. Such a decrease in costs has already been seen with isoflurane. Similarly, costs of propofol, alfentanil, and sufentanil should decrease within a few years once other pharmaceutical companies market generic versions of these drugs.
Blood products. Another example of the effect of education in cost-containment was provided by Nightingale et al., who evaluated blood product use in cardiac surgery before and after an educational program designed to minimize use of these products.  Although these authors claimed cost savings as a result of these efforts, they noted that the use of blood salvage devices was associated with cost savings only if packed red cell infusions were reduced by two or more units. These cost estimates were conservative because the authors did not place values on the lower risks of disease transmission, sensitization against blood products (platelet and leukocyte antigens), and transfusion reactions with autologous compared with homologous blood transfusions. 
Limitations of cost-minimization. Decisions based on the acquisition cost of drugs, without considering personnel costs, may fail to achieve the desired cost savings. For example, a time and motion study  of the procedures involved in administration of opioid analgesics revealed that despite higher acquisition costs of controlled release morphine tablets (MS Contin(R)), there were overall cost savings compared with equipotent oral morphine or a continuous subcutaneous morphine infusion (Table 2). These cost savings came from a significantly lower pharmacy staff, nursing, and ancillary supply costs.
Cost analysis of the financial consequences of a medical intervention requires a careful assessment of patient outcome because the duration of action and incidence of side effects differ. In the example of neuromuscular blockade mentioned previously,  it was unclear if the cost savings in the pharmaceutical budget from the use of pancuronium exceeded the increased expenditure from managing postoperative airway obstruction and hypoventilation from residual paralysis.
Definitions and methods of analysis. A cost-benefit analysis defines the monetary value of the benefits obtained for the money spent. In this analysis, the specific intervention or therapeutic program is identified along with the resources consumed and benefits obtained from the intervention. A monetary value is assigned to these resources and benefits, and the net benefits (total benefits-total costs) or the ratio of benefits-to-costs is then calculated. The need to discount benefits that accrue during a number of years is uncertain, but discounting is irrelevant for anesthetic drugs because the duration of action is brief.  Benefits may result from decreased use of a drug, a lower incidence of adverse drug interactions, or a decreased use of resources to manage side effects. In the previous comparison of an oral solution versus controlled release of morphine, the cost-benefit ratio was in favor of the controlled release formulation (Table 2).
Two different approaches have been taken for assigning the monetary value of the benefits resulting from a health care intervention. In the “human capital” approach, it is assumed that humans (like capital equipment) have a flow of productive activity in the years ahead and that the benefits of an intervention can be measured in terms of its potential interruption of this flow of income.  Criticisms of the human capital approach include 1) ethical objections to placing a monetary value on human life, 2) placing a value of health benefits based on income ignores the value of benefits to those individuals who do not receive financial compensation for their efforts (e.g., housewives, retired people, temporarily unemployed), and 3) it does not account for the individual's view of the benefits resulting from an intervention. An affluent individual may consider costs of a drug relatively favorable compared with its benefits, whereas an unemployed person may not.
The second approach involves asking patients how much they are willing to pay to avoid the consequences of an illness. Sawaki et al.  noted that 94% of patients stated that they would chose patient controlled analgesia (PCA) for the management of acute postoperative pain if they required surgery in the future; however, only 40% believed an additional cost was justified. The questions asked in the willingness-to-pay method may be open-ended or discrete.  In the open-ended model, the questioner makes a starting bid, which the subject can accept or reject. Bids are increased or decreased until the maximum willingness-to-pay is reached. For example, a subject may be asked “Would you be prepared to pay 100 dollars for a drug that will reduce your chance of postoperative nausea and vomiting from 70% to 20%? What if the costs were 5 dollars? How about 50 dollars or 25 dollars?” Unfortunately, the results from this simple method may depend on the initial bid. In the discrete method, patients are asked to answer yes or no to a series of prices. Regardless of the method of asking questions, answers may vary with the patient's income level, education, and previous experiences of a complication. Perhaps a better approach would be to ask how much a subject is willing to pay for additional insurance to indemnify the costs of a therapy to avoid a particular side effect (e.g., nausea, pain, fatigue, dental damage), or complication (e.g., pulmonary embolus, renal failure).
The difficulty in assigning a monetary value on benefits is a major limitation of cost-benefit studies. One advantage of a cost-benefit analysis compared with the other techniques is that it permits administrators to decide between different programs with unrelated outcomes when resources are limited. For example, an anesthesiology department may need to decide between opening a pain management clinic or providing additional coverage for the intensive care unit.
Examples of cost-benefit analysis. Operating room pharmacy. The benefits from establishing a satellite OR pharmacy in a community hospital were estimated to exceed the costs.  In this study, costs included the salary of the pharmacist assigned to the area, whereas the benefits included recovery of lost revenue by better documentation of drug charges, decreased drug wastage, lower drug inventory, and better regulation of controlled drugs.  Although it was difficult to place a monetary value on the consequent improved communications between the OR and pharmacy staffs, this was presumably an added intangible benefit.
Capnographs and pulse oximeters. Roizen et al.  compared the costs of installing capnographs and pulse oximeters in all operating rooms with the benefits resulting from the decrease in arterial blood gas measurements. These investigators noted that savings from the 44% reduction in the number of blood gas determinations exceeded costs of the new technology. Although not clearly stated in the study, the perspective was limited to that of the health care institution because there were no assumptions regarding benefits from potential lives saved or reduced morbidity from early detection of untoward anesthetic-related events, such as airway obstruction, circuit disconnects, inadvertent esophageal intubation, one-lung ventilation, or unplanned tracheal extubation. Limitations of the study included a lack of assumptions regarding additional costs of managing “false-positive” alarms from the monitors. 
Regional versus general anesthesia. In a study of high-risk patients undergoing major surgery, Yeager et al.  noted that intraoperative and postoperative epidural analgesia reduced mortality and morbidity compared with intravenous opioids. These investigators suggested that epidural analgesia was associated with a 30% reduction in hospital charges.  A more complete cost-benefit study from the perspective of society as a whole would require estimates of the true hospital costs, not charges, along with a monetary value assigned to the life years saved and to the increased productivity from reduced morbidity.  In a randomized study of patients undergoing colon surgery, Liu et al.  demonstrated that epidural analgesia with bupivacaine and morphine provided the best balance of analgesia and freedom from side effects, whereas accelerating recovery of gastrointestinal function and fulfillment of discharge criteria. However, more recent studies have questioned these conclusions regarding reduced perioperative mortality with regional anesthesia in a patient population at high risk for cardiac complications.  In a study of 423 patients randomly assigned to epidural, spinal, or general anesthesia for femoral-distal vessel bypass surgery, there were no statistically significant differences with regard to in hospital mortality, nonfatal myocardial infarction, angina, and congestive heart failure.  Epidural analgesia has been associated with improved pain relief, reduced perioperative physiologic alterations, and decreased coagulability (with a significant reduction in venous and arterial thromboses). [72,73,75] Unfortunately, these studies have provided few (if any) data on the times patients return to work. It is easier to claim economic benefits of epidural anesthesia and analgesia if their use was associated with an earlier return to work. Future studies should be designed to provide this information.
Limitations of cost-benefit analysis. A major problem with cost-benefit studies is the difficulty in placing relative monetary values on the desired therapeutic effect and the unwanted side effects of a drug. It is often assumed that the value of avoiding a side effect is the same as the value of achieving a desired effect. For example, in a comparative study of the pharmacoeconomics of ondansetron, droperidol, and metoclopramide,  the authors recommended the preferential use of droperidol over the other two antiemetics for the prophylaxis of PONV. However, monetary values were not placed on anxiety and restlessness, delayed (post-discharge) side effects of prophylactic droperidol during ambulatory surgery.  The monetary value placed on avoiding a night of restlessness and intermittent nausea “clearly depends upon who is at the receiving end.”
Neuromuscular blocking drugs. Many studies use the terms cost-effectiveness and cost-benefit incorrectly.  For example, Donati identified the least expensive strategy for the establishment, maintenance, and antagonism of neuromuscular blockade during surgical procedures of varying duration.  He assumed that residual nondepolarizing neuromuscular blockade would be antagonized by atropine and neostigmine. Although the study was alleged to be a cost-benefit analysis, it did not provide monetary values for any benefits and hence was a cost-minimization study. For a 60-min case, the least expensive strategy was to use a loading dose of succinylcholine followed by a continuous infusion (total cost Canadian 8.77 dollars) compared with the use of vecuronium or atracurium (18.99 dollars if no succinylcholine is used, or 12.54 dollars if succinylcholine-facilitated tracheal intubation was followed by a nondepolarizing agent).  In contrast, for a 120-min case, Donati  calculated that it was more economical to use succinylcholine followed by d-tubocurarine or pancuronium (4.06 dollars and 12.24 dollars, respectively) compared with vecuronium, atracurium, or doxacurium (22.28 dollars, 28.74 dollars, and 16.08 dollars, respectively). However, total costs may be higher than stated because the author did not identify costs associated with drug wastage and with management of side effects such as prolonged blockade with the longer-acting agents, potential myocardial ischemia from pancuronium-induced tachycardia, or hypotension from curare-induced histamine release. In another study comparing use of doxacurium, pipecuronium, and pancuronium in patients undergoing coronary artery grafts, no episodes of myocardial ischemia occurred in the pancuronium group despite the increase in heart rate.  Thus, these authors concluded that the purported benefits from doxacurium and pipecuronium did not justify their increased costs over pancuronium.
Preoperative laboratory tests. The value of routine screening preoperative laboratory testing has been examined. A false-positive result from these screening tests will require additional investigation, time, and expense and may lead to invasive diagnostic tests with a potential for directly harming the patient. [81,82] In contrast, a positive screening test may lead to early discovery of a potentially treatable illness, with a consequent savings to the patient, society, and presumably to the health insurance company. However, studies have shown that a battery of routine screening tests before elective surgery in healthy patients does not contribute to better perioperative management compared with tests based on a patient's history and physical examination.  In 3,782 healthy patients at the Mayo Clinic, routine preoperative blood testing revealed abnormalities in 160 patients, but only one patient benefited by initiation of new medical treatment as a consequence of screening laboratory tests.  Because this study used satisfactory perioperative outcome as the end-point instead of defining economic benefits from early detection and treatment, it was appropriately termed a cost-effectiveness study and not a cost-benefit study.
Capnographs and pulse oximeters. The limitations of cost-benefit analysis are well demonstrated by the difficulty in establishing the value of pulse oximetry from a societal perspective as opposed to the institutional perspective used in the study by Roizen et al.  There is a widespread belief that pulse oximetry enhances patient safety and may be an important factor in the recent decrease in anesthetic-related morbidity and mortality when used in every OR and PACU, as recommended by the American Society of Anesthesiology (ASA):  Although there is little doubt that pulse oximetry detects hypoxemia earlier than clinical observation,  the relationship between its use and improved patient outcome has not been firmly established.  Although unplanned admissions to the ICU have been reported to be decreased in patients who underwent intraoperative pulse oximetry monitoring, other confounding factors in the study prevented definitive conclusions about its beneficial effects. [88,89] A clinical trial of more than 20,000 patients failed to detect differences in the incidence of adverse events after pulse oximetry monitoring. [90,91] This study had a 90% chance of detecting a 25% reduction (from > 3% to 2.25%) in the incidence of major cardiorespiratory adverse events such as perioperative pneumonia and cardiac arrhythmias. However, less common complications such as cardiac arrest, brain damage, and death would require a sample size of approximately 2 million. In the absence of such data, it is difficult to prove pulse oximetry monitoring is associated with an economic benefit, viz. reduced mortality and morbidity. Yet, it is highly likely that pulse oximetry monitoring will remain a standard of care regardless of demands that widespread adoption of new technology should be limited to devices and procedures with a demonstrated superior cost-benefit ratio. [85,92]
The absence of reliable data on the incidence and costs of managing side effects of anesthetic-related interventions is another limitation of cost-benefit analysis. Hence, techniques such as cost-effectiveness and cost-utility analysis are more commonly used.
Definitions and methods of analysis. Cost-effectiveness analysis examines costs associated with a unit of success without placing a monetary value on the success.  To perform a cost-effectiveness analysis, it is essential to define the measure of effectiveness. For example, it may be expressed as mmHg reduction in blood pressure, years of life saved, number of successfully diagnosed cases, or number of patients free from a specific complication. Jolicoeur et al  have described a 10-step approach to ensure that a complete economic evaluation is performed. These steps include a determination of the perspective of the person performing the evaluation, followed by a definition of the problem, outcomes and alternative solutions, and the allocation of the probabilities for each outcome. The health care and non-health care resources consumed in each pathway are listed, and a monetary value assigned; the unit of effectiveness is then defined, and data are analyzed. Finally, the effect of varying the probabilities of outcome and the assigned costs on the overall conclusions is examined as part of a sensitivity analysis.
The concept of cost-effectiveness is most useful when comparisons are made between new therapeutic approaches and older established drug regimens. If a cost-effectiveness study of a new antiemetic drug (e.g., ondansetron) finds that it will cost 50 dollars per emesis-free patient, the practitioner will have to make a subjective judgment to determine whether the use of the drug is appropriate. However, if a comparative study also finds that use of a less expensive and less effective antiemetic costs more than 50 dollars per emesis-free patient, a stronger case can be made for the new drug. This approach is termed a marginal cost-effectiveness analysis and has the following equation:
Marginal cost effectiveness Equation 1.
Traditionally, the pharmaceutical industry has not obtained these data during the rigidly controlled, phase III patient trials conducted before the release of a new drug.  Because controlled clinical trials are the best source of efficacy and side effects data, it has been suggested that economic analysis be included as part of the early clinical trials involving investigational new drugs and medical technology.  In some countries (e.g., Australia), regulatory agencies will approve a safe efficacious drug, but require that manufacturers submit economic analyses in support of a request for registration of the product on the national formulary.  Drugs from this formulary are available for little charge to the patient. Canadian authorities have proposed guidelines for cost analysis that will be required in future applications for approval for formulary listing and reimbursement.  At the present time, the US FDA requires only demonstration that a new drug is efficacious and has a satisfactory side effect profile before it is released for clinical use. It has been suggested by some health care economists that the FDA should require economic evaluations as part of the approval process.** However, costs and practices vary from institution-to-institution in the United States, making it controversial to perform a cost analysis that will apply to the entire country. Regardless of regulatory agency approaches, market factors in the US managed health care environment have increased the importance of establishing the cost-effectiveness of new pharmaceutical products (e.g., desflurane, rocuronium, sevoflurane, remifentanil) and surgical techniques (e.g., laparoscopic herniorrhaphy, endoscopic carpal tunnel release) to determine their eventual role as a replacement for existing drugs and techniques. Pharmaceutical companies have realized that emphasis is being placed on costs by Pharmacy and Therapeutics committees*** and have started including such material in their marketing presentations (e.g., cis-atracurium product brochure). Because of an anticipated increase in such presentations, the FDA Division of Drug Marketing, Advertising and Communications has developed a draft of the principles to be followed in its review of pharmacoeconomic claims in promotional material.****
Decision analysis. The techniques used in cost-effectiveness analysis are based on a derivative of operations research and game theory called decision analysis. [95,96] Unlike the traditional randomized clinical trial that attempts to determine efficacy of one drug versus a placebo or another drug, decision analysis involves a systematic approach to decision-making under conditions of uncertainty. The temporal and logical consequences of actions and events after a particular choice can be represented in a diagram called a decision analysis tree (Figure 1). Assigning the probability of each outcome requires a large scale prospective study. In the absence of prospectively obtained data, the probability of each outcome may be based on a meta-analysis of multiple smaller studies or on Delphi surveys (i.e., consensus of opinion among experts in the field).  Each of these methods introduces a potential for error in the conclusions. In cost-effectiveness analysis, costs are assigned to each endpoint or node of the decision analysis tree, and the product of costs and the probability of the patient arriving at that endpoint provides the nodal costs. The sum of the nodal costs will provide the total costs associated with a decision. When this sum is divided by a measure of efficacy (or effectiveness), it provides the cost-effectiveness ratio (i.e., the costs for a single unit of effectiveness). Computer programs are available to simplify the calculations and to provide the degree of confidence attached to the conclusion. [97,98]
The decision analytic approach was used to examine the risks and benefits of performing abdominal aortic surgery alone or performing preoperative cardiac testing followed by coronary artery revascularization (or angioplasty) before abdominal aortic aneurysm repair.  This study suggested that either strategy was equally justified, given the reported range of mortality associated with the coronary disease and the aortic surgery without previous coronary surgery. A similar approach was used to compare the cost-effectiveness of droperidol, metoclopramide, and ondansetron in the prophylaxis of PONV.  This analysis suggested that droperidol was more cost-effective than ondansetron, and both were more cost-effective than metoclopramide. Moreover, prophylactic use of any of the three drugs was cost-effective compared to treatment of PONV, but only in out-patients at high-risk of developing these symptoms.
Sensitivity analysis. A number of critical assumptions have to be made in order to calculate these cost-effectiveness ratios. Often a risk-averse strategy is taken where the assumptions are loaded against the new technology or therapy, and if the new therapy still has a better cost-effectiveness ratio than the older (standard) therapy, it is clearly the one to be preferred.  A more reasonable test of the robustness of the conclusions drawn from a comparative cost-effectiveness analysis is achieved by sensitivity analysis. This type of analysis involves examining the effects of varying underlying key assumptions (e.g., the probability of an outcome at a branch point in the decision analysis tree), to determine the range over which the conclusions remain unchanged. If the conclusions are not changed over a “best case-worst case” scenario, it is difficult to argue against validity of the relative cost-effectiveness analysis. Sensitivity analysis is the most important part of any cost-effectiveness analysis, and failure to include a sensitivity analysis can limit validity of the conclusions of cost-effectiveness comparisons. 
Examples of cost-effectiveness analysis. By definition, cost-effective treatments are not necessarily the least expensive options available. An expensive therapeutic regimen may be the most cost-effective if associated with a lower incidence of side effects. For example, the cost of performing a laparoscopic cholecystectomy is greater than an open cholecystectomy because the surgical procedure takes more time and involves use of expensive equipment and supplies. Yet, more rapid recovery, shorter hospitalization, and earlier resumption of normal activities after a laparoscopic surgical approach make it more cost-effective.  In a study of the cost-effectiveness of vecuronium, and pancuronium, DeMonaco and Shah defined the measure of efficacy as the number of myocardial ischemic episodes related to heart rate increases that were avoided by the use of vecuronium.  The authors calculated the relative costs of preventing each ischemic event by first determining the differences in costs of pancuronium and vecuronium, and dividing this difference by the probability of a perioperative ischemic event as estimated from the Goldman cardiac risk index.  In this study, the relative costs for preventing an ischemic episode by using vecuronium in preference to pancuronium was $ 2,450 in a 62 year old patient with diabetes and exercise-induced angina undergoing a colectomy. The costs of avoiding such an episode in an otherwise healthy 41 year old undergoing a femoral-popliteal bypass graft were estimated to be $ 175,000. Although all costs were not considered, this study demonstrated the need for individualizing decisions on drug usage based on the patient's underlying medical condition and the type of surgical procedure.
Propofol versus inhalation agents. It has been suggested that use of propofol regimens for outpatient anesthesia is cost-effective as it results in more rapid recovery, decreased PONV, and/or a shorter time to being judged “fit for discharge”.  Sung et al.  reported that patients receiving propofol spent 15 min less in the labor-intensive phase I recovery area compared to those receiving a standard thiopental-isoflurane regimen. However, these authors did not include the costs of wasted propofol, infusion equipment, and supplies, and the potential decrease in costs of the inhalation regimen with the use of low fresh gas flows. [101,102] Marais et al. used a computer simulation of the flow of patients through a recovery room area and claimed that preferential use of propofol over a thiopental-isoflurane regimen would reduce nursing costs 25%.***** These comparisons were based on theoretical rather than actual savings (i.e., decreased overtime or decreased hiring of OR/PACU personnel). A definitive study is still required to confirm that propofol usage is associated with real cost savings. This type of study must go beyond merely demonstrating earlier recovery in the operating room and PACU with propofol, but must show that its use is associated with decreased times to actual discharge, reduced hiring of nurses and support personnel, or the completion of at least one additional case in the same OR session. In a study by Suver et al.,  the theoretical total costs for surgery were reduced in an HMO-owned hospital when propofol was used for induction and maintenance of anesthesia compared to a thiopental/isoflurane regimen. Yet, this did not translate into decreased discharge times.
In contrast to studies of propofol costs in outpatient anesthesia, more reliable data are available on the clinical effects, safety, and economic costs of propofol and midazolam for sedation of critically-ill patients requiring mechanical ventilation in the ICU.  In a randomized study, a group of investigators noted that patients could be maintained at a satisfactory level of sedation for a larger percentage of time with propofol than midazolam (mean hours of adequate sedation as a percentage of total infusion time, 93% for propofol vs. 82% for midazolam, P < 0.05). In spite of the more constant level of sedation with propofol, recovery times to tracheal extubation and discharge from the ICU after discontinuation of the sedative medication were shorter and less variable (Table 3). Compared to the midazolam group, the higher costs of propofol were exceeded by the benefits from the reduced costs of post-sedation care if short-term sedation (< 24 h) was required. However, if duration of sedation exceeded 6 days, the larger costs of propofol exceeded the savings from a reduced ICU stay due to the more rapid weaning from mechanical ventilatory support.
Postoperative analgesia. Another example of cost-effectiveness analysis is provided in a cost comparison of morphine via the epidural, intrathecal,, and intravenous patient-controlled analgesia (PCA) routes of administration and “on demand” intramuscular (IM) administration (Table 4). Data in the table are based on multiple published studies, including a meta-analysis of randomized clinical trials of PCA. [105–109] While total costs were lowest in the IM group, patient satisfaction with analgesia therapy was also the lowest. If therapeutic effectiveness is defined the percentage of as patients who are very satisfied with the quality of analgesia, epidural morphine was associated with the lowest cost per “very satisfied” patient (table 4), and this would represent a cost-effectiveness study. These complex issues of effectiveness, costs, side effects, and patient satisfaction with different pain control techniques still need to be examined in greater detail in large, prospective, randomized studies.
Cost-effectiveness analyses of analgesic medication also appear to differ depending on the type of surgery and the compared drug. Trotter et al. suggested that use of ketorolac as a primary postoperative analgesic was associated with decreased overall resource costs per patient compared to meperidine or morphine after open cholecystectomy but not after joint replacement surgery.  Thus, conclusions about the cost-effectiveness of a drug after one surgical procedure may not apply to all procedures, and arbitrary limitations on the postoperative use of ketorolac without consideration of the type of operation may not achieve the desired cost savings. Similarly, beneficial effects of epidural anesthesia on return of bowel function have been noted after colon surgery but not after radical prostatectomy. [73,111]
Spinal versus epidural. A recent cost-effectiveness analysis compared spinal versus epidural anesthesia for elective cesarean section.  Hospital costs (including costs of supplies, drugs, and time spent in the operating room, PACU, and resources used in managing complications) were compared between the two groups (Table 5). Although the perspective of the study was not explicitly stated, it appeared to be limited to that of the hospital. Patients in the epidural group spent significantly more time in the operating room because of the increased time from start of anesthesia until skin incision compared to the spinal anesthesia group. In addition, supplemental intravenous analgesic and sedative use was greater, and the complication rate was higher in the epidural group, while times spent in the PACU were similar. The authors concluded that spinal anesthesia was a better choice in this patient population because it was faster to perform, resulted in a more comfortable patient and utilized fewer resources. The cost differential would have been even greater if the authors had added costs for the time of the anesthesiologist and surgeon, as well as the costs for managing side effects. In a study of younger adults (< 50 years of age) undergoing urological, orthopedic, vascular or plastic surgery, Seeberger et al. also concluded that spinal anesthesia was preferable to epidural anesthesia, based on the decreased time to achieve a block and a lower incidence of incomplete blocks, pain during surgery, and post-lumbar puncture backache. 
Anesthetic related cost-effectiveness studies with methodological limitations. There has been a dramatic increase in published economic assessments in health care,  but the quality of these economic evaluations varies widely.  An inappropriate specification of costs is a frequent methodological problem in these analyses.  Another continuing problem is misinterpreting cost savings and cost-minimization as cost-effectiveness or cost-benefit studies. [116,117] Sacristan et al.  have published a check list to evaluate quality of economic studies, with particular emphasis on identifying the study aim, perspective, benefits, costs, and ethical issues. Anesthesiologists may be interested in economic evaluations in two areas: (a) comparisons of diagnostic and therapeutic procedures performed with or without the presence of an anesthesiologist, and (b) comparisons of general, regional, and local anesthesia with sedation (monitored anesthetic care-MAC). Unfortunately, the quality of many studies of economic evaluations on these topics have not met the standards described above. [96,115]
Sedation by anesthesiologists versus surgeon. Interest in the role of the anesthesiologist in diagnostic procedures has assumed greater importance with the development of fiberoptic scopes and specialized catheters for use in minimally-invasive surgical and radiological procedures. Many disorders that traditionally required major, open surgical procedures can now be managed with minimally invasive procedures. Claims of cost-effectiveness of the newer surgical techniques are based on the more rapid recovery of the patient and on the ability to perform these procedures with local anesthesia and sedation. These claims have been made in studies of arthroscopy, [118,119] hernia repairs,  laparoscopic tubal ligation, [121,122] breast biopsy,  endoscopic gastrostomy , and tonsillectomy.  There have also been claims of additional savings if an anesthesiologist is not involved in providing MAC,  and further savings if the procedure is performed outside the operating rooms.  However, different critical aspects of the research design for economic evaluation were not met in these reports. These analyses may be faulted for: (a) not explicitly identifying the perspective of the study, (b) using hospital charges rather than costs, [120,123,124](c) analyzing retrospective data, [120,123](d) having small group sizes that fail to detect rare but life-threatening complications, which require a considerable expenditure of resources to manage, (e) failing to ensure homogeneity of the comparative groups, [121,125] and (f) not considering personnel costs and inconvenience to patients and staff if the sedative regimen failed. [120,124]
Harshfield et al.  contended that extra time and costs involved with use of epidural anesthesia during interventional biliary tract procedures were justified in view of the better pain control, fewer complications, and decreased distractions of the radiologist caused by variations in the patient's status during the procedure. While it is difficult to place a monetary value on increased pain and less-than-optimum intraoperative conditions when a local analgesic-sedative regimen is used, it is clear that adjustments in costs should be made for managing adverse effects and for failure to complete a procedure under sedation. In a study of arthroscopy performed under surgeon-administered local anesthesia and sedation, 15% of procedures had to be abandoned because of inadequate pain control.  When an anesthesiologist was present to provide monitored anesthesia care, the technique had to be changed to general anesthesia in 12% of the cases. The conclusions regarding the value of having an anesthesiologist present for all sedation cases would depend on the frequency of failure and side effects of a surgeon-directed sedation-analgesia regimen. Monetary values will also need to be placed on the time and resources spent by the physician and patient when the regimen failed.  The paucity of reliable data on the incidence of sedation-related problems in the absence of an anesthesiologist makes this economic analysis extremely difficult.
Use of general versus regional versus local anesthesia with sedation. Methodological problems are also noted in comparing cost-effectiveness of various anesthetic techniques, such as general, regional or local anesthesia with sedation. Although there are some data available on the costs associated with each anesthetic technique, [34,127] comparisons of their relative cost-effectiveness may not be valid because patients in each anesthetic group had different medical conditions, different operations, and were not prospectively assigned at random to an anesthetic technique. For example, in a non-randomized, retrospective study of 248 surgical procedures performed in a teaching hospital in France, the total cost (drugs and personnel) of a general anesthetic was 2.5 times higher than the costs of a local/regional technique, but this could be explained by differences in the complexity and duration of the surgical procedure and the severity of underlying illness.  In a similar study, Dexter and Tinker noted an apparent relationship between anesthetic technique and time to discharge.  However, these authors cautioned that retrospective data of patients undergoing different types of surgery “should not be used to conclude that using monitored anesthesia care would decrease time to discharge and thus save money.” It is important that data on the effect of anesthetic technique on costs be collected prospectively after a large number of patients undergoing similar operations are randomized to one of three anesthetic regimens-general, regional or MAC. Unfortunately, no such study is readily available, although a number of outpatient procedures such as inguinal herniorrhaphy,  tubal sterilization, [121,122] arthroscopy,  extracorporeal shock wave lithotripsy [128,129] can be performed with any one of the three techniques.
The inadequacy of cost data in healthy patients does not permit definitive conclusions in favor of any one anesthetic technique in that patient population, although opinions regarding their relative cost-effectiveness of have been vigorously espoused. [120,125,130] Claims for decreased costs are based on decreased complication rates and shorter times spent in the more labor intensive areas-the operating room and the PACU. Proponents of regional anesthesia contend that it is associated with decreased time from the end of surgery to arrival in the PACU, less postoperative pain, sore throat, emesis, and a lower risk of life-threatening complications compared to general anesthesia. [131,132] Surgeons are more concerned about technical failures of regional anesthesia and delays waiting for the block “to begin working”.  Perhaps differences in opinion can be explained by institutional differences in clinical practices and costs. If a practitioner has to complete the anesthetic care of one patient before being available for the next patient, costs will differ from those in an institution where the nerve block can be placed outside the operating room (e.g., in the preoperative holding area, induction area or PACU) while surgery is being completed on another patient. Others believe there are significant cost advantages to using local anesthesia with sedative-analgesic medications (MAC) in selected cases. [120,122,125,133] The relative lack of major side effects and the ability for the patient to be transported directly from the operating room (OR) to a phase II step-down observation area, without first being observed in the PACU will contribute to decreased costs in the MAC group. In addition, residual analgesia from the local anesthetic facilitates earlier discharge after ambulatory surgery.  However, if institutional policies mandate a minimal period of observation in the PACU, economic benefits of MAC may not be obtained.
There have been few randomized trials of general anesthesia versus MAC. Bordahl reported a decreased use of drugs and disposable anesthetic supplies in a study of 125 women who underwent laparoscopic tubal sterilization during a midazolam-alfentanil MAC compared to a propofol-alfentanil-atracurium general anesthetic ($ 21 vs. $ 46).  Although preoperative and operative times were similar, post-surgery time was significantly greater in the general anesthesia group (8 +/- 3 vs. 3 +/- 1 min). In addition, the incidence of postoperative sore throat (70% vs. 3%) and pain (80% vs. 33%) were higher in the general anesthesia group along with the need for analgesic therapy (33% vs. 14%). Petersen et al. confirmed these findings in another study and added that the incidence of emesis was also higher in the general anesthesia group (19% vs. 14%).  In comparing general and local anesthesia for inguinal herniorrhaphy, Behnia et al. noted decreased costs from the perspective of the health insurance company because of the elimination of fees for anesthesia and a decreased recovery room fee.  Similar cost savings were claimed for local anesthesia during tonsillectomy.  Bredenkamp et al.  also contended that decreased blood loss was an additional benefit in the local anesthesia group.
General anesthesia has been compared with epidural anesthesia for knee arthroscopy.  Increased sore throat, muscle aches, dizziness, nausea, emesis, analgesic requirements and longer mean discharge times were reported in the general anesthesia group. In a retrospective study of 278 patients undergoing knee arthroscopy, Erickson et al. noted a greater failure rate and patient dissatisfaction with spinal anesthesia compared to general anesthesia and surgeon-directed local anesthesia with sedation.  While 11% of patients who received a spinal anesthetic required general anesthesia for supplementation, no patient who received the sedative-local anesthesia regimen required a general anesthesia. It is unclear if these results reflect proper patient selection or inadequate anesthetic management.
In other studies, patient acceptance of the sedative-analgesic regimen has been reported to similar to that of general or central neuraxial anesthesia. [132,133,135] Sedative regimens with local anesthesia have proved successful in cataract surgery.  Because patient cooperation and a gentle surgical technique are essential for success of intravenous sedation-analgesic techniques, the decision to use MAC in place of general or regional anesthesia should be made only after carefully assessing patient and surgeon preferences, as well as any associated medical conditions (e.g., obesity, COPD, heart disease, previous surgery, and anesthesia). However, the potential for serious airway-related problems with deep sedation mandates the need for careful patient selection and monitoring if an anesthesiologist is not available. The American Academy of Pediatrics guidelines for sedating pediatric patients requires the presence of an individual whose sole function is to monitor the patient during the procedure. 
Laryngeal mask versus tracheal tube. Cost-effectiveness analyses in anesthetic practice have not been limited to drug therapy, but have increasingly been extended to anesthetic devices. Joshi et al. compared direct costs associated with the use of a laryngeal mask airway (LMA) and a tracheal tube and concluded that the LMA device was more cost-effective than a tracheal tube if vecuronium was used to facilitate intubation and if the LMA was reused at least 10 times.  These authors also concluded that succinylcholine-facilitated tracheal intubation and spontaneous ventilation was more cost-effective than the LMA unless the LMA could be reused 150 times. In this study it was assumed that anesthetic requirements would depend on the surgical procedure and hence would be the same in all three airway management groups.
In a similar study, Macario et al. compared hypothetical costs of four airway management techniques during isoflurane anesthesia: (a) LMA with spontaneous ventilation, (b) face mask with spontaneous ventilation, (c) tracheal intubation after succinylcholine, followed by spontaneous ventilation, and (d) tracheal intubation after non-depolarizing neuromuscular blockade and controlled ventilation.  These authors concluded that a face mask was the most cost-efficient airway device for short cases. If the LMA was used 40 times before being discarded, it was the most cost-efficient device for cases longer than 40 min (Table 6). The costs of neuromuscular blockade did not affect the case duration at which the LMA became the most cost-effective strategy, provided the use rate of the LMA was > 25.  This study explicitly identified the perspective from which the economic analysis was performed, specified the available alternatives, assigned costs for equipment, drug, and personnel use and for management of complications, and provided a sensitivity analysis. However, data were not collected prospectively, and the conclusions depended on published data from other institutions and on a number of key assumptions. Critical cost assumptions included (a) higher fresh gas flow rates in the face mask group, and (b) increased perioperative anesthetic and analgesic requirements in the tracheal tube group. The last assumption was supported by a study of 44 patients, where anesthesiologists used higher isoflurane concentrations in the tracheal tube group immediately after airway insertion and skin incision, but not by the end of surgery.  Macario et al. assumed that differences in the consumption of inhalation agents between the groups would not change during the operation, but it is likely that any differences would decrease after skin incision.
The validity of assigning costs from one institution to published data on outcomes from another institution may be questioned as practice patterns differ between institutions and may not be identified in the study design. The Task Force For Economic Analysis of Health Care Technology has recently stated that the study design with the highest internal validity for evaluating both clinical and economic effects is a randomized controlled trial.  There are also suggestions that the same statistical principles used in clinical trials to calculate confidence intervals and sample size should be applied to health economic studies.  A well-designed prospective study is still essential to determine cost-effectiveness of the laryngeal mask against other methods of airway maintenance. For example, data obtained prospectively on durability of the LMA, the actual costs of inhalation agents and resources used in managing side effects, along with the incidence of successful use of the LMA to maintain the airway when tracheal intubation failed, are needed. These data can be used in the decision analysis model presented in Figure 1to calculate cost-effectiveness ratios for various airway management strategies, along with their confidence intervals.
Limitations of cost-effectiveness analysis. Cost-effectiveness analyses in anesthetic practice depend critically on reliability of data regarding the incidence of complications and the costs assigned to their management. The most reliable data for efficacy and safety of a drug comes from the rigidly controlled environment of a phase III clinical trial before the release of a drug. Data form these trials have commonly been used to evaluate the economic effects of introducing a new drug. However, the economic aspects of drug usage should be based on answering questions on drug effectiveness (what happens under actual conditions of use in clinical practice) rather than on efficacy (what happens under ideal conditions).
Short acting drugs for ambulatory surgery. Although randomized, controlled trials have established that emergence from anesthesia after surgery is more rapid with use of desflurane, sevoflurane or propofol compared to older agents, actual cost savings achieved by using these more expensive drugs has not been established.  It is difficult to use these data from published reports in cost comparisons of the newer drugs, because of the reasons listed below:
1) These randomized controlled trials are performed on highly selected patients, and consume more resources as the protocols often require additional laboratory tests, a longer minimum hospital stay, and more intense monitoring than in routine practice.  In addition, these protocols may call for perioperative management that may not reflect common clinical practices. For example, some published recovery data on newer anesthetic drugs are based on recovery from end-tidal concentrations equivalent to a fixed multiple of MAC, [143,144] or from infusion techniques designed to maintain a desired blood level of propofol. [145,146] Because most anesthesiologists reduce anesthetic concentrations with the anticipated end of a case, the value of these data in cost effectiveness studies is questionable.
2) Many studies were designed to compare surrogate end points such as earlier awakening, return to baseline on psychometric tests, or a decreased incidence of emesis after ambulatory surgery.  In general, it requires fewer patients to establish statistically significant differences in these endpoints than in other outcomes of interest for comparisons of cost-effectiveness (e.g., duration of recovery room stay, incidence of unplanned hospital admissions, resumption of normal activities, and patient satisfaction). Hence, there may be a type II error in the conclusions of a lack of significant differences in these non-surrogate outcomes. While pooling of data from multiple studies in a meta-analysis can increase the power of the analyses, meta-analysis requires careful assessment of the quality of each study, with emphasis on blinding, randomization and homogeneity of the patient populations. Potential problems with meta-analysis have been extensively reviewed elsewhere.  Nevertheless, meta-analyses have confirmed the association between propofol and decreased emesis,  and the efficacy of PCA devices compared to conventional IM opioid therapy. 
3) The true incidence of side effects is essential for economic analyses of drug usage, but the incidence detected in phase III controlled trials of anesthetic drugs, where a limited number of healthy patients are usually recruited and cared for under ideal conditions, may not reflect what happens during actual conditions of use by practitioners. In addition, some rarer (but resource consuming) side-effects may be noted only in sicker patients, and only after more widespread use. For example, the effects of etomidate on the adrenal axis were not detected in phase III trials, but only after the drug was used as an infusion for sedating ventilated patients in the ICU. Given inter-institutional variability in the amount and costs of resources used to manage these side effects, it is difficult to establish both statistically significant and clinically important differences for the estimates of total costs between alternative management regimens. Cost comparisons of anesthetic techniques (e.g., comparisons of general, central neuraxial or local anesthesia with sedation) can also be criticized for their failure to provide confidence limits of their conclusions. [104,105,109,116]
4) There are also limited data on delayed recovery after discharge and the time for patients to return to work after ambulatory surgery.  These data are important in calculating costs from the perspective of the patients because earlier awakening, return to baseline of psychometric tests and discharge from hospital may not translate into a “real world” increase in economic benefits to the patients unless they are able to return to work earlier. Even if the analysis is performed from the perspective of a hospital in a managed care environment, earlier awakening would be associated with decreased costs only if the time saved could be used to complete additional cases during the OR session or overtime personnel costs were reduced.  This is possible only in an institution with a very high OR utilization rate or long waiting lists. Otherwise, a more rapid emergence from anesthesia may be associated with more time when the operating room personnel were not being utilized. In institutions where delayed discharge from the PACU does not prevent continued use of the OR for succeeding cases or necessitate overtime OR/PACU nursing coverage, it will be more difficult to demonstrate that earlier awakening is associated with decreased real life or “bottom line” costs from the perspective of the institution. Recently, Dexter and Tinker concluded that PACU economics are affected by the peak number of patients admitted to the unit rather than the choice of anesthetic drugs.  Thus, if the patient can go directly from the operating room to the phase II recovery unit (which usually has a larger patient: staff ratio), there is a potential for cost savings.
5) Finally, the evaluation of costs in a cost-effectiveness analysis will vary with differing work patterns in different institutions. Minor changes in work patterns may have a greater effect on the appropriate use of an OR/PACU facility than the differences in prices between the two anesthetic drugs. For example, a study of practices in outpatient surgery centers has determined that time between the end of surgery in one case and the surgical incision in the next case varies with institutional practices.****** Less time is spent between cases in institutions where preparation of the operating room, induction of anesthesia and the surgical scrub and gowning processes are performed simultaneously compared to institutions where these activities are performed sequentially. Examples of work patterns influencing costs may also be noted in institutions that follow protocols mandating a minimum duration of stay in the PACU regardless of the patient's discharge readiness, and in institutions that do not have a separate phase II recovery area where nursing labor requirements are lower. Personnel costs constitute a major portion of the overall costs (even in countries with national health care systems). 
The limitations of cost-effectiveness analyses of anesthetic practices should not discourage practitioners from performing these studies. However, readers of these studies should examine carefully the assumptions on which comparisons are based to determine if conditions in their institution are comparable before accepting the conclusions. It may be more helpful to use models described in various studies as templates to recalculate cost-effectiveness ratios based on actual data from the individual hospital's personnel costs, complication rates, duration of stay in the PACU and phase II recovery areas.
Monitoring-use of central venous versus pulmonary artery catheterization. Examination of anesthetic costs should not be limited to the use of pharmaceutical products and airway devices, but should be extended to other anesthetic practices, including the reuse of airway circuits, face masks, esophageal stethoscopes, as well as physiological and anesthetic monitors  to determine if costs can be reduced without increasing complications. In a theoretical evaluation of the cost-effectiveness of pulmonary artery (PA) catheters, Spackman noted that the routine use of a PA catheter would be cost-effective if the incidence of mortality was reduced by 0.2%(from 3% to 2.8%) compared to a group where only the central venous pressure (CVP) was monitored.  However, the theoretical approach was criticized because it oversimplifies the cost and benefit parameters.  One suggestion has been to choose a subset of patients for PA pressure monitoring (e.g., patients with low ejection fraction, triple-vessel disease, recent preoperative myocardial infarct) rather than use it routinely for all patients undergoing a specific type of surgery. If the end-point to determine the cost-effectiveness of CVP versus PA pressure monitoring was limited to demonstrating reduced mortality, the number of patients required would make the study prohibitively expensive. However, if morbidity and the quality of life in the additional years gained by the preferential monitoring of PA pressures were used as the endpoints, the study would be more feasible.  Rapid adoption of perioperative transesophageal echocardiography may make the question of PA versus CVP monitoring moot. In turn, cost-effectiveness of the transesophageal echo device (versus invasive monitoring) may become a more important issue in the future.
We contend that anesthesiologists should have a voice in the choice of capital expenditures in the operating and recovery rooms, as well as other hospital-based devices (e.g., infusion pumps, PCA devices), and that purchasing decisions should be based on a careful examination of costs and benefits of each piece of equipment (including maintenance costs). [67,154] These examinations of costs and benefits of drugs, devices, and monitoring equipment will require collecting large amounts of data. The availability of computerized anesthesia records and patient care data bases can assist physicians in obtaining important data regarding the quality of patient care.  With the appropriate environment and a non-threatening focus on processes and systems of care, continuous quality improvement (CQI) can be instituted in the operating room suites.  However, the potential misuse of these automated data bases for economic credentialling of practitioners by managed care organizations may lead to development of a “Big Brother is watching you” attitude among practitioners.
While the cost of a drug should enter into the quality-of-care decision, it should not be the sole criterion for drug usage, as the preferences of the patients should also be considered. In a cost-utility analysis, the outcome of a procedure is expressed as a “utility”(i.e., a subjective assessment of the level of well-being in different health states). Utilities are values assigned to a health state using a scale where a value of 1.0 is the healthy state and a value of 0.0 represents death. There are three methods for obtaining utility values for health states in cost-utility analyses: (1) judgment of the analyst, (2) values from the medical literature, and (3) values from measurements on a cohort of subjects.  Although the use of the analyst's judgment is the simplest method of assigning utility values, this approach is associated with considerable interindividual variability. Because there are few published data on the “quality of life” after anesthetic-related interventions, these data can only be obtained from measurements in a sample of subjects. 
In a cost-utility analysis, the benefit of a health service is stated in terms of the number of quality-adjusted life years (QALY), a composite index that measures the number of additional years of life obtained by a medical intervention and adjusts it for the quality of life during those years.  In this type of analysis, a year of poor health is less desirable than a year of good health, and some health conditions are considered worse than death.  Most studies on QALY involve chronic conditions such as rheumatoid arthritis, cystic fibrosis, cancer and chronic renal failure requiring dialysis. [22,158] Unfortunately, there are few data on alterations of the quality of life by anesthetic-related complications (e.g., dental damage, embolic phenomena after central venous line placement, paralysis after central neuraxial or peripheral nerve blockade), but the methods used in studies of patients with chronic medical problems can be adapted to the postoperative period. [22,158] Both general and disease-specific measures of the quality of life are available.  General measures include the sickness impact profile, and the quality of well-being scale, while disease-specific measures of the quality of life have been designed for rheumatoid arthritis, renal failure, and lung transplantation.  However, these disease-specific scoring schemes cannot be easily translated into a single measure of quality of life that can be used to compare outcomes between different management programs for different diseases.
One of the earliest and most widely used measure of the quality of life is the Rosser scale, which describes health status in terms of one of eight disability categories and one of four distress categories (Table 7).  These two category scores (distress and disability) are converted into a quality of life score (Table 7). [22,162] These Rosser scales can be used to obtain data on the quality of life after surgery. If the same surgical procedure was performed using different anesthetic techniques, the Rosser scales can be used to compare the effects of these techniques on the quality of life in the early postoperative period. Subsequently, the costs to obtain one QALY (or its equivalent) with each anesthetic technique can be compared.
There are few data on patient preferences for anesthetic-related outcomes. The ASA has established a task force to determine the best anesthetic outcome at a reasonable cost. In the USA, the Agency for Health Care Policy and Research (AHCPR) has been examining the effectiveness of medical interventions in terms of outcomes of importance to patients cared by typical health care providers. The AHCPR-funded patient outcome research teams have used a questionnaire developed from the Rand Corporation's health insurance study. From the original list of 149 questions developed and tested on a population of over 22,000 patients as part of a medical outcome study, 36 items were selected.  This SF-36 form measures 8 multi-item variables: (1) limitations in physical activities (10 items), (2) social functioning (2 items), (3) role limitations due to physical problems (4 items), (4) bodily pain (2 items), (5) role limitations due to emotional problems (3 items), (6) mental health (5 items), (7) energy and vitality (4 items) and (8) general perception of health (5 items). There is also a single unscaled item on changes in health over the previous year. This questionnaire has been tested for internal consistency, construct validity, and clinical validity, and normative data are available for different patient populations. [164,165] Although there are no data on its use to measure short-term changes in patient well-being after surgery, there are data on using the SF-36 form to measure the long term success of total hip and knee replacement in patient terms.  This questionnaire may be useful in determining the effect of anesthetic interventions on patient well-being as judged by an individual patient.
Methods of assessing patient preferences. Anesthetic techniques, like other medical interventions, may be associated with undesirable side effects, and the choice of a specific technique often involves a trade off between desirable and undesirable effects. Orkin asked subjects to rank structured postoperative scenarios drawn from a combination of factors such as mental acuity, fatigue, pain, and emetic symptoms.  A conjoint analysis (a multivariate statistical method) was then used to determine preferences for various possible outcomes after anesthesia. Subjects in this study stated they were willing to tolerate moderate pain and decreased mental acuity to obtain relief from PONV.
There are few (if any) data regarding individual patient preferences for anesthetic related outcomes, other than the report by Orkin. However, patient preferences for non-anesthetic related outcomes have been determined using the rating scale, category rating, “standard gamble”, “time trade-off”, and “person trade-off” models. [22,115,158] Readers are referred to the review article by Torrance for details of these methods.******* While each model has its proponents, the relationships between different methodologies are complex. 
It may be possible to apply these methods to determine patient preferences for anesthetic-related outcomes. However, it would be necessary to first establish reproducibility, content and construct validity, and reliability of these instruments in anesthesia,  because the effects of anesthesia are short-lived and rarely affect long-term outcome. In addition, risk aversion and previous experiences with a complication (e.g., previous history of PONV) may lead to significant inter-individual differences in the values placed on a given post-anesthetic complication.
In spite of limitations of these analytic instruments, there is clearly a need for studies that examine patient preferences for a given outcome rather than the surrogate outcomes being reported in most studies of anesthetic drugs and techniques.  With the growth in capitated reimbursement, there will be increasing pressure to utilize less costly drugs and equipment while adhering to more rigid guidelines or treatment protocols.  Use of more expensive anesthetic drugs can only be justified if they have a better safety profile, improve patient comfort (as assessed by the patient), and/or facilitate the recovery process.  As physicians, it is our duty to ensure that each patient's values and preferences are considered in the decision-making process, as it is our primary responsibility to provide the best possible care for our patients. 
It is also possible that in many circumstances, a patient has no preference for a specific anesthetic technique and there are few clinically important differences in outcome between two or more anesthetic techniques. For example, there were no differences in the incidence of new postoperative neurological deficits, or in the total hospital stay and charges, when elective supratentorial craniotomy was performed with propofol/fentanyl, isoflurane/nitrous oxide or fentanyl/nitrous oxide anesthesia.  Similarly, in a study of 1094 patients undergoing coronary artery revascularization, Tuman et al. found no differences in serious pulmonary, renal, neurologic outcomes or in the incidence of perioperative myocardial infarction and mortality between 5 anesthetic techniques.  In this study, anesthesiologists chose freely between high-dose fentanyl, moderate dose fentanyl, sufentanil, diazepam-ketamine or thiopental-halothane as the primary anesthetic. More recent data have also failed to demonstrate differences in cardiac outcome after peripheral vascular surgery under regional or general anesthesia. 
It has been suggested that the least expensive anesthetic drugs and techniques should be used when there are no differences in outcome. This is based on the principle that the preferred health care intervention is the one that provides more benefits per dollar.  These recommendations are based on the assumption that all practitioners will have the same outcome with a technique. Yet, there are data to suggest outcomes vary between practitioners. [174,175] Thus, the best outcome for an individual anesthesiologist may occur with the use of the most familiar technique. In the absence of clear-cut patient preferences or differences in outcome or cost, it would be reasonable for two anesthesiologists to choose different techniques based on their previous experiences.
Anesthesiologists, like all other specialists, need to examine carefully their clinical practices so that excessive costs and waste can be reduced without compromising patient care or safety.  While costs of drugs used for anesthesia constitute only a small fraction of total health care cost, they are highly visible costs which are easy for administrators to scrutinize.  Although cost savings in an individual case may be small, the total savings may be impressive because of the large volume of cases performed.  In a recent analysis of strategies to decrease PACU costs, Dexter and Tinker found that anesthesiologists have “little control over PACU economics via the choice of anesthetic drugs.” Greater savings could be achieved by timing the arrival of patients into the PACU to reduce the peak requirement of nursing personnel. Hospital and operating room management would be better served by concentrating on these simple measures to improve efficiency rather than forcing anesthesiologists to base drug usage on acquisition costs. [35,127] Even in countries that have nationalized health services, salaries make up the largest part of the costs, and the expenses in delaying an operation by 30 min exceeds the costs of a 2 h propofol infusion. 
It is becoming increasingly apparent that attempts at better scheduling of cases,  more efficient processing of patients in the PACU to optimize admission rates,  and reduced wastage of anesthetic and surgical supplies  lead to greater savings than reducing anesthetic-related drug costs.  Nevertheless, it is still important for anesthesiologists to participate in the ongoing effort to reduce medical costs without affecting the quality of patient care. Quality care and fiscally sound decision-making are not necessarily mutually exclusive.
Simple, effective cost containment measures that all anesthesiologists can practice include using low fresh gas flow rates with inhalation agents and opening sterile packages and drug ampules only if the contents will be used. [35,45] The choice of an anesthetic agent for routine use depends not only on its demonstrated efficacy and side effect profile, but also on economic factors. It is important to perform careful pharmacoeconomic evaluations of these newer drugs, including assessing all associated costs and benefits for subsets of patients undergoing different types of surgical procedures. These evaluations should also include input from patients regarding their personal preferences. Excessive emphasis on the acquisition costs of drugs may lead to blanket bans on the use of new drugs because of their higher costs rather than permitting physicians to individualize therapy according to their clinical experience and the perceived needs of a given patient.  Institutional and individual variations in clinical practices, their associated costs and outcomes may alter conclusions about acceptability and economic evaluation of a particular drug or technique. The information in this review can be used to provide a rational basis for incorporating cost considerations into the decision-making process regarding the drugs, devices and techniques used in anesthesiology. 
*Tuman KJ: Value-based anesthesia management: Choice of anesthetic techniques. ASA Newsletter June 1995; 59(6):28–30.
**Bootman JL: Pharmacoeconomics and outcomes research. Am J Health-Syst Pharm 1995;52(Suppl 3):S16-S9.
***Langley PC: Pharmacoeconomics and the quality of decision-making by pharmacy and therapeutics committees. Am J Health-Syst Pharm 1995;52(Suppl 3):S24-S6.
****Talley CR: Pharmacoeconomic principles. Am J Health Syst Pharm 1995;52:1871.
*****Marais ML, Maber MW, Wetchler BV, Kortila K, Apfelbaum JL: Reduced demands on recovery room resources with propofol (Diprivan) compared to thiopental-isoflurane. Anesthesiology Review 1989;16:29–40.
******Cattalini D, Chin KM, Partridge JS, Kirstein B: Ambulatory surgery best practices. Arthur Anderson & Co. SC 1992.
*******Torrance GW: Measurement of health state utilities for economic appraisal. A review. Journal of Health Original Investigations 1986;5:1–30.