Methodology of cost-effectiveness studies

There are several different ways health costing studies can be considered and quantified. Approaches to analyses typically include the following:

The perspective of a cost analysis refers to who is bearing the costs. This may be the patient, the hospital, a third-party payer or society overall. It is generally recommended that all costs and outcomes should be considered from a societal perspective, because the patient or anesthesiologist may not appreciate what is truly cost-effective, and they or the hospital are not usually responsible for all cost outlays.

In cost-effectiveness analysis, it is conventional to distinguish between the direct costs and the indirect costs associated with the treatment, perhaps including intangibles, which may be difficult to quantify, but are often consequences of treatment or opting for no treatment. Direct costs might include drug and disposables costs, staff time to deal with PONV, cleaning and linen, and any other patient expenses associated with PONV. Indirect costs include family burdens and productivity losses; intangibles might include patient distress and other adverse effects.

In 1994, it was estimated that each episode of emesis delays discharge from PACU by approximately 20 min[24]. These costs are likely to be a lot higher in contemporary settings.

The incremental cost-effectiveness ratio (ICER) is the ratio of the change in costs to incremental benefits of a therapeutic intervention or treatment, calculated as[22]:

where:

Costs are described in monetary units and benefit/effect on health status is usually measured in terms of QALYs gained or lost. That is, the numerator measures treatment costs and the denominator places a monetary cost on the health consequences. Increased length of life, and/or quality of life, will increase QALYs. One QALY is equal to 1 year of life lived in perfect health. A medical complication leading to a 25% reduction in quality of life but with no effect on longevity will result in a 0.75 QALY. QALYs can be used to rank any number of competing treatment options or to compare two treatment options.

Most national drug approval or funding agencies expect cost-effectiveness to be expressed as a cost per QALY gained or lost. That is, in terms of society’s willingness to pay for an additional unit of health gain.

In the UK, the National Institute for Health and Clinical Excellence (NICE) uses QALYs as the “common currency” to enable comparisons across therapeutic areas. The ICER is expressed as the cost per QALY gained and can be compared with those of other interventions or with a national threshold value of what is considered to represent cost-effectiveness. NICE has a range of acceptable cost-effectiveness of around £25,000 (US$55,000) per QALY[10].

Another approach that has been used in the PONV literature is the willingness to pay method. For example, several studies have found that patients are willing to pay approximately $100 to avoid experiencing PONV[25,26], and another study found parents are willing to spend approximately $80 to prevent postoperative vomiting (POV) in their children[27]. The threshold for the willingness to pay method is about $50–100,000 in the USA[28]. A willingness to pay rate of $100 per case avoided makes PONV prophylaxis cost-effective in clinical settings with a baseline incidence of PONV of 40%.

The QALY approach can be modified to suit a perioperative, particularly ambulatory surgery setting, such as if a proposed prophylactic treatment costs $11.00 and on average increases a person’s quality of recovery score 2 points on a 0–10 point scale, say from 6 to 8, for the next 24-h period and they otherwise recover fully at home (i.e., they will eventually recover full health), we can calculate a QALY as:

Estimates of cost-effectiveness can vary markedly because of different assumptions relating to the cost of treatment and impact on “outcomes,” particularly measures of resultant health status. Costing studies should include some sensitivity analysis, to explore the effect of changes in the underlying assumptions of the costs and benefits. For example, if the drug acquisition and disposables cost were twice as much or the benefit of reduced PONV on overall health status was less.

The practicality of developing evidence on drug cost-effectiveness has been addressed in other specialties[29], and can be explored in perioperative practice. It is not too difficult to obtain cost data on types of treatments available, dosages, and sundry equipment and staffing. It is therefore possible to estimate incremental cost-effectiveness to facilitate informed decision-making by both payers and physicians. This can improve quality of care and enhance the efficient allocation of resources.

Patient-centered outcome measures and PONV

In the USA, the Patient Protection and Affordable Care Act was devised with the aim of improving quality of healthcare, outcomes and cost-efficiency. The Patient-Centered Outcomes Research Institute (PCORI) was established to encourage and support more comparative effectiveness research[30]. These changes place much greater emphasis on outcomes research, knowledge translation, and the need to incorporate patient preferences in clinical decision-making. Patient outcomes are also related to the level of communication with and trust in the doctor[31,32].

Patient-reported outcomes are events directly reported by patients or their surrogates about experiences with care, including symptoms, functional status or quality of life[33]. In the perioperative setting, this would include quality of recovery[34–37]. These are necessary and important outcome measures used by drug registration agencies, as well as to provide clinicians and the patients they treat with clinically useful information. This is in contrast to studies focusing on surrogate outcome measures[38,39]; many are of questionable significance and often have no convincing relationship with patient outcome.

Whilst some have argued that PONV is not necessarily an adverse patient outcome[40]; there is little doubt that many PONV episodes are clinically important[41,42]. It is, however, necessary to include some measure(s) of patient outcome over and above the incidence of PONV in perioperative research. This may include quantifying the clinical importance of PONV[41], or including one or more measures of patient quality of recovery or satisfaction with care[40].

Several quality of recovery (QoR) scores are available[32,43–47], but the most stringently studied has been the QoR-40[48]. For example, several groups have identified interventions that not only reduce PONV but also the resultant effect on QoR[49,50]. Quantitative measures of quality of recovery provide a numerical value for health status, and so can be used to calculate QALYs and ICER.

Myles and Wengritzky[41] developed a simple-to-use measure of the intensity and clinical impact of PONV. They found that around one in five patients with PONV had features that could classify them as having clinically important PONV; that is, where the PONV episode had a demonstrable effect on measures of health status. Patients with clinically important PONV had a much poorer quality of recovery (P < 0.0005), needed more antiemetic administrations for treatment (P < 0.0005), and were more likely to have consequences and complications of PONV (P < 0.01) when compared with those with lesser degrees of PONV.

PONV cost-evaluation studies

Whilst prophylactic antiemetic therapy reduces the incidence of PONV, it is unclear whether there is net benefit (less patient discomfort, delayed recovery, less staff resources) or harm (drug administration costs, side effects). Most patients do not benefit when given prophylactic antiemetic therapy, but all are exposed to potential harms.

Carlisle and Stevenson[51] performed a systematic review and meta-analysis of randomized controlled trials, which compared PONV prophylaxis with placebo. They identified 737 trials involving 103,237 patients and found that eight drugs were efficacious in preventing PONV when compared with placebo: dexamethasone, droperidol, metoclopramide, ondansetron, tropisetron, dolasetron, granisetron and cyclizine. The relative risks (RR) were between 0.60 and 0.80, depending upon the drug and outcome, indicating a 20–40% reduction in risk. Most of the original studies did not report side effects, but there was evidence of excess sedation with droperidol (RR 1.32) and headache with ondansetron (RR 1.16). The authors concluded that for each 100 patients given one of the above drugs, of whom 30 would report PONV without prophylaxis, 10 people would benefit from treatment and 90 people would not.

Failure to treat and overtreatment are two ends of a poor cost-effectiveness spectrum. Risk scores can identify patients at increased risk of PONV, and may therefore aid cost-effectiveness. An electronic decision-support clinical information system can assist[52]. Quality metrics to improve the cost-effective management of PONV have been proposed[53].

Pierre et al.[54] triaged surgical patients into three groups: (i) those at low risk of PONV that did not receive any antiemetic prophylaxis; (ii) those at moderate risk received low-dose (0.625 mg) droperidol or propofol-based intravenous anesthesia without droperidol; and (iii) those in the high-risk group received propofol-based intravenous anesthesia, dexamethasone 4 mg and droperidol 0.625 mg. Rates of PONV were reduced from 50% to 14% (P < 0.001). Such risk-guided antiemetic regimens are likely to be a cost-effective strategy. This concept has further support from another continuous quality improvement (QI) program aimed at lowering PONV in the PACU[55]. A multimodal PONV management protocol that standardized the anesthetic technique and antiemetic regimen was used, according to the Apfel PONV risk scoring system[56]. There were three stages to the QI program: (1) a prospective analysis of existing practice; (2) protocol implementation; (3) active feedback to staff and evaluation of guideline compliance. They found that 37/395 (9.4%) and 151/3,864 (3.9%) patients experienced PONV in the PACU before and after protocol implementation, respectively (P < 0.001), demonstrating a successful QI program.

A QI program introduced at the University of Texas MD Anderson Cancer Center and covering 23,279 anesthetics[57] found that compliance was not significantly influenced by a brief education program or in low-risk patients (the latter because of ingrained habits of universal antiemetic prophylaxis), but some improvement occurred when anesthesiologists were provided with brief performance data (59% versus 54%, P < 0.001) and most strongly when ongoing compliance data were presented (65% versus 59%, P < 0.001). Beneficial effects were strongest for those patients who had at least three risk factors for PONV.

Hill et al.[58] found prophylaxis was more cost-effective than placebo in high-risk patients because of the increased costs associated with nausea and vomiting[59]. The additional costs associated with PONV in placebo patients were up to 100 times higher compared with prophylaxis with a generic antiemetic, and the cost of treating vomiting was three times greater than the cost of treating nausea. Other studies have reported similar findings, but prophylaxis is only marginally more effective than treatment. Tramer et al.[60] did a modeling study and found that treatment of established PONV with ondansetron is more cost-effective and safer than prophylaxis with the same drug when effective doses are used. This was in part due to the weak antinausea effect of prophylactic ondansetron. These findings, however, might not be currently applicable since ondansetron became generic.

Paech et al.[61] included a cost–benefit analysis in a comparison of three 5-HT₃ receptor antagonists in 118 patients undergoing major gynecologic surgery. They found no significant differences between groups, and concluded that the choice between these agents should be based on the lowest available acquisition cost for each agent.

Pueyo et al.[62] compared the cost-effectiveness of three combinations of antiemetics in the prevention of PONV in 90 women undergoing major gynecologic surgery. A decision analysis tree was used to divide each group into nine mutually exclusive subgroups, depending on the incidence of PONV, need for rescue therapy, side effects and their treatment. Direct cost and probabilities were calculated for each subgroup, and then a cost-effectiveness analysis was conducted from the hospital point-of-view. They found that the combination of ondansetron–droperidol was cheaper than and at least as effective as ondansetron–dexamethasone, and was more effective than dexamethasone–droperidol with minimal extra cost.

Chang et al.[15] compared the cost-effectiveness of ondansetron and prochlorperazine in 78 patients undergoing hip or knee arthroplasty. They measured the cost-effectiveness ratio for each antiemetic. Prochlorperazine was more effective at reducing PONV in this study population. The mean total costs of PONV management per patient in the prochlorperazine and ondansetron groups were $13.99 and $51.98, respectively. The cost of successfully treating one patient with prochlorperazine and ondansetron was $31.87 and $275.01, respectively. They concluded that prochlorperazine is a more cost-effective antiemetic compared with ondansetron for the prevention of PONV in their study population. However, ondansetron has since become generic and so those findings might not be applicable to current practice.

There are few pediatric studies evaluating cost-effectiveness of PONV management.

An excellent study by Sennaraj et al.[63] compared the cost-effectiveness of prophylactic ondansetron with early ondansetron treatment in the management of PONV in 150 children undergoing strabismus repair using patient-centered outcome measures. Outcome measures included PONV rates, duration of stay in the PACU, parental satisfaction scores and costs (cost to benefit a child and cost per PONV-free child). Ondansetron prophylaxis was effective in reducing early and late PONV, and resulted in a shorter duration of PACU stay and superior parental satisfaction scores. The cost to benefit a child was more than fourfold less and the cost per PONV-free child was 35% less in the PONV prophylaxis group.

Olutoye et al.[64] evaluated the smallest dose of dolasetron for the prophylaxis of POV in 204 children undergoing ambulatory surgery. Costs were calculated from the perspective of the hospital. The incidence of early (0–6 h) and 24-h emesis was more frequent in the dolasetron 45 μg/kg group compared with the dolasetron 350 and 700 μg/kg groups and with the ondansetron group. They concluded that dolasetron 350 μg/kg IV was the smallest dose that provided acceptable equivalent efficacy and patient satisfaction scores to ondansetron 100 μg/kg.

Parra-Sanchez et al.[11] determined the incremental costs of PONV/PDNV in ambulatory patients with a time-and-motion study in 100 ambulatory surgery patients. They evaluated the incidence of PONV, time staff spent with patients, use of PONV-related supplies, recovery duration, PONV rescue treatments and a quality-of-life metric up to the third postoperative morning. They found that 37% of patients experienced PONV during hospitalization; this increased to 42% by the first postoperative morning and to 49% by the third postoperative morning. Patients with PONV spent 1 h longer in PACU compared with patients without PONV (P = 0.001) and more nursing time was required (P = 0.02). The total cost of postoperative recovery was significantly greater for patients with PONV/PDNV than for those without (US$730 versus $640; ICER $75, P = 0.006). The proportion of those who rated their quality of life high on each of four domains was less in those with PONV/PDNV, 49% versus 94%, respectively (P < 0.001). The incremental cost was comparable with the cost patients are willing to pay to avoid PONV.

There are many effective, cheap (generic) antiemetic agents that have been shown to be effective in the prophylaxis and treatment of PONV and PDNV. Lower drug acquisition costs support the cost-effectiveness of PONV prophylaxis in at-risk groups. In general, the decision about whether or not to use PONV prophylaxis is determined by the drug efficacy, risk for PONV, and drug acquisition costs, and these will vary from one setting to another.