Trajectories of Recovery after Intravenous propofol versus inhaled VolatilE anaesthesia (THRIVE) randomised controlled trial in the USA: A protocol

Abstract

Introduction

Millions of patients receive general anaesthesia every year with either propofol total intravenous anaesthesia (TIVA) or inhaled volatile anaesthesia (INVA). It is currently unknown which of these techniques is superior in relation to patient experience, safety and clinical outcomes. The primary aims of this trial are to determine (1) whether patients undergoing (a) major inpatient surgery, (b) minor inpatient surgery or (c) outpatient surgery have a superior quality of recovery after INVA or TIVA and (2) whether TIVA confers no more than a small (0.2%) increased risk of definite intraoperative awareness than INVA.

Methods and analysis

This protocol was co-created by a diverse team, including patient partners with personal experience of TIVA or INVA. The design is a 13 000-patient, multicentre, patient-blinded, randomised, comparative effectiveness trial. Patients 18 years of age or older, undergoing elective non-cardiac surgery requiring general anaesthesia with a tracheal tube or laryngeal mask airway will be eligible. Patients will be randomised 1:1 to one of two anaesthetic approaches, TIVA or INVA, using minimisation. The primary effectiveness endpoints are Quality of Recovery-15 (QOR-15) score on postoperative day (POD) 1 in patients undergoing (1) major inpatient surgery, (2) minor inpatient surgery or (3) outpatient surgery, and the primary safety endpoint is the incidence of unintended definite intraoperative awareness with recall in all patients, assessed on POD1 or POD30. Secondary endpoints include QOR-15 score on POD0, POD2 and POD7; incidence of delirium on POD0 and POD1; functional status on POD30 and POD90; health-related quality of life on POD30, POD90, POD180 and POD365; days alive and at home at POD30; patient satisfaction with anaesthesia at POD2; respiratory failure on POD0; kidney injury on POD7; all-cause mortality at POD30 and POD90; intraoperative hypotension; moderate-to-severe intraoperative movement; unplanned hospital admission after outpatient surgery in a free-standing ambulatory surgery centre setting; propofol-related infusion syndrome and malignant hyperthermia.

Ethics and dissemination

This study is approved by the ethics board at Washington University, serving as the single Institutional Review Board for all participating sites. Recruitment began in September 2023. Dissemination plans include presentations at scientific conferences, scientific publications, internet-based educational materials and mass media.

Trial registration number

NCT05991453.

STRENGTHS AND LIMITATIONS OF THIS STUDY

• This will be the largest trial conducted to date comparing general anaesthetic techniques in a variety of hospital settings across the USA.

• This trial was conceptualised and created by a broad group of collaborators, with a range of relevant expertise and experiences.

• This trial uses advanced electronic information systems built upon the infrastructure of the Multicenter Perioperative Outcomes Group to integrate data.

• Dose rates for propofol and other drugs will be based on clinician discretion, as opposed to target-controlled infusion drug-delivery systems, reflecting anaesthesia practice in the USA.

Introduction

Each year in the USA, millions of patients receive general anaesthesia for surgical procedures.^{1 2} The choice to use inhaled volatile anaesthesia (INVA) and total intravenous anaesthesia (TIVA) with propofol is typically made by anaesthesia clinicians,³ often without collaborative input from patients. While certain rare contraindications, such as propofol allergies, malignant hyperthermia or severe postoperative nausea and vomiting associated with inhaled volatile agents, are well-known, there is limited evidence suggesting that one technique offers a superior patient experience over the other. Moreover, despite the incidence of unintended definite intraoperative awareness with general anaesthesia being rare (~0.1–0.2%),⁴,⁶ it is still an important concern to patients³ and there is a lack of high-quality clinical trials comparing the risks of unintended awareness between TIVA and INVA. If either anaesthetic technique were associated with a superior recovery experience from surgery, without increasing the risk of unintended awareness, it would significantly influence the decision-making process regarding anaesthetic choice. To address these critical gaps in knowledge, the following comparative effectiveness trial was conceptualised and co-designed by a broad group of collaborators, including patients, as described previously.⁷

Feasibility pilot trial

In preparation for this trial, we conducted a pilot feasibility study involving 300 patients across two US sites⁷ (ClinicalTrials.gov identifier: NCT05346588), in which we assessed (1) the feasibility of enrolling patients over 10% of approached patients into a pragmatic clinical trial comparing TIVA with INVA; (2) clinician compliance with the assigned anaesthetic technique (>80%), based on random group allocation; and (3) the successful ascertainment of primary outcomes—quality of recovery (QOR) on postoperative day 1 and incidence of intraoperative awareness—in over 90% of patients.

We demonstrated the ability to enrol 53% of patients who were approached, administer the assigned anaesthetic technique per protocol in 95% of TIVA cases and 99% of INVA cases, and obtain complete data collection for the Quality of Recovery-15 (QOR-15) and modified Brice in 93% and 100% of patients across three surgical complexity categories: major inpatient surgery (approximately 40% of the cohort), minor inpatient surgery (approximately 20%) and outpatient surgery (approximately 40%).

Objectives and endpoints

The following randomised multicentre comparative effectiveness trial will randomise 13 000 patients to determine (1) which general anaesthesia technique yields superior patient recovery experiences in any of the three surgical complexity categories: major inpatient surgery, minor inpatient surgery and outpatient surgery; and (2) whether TIVA confers no more than a small (0.2%) increased risk of definite intraoperative awareness than INVA in patients undergoing both inpatient and outpatient surgeries (see table 1).

Table 1. Objectives and endpoints summary.

Objectives	Endpoints	Justification for endpoints
Primary
Compare the early patient quality of recovery following two commonly used and established anaesthetic techniques, TIVA and INVA, in patients undergoing (1) major inpatient surgery, (2) minor inpatient surgery or (3) outpatient surgery.	Patient-reported QOR-15 score on POD1.	The QOR-15 is an internationally validated scale routinely used in perioperative clinical trials with existing literature being most mature for assessments on POD1. The construct validity, feasibility and patient acceptance of the elements have been evaluated in thousands of patients across ages, countries and languages.819 52,54
Determine whether the incidence of definite intraoperative awareness with TIVA is non-inferior to that with INVA.	Incidence of unintended definite intraoperative awareness with recall at either POD1 or POD30.	Patients prioritise avoiding intraoperative awareness,3 a rare but devastating complication with some evidence suggesting it is more common with propofol TIVA than INVA.55,57
Secondary
Compare the medium-term trajectories of patient recovery after anaesthesia and surgery following two commonly used and established anaesthetic techniques: (1) TIVA with propofol and (2) INVA in patients undergoing (a) major inpatient surgery, (b) minor inpatient surgery or (c) outpatient surgery.	QOR-15 score on POD0, POD2 and POD7. DAH30.	While not all patients undergoing major inpatient procedures can complete the QOR-15 on the evening of surgery, a subset will be able to do so. Evaluating QOR-15 on POD2 will still provide valuable insights, and POD7 is included to capture the non-linear nature of recovery. Since patients prioritise a rapid return home, DAH30 will serve as a key secondary outcome, aligned VITAL trial (ISRCTN62903453).
Determine the incidence of delirium, the incidence of subsyndromal delirium, the severity of delirium, functional status and health-related quality of life after anaesthesia and surgery following two commonly used and established anaesthetic techniques: (1) TIVA with propofol and (2) INVA in patients undergoing (a) major inpatient surgery, (b) minor inpatient surgery or (c) outpatient surgery.	Incidence and severity of delirium on POD0 and POD1 with the 3D-CAM, functional status measured with 12-item WHO WHODAS 2.0 on POD30 and POD90, and quality of life measured with the EQ-5D-5L on POD30, POD90, POD180 and POD365.	There are no large comparative effectiveness studies of intraoperative anaesthesia choice and delirium despite this being one of the most emotionally devastating complications of surgery.58 59 Consensus guidelines have established both the 12-item WHODAS 2.0 as a measure of functional outcome and EQ-5D-5L as a measure of health-related quality of life as essential patient-centred outcomes due to their established validity, reliability and patient centeredness.60 These tools have been explicitly validated in surgical cohorts through prospective observational work61 and have been used as the primary endpoint in landmark surgical interventional studies.62 Additionally, the EQ-5D-5L harmonises with the VITAL and VAPOR-C (ClinicalTrials.gov Identifier: NCT04316013) trials.
Assess patient satisfaction with anaesthesia.	Bauer questionnaire on POD2.	Patient satisfaction with with anaesthesia is an important patient-centred outcome.60 This instrument is also harmonised with the VITAL trial.
Determine the incidence of respiratory failure, kidney injury, all-cause mortality, intraoperative hypotension, moderate-to-severe intraoperative movement, unplanned admission after outpatient surgery, propofol-related infusion syndrome and malignant hyperthermia after or during anaesthesia and surgery following two commonly used and established anaesthetic techniques: (1) TIVA with propofol and (2) INVA.	Incidence of respiratory failure on POD0, kidney injury on POD7, all-cause mortality at POD30 and POD90, intraoperative hypotension, moderate-to-severe intraoperative patient movement, unplanned admission after an outpatient surgery, propofol-related infusion syndrome and malignant hyperthermia.	These secondary safety outcomes are most germane to patients undergoing surgery. There are no large-scale reliable data demonstrating a safety difference.35 63 64

DAH30, days alive and at home up to POD30; 3D-CAM, 3D-Confusion Assessment Method; EQ-5D-5L, EuroQol 5 dimension, five-level version with visual analogue scale; INVA, inhaled volatile anaesthesia; POD, postoperative day; QOR-15, Quality of Recovery-15; TIVA, total intravenous anaesthesia; VAPOR-C, Volatile Anaesthesia and Perioperative Outcomes Related to Cancer; VITAL, Volatile vs Total intravenous Anaesthesia for major non-cardiac surgery; WHODAS, WHO Disability Assessment Schedule.

Biological rationale for the primary endpoints

QOR-15 instrument measures overall patient experience across the domains of breathing, eating, rest, sleep, personal hygiene, communication, functional status and general well-being.^{8 9} The recovery experience as measured by the QOR-15 instrument is biologically plausibly impacted by (1) the type of anaesthesia (TIVA vs INVA), (2) the invasiveness of surgery and (3) the hospital experience. For patients undergoing (1) major inpatient surgery, the impact of invasive surgery and being admitted to the hospital on recovery experience might be large such that the contribution of anaesthesia type (TIVA vs INVA) might be difficult to detect. For patients undergoing (2) minor inpatient surgery, the impact of being admitted to the hospital on recovery experience might be sufficiently large such that the contribution of anaesthesia type might not be apparent. For patients having (3) minor outpatient surgery, it is plausible that the contribution of anaesthesia type to QOR will be relatively more impactful and will be easier to detect. Thus, it is important to evaluate the QOR as a co-primary endpoint in each of these three surgical groups. With ~13 000 patients in the THRIVE trial, and with an estimated minimum of 1300 (10%) of patients in each surgical category, there will be adequate statistical power to compare the QOR between patients who receive INVA and TIVA in each of these three surgical groups.

In contrast to QOR, it is unlikely that the three surgical categories described will variably impact the risk of the primary safety endpoint of unintended definite intraoperative awareness, which is a rare complication occurring during surgery. Within the elective, non-cardiac, non-obstetric surgical population, no specific surgical complexity has been identified as a risk factor for unintended definite intraoperative awareness. Thus, it is both biologically rational and statistically important to assess the entire ~13 000 patient population for this primary endpoint. With 13 000 patients, there will be adequate statistical power to evaluate whether TIVA confers no more than a small (0.2%) increased risk of definite intraoperative awareness than INVA. The statistical considerations for the analyses in relation to the trial’s primary endpoints are elaborated in the section of this protocol on ‘Statistical considerations’.

Methods

Trial design

This trial is designed in accordance with the Standard Protocol Items: Recommendations for Interventional Trials guidelines¹⁰ to establish the safety and efficacy of propofol TIVA compared with INVA in 13 000 non-cardiac surgical patients. It is a multicentre, patient-blinded, randomised controlled trial in which one group will receive propofol TIVA and the other INVA (see figure 1). The study began enrolling patients in September 2023 and is expected to conclude in June 2028. Patients will be allocated to one of two trial arms using minimisation,^{11 12} a method of unbiased allocation ensuring balance across confounding factors. This approach addresses the importance of controlling for known or suspected confounders: centre, age, sex and procedural complexity (see section ‘Definitions’).

Figure 1. Trial design. INVA, inhaled volatile anaesthesia; TIVA, total intravenous anaesthesia.

Patient and public involvement

Patients with previous lived experiences receiving general anaesthesia for surgical procedures were involved in the design and conduct of this research. These individuals participated in planning discussions for the feasibility trial and provided intellectual input to the development of this protocol.⁷

Trial setting

This trial and data collection will take place at 12 or more US hospitals.

Study population

Inclusion criteria

Each patient must meet all of the following criteria:

1. Aged 18 years or older.

2. Undergoing elective non-cardiac surgery expected to last ≥60 min requiring general anaesthesia with a tracheal tube or laryngeal mask airway (or similar supra-glottic device).

Exclusion criteria

Patients will not be enrolled if any of the following criteria are met:

1. Inability to provide informed consent in English (at all study sites) or Spanish (at sites where Spanish consent is provided as an option).

2. Pregnancy (based on patient report or positive test on the day of surgery).

3. Surgical procedure requiring general, regional, neuraxial anaesthesia administered by an anaesthesia clinician (anaesthesiologist, certified registered nurse anesthetist, anaesthesiology assistant) occurring within 30 days prior to or planned to occur within 30 days after surgery date.

4. Contraindication to propofol TIVA or INVA (eg, documented allergy to propofol, history of severe postoperative nausea or vomiting, concern for or history of malignant hyperthermia) based on self-report.

5. Surgical procedures specifically requiring either TIVA or INVA.

6. Locally approved, written protocol mandating a particular anaesthetic technique.

7. History of possible or definite intraoperative awareness during general anaesthesia based on patient self-report.

8. Planned postoperative intubation.

9. Current incarceration.

Trial procedures

Recruitment and informed consent

This trial will randomise approximately 13 000 patients aged 18 years or older who are able to participate in informed consent and are undergoing elective non-cardiac surgery requiring general anaesthesia. Recruitment strategies include (1) individualised outreach to participants at home via telephone or email, (2) in-clinic enrolment during preoperative assessment and (3) surgical patient community engagement.

After reviewing upcoming clinic or operating room schedules, research coordinators will reach out to patients via emails, phone calls and/or through patient portal messages to inform them of this trial. Patients may also be approached during surgical or preoperative clinic visits. Prior to the surgery, patients will complete informed consent via one of two mechanisms: (1) study coordinator-mediated remote eConsent using modules on a participant’s personal smartphone, tablet or website; or (2) study coordinator-mediated eConsent on a study tablet or computer. Patients will be asked a series of questions assessing their understanding of the consent document. Patients will be considered fully consented when they answer all assessment of understanding questions correctly and agree to participate (see online supplemental appendix 1).

Participants contacted by the study team for eligibility screening will provide verbal consent via a phone screening process (online supplemental appendix 1). If a participant declines to enrol after the phone screening, they will be asked to provide reasons for their decision. For participants who are screened through medical records but are either unreachable for screening consent or opt out of screening, relevant data from the electronic health record (EHR), including race and ethnicity, gender, gender identity, age, planned surgery and medical record number (MRN), will be retained for analysis. This data collection will be conducted under a waiver of consent as it involves minimal risk.

Blinding

This is a double-blinded study (participant and outcome assessor). Both treatments are administered after the patient is unconscious and discontinued before regaining consciousness. The EHR available in the patient portal does not reveal intraoperative anaesthesia details. To maintain blinding, education will be provided to surgical and recovery staff and other clinicians not involved in administering the intervention. Patient-reported outcomes collection will be completed at approximately POD365, after which patients will be unblinded and informed of the treatment they received. Anaesthesia clinicians caring for patients in the operating rooms cannot practically or ethically be blinded due to their role in administering the treatment. Outcome data collectors, those administering post-Brice questionnaires and the intraoperative awareness classification team will remain blinded to treatment allocation.

Wearables

Participation in the wearable device data collection is optional and does not affect overall trial eligibility. Wearable data collection will be included in a single informed consent document with an opt-in option. Participants may use their own compatible devices (Apple Watch or Fitbit) or, at select sites, may be offered a study-provided device. Details of procedures and statistical analysis for wearable data will be outlined separately.

Data

Outcomes

Table 2 provides a summary of the patient outcomes that will be collected.

Table 2. Patient outcomes.

Type	Outcome	Source/data system	Timepoints
Primary (effectiveness)	Quality of recovery	PRO/MDH	POD1
Primary (safety)	Definite intraoperative awareness	PRO/MDH	POD1 and 30
Secondary (effectiveness)	Quality of recovery	PRO/MDH	POD0, 2 and 7
Secondary (effectiveness)	Delirium	Coordinator assessment/MDH	POD0 and 1
Secondary (effectiveness)	Subsyndromal delirium	Coordinator assessment/MDH	POD0 and 1
Secondary (effectiveness)	Delirium severity	Coordinator assessment/MDH	POD0 and 1
Secondary (effectiveness)	Functional status	PRO/MDH	POD30 and 90
Secondary (effectiveness)	Health-related quality of life	PRO/MDH	POD30, 90, 180 and 365
Secondary (effectiveness)	Days alive and at home	PRO/MDH and coordinator review of EHR/MQUARK	POD30
Secondary (effectiveness)	Patient satisfaction with anaesthesia	PRO/MDH	POD2
Secondary (safety)	Respiratory failure	Coordinator review of EHR/MQUARK	POD0
Secondary (safety)	Stage 1 AKI	EHR/MPOG	POD7
Secondary (safety)	Mortality	EHR/MPOG and coordinator review of EHR and NDI Query	POD30 and 90
Secondary (safety)	Intraoperative hypotension (Mean arterial pressure <55 and Mean arterial pressure <65)	EHR/MPOG	During surgery
Secondary (safety)	Moderate-to-severe intraoperative movement	Clinician report/MQUARK	During surgery
Secondary (safety)	Unplanned hospital admission from a freestanding ambulatory surgery centre	Coordinator review of EHR	POD0–1
Secondary (safety)	Propofol-related infusion syndrome	Clinician report and coordinator review of EHR/MQUARK	During surgery or POD0
Secondary (safety)	Malignant hyperthermia	Clinician report and coordinator review of EHR/MQUARK	During surgery or POD0
Exploratory (trial conduct)	Patient satisfaction with participation	PRO/MDH	POD90

AKI, acute kidney injury; EHR, electronic health record; MAP, Mean Arterial Pressure; MDH, MyDataHelps; MPOG, Multicenter Perioperative Outcomes Group; MQUARK, MPOG Quality and Research Kit; NDI, National Death Index; POD, postoperative day; PRO, patient-reported outcome.

Data systems

Three distinct information systems will be used to collect patient and procedure data. These data are integrated to provide a complete trial record:

1. MPOG Quality and Research Kit (MQUARK) will be used to manage patient screening, enrolment, randomisation and serious adverse event (SAE) processes. This existing research system has been customised to the needs of the THRIVE trial and provides seamless integration with data collected from the other systems. Patient enrolment details, patient demographics, per protocol treatment delivered and clinician report of intraoperative patient movement will be captured here.

2. MyDataHelps is a patient-facing application that allows the collection of patient-reported outcome data via the administration of surveys. Surveys can be completed by dedicated smartphone application, email or web. This application can also be used by trial coordinators to document patient-reported outcomes collected by the coordinator. Additionally, wearable device data will be obtained using this application.

   1. Multicenter Perioperative Outcomes Group (MPOG) Import Manager extracts data from the EHR at each participating institution, standardises it against a common data dictionary and transfers the data to the data coordinating centre (DCC) at the University of Michigan. Perioperative EHR data will be collected via this system. The MPOG EHR interface has been modified to include protected health information (PHI) consented participants in this trial.

Data collection

Table 3 summarises the data that will be collected. Online supplemental appendix 2 provides a comprehensive overview of the data collection timepoints.

Table 3. Data collection.

Data collected	Method of data acquisition
Patient enrolment details
Demographics Screening Eligibility Attempted approach Successful approach (bidirectional communication) Eligibility after approach Ineligibility reasons Consent or refusal to participate Refusal to participate reason if provided Withdrawal post consent Reason for withdrawal post consent Randomisation success Reasons why randomisation did not occur	Completed by a research coordinator from the patient or form a review of the electronic health record
Participant-provided information (see online supplemental appendix 3)
Quality of Recovery-15 Instrument52 Modified Brice Interview36,3865 Risk Analysis Index surgical frailty assessment66,68 Ultra-brief Confusion Assessment Method69 70 3D-Confusion Assessment Method71,73 WHO Disability Assessment Scale 2.074 75 Health-related quality of life60 Patient satisfaction with the study and with anaesthesia60 Days alive and at home at postoperative day 3060 Demographics survey	Completed by patient or facilitated through a research coordinator interview
Intervention adherence and perioperative events
Adherence to randomised treatment Intraoperative undesired patient movement questionnaire Occurrence of malignant hyperthermia Occurrence of propofol-related infusion syndrome Use of processed electroencephalogram Michigan Awareness Classification Instrument76 and a structured follow-up questionnaire for participants who report a positive Modified Brice screen	Completed by anaesthesia clinicians via the clinician checklist (online supplemental appendix 3) or by a designated healthcare worker±-the intraoperative awareness adjudication team (online supplemental appendix 4)
Inhaled volatile or nitrous oxide inspired and expired concentration Medications administered between anaesthesia start and stop: total dose (including boluses and infusions) and infusion time (in minutes) of propofol, dexmedetomidine, remifentanil, fentanyl, sufentanil, hydromorphone, morphine, lidocaine, ketamine, magnesium, midazolam and lorazepam Processed electroencephalogram device data Surgical duration Anaesthesia duration Intraoperative mean arterial pressure Serum creatinine value (preoperative and postoperative, if available per routine clinical process) Timing of prolonged mechanical ventilation or reintubation Occurrence of malignant hyperthermia Occurrence of propofol-related infusion syndrome Hospital admission within 24 hours after outpatient surgery at a freestanding ambulatory surgery centre All-cause mortality	Obtained directly through the electronic health record or through a coordinator review of the electronic health record

Definitions

Table 4 provides definitions for per-protocol treatment, procedural complexity and adverse events (AEs) relevant to this study.

Table 4. Definitions.

Per-protocol treatment
TIVA	Patient receives intravenous propofol and does not receive inhaled anaesthetics (sevoflurane, isoflurane, desflurane, nitrous oxide).*.
INVA	Patient receives an inhaled volatile anaesthetic agent (sevoflurane, isoflurane, desflurane) at a concentration of at least 0.5 age-adjusted Minimum Alveolar Concentration (MAC) during the surgery.†.
Patients in both groups may receive additional intravenous adjuncts, including propofol, as deemed appropriate by the clinical team. All other interventions are at the discretion of the treating anaesthesia clinicians.
Procedural complexity
Outpatient	Prior to surgery, the patient is scheduled to go home (or to the location they came from, such as a skilled nursing facility) the same calendar day as their procedure.
Minor inpatient	Prior to surgery, the patient is scheduled to spend one midnight in the hospital.
Major inpatient	Prior to surgery, the patient is scheduled to spend two or more midnights in the hospital.
Adverse events‡
31–90-day all-cause mortality	Mortality that occurred between postoperative days 31–90.
Acute kidney injury	Serum creatinine increase of 50% of 0.3 mg/dL from preoperative baseline within 7 days of surgery.77
Intraoperative hypotension	Cumulative duration of mean arterial pressure<55 mm Hg for 20 min or greater and <65 mm Hg for 20 min or greater.78 79
Moderate or severe intraoperative undesired patient movement	Moderate: movement that impacted the surgery (eg, required a pause in the surgery for coughing or straining); severe: movement with a marked negative impact on the surgery (eg, a patient injury, loss of sterility of the surgical field, other surgical complication).
Unplanned admission after outpatient surgery	Hospital admission no later than 24 hours postoperatively after outpatient surgery in a free-standing ambulatory surgery centre.
Serious adverse events‡
Definite intraoperative awareness	Explicit episodic recall during a period of intended general anaesthesia.
Respiratory failure	Unplanned postoperative intubation or reintubation or continued mechanical ventilation>6 hours postoperatively. Reintubation due to reoperation is excluded.
30-day all-cause mortality	Mortality that occurred by postoperative day 30.
Propofol-related infusion syndrome	Acute refractory bradycardia in the presence of metabolic acidosis, and at least one of the following: rhabdomyolysis, acute kidney injury or hypertriglyceridaemia,80 81 occurring after the start of propofol intraoperatively and within 6 hours postoperatively.
Malignant hyperthermia	Unexplained muscle rigidity, tachycardia, hypercapnia and rapidly increasing temperature leading to metabolic acidosis, rhabdomyolysis, disseminated intravascular coagulation and ventricular arrhythmias,43 82 occurring after the start of an inhaled volatile agent intraoperatively and within 6 hours postoperatively.

^*< 5 min of inhaled volatile end tidal concentration detected by the automated MPOG data interface will be considered compliant with TIVA. This can occur during the administration of non-anaesthetic inhaled medications erroneously measured as such (eg, albuterol) or due to inadvertent activation of the volatile vaporiser which is immediately detected and corrected.

^†Inhaled agent(s) will be chosen at the discretion of the administering clinician.

^‡All adverse events and serious adverse events will be collected after treatment with anaesthesia has begun. Time of arrival in post-anaesthesia care unit (or if arrival time is not documented, anaesthesia end) will define the beginning of the postoperative period.

INVA, inhaled volatile anaesthesia; MAC, Minimum Alveolar Concentration; MPOG, Multicenter Perioperative Outcomes Group; TIVA, total intravenous anaesthesia.

Processed EEG monitoring

Each site will be expected to determine the method of ensuring processed EEG (pEEG) monitoring is consistent in both treatment arms. Guidelines^{13 14} strongly recommended pEEG monitoring be used when TIVA is administered with neuromuscular blockade, encourage use when TIVA is used alone and suggest a role for reducing accidental awareness risk in patients receiving INVA. Additionally, anaesthesia clinicians should be trained in the use of pEEG monitors and familiar with the principles, interpretations and limitations of such.^{13 14} Therefore, it is strongly recommended that EEG monitoring is used in patients randomised to both TIVA and INVA in this trial.

Statistical considerations

Statistical and analytic plans

A statistical analysis plan (SAP) will be written for the trial that contains detailed descriptions of the analyses to be performed. The SAP will be finalised prior to the data lock.

Statistical hypothesis

There are four primary hypotheses to be evaluated in this study: (1) the superiority of TIVA relative to INVA for effectiveness in patient recovery, as measured by QOR-15 score on POD1, in each of three procedural complexity groups: (a) major inpatient surgery, (b) minor inpatient surgery or (c) outpatient surgery, and (2) the non-inferiority of TIVA relative to INVA for safety, as measured by the incidence of unintended definite intraoperative awareness with recall at either POD1 or POD30 in all participants. Statistically, the following null and alternative hypotheses will test superiority and non-inferiority of TIVA vs INVA, respectively, for the effectiveness and safety co-primary endpoints:

Superiority hypotheses for effectiveness

For participants undergoing major inpatient surgery:

• H01a: μ TIVA – μ INVA=0

• HA1a: μ TIVA – μ INVA ≠ 0

For participants undergoing minor inpatient surgery:

• H01b: μ TIVA – μ INVA=0

• HA1b: μ TIVA – μ INVA ≠ 0

For participants undergoing outpatient surgery:

• H01c: μ TIVA – μ INVA=0

• HA1c: μ TIVA – μ INVA ≠ 0

where μ reflects the adjusted mean QOR-15 score on POD1.

Non-inferiority hypothesis for safety

• H02: π TIVA – π INVA ≥ δ

• HA2: π TIVA – π INVA < δ

where π reflects the incidence of unintended definite intraoperative awareness of recall at either POD1 or POD30 and δ reflects the non-inferiority margin of 0.2%.

Sample size determination

In order to have sufficient evidence to influence patient and clinician practice in the selection of anaesthesia, we sized THRIVE to test the superiority of TIVA relative to INVA for effectiveness in patient recovery in at least one procedural complexity group (outpatient, minor inpatient and major inpatient), and to test the non-inferiority of TIVA relative to INVA for the safety endpoint of intraoperative awareness. Because of the low estimated incidence of definite intraoperative awareness and the importance to patients of small increases in this safety endpoint, the targeted sample size of 13 000 is driven by this safety aim. The overall type I (α) error for the study is controlled for the three tests of effectiveness in the procedural complexity groups by using a Bonferroni adjustment ($\alpha /3)$). Because both effectiveness and safety are required to recommend TIVA, there is no additional adjustment of the type I error for these comparisons. If success is achieved for at least one of the procedural complexity groups for the co-primary effectiveness endpoint and the co-primary safety endpoint, we will use a gatekeeper approach to then test the superiority of TIVA in patient recovery for each additional subgroup (age, sex, race/ethnicity).¹⁵ For this family of tests, we adjust for multiplicity in planned effectiveness subgroup analyses using a Bonferroni adjustment approach. We also use a conservative approach to power calculations by using simplifications of the analysis plans below and ignoring covariate adjustment. Since both effectiveness and safety are required for success, we note that the impact on type II error is capped at 14.5% (ie, 0.90×0.95) in the unlikely situation that effectiveness and safety outcomes are independent.^{16 17}

The non-inferiority approach¹⁸ described in figure 2 shows the underlying hypothesis testing threshold for superiority (ie, the treatment difference of intraoperative awareness for TIVA intraoperative awareness for INVA=0) using CIs. The x-axis represents the absolute treatment difference between TIVA and INVA groups. Adding the dashed blue line at the non-inferiority margin value, Δ, and shading in the non-inferiority region helps illustrate the analysis using CIs and the interpretation of various scenarios. For example, if the entire CI is completely to the left of 0, then TIVA is non-inferior (indeed superior) to INVA, and if the entire CI is completely to the right of 0, then INVA is superior to TIVA (or TIVA is not non-inferior). If the upper confidence bound is less than Δ, then TIVA is non-inferior to INVA.

Figure 2. Diagrammatic representation of the non-inferiority approach. INVA, inhaled volatile anaesthesia; TIVA, total intravenous anaesthesia.

The planned sample size was based on achieving at least 90% power using a one-sided type I error of 0.025 to test for non-inferiority (two-sample binomial test). Based on prior work, we assumed a 10% crossover or loss to follow-up.¹⁹ Using an estimated incidence of definite intraoperative awareness of 0.1% with INVA²⁰ and choosing a relatively small non-inferiority margin of 0.2%, we need to randomise 11 664 patients. A total trial population of 12 500 was initially planned.

Early in trial conduct (prior to 25% enrolment), an alternative sample size calculation considering the rare event status of the safety outcome analysis using continuity correction for rare event analysis was proposed. An approximate sample size calculation used a continuity correction to the one-sided binomial test, resulting in 12 418 total patients. To account for a crossover or loss to follow-up rate of 2–5%, we considered inflating this number to 13 000 patients. We then ran simulations applying the Newcombe/Wilson interval to calculate the upper 97.5% confidence limit and see that we have 87–89% power to declare non-inferiority (table 5).

Table 5. Power of the safety endpoint to determine non-inferiority with a 0.2% margin.

Total sample size	Crossover and loss to follow-up rate (%)	Total sample size	Average N/arm	Power with Newcombe interval
13 000	2	12 740	6370	0.888
	3	12 610	6305	0.888
	4	12 480	6240	0.872
	5	12 350	6175	0.876

For the effectiveness endpoint of QOR-15 at POD1, the clinically important difference for QOR-15 ranges between 3 and 8 depending on the population and the SD ranges between 15 and 20.¹⁵ With 13 000 patients, 10% dropout (due to POD1 QOR-15 primary effectiveness outcome loss to follow-up or crossover), 96% power and a two-sided Bonferroni-adjusted type I error of 0.0167 (ie, 0.05/3), we can detect an extremely small effect size (treatment difference/SD) of 0.241 if the size of the smallest procedural complexity group is 10% of the total (two-sample t test, R V.4.4.1). This translates to a mean treatment difference of 4.82 with SD=20 or 3.62 with SD=15, below the clinically important difference, assuming our assumptions hold. If the size of the smallest procedural complexity group is 20% or 30% of the total, the respective effect sizes are 0.171 and 0.139. These translate into mean treatment differences of 3.42 and 2.78, respectively, with SD=20 and 2.56 and 2.08, respectively, with SD=15. As with the non-inferiority approach, CIs will be calculated to provide assurance that a clinically meaningful, as well as statistically significant, improvement in recovery is demonstrated with TIVA (ie, the 95% CI includes 3) in at least one of the three procedural complexity groups.

We also calculated the effect size we can detect for subgroup analyses of effectiveness using a Bonferroni adjustment for nine comparisons (three planned subgroup comparisons (three age groups, two sex groups and four race/ethnicity groups)). In addition to these features, the effect size is dependent on the proportion of patients in each subgroup and the treatment allocation ratio. For example, we anticipate a 1:1 ratio of female and male patients treated with TIVA and INVA. For other subgroups, we assume that the ratio would be no more than 1:2. Table 6 describes the effect size for various subgroup sample sizes and allocation ratios (two-sample t test, R V.4.4.1). The effect sizes are equivalent to mean treatment differences that are consistent with clinically important differences.

Table 6. Subgroup sample size and effect size analysis.

Patients in subgroup (%)	Total	Allocation ratio	Group 1	Group 2	Effect size
	N		N	N
20	2340	1:1	1170	1170	0.183
15	1756	1:1	878	878	0.211
10	1170	1:1	585	585	0.259
5	586	1:1	293	293	0.366
20	2340	1:2	780	1560	0.194
15	1756	1:2	585	1170	0.224
10	1170	1:2	390	780	0.275
5	586	1:2	195	390	0.389

Next, the minimal clinically important difference for the change from baseline in EQ-5D-5L is 0.063 with a reported SD of 0.013 ²¹. With a two-sided type I error of 0.05, 10% dropout, there is >99% power to detect this magnitude of effect.

Populations for analysis

For the primary effectiveness endpoint of QOR-15 at POD1 in the three procedural complexity groups, we will use the intention-to-treat (ITT) analysis set in the superiority test setting to avoid overoptimistic estimates of effectiveness resulting from a per-protocol analysis and make full use of the randomisation. The ITT analysis set includes all randomised participants. We will use multiple imputation methods to handle missing data.^{22 23} Sensitivity to missing data assumptions will be assessed based on flexible, semi-parametric approaches.²⁴

For the primary safety endpoint of definite intraoperative awareness, regulatory guidances^{25 26} and the clinical trials literature^{27 28} note that the conventional wisdom has been that the per-protocol analysis set may be preferable to the ITT analysis set in the non-inferiority trial setting. The per-protocol population is defined as all randomised participants who undergo the assigned propofol TIVA or INVA treatment, have no major protocol deviations and provide the endpoint. The rationale is that the ITT analysis tends to be more liberal by including those who do not complete the full course of treatment, resulting in a bias towards making the two treatments look similar. However, there is debate about this opinion. Thus, we will assess the non-inferiority of the primary safety endpoint using the per-protocol population and use the ITT population as a secondary analysis. Both results will be considered in assessing the success of this objective. Finally, no adjustments for multiplicity will be made for secondary and other exploratory endpoints. Thus, interpretation of these endpoints will rely on CIs and the magnitude of the treatment differences.

The ITT analysis set will be used for all tests of secondary and exploratory endpoints.

Statistical analyses

Primary analysis of the primary endpoints

Two-sample t tests will assess the superiority of TIVA vs INVA on the QOR-15 at POD1 in the ITT analysis set. For the QOR-15 comparison, the two-sided p value<0.0167 (=0.05/3; Bonferroni adjustment for test of the co-primary effectiveness endpoint in three procedural complexity groups) will be considered statistically significant. Simple unadjusted treatment means, SEs and 95% CIs will be reported for each procedural complexity group.

The primary safety endpoint will be assessed using the upper 97.5% CI derived from the Newcombe/Wilson score interval on the difference of proportions between arms.²⁹ Non-inferiority of TIVA relative to INVA will be shown if the upper limit of the two-sided 95% CI (equivalent to the one-sided 97.5% upper confidence bound) of the difference (TIVA – INVA) in the definite intraoperative awareness incidence risk is 0.2% or less in the per protocol analysis set.

Secondary analyses of the primary endpoints

The same analysis described above for definite intraoperative awareness will be performed using the ITT analysis set.

Secondary multivariable analyses of QOR-15 at POD1 incorporate prespecified baseline prognostic information (eg, baseline QOR-15, EQ-5D-5L, WHO Disability Assessment Schedule (WHODAS), frailty assessment and specific comorbidities; detailed surgery type, surgical duration and stratification factors) to boost statistical precision for the same causal estimands.³⁰ For the primary effectiveness endpoint, linear models and for the primary safety endpoint, a generalised linear model (GLM) for binary endpoints, will be used with terms for the treatment and other prognostic baseline information³¹ to provide adjusted estimates of treatment differences.

Subgroup analyses of QOR-15 at POD1 will be performed using two-sample t tests to assess treatment differences within each subgroup. For prespecified marginal subgroup-specific comparisons, two-sided p values<0.0056 (=0.05/9; Bonferroni adjustment) will be considered statistically significant for testing treatment effects.

Analysis of secondary endpoints

The analytic approach for secondary endpoints is testing for superiority between TIVA and INVA at the two-sided 0.05 level using the ITT analysis set. Continuous effectiveness endpoints will be analysed with linear models (as for secondary analyses of the primary effectiveness endpoint), with terms for the treatment arm, baseline endpoint and additional baseline prognostic information. For binary safety endpoints (eg, kidney injury, respiratory failure, mortality at POD30 or POD90), GLM with the same independent variables as in the linear models will be used. Assumptions of each model will be assessed. Missing data will be handled using multiple imputation methods, as described for the primary analysis. For secondary effectiveness endpoints that are repeatedly assessed at multiple postoperative occasions (eg, QOR-15 POD0, POD2 and POD7, WHODAS POD30, POD90, EQ-5D-5L POD30, POD90, POD180 and POD365), statistical analysis will use the generalised estimating equations (GEE) method.³² GEEs will consider an unstructured correlation matrix (where other working covariance structures will be explored if there are convergence issues), and linear link for continuous endpoints; robust SEs will be used for inference. GEE models will adjust for independent variables. Findings here will be interpreted based on CIs and the magnitude of the effect estimates.

Planned interim analyses

There are no planned interim analyses to stop for benefit or futility. The final analysis will be performed approximately 2 months after the last participant’s last visit.

Monitoring

Adverse event reporting and safety monitoring

There is no single standard or consensus definition regarding what constitutes a safety or AE in a clinical trial. In documents focused on AEs from the US Office for Human Research Protections and the US Food and Drug Administration, the following broad definition is provided: a safety or AE is any untoward or unfavourable medical occurrence in a human subject, including any abnormal sign, symptom or disease, temporally associated with the subject’s participation in the research, whether or not considered related to the subject’s participation in the research.^{33 34} The safety profiles of both commonly used general anaesthetic techniques in this trial, propofol TIVA and INVA, are well-recognised and can be attributed as low-risk in a controlled intraoperative setting. As such, we collaborated with our patient partner panel to identify relevant AEs and SAEs for this trial that would be considered life-threatening or have a potential negative long-term impact on participants. Unplanned hospital admission after outpatient surgery in a free-standing ambulatory surgery centre setting, acute kidney injury,³⁵ intraoperative hypotension,³⁵ moderate-to-severe undesired intraoperative patient movement and all-cause mortality between POD31–90 will be considered AEs. Unintended definite intraoperative awareness,³⁶,³⁹ respiratory failure,⁴⁰ all-cause mortality at POD30,⁴¹ propofol-related infusion syndrome⁴² and malignant hyperthermia⁴³ are all rare^{4 5 44 45} and will be considered SAEs (see table 4 for definitions). All SAEs will be reported according to an established trial reporting process (see online supplemental appendix 5).

As part of the informed consent process for this trial, patients will be informed of safety and AEs and the likelihood of them occurring based on the evidence available for these events. The research team at each participating site will monitor the trial for all safety and AEs or any unanticipated problems involving risk to the patients or others. Prompt reporting of any unanticipated events at enrolling sites will be reported promptly to the Clinical Coordinating Center for relevant reporting to regulatory entities. SAEs will be reported to the Institutional Review Board (IRB), the principal investigator (PI) at each site and an independent safety officer. The role of the safety officer is to provide prompt clinical review of each SAE reported to them and provide a determination of whether the event was causally related to the participant’s involvement in the study. This safety officer is a clinically trained anaesthesiologist and clinician scientist uninvolved in the conduct of the study and without any relevant conflicts of interest.

This trial will be advised by a Data Safety and Monitoring Board (DSMB). The DSMB will review accumulating clinical trial data to provide independent safety oversight of this trial, as well as the general conduct of the trial.⁴⁶ The DSMB will comprise 3–6 independent, multidisciplinary experts from multiple institutions. One of the members of the DSMB will include the safety officer who will help guide the DSMB in evaluating AEs and SAEs. The DSMB members will have the requisite expertise to examine accumulating data, protect and enhance the scientific integrity of the trial, and protect trial participants and future patients by ensuring that they are not unduly or unfairly at risk for harm by trial interventions or denial of effective interventions.^{47 48} The DSMB members will advise the trial sponsor regarding trial continuation, modification, or termination. These members will not be involved in other aspects of the trial and will not have financial, proprietary or professional conflicts of interest, which may affect the impartial, independent decision-making responsibilities of the DSMB.^{46 49 50}

Premature discontinuation

Patients will be withdrawn if the investigator decides that discontinuation is in the best interest of the patient or the patient requests withdrawal from the trial. Early stoppage will be based on safety concerns only, which are not anticipated given that both anaesthetic techniques are in regular, routine clinical practice. There will be no prespecified interim analysis.

We will discontinue collection of any new data after the request has been processed; however, data collected prior to the date of withdrawal can be used for research initiated after the date of withdrawal.

Potential risks

The risks to patients in this trial are anticipated to be no greater than the risks associated with the planned surgery and general anaesthesia. There is a small risk of breach of confidentiality.

Patients will not incur any trial-related expenses. Both treatment allocations are in routine use and have similar technical charges associated with them. The anaesthesia clinician professional charges are identical with each treatment option. In routine care, there is no discussion of cost differential between the two options given this similarity.

If a patient is provided a study wearable device, there is a small chance that they may experience local reactions to materials in the wearable device (Apple Watch, Fitbit, etc) due to allergies, environmental factors, extended exposure to irritants like soap or sweat and other causes. Patients will be advised to remove their wearable device and consult their physician if they experience redness, swelling, itchiness or any other irritation. Patients may also be encouraged to clean the band or change the type of strap they use with the wearable, though this would be at the patient’s own expense. In the event that the study team becomes aware of unexpected medical events reported from wearable devices, the patient will be advised to seek appropriate medical care for diagnosis and treatment.

Procedures to minimise potential risks

Data protections will be consistently applied to trial data to minimise risk of breach of confidentiality. Patients will not be identified by name in any analyses, reports or publications. Some patients may wonder about the confidentiality of the information collected from the surveys and other data. The PIs, co-investigators and trial personnel have been trained in the protection of patient confidentiality and will be able to reassure the small number of anticipated patients who might raise concerns. Participation in the trial will be voluntary, and the trial procedures will be described in the consent process. All trial staff have or will receive training in the responsible conduct of research prior to their involvement in the trial.

Ethics and dissemination

This study is approved by the ethics board at Washington University (IRB# 202304082), serving as the single IRB for all participating sites.

Protocol amendments

All protocol modifications made during the course of this trial will be communicated to the IRB, DSMB and Patient-Centered Outcomes Research Institute (PCORI).

Protection of patients

Patients will provide informed consent. Patients will undergo the standard preoperative anaesthesia assessment and will be enrolled in the trial prior to surgery. Both interventions in this trial are established, routine standards of care. Thus, participation in this trial is not considered to have the potential for increased risk.

Sources of materials

Patient research outcomes will be obtained from the EHR at each participating institution (including the MPOG database) in addition to survey data collected by blinded research assistants, and data from wearable devices (Apple Watch or Google Fitbit) using the MyDataHelps application.

List of PHI collected for study

In order to facilitate follow-up, compensation for participation and linkage to vital records data, we will collect individual identifiers, including name, birthdate, social security number, MRN, addresses and telephone numbers. Access to PHI will be restricted to study personnel in roles directly requiring it for trial operations or required in the analysis and interpretation of study data.

Data management

The potential risk of disclosure of confidential information is guarded against by maintaining data on a secure server with access limited to the key research personnel. The primary database server and all information system servers will be housed at the DCC (University of Michigan), compliant with enterprise information assurance requirements (firewall, VPN, intrusion detection). All data stored electronically will be encrypted at rest. In addition, datum level audit trails, role-based access, two-factor authentication and minimal necessary use of identifiers will be implemented. While the study aims to be as paperless as possible, study sites do collect a paper clinical checklist that is then transcribed into the study Electronic Data Capture (EDC). In the case of system/software downtime, procedures may include the temporary use of paper records. Any physical research materials containing PHI will be stored in a locked cabinet inside a locked research office in case of a software downtime paper process. We will customize and deploy the existing MQUARK application to support this trial. This web-based application, hosted at the University of Michigan, will be the primary interface for the study sites. Sites will use this to document patient screening, approach, consenting and enrolment electronic case report forms (eCRFs). MQUARK includes customisable, data-driven eCRFs to capture data gathered by research coordinators at each site. MQUARK has been used to document clinical quality projects and prospective observational research for more than 10 000 patients across MPOG sites and has met strict medicolegal, audit trail, electronic signature and disaster recovery requirements across federal and state regulations. Only de-identified data will be sent out to research team members and data analysts for further data analysis. All persons involved in recruitment and data collection will undergo training in Human Subjects Research and Health Insurance Portability and Accountability Act (HIPAA).

Dissemination policy

The THRIVE team will disseminate the protocol and its contents through various channels, including peer-reviewed publication, media, blogs and plain language summaries on our website. We will present the protocol at relevant international scientific meetings. Our team will communicate progress in the trial to relevant stakeholders (eg, clinicians, hospital leaders, sponsor) and relevant updates will be appropriately communicated on social media platforms such as LinkedIn, Twitter and Instagram. The final results of the trial will be presented at scientific meetings, published in a peer-reviewed publication, included on ClinicalTrials.gov, shared with patients who participated in the trial and disseminated on relevant media and social media platforms.

Discussion

Overview

With a target inclusion of approximately 13 000 adult patients undergoing non-cardiac surgery, the THRIVE trial is the largest comparative effectiveness trial of the two main techniques of delivering general anaesthesia: TIVA and INVA. As such, the THRIVE trial has notable strengths and will complement the findings of the ongoing VITAL (ISRCTN registration number: ISRCTN62903453) and VAPOR-C trials (ClinicalTrials.gov Identifier: NCT04316013), which are also comparing TIVA with INVA. The THRIVE trial is designed primarily to address QOR, a patient-centered outcome reflecting perioperative experience and well-being. A recent systematic review and meta-analysis ⁵¹ found that patients receiving TIVA-reported improvements in QOR scores compared with those receiving INVA. These findings underscore the clinical relevance of this outcome and highlight the importance of high-quality, adequately powered trials to further define these effects in diverse surgical populations. The THRIVE trial is also sufficiently large that it is designed to address whether TIVA is non-inferior to INVA with respect to the safety endpoint of definite intraoperative awareness, the avoidance of which is consistently prioritised by surgical patients who are undergoing general anaesthesia. Notably, TIVA has been compared with INVA in patients undergoing cardiac surgery, and no significant differences were found between groups in postoperative complications, mortality or other outcomes.⁸⁰

There are concurrent trials that share overlapping endpoints with THRIVE. The VITAL trial is based in the UK and will include 2500 non-cardiac surgical patients older than 50. The primary endpoint of the VITAL trial is days alive and at home at 30 days after surgery. The VITAL trial has several secondary endpoints that are overlapping with THRIVE, including QOR; patient satisfaction with anaesthesia; intraoperative awareness and postoperative complications. The VAPOR-C trial is based predominantly in Australia and will include 5736 adult patients undergoing surgery for colon cancer, rectal cancer or non-small-cell lung cancer. The primary endpoint of the VAPOR-C trial is disease-free survival up to 3 years after surgery. Similar to VITAL, the VAPOR-C trial has several secondary endpoints that are overlapping with THRIVE.

Strengths

The THRIVE trial benefits from the clinical trials’ infrastructure of MPOG (www.mpog.org). MPOG is a unique resource that uses health data to analyse the interplay between patient characteristics, healthcare systems, surgical procedures, perioperative care and postoperative outcomes. MPOG creates an environment of innovation and collaboration using priorities driven by its participating members across 63 participating sites. MPOG conducts over 20 simultaneous prospective, retrospective and enhanced observational trials at any given time. Over the last decade, MPOG has built a perioperative patient registry of more than 30 million anaesthetic cases integrated across >70 health systems representing the most comprehensive and detailed global perioperative anaesthesiology registry. By embedding the trial in this ongoing data collection platform, it will enable post hoc analyses for important questions such as choice of adjuvant agents, environmental impact of care and practice change emerging from trial findings.

The THRIVE trial will be conducted in a variety of hospital settings across the USA and will therefore have broad applicability to US anaesthesia practice. This trial incorporates standard-of-care interventions that clinicians are trained to administer and examines outcomes that are important to patients, healthcare workers and the community. The findings from this trial will provide evidence to clinicians and patients that can inform future anaesthetic decision-making processes.

The THRIVE trial has been designed by diverse stakeholders, including anaesthesia clinicians, surgical patients, surgeons, hospital administrators and leaders of representative organizations. As such, it addresses outcomes that are relevant to all these groups (eg, clinical outcomes, patient experience, safety and efficiency) and is patient-centred in its conceptualisation. This approach is intended to empower patients and other stakeholders to share in decision-making regarding important perioperative decisions, in this case to be able to select the anaesthetic technique that is most appropriate for each individual patient.

Limitations

In many countries, the administration of propofol as part of TIVA is guided by a target-controlled infusion (TCI) algorithm. TCI algorithms for propofol have not been approved in the USA and dose rates for propofol and other drugs (eg, opioids) will be empiric based on clinician discretion in the THRIVE trial. While this can be considered a limitation, it does reflect US anaesthesia practice, which means that this approach will remain relevant and generalisable in the USA. Although in many countries both INVA and TIVA are commonly used for general anaesthesia, the overwhelming majority (>90%) of general anaesthetics in the USA are currently INVA rather than TIVA. Thus, for many clinicians participating in the THRIVE trial, there might be a learning curve for TIVA administration. To account for this, we plan to do a sensitivity analysis comparing the results towards the end of the trial (last 2 years) with the results obtained earlier in the trial (first 2 years). Electroencephalography (EEG) monitoring is not typically used for INVA in the USA, and since the majority of general anaesthetics are INVA, anaesthesia clinicians have limited experience with EEG monitoring. Therefore, our team has created EEG monitoring educational resources (videos, lectures, tip sheets), widely accessible to all sites, in order to strengthen clinician knowledge and confidence in use. In contrast, EEG monitoring is often used to assist with TIVA administration. To ensure that the use of EEG monitoring does not confound the findings of the THRIVE trial, the use of EEG monitoring at participating sites in the THRIVE trial will be standardised between groups.

All statements in this report, including its findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee. Although Blue Cross Blue Shield of Michigan/Blue Care Network and Multicenter Perioperative Outcomes Group work collaboratively, the opinions, beliefs and viewpoints expressed by the authors do not necessarily reflect the opinions, beliefs, and viewpoints of Blue Cross Blue Shield of Michigan/Blue Care Network or any of its employees.

Data availability statement

A statement regarding the availability of data will accompany the manuscripts arising from trial conduct. The THRIVE Data Sharing Plan will be consistent with PCORI’s Policy for Data Management and Data Sharing, which requires awardees’ to share their data sets and documentation for reanalysis and reuse. THRIVE will upload a deidentified dataset to the Patient-Centered Outcomes Data Repository (PCODR), which was created and is hosted by the Inter-University Consortium for Political and Social Research (ICPSR). The THRIVE study, as the awardee, will enter into a Data Contributor Agreement (DCA) with ICPSR. The DCA governs the data deposition and establishes the awardee’s rights and obligations. The data package will be transferred to the data repository by 1 October 2028.