Cost-effectiveness of “aiming for double zero”: strategies to avoid both PONV and downstream High Abuse-Liability Opioid use in bariatric anesthesia and surgery contexts

1 Introduction

1.1 Big little problems and public health relevance of aiming for zero high abuse-liability opioid exposure for primary prevention of new persistent opioid use

In anesthesia and surgery, a “big little problem” is (1, 2) both “frequently encountered” and “not linked to life-threatening outcomes.” (2) Postoperative nausea and vomiting (PONV) has been long-considered a “big little problem.” (1, 2) Meanwhile, perioperative opioid selection and downstream opioid-related sequelae could be a “big little problem,” or an underappreciated “big big problem,” in that it is unclear how frequently perioperative exposure to opioids such as fentanyl, oxycodone, and/or hydromorphone enhance new persistent opioid use (NPOU) risk, leading to a homegoing oxycodone prescription that is associated with a 6% (3) NPOU rate.

We coined the term HALO, for high abuse-liability opioids (4, 5), to describe hospital-based medications that likely carry equal or greater risk of NPOU as does oxycodone [ascribed to posthospitalization prescriptions (3, 6)]. Patients receiving usual perioperative and postoperative opioids (fentanyl, remifentanil, hydromorphone, oxycodone) would also seem to be at high risk for developing NPOU in ∼6% of cases (3) in both major and minor surgical and anesthetic contexts, even though no prior study has addressed NPOU from opioids administered as part of the anesthetic care plan. However, NPOU risk after single-dose intrathecal morphine (ITM) without any other perioperative opioid exposure is not yet described. Therefore, ITM may not necessarily generate a worrisome NPOU outcome. We are unaware of research addressing NPOU prevention by strategically avoiding exposure to both usual intra- and postoperative HALO, such as fentanyl, remifentanil, and/or hydromorphone (from the anesthesia care team), and hydromorphone and oxycodone/hydrocodone (from the surgical care team). For narrative convenience, we will therefore cluster these five italicized opioids into the descriptive phrase of HALO (4).

To our knowledge, there are also no systematic recommendations addressing NPOU primary prevention (4) by eliminating HALO exposure at the level of the anesthesia care team. Meanwhile, the thrust of NPOU mitigation from the surgical perspective has entailed reduction of, most commonly, homegoing oxycodone prescription pill counts (7), including efforts prompted by limits set by insurance carriers [exemplified in (8)]. Efforts to reduce postoperative opioid consumption have likely already been potentially burdened by earlier (typically hospital-based) exposure to HALO. In our view, HALO exposure at any time (anesthesia, surgery, hospitalization, convalescence/recovery at home) may carry NPOU risk via euphoria, acute opioid-induced hyperalgesia, and/or development of tolerance. NPOU risk is likely realized either via a legitimate medical prescription or secondary opioid market acquisition by patients. The likely basic science mechanism appears to be related to brief opioid exposure augmenting primary afferent input to spinal excitatory neurons (9). Further, recent mechanistic clarification (10) by the same basic laboratory (9) entails NMDA receptor-driven tolerance and hyperalgesia after short-duration opioid analgesia dose being related to the G-protein Gα_q signaling in primary sensory neurons (in dorsal root ganglia). Hereunder, we approach this HALO conundrum with a similar “Aim for Zero” perspective as we did with “Aiming for Zero” PONV prevention with 5-MMAEPPx in our previous report (11), hence, the dual purpose of “Aiming for Double Zero.”

1.2 Rationale for cost-effectiveness analyses considering both NPOU and PONV based on recent cost-benefit and cost-minimization analyses of PONV prevention

We recently published cost-minimization and cost-benefit perspectives (12) related to PONV prevention strategies, describing an innovative “5 + 3 Prevention + Booster” technique, entailing 5-drug multimodal antiemetic prophylaxis (MMAEPPx: palonosetron, perphenazine, aprepitant, diphenhydramine, dexamethasone), and 3-drug boosters (with these italicized agents). This “5 + 3” strategy was associated with an approximate €2,000/patient cost savings, compared with cost projections related to ongoing PONV consensus guideline use (2003–2020), along with consensus-guided breakthrough/rescue PONV cyclical management and length-of-stay sequelae (12). “Preventing preventable PONV” may also facilitate non-opioid analgesic gastrointestinal absorption, obviating the need for further “usual postoperative opioids,” and avoiding a “vicious cycle” of both PONV and NPOU risk escalation. In other words, the authors raise a practical question: if an enhanced recovery protocol is built around aggressive antiemetic prophylaxis and intrathecal morphine, can one meaningfully cut both PONV and downstream opioid misuse, while also showing significant health system cost savings over a 5-year time horizon?

1.3 Preoperative intrathecal morphine and magnesium for downstream opioid avoidance

We also recently observed and published (4) postoperative opioid avoidance (in ∼57% of major abdominal surgery cases) when 40-h analgesia was achieved with combined intrathecal morphine and magnesium (ITM + Mg⁺). Magnesium sulfate (as a preservative-free physiologic ion) was introduced intrathecally at our institution based on safety and multidrug compatibility in a literature review (13–17). If an enhanced recovery clinical pathway integrated to prevent both PONV (big little problem) and opioid exposure (big big problem) were implemented, the previously forecasted cost-minimization and cost-benefit may represent significantly improved epidemiologic priority from our earlier report (12). Such an improvement, measured by cost-effectiveness analysis of the described intervention, would primarily occur via preventing a significant proportion of the expected 6% incidence of NPOU (3) by avoiding surgeons' usual use of oxycodone/related prescriptions.

An NPOU incidence of 6% (3) is accepted (2017) as the expected NPOU risk after traditional exposure to oxycodone in postoperative discharge prescriptions. Bariatric-specific NPOU was also reported (2023) as 6% (18); however, the Michigan Bariatric Surgery Collaborative (2022) (6) reported a 3.5% NPOU risk when postoperative oxycodone prescriptions were reduced from discretionary milligram morphine equivalents (MME; entailing an associated 10.6% NPOU rate in 2006) to pill counts not exceeding 75 MME. We believe there is value in restricting opioid exposure as feasible to well-before the hospital discharge oxycodone prescription, such that considering both PONV and “usual IV/PO opioid avoidance” are inter-related high priorities, that when considered in tandem, may hold promise in addressing both of these anesthesia-centered public health issues, while still providing approximately 42-h analgesia (4) with ITM + Mg⁺.

1.4 Aiming for zero PONV

“Aiming for Zero” PONV now appears to be a 90%–95% achievable goal [on postoperative days (POD)#0–1], based on recent paradigm challenges (11, 12, 19–21) and clinical observations (4, 11, 21) reported in the British Journal of Anaesthesia (11) and Frontiers in Anesthesiology (4, 20–22). Aiming for Zero PONV appears promising for patients receiving procedural anesthesia and 5-MMAEPPx (4, 11), using all the off-patent agents listed above. 5-MMAEPPx has shown associated success, particularly in combination with usually emetogenic and opioid-sparing ITM without (22) or with (4) spinal Mg⁺, with ITM being associated with twice the PONV risk as that generated by usual perioperative opioids in cases not involving ITM (11).

1.5 Patient preference precedent

The patient-centered importance of avoiding PONV, in particular, was reported by Macario et al. in their 1999 willingness-to-pay study, where they evaluated patient priorities and their relative allocations from a theoretical US$100 to avoid unwanted effects of general anesthesia (23). The importance of PONV was clear, as patients allocated the greatest portion of their allowance (US$30) to PONV prevention (23). Preventing NPOU, and how much of the same theoretical US$100 would now be allocated, would require new research.

Recently (11) and in follow-up (4), 5-MMAEPPx (employing five or more differing pharmacologic mechanisms) led to ∼95% successful PONV prophylaxis on postoperative day zero (POD#0), while only costing US$23, which is lower than the US$30 sum from Macario et al. (23) 5-MMAEPPx (4, 11, 22) was used in a single preoperative encounter of pan-prophylaxis [without any subsequent booster dosing of palonosetron/perphenazine/aprepitant, 2 days of which separately costing $32, in another study (12)].

1.6 Intrathecal morphine and magnesium as potentially central to HALO avoidance

With ITM use (11), particularly combined with Mg⁺ (4), significant associated reductions in HALO requirements postoperatively were observed compared to historical controls not receiving ITM. Associated HALO avoidance was (i) 17% successful after ITM (sans Mg⁺) and an enhanced recovery protocol (22), and (ii) 57% successful after both ITM + Mg⁺ and further recovery protocol enhancements (4).

1.7 HALO avoidance combined with ITM + Mg⁺, 5-MMAEPPx, and 3-booster MMAEPPx

We believe that the HALO-sparing benefits of ITM (22) and particularly ITM + Mg⁺ (22) can yield an informed cost-effectiveness model. This would be by incorporating published observations (4, 11, 22) regarding 5 + 3 MMAEPPx [i.e., with newly recommended three booster antiemetics: daily-repeated perphenazine and aprepitant, along with repeated palonosetron every 40 h (4)]. Doing so would elevate treatment goals to not only include “Zero PONV” (24) for patients undergoing any anesthetic technique, but also “closer-to-Zero HALO exposure.” Further, this model, incorporating newly published data incorporating the described antiemetic boosters after initial 5-MMAEPPx (4, 22), would better address public health implications of HALO exposure by determining PONV associations (from earlier HALO exposures) with subsequent HALO exposures (in-hospital) for pain management. In other words, we forecast that ITM + Mg⁺, both preceded by 5-MMAEPPx, and followed by “3-AEPPx boosters” [“5 + 3,” with boosters POD#1 and thereafter (4, 22)], may (i) facilitate more reliable adherence to an oral/enteral non-opioid analgesic regimen and (ii) extend avoidance of HALO exposure, as public health goals.

2 Methods

2.1 Overview of the model

We built a Markov decision analysis model (Figures 1, 2) comparing economic and clinical outcomes over a 5-year time horizon after a 2–4 day surgical encounter for elective bariatric surgery [day of surgery (DOS), and 1–3 postoperative days (POD#1–3)] in an enhanced recovery after surgery (ERAS) program. This hypothetical cohort is derived from published (4, 11, 22) quality improvement (QI) outcome data addressing ITM ± Mg⁺ and 5-MMAEPPx with two AEPPx boosters (vs. applicable case-matched historical controls). For this new analysis, we assumed a third AEPPx booster (palonosetron, every 40 h after initial dosing preoperatively) beyond the two booster medications previously used, perphenazine and aprepitant (4, 12). We also assumed meaningful improvements in PONV prevention [and better ITM + Mg⁺-induced pruritus mitigation (25)] on POD#1–2 with the three-booster technique (vs. only two boosters).

2.2 Rationale for bariatric surgery modeling

We selected bariatric surgery due to (i) its favorable standardization profile of surgical invasiveness, (ii) ease of coding capture [current procedural terminology (CPT) code 43644 for Roux-en-Y (RnY) gastric bypass, and code 43775 for sleeve gastrectomy], and (iii) existing/published (4, 11, 22) data from our center including ITM ± Mg⁺ and 5-MMAEPPx/2-booster use, along with applicable, case-matched historical controls. Our RnY dataset, in particular, was enhanced by our center's other ITM QI data caseload [similarly published here (4, 11, 22)] involving general/colorectal/invasive genitourinary cancer surgery, and our center's associated consensus-guided matched cases with historical controls. Previous reports state that postdischarge opioid consumption does not differ between minimally invasive vs. open procedures of the abdomen/trunk (26). Therefore, we assumed that NPOU risk [entailing a 6% (3) point estimate] after any HALO exposure, including from anesthesia services, and not just from posthospitalization oxycodone prescriptions. The usual target POD discharge date in our institution is 2 days after bariatric surgery.

2.3 Cost determination

For direct medical costs, we followed the recommendations of the Panel on Cost-Effectiveness in Health and Medicine (27). We discounted future benefits and costs at 3% per year (28), and applied inflationary adjustments from cost projections of relevant past publications (29–31), adjusting all costs to 2023–2024 by using the medical care component of the Consumer Price Index.¹ The model was constructed using TreeAge Pro software (Healthcare version 2024, Williamstown, MA, USA).

2.4 Treatment alternatives

Patients (Table 1) either received (i) partial ERAS: ITM ± Mg⁺ [using a dosing algorithm published elsewhere (22), based on clinical, gender, and anthropomorphic characteristics], 5-MMAEPPx before surgery, preprocedural sedation including fentanyl, discretionary postoperative AEPPx (2-drug booster with perphenazine/aprepitant, with or without rescue treatment), and rescue analgesia entailing HALO such as oxycodone and/or hydromorphone being delayed until after acetaminophen/dextromethorphan/celecoxib; (ii) comprehensive ERAS: enhanced recovery protocol upgrade entailing ITM + Mg⁺, 5 + 3 MMAEPPx, preprocedural IV methadone (∼2.5 mg) instead of fentanyl, and a fixed multimodal analgesia regimen using daily acetaminophen/dextromethorphan/meloxicam before any HALO exposure; or (iii) pre-ERAS: discretionary intraoperative HALO (with no ITM) and assumed consensus-guided AEPPx that was part-discretionary and part-contemporaneous to the year of the historical control case, since newer consensus guidelines (32, 33) were published over the historical control period from our hospital electronic medical records. The overall date range for the analyzed cases incorporated into the model was 31 May 2016 through 22 March 2022 (11), with a start date of 9 March 2021 for Partial ERAS, and a start date of 1 February 2024 for Comprehensive ERAS, with no case data collected or analyzed after 9 October 2024 (4).

Table 1. Three treatment alternatives and potential outcomes modeled.

ERAS, enhanced recovery after surgery; ITM, intrathecal morphine (not considered a high abuse-liability opioid); AEPPx, antiemetic prophylaxis; MM, multimodal; HALO: high abuse-liability opioids, which in the postoperative context would include PO oxycodone and/or IV hydromorphone, or an intravenous (IV) infusion of fentanyl while intubated in the intensive care unit (only applicable to comprehensive ERAS); ICU, intensive care unit; NPOU, new persistent opioid use. Differences between alternatives are shown in bold.

2.5 Nuances of ITM dosing compared to standard-of-care ITM dosing and rationale

The ITM dose used (22) in partial ERAS and comprehensive ERAS was higher than the lower dose recommendations published elsewhere (34). This is because we had previously observed (35) cases with no meaningful analgesia from small-dose (22) ITM, since surgical trespass was likely cephalad (abdomen/trunk/axial skeletal dermatomes and sclerotomes) to the “reach” of the injected small-dose ITM. We observed significant cephalad abdominal pain postoperatively in the region of bariatric surgical trespass with ITM doses of 200 µg or less, so we used doses typically starting at 250 µg for women and 300 µg for men. Because Mg⁺ is hyperbaric, the spinal injectate was diluted with sterile water to a total volume of 8 mL [with the patient assuming/maintaining a sitting position after the spinal procedure in the block room before operating room (OR) entry, detailed elsewhere (22)].

2.6 Historical control (pre-ERAS) assumptions

The discretionary HALO care plans in the pre-ERAS historical controls did not involve ITM ± Mg⁺ use. ITM patients not achieving satisfactory analgesia during or after ITM use/dissipation of effect (despite non-opioids including acetaminophen, celecoxib or meloxicam, and/or dextromethorphan) were assumed to be treated with HALO (either IV hydromorphone or PO liquid/tablet oxycodone). ITM + Mg⁺ comprehensive ERAS cases were assumed to routinely include pre-/postoperative celecoxib (36)/meloxicam and dextromethorphan (37), along with intra/postoperative acetaminophen. No use of suzetrigine, a patent-protected non-opioid that was approved after the QI observational period, was modeled. Pre-ERAS historical control cases had perioperative discretionary use of non-opioid oral vs. IV analgesic adjuvants (these being tracked in the case-matching process), with IV acetaminophen use being most common, and other agents (dextromethorphan or celecoxib/meloxicam) rarely used in the pre-ERAS historical controls.

2.7 Other clinical assumptions, including those specifically addressing bariatric procedure and postbariatric PONV

All bariatric cases were performed by the same surgeon, including historical control bariatric cases. However, RnY case data for the model were augmented by our center's non-bariatric case-matched data (performed by different surgeons) across all three treatments, which has been presented in our previous reports (4, 11, 20–22), particularly to strengthen the accuracy of the PONV and pain rates (since the RnY cases did not differ from the majority of the other ITM-based and historical control cases in their relative PONV risks). RnY, however, was one of two bariatric case types collated, the other being the particularly emetogenic (38) sleeve gastrectomy (ESG). In other words, for more accurate PONV projections, our existing non-bariatric partial ERAS and comprehensive ERAS data (and their corresponding pre-ERAS historical control cases) were clustered with RnY cases, but not with ESG cases. RnY gastric bypass cases are acknowledged as slightly more painful (38) but less emetogenic than ESG. Phrased colloquially, ESG was considered “an orange among apples,” based on its extreme emetogenic potential, both previously reported (38) and observed (4, 11).

We made other assumptions. First, severe cardiopulmonary (from pain physiology and/or opioid dosing and its sequelae) or gastrointestinal (e.g., esophageal perforation from PONV) sequelae were not assumed to occur as possible outcomes. These assumptions were adopted because such events were not observed in our tracked ITM ± Mg⁺/5-MMAEPPx bariatric cases (4, 11, 22). Next, Comprehensive ERAS cases were associated with a 2.7% rate of brief postoperative intensive care unit (ICU) management for delayed extubation due in part to in-OR care team variances, with the ICU stay in such cases encompassing the rest of the OR day and slightly into the following day. There was no apparent relationship between delayed extubation and prolonged length of stay (LOS) beyond POD#2 (data not shown). Variances delaying extubation could have included late-case doses of dexmedetomidine against medical direction, anesthetic depth maintenance at bispectral index monitoring values below the requested thresholds (typically 50–65), insufficient maintenance of normothermia, acid-base complexities related to coexisting medications, and/or other variant events leading to failure to extubate or need to reintubate. Probabilities for PONV, postoperative pain, LOS, and adverse events (e.g., pruritus if ITM ± Mg⁺) depended on the treatment received and the type of bariatric procedure. Probabilities for NPOU depended on the probability of avoiding any HALO exposure based on the type of treatment received (6% if any HALO, zero if no HALO). ITM ± Mg⁺ was not considered HALO, and methadone as preprocedural analgo-sedation (before the spinal) was not considered HALO exposure.

2.8 Base case estimates for relevant clinical parameters

Base case point estimates and associated ranges were established for clinical parameters based on external literature review when previous case series data from our institution (4, 11, 22) did not cover the specific parameters. Utilities/quality of life measures and costs were similarly established. All are summarized in Table 2. The state transition model is shown in Figure 2, while Figure 3 shows six plots (three treatments each for ESG vs. RnY) depicting the modeled proportions of common tracked symptoms or dispositions [asymptomatic, pain requiring IV opioids, PONV, adverse event (typically pruritus), discharge home, and unplanned delayed extubation (2.7% risk) only in the comprehensive ERAS-treated cases].

Table 2. Model parameter base case values and ranges examined in the sensitivity analyses.

See Figure 3 for six plots encompassing POD#1–3+ (three treatments each, for ESG versus RnY), incorporating proportions (after POD#0 above) of the common tracked symptoms or dispositions [asymptomatic, pain requiring IV opioids, PONV, adverse event (typically pruritus), and discharge home].

QI, quality improvement; CI, confidence interval; PONV, postoperative nausea/vomiting; IV, intravenous; ERAS, enhanced recovery after surgery (treatment and prophylaxis/booster protocol); RnY, Roux-en-Y gastric bypass; ESG, emetogenic sleeve gastrectomy; NPOU, new persistent opioid use; QALY, quality-adjusted life-year; AE, adverse event (i.e., itching on POD#1 with no other symptoms); POD, postoperative day; VA, Veterans Affairs (United States); OPAL, Office of Procurement, Acquisition, and Logistics, United States Federal Supply Service (https://www.va.gov/opal/nac/fss/pharmPrices.asp, accessed 24 July 2024).

When modeling for very low probability events [such as 2.7% chance of delayed extubation and brief ICU stay in the Comprehensive ERAS group (with an assumed-plausible high range probability of 20%) or 6% chance of NPOU if exposed to any HALO during hospitalization], the authors created lowest common denominator integers (for cross-group comparisons) that led to numerators with decimal values where indicated with “*”. Therefore, raw clinical data from previous citations that did not entail delayed extubation or NPOU at this point of the model calculations are not shown in the numerator and denominator integer data presented.

Binomial 95% CIs are derived from https://statpages.info/confint.html. For the Comprehensive ERAS Asymptomatic scenarios (left column), the same denominators for ESG (denominator=90) and RnY (denominator=51) were applied as they were for the other clinical parameters (PONV, Pain/IV Opioid, and Adverse Event). For costs, ICU through NPOU were “± 25%” from the point estimate.

The primary adverse event was pruritus (from intrathecal morphine, less frequent and intense without magnesium, while more frequent and intense with coadministered intrathecal magnesium). Adverse events in the absence of other symptoms prolonging the length of stay are maintained through both Day 0 and Day 1. Overall itching was 26.2% overall, or 37/141, after Comprehensive ERAS, and was 13.7% overall, or 21/153, after partial ERAS. Patients with pruritus were assumed to have no crossover pain or PONV on these days, for modeling purposes, but patients with pruritus on POD #1 may have had (for example) POD#0 pain (and were categorized as such above) before transitioning to POD#1 itching. There was no assumed pruritus after pre-ERAS, which did not entail intrathecal morphine use. Patients with pruritus on POD#0 did not incur hospital costs (until POD#1, see text), but did incur disutilities on both POD#0 and #1 in the model. Pruritus on POD#0 and #1 in our observations did not occur throughout these entire days, and the listed disutilities/costs are not prorated for the portions of the days that pruritus occurred.

Probabilities were derived as follows. For ESG vs. RnY, pain requiring IV opioids, PONV, and adverse event (e.g., pruritus after ITM) proportions were separately ascribed for POD#0 (see Table 2), POD#1, POD#2, and POD#3+ (Figure 3), adapted from our previous data (4, 11, 22). When patients encountered more than one symptom, the model required case-by-case determination of state-to-state transitions (from one postoperative day to the next), under the assumption that some patients had multiple symptoms, but were categorized as having one symptom that led to an additional overnight hospital stay. Patients (rarely) requiring a 3 + day postoperative stay were assumed to have either PONV or pain requiring specifically IV HALO on POD#2, even though “real” patients may have had both symptoms. Patients who did extremely well and were discharged on POD#1 may have had some postoperative pain, but not severe enough to require IV HALO. All patients who had any in-hospital HALO during preoperative or OR time, or after OR exit, were assumed to have a 6% chance of NPOU. Preprocedural fentanyl (for spinal in the block room) was ascribed a 6% chance of NPOU, but preprocedural IV methadone (for spinal in the block room), followed by no other HALO opioids, was not ascribed an NPOU risk. Pruritus (as the primary adverse event) did not incur a cost if it occurred on POD#0 (since the patient was already hospitalized for other reasons), but pruritus did incur a day of disutility (see below) even when only a fraction of the day involved pruritus that was treated. Pruritus that occurred on POD#1 incurred both costs (prolonged hospital admission to POD#2 if no other symptoms extended LOS) and disutility (from the patient’s perspective).

2.9 Utilities and disutilities

We assigned utility values as listed in Table 2, based largely on established utility value benchmarks (39). Utilities are preferences for health states on a scale of 0 (death) to 1 (full health). The model tracked utility loss (or disutility) resulting from health states as described below. In-hospital pruritus (e.g., from ITM ± Mg⁺, as a skin manifestation) decreased utility by 20% [“itching skin” (39)] from the baseline hospitalization value, and resulted in additional LOS (POD#1 into POD#2) despite the absence of pain/PONV. PONV decreased utility by 32% [“other gastrointestinal disorders” (39)]. Pain requiring any HALO decreased utility by 30% (39). NPOU was ascribed a disutility of 0.3 (40), that is, a “1” value for full health utility postoperatively at home was downgraded to 0.7 (for the 5-year NPOU duration) in the model.

2.10 Costs

We used 2023–2024 Medicare (US) costs (Table 2). For pre-ERAS, partial ERAS, and comprehensive ERAS costs, we used Federal Supply Service data² for pharmaceutical prices (for comprehensive ERAS) of aprepitant 40 mg (three doses, US$33.50), perphenazine 8 mg (three doses, US$1.65), dexamethasone 8 mg (two doses of 4 mg, US$0.87 per dose), diphenhydramine (one dose from a 50 mg injectable vial, US$0.59), and palonosetron (one dose ranging from 150 to 250 µg, from a 250 µg syringe, US$10). Subsequent booster doses of palonosetron did not occur in the pre- or partial ERAS groups but were incorporated into the comprehensive ERAS group model [for both PONV prophylaxis and pruritus mitigation (6)]. For partial and comprehensive ERAS, ITM drug costs were US$8.34 per patient, and ITM equipment (spinal trays from institutional inventory) cost US$12 per kit. When rounded, the partial ERAS cost was US$67, and the comprehensive ERAS cost was US$78. Pre-ERAS costs assumed cost-minimization and no ITM. Therefore, a US$3 cost per patient was assumed (US$0.45 for single-dose 4 mg IV ondansetron, US$0.87 for single-dose 4 mg IV dexamethasone, and 1–2 mg IV haloperidol, US$1.64 for a 5 mg vial, rounded up to US$3). Off-patent HALO used in the OR was not ascribed a pharmacy cost.

Downstream costs (Table 2) reported by Oderda et al. (29–31) reflect not only associated drug costs, but also overall hospital costs for symptom rescue, additional LOS, diagnostic workups, and changes in care level (e.g., medical-surgical regular floor bed vs. critical care bed). Differing years of publication from Oderda et al. (29–31) required the aforementioned inflationary adjustments to equalize cost estimates. Drug cost assumptions for postoperative treatment were subsumed within total costs per symptom encountered to avoid double-counting. Pain requiring at least one dose of an IV opioid was ascribed a US$418 cost per day [US$284 representing half of the cost increment in publication year 2007 associated with IV opioid requirement in postoperative patients encountering an adverse drug event (29), to which inflationary adjustment to year 2023–2024 was applied], while PONV had a US$2,023 cost increment per day of symptom (using the most conservative value of US$1,700 cost increment in publication year 2019, adjusted for inflation). No costs were ascribed to prolonged LOS; rather, LOS extensions were assumed to be captured in these costs. Our model, therefore, likely underestimates actual cost savings. Finally, long-term NPOU costs were derived from Schackman et al., as an inflation-adjusted sum of US$17,048 over the 5-year time horizon of the model (41).

We determined the cost of postspinal pruritus (i.e., hospital cost of our most common “adverse event”) to be US$346 (30), based on a US$209 cost published in 2003, and inflationary adjustments since then, applied on POD#1 (as described above in Section 2.8). For costs of delayed extubation and ICU stay into the next morning, we ascribed a $5,104 cost, although we acknowledge a €5,700 cost-for-ICU citation (not used in our model) from a 2020 study (42).³ Because our spinal needles for preoperative ITM + Mg⁺ were pencil-point in shape, and 24–25 gauge, postdural puncture headaches were assumed not to occur, since we did not observe any in our data set.

The analysis took a healthcare perspective; thus, indirect costs (such as patient transportation expenses for routine medical provider visits and the costs for patient time or lost productivity) were not incorporated into the model.

2.11 Analyses

2.11.1 Procedural comparisons, point estimates, one-way sensitivity analyses, and willingness-to-pay thresholds

For each of the three treatment strategies, our model calculated 5-year benefits [avoided loss of quality-adjusted life years (QALYs)] and costs. We separately analyzed non-ESG cases (RnY cases resembling PONV in general abdominal surgery) from ESG cases (representing cases with the highest a priori PONV risk available in our data). We conducted one-way sensitivity analyses for all model parameters to assess the effects of varying baseline estimates within clinically plausible ranges, as listed in Table 2. In addition, day-by-day health state transition probabilities that resulted in state membership likelihoods as depicted in Figure 3 were varied by a relative 15% in these analyses. The favorability of treatment strategies was assessed using a conservative $0/QALY gained willingness-to-pay threshold (i.e., less expensive and more effective). We acknowledge that for most anesthesia-related contexts (based on hospital-related administrative pressures), the allowable cost/QALY is likely to carry a willingness to pay value of zero, as was assumed in our recent (12) cost-minimization and cost-benefit PONV-specific report.

2.11.2 Probabilistic sensitivity analyses

Probabilistic sensitivity analyses were performed, simultaneously varying all parameters 5,000 times over distributions fitted to their respective ranges as described above. Beta distributions were used for probabilities; uniform distributions were used for costs and utilities (uniform distributions maximize parameter variability and stringently test the robustness of model results).

2.11.3 CHEERS 2022 checklist

We completed the Consolidated Health Economic Evaluation Reporting Standards 2022 [CHEERS 2022 (43)] checklist, which is included as Supplementary Content Item 1.

3 Results

3.1 Base case analysis

3.1.1 Emetogenic sleeve gastrectomy

The comprehensive ERAS strategy cost less and resulted in fewer QALYs lost than the other treatment strategies. Thus, it was the dominant strategy, i.e., less costly and more effective (Table 3, Panel 1).

Table 3. Cost-effectiveness analyses of the Comprehensive ERAS strategy incorporating the (i) ITM + Mg⁺ and (ii) 5 + 3 MMAEPPx strategies of antiemetic prophylaxis, antiemetic boosters, and non-opioid analgesic boosters.

Long-term cost entails new persistent opioid use. Short-term cost is related to hospitalization.

QALY, quality-adjusted life year; ICER, incremental cost-effectiveness ratio; Quality adjusted days lost, incremental days gained compared to comprehensive ERAS (QALY×365).

Treatment strategy is more costly and less effective than comprehensive ERAS. ERAS, enhanced recovery after surgery clinical pathway; ITM + Mg⁺, intrathecal morphine [per cited (4, 22) dosing] plus magnesium [25 mg (4)]; 5 + 3 MMAEPPx, multimodal antiemetic prophylaxis for comprehensive ERAS with perphenazine, palonosetron, aprepitant, dexamethasone, and diphenhydramine all before surgery (e.g., administered by preoperative nursing staff), boosters with perphenazine (4–8 mg PO) and aprepitant (40 mg PO) every 24 h, and boosters with palonosetron (150–250 µg IV) every 40 h. Previously published (4) prophylaxis doses in the ITM + Mg⁺ context are perphenazine PO 8 mg (assuming ≤75 years and ≥75 kg), palonosetron IV 250 µg, aprepitant PO 40 mg, dexamethasone IV 8 mg, and diphenhydramine IV 20 mg. Ondansetron is not used due to theoretical concerns of rebound nausea (58, 59) and/or breakthrough nausea.

3.1.2 Less-emetogenic RnY cases

For the RnY bariatric and similar non-bariatric cases, Comprehensive ERAS again dominated the Partial ERAS and pre-ERAS strategies, with Comprehensive ERAS having lower costs and fewer QALYs lost than the other strategies (Table 3, Panel 2).

3.2 One-way sensitivity analyses

3.2.1 Sleeve gastrectomy

For ESG cases, individual variation of all parameters over ranges listed in Table 2 did not change the dominance of comprehensive ERAS over partial ERAS and pre-ERAS. For ESG cases, to further test result robustness, one-way sensitivity analyses of PONV risk on POD#0, delayed extubation risk, and comprehensive ERAS cost going beyond the listed ranges were performed. In these analyses, comprehensive ERAS was no longer dominant if PONV risk on POD#0 was varied to ≥55.8% (base case 11.1%), ICU admission risk was ≥23.5% due to delayed extubation (base case 2.7%), or comprehensive ERAS cost was ≥$1,285 (base case $78).

3.2.2 RnY

As in the ESG analyses, Comprehensive ERAS in RnY cases remained dominant with individual variation of all parameters over the model's default ranges as listed in Table 2 with the exception of NPOU risk, where comprehensive ERAS was no longer dominant if NPOU risk was ≤1.3% (base case 6%). Going beyond the ranges in Table 2, one-way sensitivity analyses found that comprehensive ERAS was no longer dominant if ICU admission risk was ≥11.3% (base case 2.7%), comprehensive ERAS cost was ≥$598 (base case $78), or PONV risk on POD#0 was ≥25.3% (base case 7.8%).

3.3 Probabilistic sensitivity analyses

Probabilistic sensitivity analyses, simultaneously varying all parameters over distributions 5,000 times, found that in the ESG scenario, 100% of model iterations favored the comprehensive ERAS strategy at a $0 per QALY gained willingness-to-pay threshold, and continued to be favored in 100% of iterations at higher cost per QALY gained thresholds. In the RnY scenario, comprehensive ERAS was favored in 96.6% of model iterations, and in 100% of iterations at higher thresholds.

4 Discussion

4.1 Key results regarding cost savings and QALYs saved (i.e., more effective)

Our results suggest that the described comprehensive ERAS is cost-saving and more effective than partial and pre-ERAS for both emetogenic sleeve gastrectomy and less-emetogenic RnY bariatric procedures, when incorporating NPOU costs in 6% of patients receiving any HALO, but acknowledging the risks of temporary pruritus from ITM ± Mg⁺. The cost savings per case for comprehensive ERAS compared to partial ERAS were $1,174 (ESG, entailing a small clinical sample of particularly emetogenic cases) and $506 (RnY, entailing these cases along with a larger non-bariatric data set from our institution that further informed PONV and HALO exposure rates).

4.2 Tradeoff of accepting ITM-induced pruritus in exchange for NPOU prevention (by minimizing HALO exposure after ITM + Mg⁺)

Spinal anesthesia can be associated with bleeding, infection, nerve damage, and postdural puncture headache. However, in our observational period, we used 24–25-gauge pencil-point spinal needles through introducer needles to effectively eliminate the spinal headache risk. If cutting needles with larger diameters were used in a clinical practice (e.g., 22-gauge Quincke), spinal headache would be another adverse outcome to add to the model. Similarly, we have observed anesthesiologists refusing to deploy ITM techniques due to postprocedural pruritus risk. The analysis shows that the 5-year NPOU cost (∼US$17,000) vs. a risk of pruritus for part of 1–2 days (estimated cost ∼$350) seems to be a surprising public health trade-off, with respect to meaningful opportunity to avoid HALO exposure and its 6% NPOU risk vs. avoiding a nuisance symptom (for perhaps parts of 1–2 days), for which symptom prevention/reversal (palonosetron 250 µg/nalbuphine or butorphanol) is available.

4.3 Cost offsets for other expensive but routine clinical interventions

The higher pharmacy costs of comprehensive ERAS ($78 vs. $3 for pre-ERAS) were decisively offset by the lower costs of avoiding NPOU (by avoiding HALO exposure in-hospital). The cost difference would accommodate the single-dose cost for patent-protected drugs [such as sugammadex (44) for reversal of non-depolarizing neuromuscular blockade, at a cost of US$60 for 200 mg] or partially offset costs for longer-course non-opioid analgesics (such as suzetrigine). Thus, despite the higher-observed efficacy of routine 5-MMAEPPx (and projected efficacy of 3-MMAEPPx boosters for PONV and pruritus) with off-patent palonosetron and aprepitant (each costing at US$∼10/dose), concerns regarding “overshooting” PONV prophylaxis can be appeased when routine HALO is instead replaced intraoperatively by the more emetogenic ITM + Mg⁺. Other non-HALO such as buprenorphine (45) or methadone (46–48) (which likely carry a higher PONV burden than HALO) would seem to similarly benefit from the 5 + 3 MMAEPPx strategy, without the technical burden of ITM + Mg⁺ procedural logistics, as well as likely less pruritus than ITM ± Mg⁺.

4.4 Non-HALO and PONV risks approaching those of ITM, while possibly reducing NPOU risks compared to HALO

The associated expenses of IV opioids that may be “non-HALO” in their characteristics (specifically methadone and buprenorphine) may more be related to potent emetogenic effects than their HALO counterparts than related to acquisition costs, per se. Therefore, 5 + 3 MMAEPPx may ultimately carry (with partial ERAS or comprehensive ERAS) the need to be coadministered preventatively before IV/PO methadone, or whatever route of buprenorphine delivery [e.g., nerve block (49–57) or otherwise (45)] is chosen. Future research is needed regarding NPOU after only-methadone or only-buprenorphine exposure in these contexts, and whether their possibly lesser-hyperalgesia, tolerance, and euphoric effects (compared to usual HALO drugs) yield a “zero” or “higher than zero but less than 6%” risk of NPOU.

4.5 Robust sensitivity analyses and ITM + Mg⁺ pain-free status suppressing respiratory drive

Sensitivity analyses demonstrated that our results were robust over a wide range of values for the specific model parameters addressing cost of the comprehensive ERAS regimen ($75 more than the pre-ERAS cost), probability of PONV after 5 + 3 MMAEPPx vs. consensus-guided pre-ERAS, and probability of delayed extubation related to the profound analgesia from ITM + Mg⁺. Team-wide and cross-team education is likely required regarding avoidance of sequelae that may be otherwise be well tolerated en route to spontaneous respiratory effort pre-extubation. Such sequelae include core body temperature/normothermia, avoided acidosis status, confident neuromuscular blockade reversal status, and common pre-emergence medications not given before extubation as routine, such as dexmedetomidine or lidocaine. These elusive sequelae need to be carefully mitigated with a progressive mindset of “more is less” for PONV prevention and HALO avoidance, concomitantly with a “less is more” mindset in the setting of HALO-related expectations (e.g., emergence agitation and enhanced respiratory effort from associated acute HALO-induced hyperalgesia).

4.6 Limitations

Since this analysis takes a United States perspective, its results may not apply in countries where drug and related costs are lower. There are other limitations. First, we assumed that no HALO exposure in-hospital would ensure no HALO exposure upon homegoing, based on recent bench science work addressing short-term opioid exposure triggering both hyperalgesia and tolerance (9, 10). Deliberate omission of a HALO prescription after achieving “no HALO in-hospital” would likely only be true if the discharging care team involved (e.g., surgeons or transitional pain service) had the necessary confidence and contingency planning capacity to allow backup mechanisms for patients to access HALO legitimately if the necessity arose once home, along with adequate compliance with non-opioid regimens (such as the described acetaminophen/meloxicam/dextromethorphan, with or without suzetrigine, pending future research). As a frame of reference, the Michigan Bariatric Surgery Cooperative reported that 26.6% of postbariatric patients statewide did not fill an oxycodone prescription after hospital discharge, as of 2021 (6). Second, some assumptions may potentially bias our analysis toward Comprehensive ERAS. However, we strategically offset this potential bias by (i) assuming low costs for pain and PONV rescue treatment in-hospital in the pre-ERAS group, (ii) modeling conservative (lower) cost estimates of PONV complications as previously reported in gynecologic surgery [$1,698/case (31)] as opposed to modeling triple-the-cost PONV complications in general/colorectal/cancer surgery [$5,280/case (31)], and (iii) not reporting increments between pre-ERAS and Comprehensive ERAS as a result of performing incremental analyses between strategies ranked by cost [as recommended (28)], thus avoiding the possible risk of overinflating the described incremental benefits of Comprehensive ERAS. Third, comprehensive ERAS would be less favorable if PONV risk or MMAEPPx costs were substantially, perhaps implausibly, greater. Fourth, our probability estimates for adverse events entailed only pruritus, because our single-hospital case series did not yield any other complications that occurred with sufficient frequency (e.g., no spinal headaches or QT_c electrocardiogram issues) or with sufficient escalations in care costs (e.g., no epidural blood patch; no ICU stay because of QT_c concerns, as opposed to some with brief ICU stay for delayed extubation). Thus, we implicitly assumed that both Comprehensive ERAS and pre-ERAS are equally “safe” based on QT_c concerns. Little evidence exists to support or refute this assumption, particularly since Apfel has declared palonosetron as a non-factor related to QT_c concerns (58, 59). Next, our ERAS strategies did not include other commonly considered non-opioid analgesic agents. This was largely related to observed delays in extubation when somewhat sedating analgesic adjuvants were coadministered with ITM (e.g., dexmedetomidine, IV lidocaine). Also related was that we did not include ketamine in our data capture, since we already had verified pre- and postoperative NMDA-related (along with non-steroidal analgesia) mechanistic coverage in the comprehensive ERAS group (with sustained-release dextromethorphan and celecoxib/meloxicam, respectively), along with a small portion of the partial ERAS group having received all these non-opioid analgesics (along with acetaminophen). Finally, it may be acceptable to trust our basic science colleagues (9, 10) who have recently reported any exposure to any short acting opioid (in the rat laboratory) rendering real patients at real risk for 6% NPOU, not just after any homegoing oxycodone prescription but after any preceding HALO, appears to lead to a “thirst” for any prescription oxycodone (or HALO from the secondary marketplace). In our NPOU-related probabilistic analyses, the plausible range of NPOU was input as 0.6%–6%, which includes the latest available NPOU-bariatric rate of 3.5% from the Michigan Bariatric Surgery Collaborative (6).

4.7 Conclusion

In conclusion, reiterating the prescient (1993) statements that “economic considerations must be part of clinical decision making,” and “consideration of costs is neither unethical nor immoral” (60), the off-patent status of the antiemetics palonosetron and aprepitant has changed the equation not only for PONV, but also for HALO avoidance. Based on available observational evidence and decision analysis modeling, the comprehensive ERAS strategy entailing a sufficient ITM + Mg⁺ dose, along with 5 + 3 MMAEPPx, may have potentially profound clinical and economic benefits via reducing HALO exposure, which had been administered routinely in pre-ERAS care. The primary cost tradeoff appears to be avoiding years of NPOU disutility and cost vs. the risk and cost from 1 or 2 days of mild (ITM without Mg⁺) or more intense (ITM with IT Mg⁺) pruritus [somewhat preventatively mitigated by 250 µg palonosetron (25) and treatable with agents such as butorphanol or nalbuphine]. Future research could explore methadone (46–48) or buprenorphine (45) in tandem with comprehensive ERAS (i.e., replacing complexities related to ITM + Mg⁺), possibly avoiding the observed 2.7% delayed extubation risk after ITM + Mg⁺ (along with its higher itching intensity), while maintaining fiscal/economic responsibility.

Generative AI Statement

The authors declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fanes.2026.1804960/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Lucas Ferreira Gomes Pereira, Universidade de Sao Paulo Anestesiologia, Brazil

Reviewed by: Christian Bohringer, UC Davis Medical Center, United States

Anterpreet Dua, Augusta University Medical Center, United States

Footnotes

¹From the U.S. Bureau of Labor Statistics web site, available at: http://www.bls.gov/data/. Accessed 24 July 2024.

²https://www.va.gov/opal/nac/fss/pharmPrices.asp. Accessed 24 July 2024.

³The cost savings for ICU avoidance in this (42) report was €5743, but was also associated with a 7-day hospital stay instead of a 12-day hospital stay; therefore, this cost increment may be an overestimate when addressing ICU stays of less than one day, en route most commonly to a day-2 discharge. We instead used $5104 from https://www.sccm.org/Communications/Critical-Care-Statistics. Accessed 24 March 2025.