1069225 research-article20222022 CJKXXX10.1177/20543581211069225Canadian Journal of Kidney Health and DiseaseGarg et al Journal canadien de la santé et de la maladie rénale Clinical Research Protocol Canadian Journal of Kidney Health Effect of a Perioperative and Disease Volume 9: 1­–9 © The Author(s) 2022 Hypotension-Avoidance Strategy Versus Article reuse guidelines: sagepub.com/journals-permissions a Hypertension-Avoidance Strategy on DOI: 10.1177/20543581211069225 https://doi.org/10.1177/20543581211069225 journals.sagepub.com/home/cjk the Risk of Acute Kidney Injury: A Clinical Research Protocol for a Substudy of the POISE-3 Randomized Clinical Trial Amit X. Garg1 , Meaghan Cuerden1, Hector Aguado2, Mohammed Amir3, Emilie P. Belley-Cote4, Keyur Bhatt5, Bruce M. Biccard6, Flavia K. Borges7, Matthew Chan8, David Conen7, Emmanuelle Duceppe7, Sergey Efremov9, John Eikelboom4, Edith Fleischmann10, Landoni Giovanni11, Peter Gross4, Raja Jayaram12, Mikhail Kirov13, Ydo Kleinlugtenbelt14, Andrea Kurz15, Andre Lamy7, Kate Leslie16, Valery Likhvantsev17, Vladimir Lomivorotov18, Maura Marcucci4, Maria José Martínez-Zapata19, Michael McGillion4, William McIntyre7, Christian Meyhoff20, Sandra Ofori7, Thomas Painter21, Pilar Paniagua22, Chirag Parikh23, Joel Parlow24, Ameen Patel4, Carisi Polanczyk25, Toby Richards26, Pavel Roshanov27, Denis Schmartz28, Daniel Sessler29, Tim Short30, Jessica M. Sontrop27 , Jessica Spence4, Sadeesh Srinathan31, David Stillo7, Wojciech Szczeklik32, Vikas Tandon4, David Torres33, Thomas Van Helder34, Jessica Vincent7, C. Y. Wang35, Michael Wang4 , Richard Whitlock7, Maria Wittmann36, Denis Xavier37, and P. J. Devereaux4 Abstract Background: Most patients who take antihypertensive medications continue taking them on the morning of surgery and during the perioperative period. However, growing evidence suggests this practice may contribute to perioperative hypotension and a higher risk of complications. This protocol describes an acute kidney injury substudy of the Perioperative Ischemic Evaluation-3 (POISE-3) trial, which is testing the effect of a perioperative hypotension-avoidance strategy versus a hypertension-avoidance strategy in patients undergoing noncardiac surgery. Objective: To conduct a substudy of POISE-3 to determine whether a perioperative hypotension-avoidance strategy reduces the risk of acute kidney injury compared with a hypertension-avoidance strategy. Design: Randomized clinical trial with 1:1 randomization to the intervention (a perioperative hypotension-avoidance strategy) or control (a hypertension-avoidance strategy). Intervention: If the presurgery systolic blood pressure (SBP) is <130 mmHg, all antihypertensive medications are withheld on the morning of surgery. If the SBP is ≥130 mmHg, some medications (but not angiotensin receptor blockers [ACEIs], angiotensin receptor blockers [ARBs], or renin inhibitors) may be continued in a stepwise manner. During surgery, the patients’ mean arterial pressure (MAP) is maintained at ≥80 mmHg. During the first 48 hours after surgery, Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution- NonCommercial 4.0 License (https://creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Canadian Journal of Kidney Health and Disease some antihypertensive medications (but not ACEIs, ARBs, or renin inhibitors) may be restarted in a stepwise manner if the SBP is ≥130 mmHg. Control: Patients receive their usual antihypertensive medications before and after surgery. The patients’ MAP is maintained at ≥60 mmHg from anesthetic induction until the end of surgery. Setting: Recruitment from 108 centers in 22 countries from 2018 to 2021. Patients: Patients (~6800) aged ≥45 years having noncardiac surgery who have or are at risk of atherosclerotic disease and who routinely take antihypertensive medications. Measurements: The primary outcome of the substudy is postoperative acute kidney injury, defined as an increase in serum creatinine concentration of either ≥26.5 μmol/L (≥0.3 mg/dL) within 48 hours of randomization or ≥50% within 7 days of randomization. Methods: The primary analysis (intention-to-treat) will examine the relative risk and 95% confidence interval of acute kidney injury in the intervention versus control group. We will repeat the primary analysis using alternative definitions of acute kidney injury and examine effect modification by preexisting chronic kidney disease, defined as a prerandomization estimated glomerular filtration rate <60 mL/min/1.73 m2. Results: Substudy results will be analyzed in 2022. Limitations: It is not possible to mask patients or providers to the intervention; however, objective measures will be used to assess acute kidney injury. Conclusions: This substudy will provide generalizable estimates of the effect of a perioperative hypotension-avoidance strategy on the risk of acute kidney injury. 1 London Health Sciences Centre, ON, Canada 2 Hospital Clínico Universitario de Valladolid, Spain 3 Shifa International Hospital (STMU), Islamabad, Pakistan 4 McMaster University, Hamilton, ON, Canada 5 SIDS Hospital & Research Centre, Guntur, India 6 Groote Schuur Hospital, Cape Town, South Africa 7 Population Health Research Institute, Hamilton, ON, Canada 8 The Chinese University of Hong Kong, Shatin, Hong Kong 9 Saint Petersburg State University, Russia 10 Medical University of Vienna, Austria 11 Istituto Scientifico Universitario San Raffaele, Milan, Italy 12 University of Oxford, UK 13 Northern State Medical University of the Ministry of Healthcare of the Russian Federation, Arkhangelsk, Russia 14 Deventer Ziekenhuis, The Netherlands 15 Medical University of Graz, Austria 16 The Royal Melbourne Hospital, VIC, Austria 17 I. M. Sechenov Moscow Medical Academy, Russia 18 E. Meshalkin National Medical Research Center and Novosibirsk State University, Novosibirsk, Russia 19 Centro Cochrane Iberoamericano, Barcelona, Spain 20 University of Copenhagen, Denmark 21 Royal Adelaide Hospital, SA, Australia 22 Hospital de la Santa Creu i Sant Pau, Barcelona, Spain 23 Johns Hopkins University, Baltimore, MD, USA 24 Kingston General Hospital, ON, Canada 25 Hospital de Clinicas de Porto Alegre, Brazil 26 The University of Western Australia, Perth, Australia 27 Western University, London, ON, Canada 28 CHU Brugmann, Brussels, Belgium 29 Cleveland Clinic, OH, USA 30 Auckland District Health Board, New Zealand 31 University of Manitoba, Winnipeg, Canada 32 Jagiellonian University, Krakow, Poland 33 Clinica Santa Maria, Santiago, Chile 34 Hamilton Health Sciences, ON, Canada 35 University of Malaya, Kuala Lumpur, Malaysia 36 Universitätsklinikum Bonn, Germany 37 St. John’s National Academy of Health Sciences, Bangalore, India Corresponding Author: Amit X. Garg, London Health Sciences Centre, Westminster Tower, 800 Commissioners Road East, ELL-215, London ON N6A 5W9, Canada. Email: amit.garg@lhsc.on.ca Garg et al 3 Abrégé Contexte : La plupart des patients qui prennent des médicaments antihypertenseurs continuent de les prendre le matin d’une intervention chirurgicale et pendant la période périopératoire. De plus en plus de preuves suggèrent que cette pratique pourrait entraîner l’hypotension périopératoire et augmenter le risque de complications. Ce protocole décrit une sous-étude sur l’insuffisance rénale aiguë (IRA) découlant de l’essai Perioperative Ischemic Evaluation-3 (POISE-3). Cet essai teste l’effet d’une stratégie d’évitement de l’hypotension périopératoire par rapport à une stratégie d’évitement de l’hypertension chez des patients qui subissent une chirurgie non cardiaque. Objectifs : Cette sous-étude de l’essai POISE-3 vise à déterminer si une stratégie d’évitement de l’hypotension périopératoire réduit le risque d’IRA comparativement à la stratégie d’évitement de l’hypertension. Type d’étude : Essai clinique randomisé à répartition 1:1 au groupe intervention (stratégie d’évitement de l’hypotension périopératoire) ou au groupe témoin (stratégie d’évitement de l’hypertension). Groupe intervention : Si la pression artérielle systolique (PAS) avant l’opération est <130 mmHg, tous les médicaments antihypertenseurs sont suspendus le matin de la chirurgie. Si la PAS est ≥130 mmHg, certains médicaments (excluant les inhibiteurs de l’enzyme de conversion de l’angiotensine [IECA], les antagonistes du récepteur de l’angiotensine [ARA] ou les inhibiteurs de la rénine) peuvent être poursuivis de façon graduelle. Pendant la chirurgie, la pression artérielle moyenne (PAM) du patient est maintenue à ≥80 mmHg. Dans les 48 heures suivant l’intervention chirurgicale, certains médicaments antihypertenseurs (excluant les IECA, les ARA ou les inhibiteurs de la rénine) peuvent être réintroduits par étapes si la PAS est ≥130 mmHg. Groupe témoin : Les patients reçoivent leurs médicaments antihypertenseurs habituels avant et après la chirurgie. La PAM du patient est maintenue à ≥60 mmHg de l’induction de l’anesthésie à la fin de l’intervention chirurgicale. Cadre : Recrutement à partir de 108 centres dans 22 pays entre 2018 à 2021. Sujets : Des patients (~6 800) âgés de 45 ans et plus atteints d’athérosclérose, ou présentant un risque de l’être, devant subir une chirurgie non cardiaque et prenant des médicaments antihypertenseurs sur une base régulière. Mesures : Le principal critère d’évaluation de cette sous-étude est une IRA postopératoire définie par une hausse d’au moins 26,5 μmol/L (≥0,3 mg/dL) de la créatinine sérique dans les 48 heures suivant la randomisation ou d’au moins 50 % dans les 7 jours suivant la randomisation. Méthodologie : L’analyse primaire (par intention de traiter) examinera le risque relatif d’une IRA et l’intervalle de confiance à 95 % dans le groupe intervention par rapport au groupe témoin. Nous répéterons l’analyse primaire en utilisant d’autres définitions de l’IRA et nous examinerons la modification de l’effet en présence d’une insuffisance rénale préexistante (définie par un DFGe prérandomisation <60 ml/min/1,73 m2). Résultats : Les résultats de cette sous-étude seront analysés en 2022. Limites : Il n’est pas possible de procéder à l’insu des patients ou des prestataires de soins pour cette intervention; des mesures objectives seront toutefois utilisées pour évaluer l’IRA. Conclusion : Cette sous-étude fournira des estimations généralisables de l’effet d’une stratégie visant à éviter l’hypotension périopératoire sur le risque d’insuffisance rénale aiguë. Keywords acute kidney injury, antihypertensive medication, hypotension, mean arterial pressure, noncardiac surgery Received August 25, 2021. Accepted for publication November 30, 2021. What was known before Introduction Both perioperative hypotension and hypertension are inde- Most patients who take antihypertensive medications con- pendently associated with adverse outcomes; however, tinue taking them on the morning of surgery and during the hypotension is more common and may be more dangerous. perioperative period.1 However, growing evidence suggests that this practice may contribute to perioperative hypoten- sion and a higher risk of complications.1-7 This has led to What this adds debate about optimal perioperative blood pressure targets We are conducting an acute kidney injury substudy of and whether antihypertensive medications should be with- Perioperative Ischemic Evaluation-3 (POISE-3), a multina- held during the perioperative period and how they should be tional randomized clinical trial of patients on antihyperten- reintroduced postoperatively.1,8 The effect of a perioperative sive medications having noncardiac surgery; the trial is hypotension-avoidance strategy versus a hypertension- testing the effect of a perioperative hypotension-avoidance avoidance strategy on cardiovascular outcomes is currently strategy versus a hypertension-avoidance strategy. being tested in the Perioperative Ischemic Evaluation-3 4 Canadian Journal of Kidney Health and Disease (POISE-3) trial, a partial 2 × 2 factorial trial (factor 1 is test- began in June 2018 and will continue until December 2021. ing the effect of tranexamic acid vs placebo). The trial started POISE-3 is designed as a partial 2 × 2 factorial trial. In fac- recruitment in June 2018 and is expected to enroll ~7000 tor 1, the tranexamic acid factorial, 10 000 patients ≥45 patients aged ≥45 years undergoing noncardiac surgery with years having noncardiac surgery with an overnight hospital an overnight hospital stay who have or are at risk of athero- stay who have or are at risk of atherosclerotic disease (the sclerotic disease and who routinely take antihypertensive full list of inclusion criteria is provided in Appendix I in the medications. Supplement) will be randomized to receive intravenous tranexamic acid or placebo during surgery. In factor 2, the blood pressure management factorial, an estimated 7000 A Kidney Substudy of POISE-3 patients who are routinely taking at least 1 antihypertensive Approximately 10% of patients who undergo noncardiac sur- drug will also be randomized to receive a perioperative gery will develop perioperative acute kidney injury, with hypotension-avoidance strategy or a hypertension-avoidance 0.5% receiving dialysis as a result.6,9,10 Acute kidney injury strategy. The acute kidney injury substudy will analyze is associated with longer hospital stays, increased health care patients enrolled in factor 2. costs, and death.10,11 Perioperative hypotension can cause ischemia-reperfusion injury to the kidney.12 We, and others, Patient Recruitment and Informed Consent have shown that the risk for acute kidney injury increases progressively with the severity and duration of perioperative Study personnel screen the patient list in the preoperative hypotension.2,3,6,13 In a recent trial of 678 elderly hyperten- assessment clinic to identify eligible patients. A variety of sive patients having gastrointestinal surgery, patients ran- screening approaches are used to identify patients who do domized to an intraoperative mean arterial pressure (MAP) not attend the preoperative assessment clinic, including of 80 to 95 mmHg had less acute kidney injury than those screening the daily surgical list in the operating room and randomized to a lower or higher target MAP (65-79 and reviewing patient lists on surgical wards and intensive care 96-110 mmHg, respectively)—the incidence of acute kidney units. At each center, the services of anesthesia, surgery, and injury in these groups was 6.3%, 13.5%, and 12.9%, respec- medicine are asked to page study personnel to notify them of tively (P = .03).9 Whether these findings generalize to a admissions through the emergency department and ward broader patient group undergoing other types of noncardiac patients requiring surgery. Study personnel approach all eli- surgery remains unknown.9 gible patients to obtain written informed consent before sur- This protocol describes a planned kidney substudy of gery. Where possible, research personnel inform patients POISE-3 to examine the effect of a perioperative hypoten- about the trial in preoperative clinics. Patients who are not sion-avoidance strategy on the risk of acute kidney injury identified or approached in the preoperative clinic are con- compared with a hypertension-avoidance strategy. We will tacted before surgery to discuss the trial and their potential also examine whether the intervention has a larger absolute participation. benefit in patients with preexisting chronic kidney disease, which is the most prominent risk factor for acute kidney Randomized Group Assignment injury.14 Randomization occurs on the day of surgery (before sur- gery) after a patient has been deemed eligible and has pro- Main Questions in the POISE-3 Kidney Substudy vided written informed consent. To ensure that the randomization sequence is concealed from participating Primary question. Does a perioperative hypotension-avoid- centers and patients, randomization is generated centrally ance strategy reduce the risk of acute kidney injury compared online via an Interactive Web Randomization System (main- with a hypertension-avoidance strategy in patients having tained by the POISE-3 Project Office at the Population noncardiac surgery who routinely take antihypertensive Health Research Institute in Hamilton, ON, Canada). The medication and who have or are at risk of atherosclerotic randomization process uses variable-block randomization, disease? stratified by center. Study personnel who randomize patients do not know the block size. In this partial factorial trial, all Secondary question. Does the presence of preexisting chronic patients are first randomized 1:1 to the tranexamic acid kidney disease modify the effect of a perioperative hypoten- intervention or placebo, which is factor 1. Those who are sion-avoidance strategy on the risk of acute kidney injury? eligible for factor 2 and consent to participate (conserva- tively estimated to be 65% of 10 000, described in the sam- Method ple size and statistical power section below) are randomized 1:1 to a hypotension-avoidance strategy or a hypertension- Overview of the POISE-3 Main Trial avoidance strategy. Patients and providers are notified of the POISE-3 (NCT03505723) is a randomized clinical trial factor 2 assignment after it is randomly assigned to a given being conducted in 108 centers in 22 countries. Enrollment patient. Garg et al 5 Intervention surgery with at least one expected night in hospital after sur- gery, those aged ≥45 years, and those who have or are at risk All patients who are eligible to participate in factor 2, the of atherosclerotic disease. To be eligible to participate in fac- blood pressure management factorial, are instructed to not tor 2 (the blood pressure management factorial), patients take their antihypertensive medication the night before and/ need also to be treated for ≥30 days in the 6 weeks before or on the morning of surgery (before coming to the hospital) randomization with at least 1 antihypertensive medication of and to bring their antihypertensive medications to the hospi- any class (ie, ACEIs, ARBs, renin inhibitors, β-blockers, cal- tal. Patients who do not have their antihypertensive medica- cium channel blockers, central α2-agonists, α-blockers, tions with them on the day of surgery, or who already took direct vasodilators, long-acting nitrates, thiazide diuretics, their antihypertensive medications on the day of surgery, are loop diuretics, or potassium-sparing diuretics). still eligible for randomization to this factorial. They con- Because tranexamic acid is mainly eliminated through the tinue to receive the intraoperative and postoperative compo- kidney and could be unsafe in patients with very low kidney nents of the trial as described below. function, patients with a creatinine clearance <30 mL/min (Cockcroft-Gault formula) or on chronic dialysis are Hypotension-avoidance strategy (intervention). On the morning excluded from POISE-3. We expect 15% to 20% of patients of surgery, all antihypertensive medications are withheld if enrolled in POISE-3 will have chronic kidney disease defined the presurgery systolic blood pressure (SBP) is <130 mmHg. by a prerandomization estimated glomerular filtration rate If the SBP is ≥130 mmHg, some medications may be contin- (eGFR) <60 mL/min per 1.73 m2 (975-1300 patients)18 as ued in a stepwise manner (see Appendix II in the Supple- assessed by the CKD-EPI equation.19 ment); however, no patient will take an angiotensin-converting In addition, we will exclude patients missing a preran- enzyme inhibitor (ACEI), an angiotensin receptor blocker domization serum creatinine because these data are needed (ARB), or a renin inhibitor, alone or in combination, on the to define acute kidney injury. Less than 2% of participants day of surgery. are expected to be excluded for this reason.6 Intraoperatively, the attending anesthetist is asked to maintain the patient’s MAP to ≥80 mmHg, from anesthetic Substudy data collection. The most recent test result for induction until the end of surgery. Methods to achieve this serum creatinine before randomization was the baseline target (fluids, vasopressors, inotropes, etc) are left to the dis- value. These data were obtained from a review of medical cretion of the anesthetist. records. The date of randomization is used to identify the During the first 48 hours after surgery, some antihyperten- start of follow-up in this substudy. We expect >95% of sive medications may be restarted (in a stepwise manner) if enrolled patients to have surgery within 3 days of random- the SBP is ≥130 mmHg (except no ACEIs, ARBs, or renin ization. In our substudy of POISE-2, the median time from inhibitors). Starting on postoperative day 3, the management randomization to surgery was 2 hours (25th, 75th percentile: of blood pressure and antihypertensive medication is left to 1-3 hours) and we expect a similar distribution in POISE-3.6 the discretion of the treating physician. All centers receive substudy funds to measure and record a daily serum creatinine value on postoperative days 1, 2, and Hypertension-avoidance strategy (control). Patients receive 3 (or until hospital discharge) in all randomized patients. their usual antihypertensive medications before and after This schedule is expected to minimize any ascertainment surgery. Intraoperatively, the anesthetist is asked to maintain bias related to the assessment of acute kidney injury (ie, if the patient’s MAP to ≥60 mmHg, from anesthetic induction the intervention alters the incidence of another event such as until the end of surgery. myocardial infarction, this could influence the likelihood of This strategy mirrors what typically occurs in routine care serum creatinine measurement). Centers are also asked to as documented in the literature.1,8,15-17 In the VISION study record all other serum creatinine measurements performed (Vascular Events In Noncardiac Surgery Patients Cohort as a part of routine care (and their dates) during the hospital Evaluation), more than 70% of patients on chronic antihy- stay. No urine output data are used in this substudy given pertensive medications continued these medications in the difficulties with accurately measuring this variable outside 24 hours before noncardiac surgery.1 Although there is no of the intensive care unit. Receipt of new dialysis for kidney standard, the existing literature documents that usual care failure is recorded at hospital discharge and 30 days after typically consists of maintaining the intraoperative MAP randomization. above 60 mmHg.15,16 This data collection schedule is informed by our experi- ence collecting kidney function data in our substudy of POISE-2 and we expect to capture most events of acute kid- Methods Used in the POISE-3 Kidney Substudy ney injury.6,18 In a prior study where we collected daily serum Patient selection. Eligibility criteria for the POISE-3 main creatinine values in the first 5 days after surgery, we noted trial are fully detailed in Appendix I in the Supplement. that 93% of postoperative acute kidney injury events occurred Briefly, eligible patients include those having noncardiac in the first 3 days after surgery.20 At the time of the final 6 Canadian Journal of Kidney Health and Disease analysis, we will compare the characteristics of patients who serum creatinine) / prerandomization serum creatinine did and did not provide at least one serum creatinine mea- × 100. surement during the week following randomization and will 7. Absolute change in serum creatinine in the first 7 examine the number of measurements by treatment group days of randomization, defined as peak postrandom- (and measurement dates) to confirm there is no differential ization serum creatinine − prerandomization serum outcome ascertainment between groups. We expect <1% of creatinine. patients will die in the operating room or within 48 hours after surgery, which may result in no serum creatinine mea- Sample size and statistical power. The main POISE-3 trial is surement after randomization.6 In the POISE-2 substudy,6 planning to enroll 10 000 patients in factor 1 (the tranexamic more than 95% of patients had at least 1 serum creatinine acid factorial) and it is expected that at least 80% of these measurement within the 7 days following surgery and the patients will be taking preoperative maintenance antihyper- median number of postoperative serum creatinine measure- tensive medications (based on estimates from the POISE-2 ments was identical in the intervention and placebo groups (3 trial).5 As a conservative estimate, at least 70% of the measurements; interquartile range [IQR] = 2-4) along with patients in factor 1 are expected to be eligible for and con- the specific postoperative days of these measurements. The sent to participate in factor 2 (the blood pressure manage- median duration of hospital stay was identical in each group ment factorial) and we expect that 98% of these 7000 (4 days; IQR = 3-7 days). patients will be eligible for inclusion in the kidney substudy, providing an expected sample size of at least 6800 patients Substudy outcomes. The primary outcome of the POISE-3 in the substudy. kidney substudy is postoperative acute kidney injury, defined For the kidney substudy, a sample of 6800 patients will as an increase in the postrandomization serum creatinine provide >80% power to detect a 20% relative risk (RR) concentration (from the prerandomization value) of ≥26.5 reduction for the primary outcome of postoperative acute μmol/L (≥0.3 mg/dL) within 48 hours of randomization or kidney injury (2-sided α = 0.05), assuming an incidence of an increase of ≥50% within 7 days of randomization.21 acute kidney injury of 10% in the control group. Secondary definitions of acute kidney injury. Seven secondary Statistical analysis plan. In the primary analysis (intention-to- definitions of acute kidney injury will be examined to assess treat), a modified Poisson regression model (accounting for whether the primary results are robust: center) will be used to estimate the RR and 95% confidence interval (CI) for postoperative acute kidney injury compar- 1. A composite outcome of acute kidney injury (pri- ing the intervention group with the control group.22,23 Miss- mary definition) or death within 48 hours of random- ing data on postoperative serum creatinine (expected for ization (to account for the potential impact of early <5% of surviving patients)6 will be imputed using multiple deaths on outcome ascertainment). imputation; parameters will be estimated using standard 2. Acute kidney injury (primary definition) for at least 2 methods while allowing for extra imputation variability.24 A days within 7 days of randomization. 2-tailed P value of ≤.05 will be considered statistically 3. Stage 2 acute kidney injury (or higher), defined as a significant. postrandomization increase in serum creatinine of ≥100% from the prerandomization value within 7 Prespecified supporting analyses. As outlined below, we will days of randomization, or an increase to an absolute conduct several supporting analyses and examine whether value of 353.6 μmol/L or more (≥4.0 mg/dL) within the results are concordant with the results of the primary 7 days of randomization (when the primary outcome analysis. We will also examine 7 secondary definitions of definition of acute kidney injury is met), or receipt of acute kidney injury and conduct a subgroup analysis of dialysis within 30 days of randomization. patients with preexisting chronic kidney disease. 4. Stage 3 acute kidney injury, defined as a postrandom- ization increase in serum creatinine of ≥200% from Complete-case analysis. We will perform a complete-case the prerandomization value within 7 days of random- analysis restricted to patients with at least one postopera- ization, or an increase to an absolute value of 4.0 mg/ tive serum creatinine measurement (expected to be ~95% of dL or more (≥353.6 μmol/L) within 7 days of ran- patients in the primary analysis). If there are centers with domization (when the primary outcome definition of <90% complete serum creatinine measurements (or periods acute kidney injury is met), or receipt of dialysis in some centers before initiating the kidney substudy), we within 30 days of randomization. will analyze only those centers (or periods of time in those 5. Receipt of dialysis within 30 days of randomization. centers) with over 90% completed measurements. 6. Percentage change in serum creatinine in the first 7 days of randomization, defined as (peak postran- Adjusted analyses. In our experience with previous kidney domization serum creatinine – prerandomization substudies, the unadjusted and adjusted results were virtually Garg et al 7 identical;18,25 nonetheless, to potentially increase the preci- a modified Poisson regression. We will also examine the sion of our estimates,26 we will use a generalized estimating between-group difference in adherence. As well, because equation approach for binary outcome data, accounting for patients who undergo urgent or emergent surgery may be within-center correlation, adjusting for 9 covariates (mea- more likely to develop acute kidney injury before random- sured before randomization) based on their known associa- ization, we will repeat the primary analyses after excluding tion with acute kidney injury: age (in years, modeled with these categories of surgery. An additional subgroup analysis restricted cubic splines), sex, cardiovascular disease (any will examine the treatment effect in patients who are taking coronary artery disease, peripheral vascular disease, or versus not taking an ACEI or ARB before randomization; we stroke), diabetes, prerandomization eGFR (as a continuous hypothesize that the hypotension-avoidance strategy may variable modeled with restricted cubic splines), a history of confer a greater benefit (both in relative and absolute terms) smoking within 2 years of surgery, urgent or emergency sur- in preventing acute kidney injury in patients taking versus gery (~6%-7% of POISE-2 participants),6 and type of sur- not taking an ACEI or ARB before randomization. Finally, gery (major vascular surgery, major thoracic surgery, or other we expect <5% of randomized patients to have delayed sur- surgery). We will also adjust for the random allocation of gery (>3 days after randomization) or not undergo surgery tranexamic acid versus placebo. Adjusted RRs and 95% CIs at all. We will examine concordance with the primary inten- will be reported. We expect missing data on these variables tion-to-treat analysis when the date of surgery (rather than to be <0.5%.6 the date of randomization) is used as the start of follow-up, excluding those who did not undergo surgery. Alternative definitions of acute kidney injury. We will exam- ine 7 secondary definitions of acute kidney injury (5 categor- Discussion ical and 2 continuous, as described above). Binary outcomes will be assessed using modified Poisson regression models, This protocol describes our planned acute kidney injury sub- and continuous outcomes using linear regression models. We study and statistical analysis plan of the POISE-3 random- will visually inspect the point estimates and 95% CIs and ized clinical trial, which is testing the effect of a perioperative assess concordance with the primary analysis. Despite the hypotension-avoidance strategy versus a hypertension- large sample size, supplementary analyses of severe acute avoidance strategy on patients ≥45 years having noncardiac kidney injury will have limited statistical power for small surgery who have or are at risk of atherosclerotic disease, effects. and who routinely take antihypertensive medications. Although both perioperative hypotension and hyperten- Subgroup analysis. Compared with patients with preserved sion are independently associated with adverse outcomes,29 kidney function, patients with an eGFR below 60 mL/min hypotension is more common, occurring in approximately per 1.73 m2 have a higher risk of acute kidney injury and may 40% of noncardiac surgeries.2,3,13,30 Intraoperative hypoten- be more vulnerable to the ischemic effects of hypotension.14 sion may also be more dangerous.7 In the last decade, both We hypothesize that the hypotension-avoidance intervention observational1-3,7,13,31-33 and experimental studies9,34 demon- may confer a larger absolute benefit to patients with preexist- strated that perioperative hypotension is an independent pre- ing chronic kidney disease, defined by a prerandomization dictor of vascular events and death. This association is seen eGFR <60 mL/min per 1.73 m2 as assessed with the 2021 for MAPs <50 to 55 mmHg (even for less than a minute), for CKD-EPI equation.27,28 To examine the presence of additive MAPs 65 to 80 mmHg (on average and/or over time), and for interaction, we will calculate the absolute difference in risk percentage drops in MAP from baseline (eg, 20% or greater). (and 95% CI) between the intervention and control groups These data suggest that the optimal perioperative blood pres- in patients with and without preexisting chronic kidney dis- sure target may be higher than previously thought, particu- ease; a P value for the interaction term will be assessed in a larly in patients with chronic hypertension, who may have regression model for binary outcome data. We will also con- less autoregulatory capacity and may be more susceptible to duct a test for multiplicative interaction between eGFR and ischemia-reperfusion injury from hypotension.29,34,35 the intervention group, where eGFR is a continuous variable By adding additional serum creatinine measurements to modeled with restricted cubic splines to allow for nonlinear- POISE-3, we can efficiently and reliably determine whether ity. We do not expect prerandomization eGFR will modify a perioperative hypotension-avoidance strategy reduces the the RR of acute kidney injury with a strategy of hypotension- risk of acute kidney injury compared with a hypertension- avoidance compared with hypertension-avoidance. avoidance strategy. Our substudy has some limitations. It will not be possible to mask patients or providers to the Additional analyses. We anticipate that the treatment intervention; however, any resulting ascertainment bias effects of tranexamic acid and the hypotension-avoidance should be minimal as patients in both groups have the same strategy will act independently, but will explore this by prespecified schedule for serum creatinine measurements. reporting cell-by-cell results (ie, for each of the 4 treatment Furthermore, as shown in POISE-2, we anticipate no differ- groups) and will test for multiplicative interaction using ence in the number of postoperative serum creatinine 8 Canadian Journal of Kidney Health and Disease measures between treatment arms.6,21 The primary outcome Consent for Publication of this kidney substudy is postoperative acute kidney injury, Consent for publication was obtained from all authors. defined as an acute rise in serum creatinine concentration from the prerandomization value.36 While many prevention Availability of Data and Materials trials of acute kidney injury follow this definition, this out- Not available. come is a surrogate endpoint that may not directly impact how a patient feels, functions, or survives. Unfortunately, Declaration of Conflicting Interests detailed information on long-term permanent kidney func- tion after hospital discharge will not available for most The author(s) declared no potential conflicts of interest with patients enrolled in POISE-3. However, even small decreases respect to the research, authorship, and/or publication of this article. in kidney function after surgery are associated with poor short- and long-term outcomes10,11 and our definition of Funding acute kidney injury follows international clinical practice guidelines and is a defensible approach for our primary The author(s) disclosed receipt of the following financial support outcome.36 We will also examine whether intervention for the research, authorship, and/or publication of this article: The effects are consistent for stage 2 and 3 acute kidney injury; Canadian Institutes of Health Research and the Australian National Health and Medical Research Council (NHMRC) provided an despite being less frequent, these events are more important operating grant for the main POISE-3 trial. The Kidney Foundation to patients and health care providers. Our definition of acute of Canada provided an operating grant for the POISE-3 kidney kidney injury does not include oliguria given the difficulty of substudy and additional financial support was provided by the accurately measuring urine output in an international setting; Department of Medicine at Western University. In addition, the however, any resulting outcome misclassification would be authors disclosed receipt of the following financial support for the expected to be nondifferential between groups. Finally, the research, authorship, and/or publication of this article: F.K.B. prerandomization measure of serum creatinine will be the holds a McMaster University Department of Medicine Career most recently obtained measure before randomization and Research Award; M.C. holds grants from the Australian NHMRC we expect it to be obtained within 1 week before surgery for (APP1162362) and the Research Grants Council of Hong Kong, most patients. Depending on why the patient needs surgery, a General Research Fund (14104419); P.J.D. was supported by a serum creatinine test obtained at the time of hospital admis- Tier 1 Canada Research Chair in Perioperative Medicine; A.X.G. was supported by the Dr Adam Linton Chair in Kidney Health sion may be unstable or elevated. Although such error may Analytics and a Clinician Investigator Salary Award from the impede the detection of an acute rise in postrandomization Canadian Institutes of Health Research; M.J.M.Z. is funded by a serum creatinine, which is needed for the identification of Miguel Servet II research contract from the ISCIII (CP1120/00023), acute kidney injury, we expect this error to be similar between Spain; T.R. is supported by the Medical and Health Research randomized groups. We will report the prerandomization Infrastructure Fund; and P.R. was supported by the Western serum creatinine concentration for each group and examine University Resident Research Fellowship Program. No funding if it differs for patients whose baseline measurement was entity had a role in data collection, statistical analysis, manuscript obtained at the time of hospital admission versus earlier. We writing, or the decision to publish. will also conduct an adjusted analysis that controls for patients’ prerandomization eGFR and receipt of urgent or ORCID iDs emergency surgery. Amit X. Garg https://orcid.org/0000-0003-3398-3114 Jessica M. Sontrop https://orcid.org/0000-0001-7784-2028 Conclusions Michael Wang https://orcid.org/0000-0002-4949-3235 Our substudy of POISE-3 will determine whether a periop- Supplemental Material erative hypotension-avoidance strategy reduces the risk of Supplemental material for this article is available online. acute kidney injury compared with a hypertension-avoidance strategy. We will also conduct a subgroup analysis by preop- References erative chronic kidney disease, which is a strong risk factor for acute kidney injury.14 Conducting this study with a sam- 1. Roshanov PS, Rochwerg B, Patel A, et al. Withholding ver- ple size of 6800 patients in 108 centers in 22 countries with sus continuing angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers before noncardiac surgery. the use of randomized trial methodology will help generate Anesthesiology. 2017;126:16-27. results that are accurate, precise, and broadly generalizable. 2. Sun LY, Wijeysundera DN, Tait GA, Beattie WS. Association of intraoperative hypotension with acute kidney injury after Ethics Approval and Consent to Participate elective noncardiac surgery. Anesthesiology. 2015;123(3): An appropriately authorized ethics committee approved the trial in 515-523. all the participating centers. Written informed consent was obtained 3. Walsh M, Devereaux PJ, Garg AX, et al. Relationship between from all the participants before enrollment. intraoperative mean arterial pressure and clinical outcomes Garg et al 9 after noncardiac surgery: toward an empirical definition of 19. Levey AS, Stevens LA, Schmid CH, et al. A new equa- hypotension. Anesthesiology. 2013;119(3):507-515. tion to estimate glomerular filtration rate. Ann Intern Med. 4. Devereaux PJ, Yang H, Yusuf S, et al; POISE Study Group. 2009;150:604-612. Effects of extended-release metoprolol succinate in patients 20. Koyner JL, Garg AX, Coca SG, et al. Biomarkers predict pro- undergoing non-cardiac surgery (POISE trial): a randomised gression of acute kidney injury after cardiac surgery. J Am Soc controlled trial. Lancet. 2008;371:1839-1847. Nephrol. 2012;23:905-914. 5. Devereaux PJ, Sessler DI, Leslie K, et al. Clonidine in patients 21. Garg AX, Devereaux PJ, Yusuf S, et al. Kidney function after undergoing noncardiac surgery. N Engl J Med. 2014;370: off-pump or on-pump coronary artery bypass graft surgery: a 1504-1513. randomized clinical trial. JAMA. 2014;311:2191-2198. 6. Garg AX, Kurz A, Sessler DI, et al. Perioperative aspirin and 22. Zou GY, Donner A. Extension of the modified Poisson regres- clonidine and risk of acute kidney injury: a randomized clinical sion model to prospective studies with correlated binary data. trial. JAMA. 2014;312:2254-2264. Stat Methods Med Res. 2013;22(6):661-670. 7. Monk TG, Bronsert MR, Henderson WG, et al. Association 23. Zou G. A modified Poisson regression approach to pro- between intraoperative hypotension and hypertension and spective studies with binary data. Am J Epidemiol. 2004;159: 30-day postoperative mortality in noncardiac surgery. Anes­ 702-706. thesiology. 2015;123:307-319. 24. Van Buuren S, Brand JPL, Groothuis-Oudshoorn CGM, et al. 8. Hollmann C, Fernandes NL, Biccard BM. A systematic Fully conditional specification in multivariate imputation. J review of outcomes associated with withholding or continu- Stat Comput Simul. 2006;76:1049-1064. ing angiotensin-converting enzyme inhibitors and angiotensin 25. Garg AX, Vincent J, Cuerden M, et al. Steroids In caRdiac receptor blockers before noncardiac surgery. Anesth Analg. Surgery (SIRS) trial: acute kidney injury substudy protocol 2018;127:678-687. of an international randomised controlled trial. BMJ Open. 9. Wu X, Jiang Z, Ying J, Han Y, Chen Z. Optimal blood pressure 2014;4:e004842. decreases acute kidney injury after gastrointestinal surgery in 26. Zhou X, Zhang Y, Teng Y, et al. Predictors of postoperative elderly hypertensive patients: a randomized study: optimal acute kidney injury in patients undergoing hip fracture sur- blood pressure reduces acute kidney injury. J Clin Anesth. gery: a systematic review and meta-analysis. Injury. 2021;52: 2017;43:77-83. 330-338. doi:10.1016/j.injury.2020.09.060. 10. Grams ME, Sang Y, Coresh J, et al. Acute kidney injury after 27. National Kidney Foundation. CKD-EPI creatinine equation; major surgery: a retrospective analysis of Veterans Health 2021. https://www.kidney.org/professionals/kdoqi/gfr_calcu- Administration Data. Am J Kidney Dis. 2016;67(6):872-880. lator/formula. Accessed November 1, 2021. 11. Biteker M, Dayan A, Tekkeşin Aİ, et al. Incidence, risk factors, 28. Delgado C, Baweja M, Crews DC, et al. A unifying approach and outcomes of perioperative acute kidney injury in noncar- for GFR estimation: recommendations of the NKF-ASN Task diac and nonvascular surgery. Am J Surg. 2014;207(1):53-59. Force on reassessing the inclusion of race in diagnosing kidney 12. Devarajan P. Update on mechanisms of ischemic acute kidney disease [published online ahead of print September 23, 2021]. injury. J Am Soc Nephrol. 2006;17(6):1503-1520. Am J Kidney Dis. doi:10.1053/J.AJKD.2021.08.003. 13. Salmasi V, Maheshwari K, Yang D, et al. Relationship 29. Palmer BF. Renal dysfunction complicating the treatment of between intraoperative hypotension, defined by either reduc- hypertension. N Engl J Med. 2002;347:1256-1261. tion from baseline or absolute thresholds, and acute kidney and 30. Mathis MR, Kheterpal S. Under pressure: evaluating the role myocardial injury after noncardiac surgery. Anesthesiology. of intraoperative hemodynamics in postoperative acute kidney 2017;126(1):47-65. injury. J Clin Anesth. 2018;44:114-115. 14. Chawla LS, Eggers PW, Star RA, et al. Acute kidney injury 31. Hallqvist L, Granath F, Huldt E, et al. Intraoperative hypoten- and chronic kidney disease as interconnected syndromes. sion is associated with acute kidney injury in noncardiac sur- N Engl J Med. 2014;371:58-66. gery. Eur J Anaesthesiol. 2018;35:273-279. 15. Mayer J, Boldt J, Mengistu AM, Röhm KD, Suttner S. Goal- 32. Sessler DI, Meyhoff CS, Zimmerman NM, et al. Period- directed intraoperative therapy based on autocalibrated arterial dependent associations between hypotension during and for pressure waveform analysis reduces hospital stay in high-risk four days after noncardiac surgery and a composite of myo- surgical patients: a randomized, controlled trial. Crit Care. cardial infarction and death. Anesthesiology. 2018;128(2): 2010;14(1):R18. 317-327. 16. Salzwedel C, Puig J, Carstens A, et al. Perioperative goal- 33. Mascha EJ, Yang D, Weiss S, Sessler DI. Intraoperative mean directed hemodynamic therapy based on radial arterial pulse arterial pressure variability and 30-day mortality in patients pressure variation and continuous cardiac index trending having noncardiac surgery. Anesthesiology. 2015;123(1): reduces postoperative complications after major abdominal 79-91. surgery: a multi-center, prospective, randomized study. Crit 34. Futier E, Lefrant J-Y, Guinot P-G, et al. Effect of individu- Care. 2013;17:R191. alized vs standard blood pressure management strategies on 17. Ling Q, Gu Y, Chen J, et al. Consequences of continuing postoperative organ dysfunction among high-risk patients renin angiotensin aldosterone system antagonists in the pre- undergoing major surgery. JAMA. 2017;318:1346-1357. operative period: a systematic review and meta-analysis. BMC 35. Aronson S, Mythen MG. Perioperative management of high- Anesthesiol. 2018;18:26. risk patients: going beyond “avoid hypoxia and hypotension.” 18. Garg AX, Kurz A, Sessler DI, et al. Aspirin and clonidine in JAMA. 2017;318:1330-1332. non-cardiac surgery: acute kidney injury substudy protocol of 36. Kidney Disease: Improving Global Outcomes (KDIGO) Acute the Perioperative Ischaemic Evaluation (POISE) 2 randomised Kidney Injury Work Group. KDIGO clinical practice guideline controlled trial. BMJ Open. 2014;4:e004886. for acute kidney injury. Kidney Int Suppl. 2012;2:1-138.