By Michael Keane, BMBS, FANZCA and Shashikanth Manikappa, MD, DNB, FANZCA
February 21, 2019
Patient safety requires a well-informed debate about the rationale of large, randomized controlled trials (RCTs) and whether, by their very nature, they can adequately consider the unique physiology of every individual patient.
RCTs must be reproducible; the fundamental premise of any scientific experiment. In their day-to-day practice, clinicians must, therefore, be willing to reproduce the same treatment of individual patients’ physiology. If clinicians would treat individual patients (including physiological outliers) differently to that in which individual patients were treated in a trial, then that trial data becomes meaningless to their practice.
The anesthesiology community needs to review its past and future trials in this context. Arguably, the pre-eminent perioperative medicine trial within the last year was the REstrictive versus LIberal Fluid (RELIEF) study.1 Indeed, the RELIEF study was analyzed for Harvard CME’s blog Trends in Medicine.2 With such a position of prestige, RELIEF requires a high level of scrutiny and critique, both to challenge take-home messages of this trial and to debate the general principles of how we deal with individual physiology in large RCTs.
Although there was no difference in disability-free survival between the groups in RELIEF, creatinine levels were doubled in an increased number of patients in the fluid-restricted group. From the point of view of an experiment, there is a major difference between utilizing a restrictive fluid paradigm, but being willing to treat patients who become overtly hypovolaemic, versus an across-the-board use of a restrictive fluid regimen regardless of individual patient circumstance.
It is in this context, that the post-operative treatment of patients who became hypotensive and/or hypovolemic in the fluid-restricted group of RELIEF, which does not represent contemporary practice, is relevant to the observed high rates of increase in creatinine. Fluid-restricted patients, even if definitely diagnosed as hypovolemic, could only be resuscitated if “evidence of hypovolaemia,” “AND” hypotension (SBP<90mmHg) (emphasis in the original) were present. Specifically, patients who were “normotensive but monitoring suggest(ed) hypovolaemia” received no fluid bolus and no vasopressors. This included those with ‘normal’ BP but significant tachycardia, possibly with high fever, or any other clinical signs that led to the diagnosis of hypovolemia.
Furthermore, fluid-restricted patients who were hypotensive but not hypovolemic received no fluid bolus, and vasopressors were only used “if in ICU/HDU.” The accuracy of this diagnosis (hypotensive but normovolaemic), made by potentially junior residents working overnight on a busy general ward, was not established.
Most importantly, management of hypotension in both ward and ICU patients was based on the directive that: “The lower limit of acceptable systolic blood pressure in the restrictive group can be further reduced by the attending anaesthetist or intensivist in order to limit fluid replacement or potentially UNNECESSARY inotropic support” (emphasis added). This directive to just “lower the limit” of acceptable blood pressure was specifically for the restrictive group. Given that fluid-restricted patients who were not hypotensive, even if severely hypovolemic, received no treatment, what systolic pressure renders intervention “unnecessary”?—85 mmHg? 80 mmHg? Diastolic pressure was not discussed.
Thus, the protocolized treatment for restrictive-fluid patients suffering from hypotension was initially to consider to “just lower the limit you will accept.” Is that contemporary care? Indeed, one of the Anesthesia Patient Safety Foundation’s Perioperative Patient Safety Priorities is “Preventing, detecting, and mitigating clinical deterioration in the perioperative period.”3
A trend towards increased sepsis [Adjusted P=0.065] in the restrictive group potentially compounded the conservative treatment of hypotension/hypovolemia, along with unknown use of angiotensin receptor blockers/angiotensin converting enzyme inhibitors (ARBs/ACEi). Pre-operatively, 40% in both groups were listed as being on ARBs/ACEi.
Treatment of established hypotension/hypovolemia in the fluid-restricted group is relevant to increased creatinine. However, point estimates of the outcomes of renal replacement therapy (13 versus 4) and pulmonary oedema (20 vs 30, worse in liberal-group) were both nonsignificant with Bonferroni correction. Over-resuscitation is, however, well documented to cause pulmonary complications, and the mechanics are well understood. Therefore, conclusions that a liberal regimen is “safer,”4 would require analysis of pulmonary edema with an adequately powered, non-inferiority test; in addition to adequate treatment of hypotension/hypovolemia in restrictive-group.
These facts raise a significant question: Unless postoperative hypotension/hypovolemia is managed according to the trial protocol, should the results of RELIEF be used to guide fluid management? What did the trial (the experiment) really show? From a scientific perspective, unless a clinician affirms that they would treat hypotension and hypovolaemia in that way, then the experiment is not relevant and should not be used to guide fluid management.
In the context of this discussion, physicians treat an individual patient’s physiology, not group means.
Interpretation of RELIEF in relation to the care of individual older, sicker patients is complicated because, although evenly distributed between groups, RELIEF did not justify the clinically-implausible homogenization of ASA I/II patients, who have a low burden of comorbid disease, (38% of participants) with ASA III/IV patients (62%), who have severe systemic disease and are at greater risk of mortality. Is a 70-year-old ASA 1 patient who runs regularly, having an incisional hernia repair, the same, physiologically, as an 85-year-old ASA 3 patient with a poor ejection fraction, significant coronary artery disease, and multiple co-morbidities, after an open AAA repair? Might the tolerance of both ultra-conservative and increased fluid administration be tolerated differently?
While RCTs can be extraordinarily valuable in elucidating difficult-to-isolate fundamental biologic and clinical relationships, many RCTs are conducted on ever-changing complex systems in which, inevitably, there is a balancing of opposing phenomena and which involve constant real-time adjustments.5
In reality, the extraordinary multitude of ways of managing fluid for individual patients with individual circumstances will be concurrently tested by incremental changes from innovators around the world in inordinate combinations.6 Importantly, science will inform us of the indicators of beneficial adjustments of individual patient physiology.
In this context, one must question whether the proposed PeriOperative ISchemic Evaluation-3 (POISE-3) trial7 should be conducted. POISE-3 will study tranexamic acid as well as comparing a: “Perioperative hypotension-avoidance strategy” versus a “Perioperative hypertension-avoidance strategy (i.e., routine care) [which] continues all antihypertensive drugs before and after surgery and an intraoperative BP strategy targeting a MAP ≥60 mm Hg.”
Is a MAP of ≥60 mm Hg routine care? On what basis? Most anesthesiologists would treat an 80-year old patient and not allow the blood pressure to remain 76/52 mmHg (which equals MAP≥60 mm Hg) for three hours. Pre-operative ARBs/ACEi, plus calcium-channel blockers combined with anesthesia, intra-operative epidural or intravenous clonidine or dexmedetomidine, could readily produce such significant hypotension unless treated. Furthermore, does routine care involve giving all antihypertensives postoperatively irrespective of pathophysiology, including significant hypovolemia/tachycardia, SIRS/sepsis, anuria/doubling-creatinine, or very low diastolic BP, on a general ward?
In relation to POISE-3, comparing the benefits of hypertension-avoidance versus hypotension-avoidance strategies, within an already prohibitively complex system, is far too vague a question to be answered by an RCT. 8,9,6 Rather than two extremes, could elements of both arms of care ultimately be utilized? If, in real-world clinical practice, elements of both arms are selectively utilized in an individualized approach to patients, then the experiment is not reproducible for clinical practice, and therefore whole study immediately loses any meaning and becomes futile.
[Learn more about Harvard Medical School’s Developing Essential Skills in Clinical Research , an online course for medical doctors who wish to pursue a career in U.S. clinical research.]
- Myles PS, Bellomo R, Corcoran T, et al. Restrictive versus liberal fluid therapy for major abdominal surgery. N Engl J Med 2018; 378:2263-2274 (Including supplemental protocol and appendix, and including https://clinicaltrials.gov/ct2/show/NCT01424150 accessed September 2018.)
- Emdin, C. Restrictive versus Liberal Fluid Therapy in Abdominal Surgery. Trends.hms.harvard.edu. (Accessed Jan 2019.)
- Anesthesia Patient Safety Foundation Patient Safety Initiatives
- Brandstrup B. Finding the Right Balance. N Engl J Med 2018; 378:2335-2336.
- Keane M, Berg C. Evidence-based medicine: A predictably flawed paradigm. Trends in Anaesthesia and Critical Care, Volume 9 , 49 – 52.
- Keane MJ, Berg C. Is science the answer? Br J Anaesth. 2017;119:1081-4.
- PeriOperative ISchemic Evaluation-3 Trial (POISE-3). Clinicaltrials.gov. (Accessed August 2018)
- Nature Editorial. Reality check on reproducibility. Nature. 2016; 533: 437
Michael Keane is a specialist physician anaesthetist (anaesthesiologist), Casey Hospital, Melbourne and fellow of the Australian and New Zealand College of Anaesthetists. He is an adjunct associate professor at Swinburne University and adjunct senior lecturer at Monash University, both in Melbourne Victoria. Dr. Keane is interested in patient safety, ethics, pharmacology, health policy and economics, evidence-based medicine and new ways of innovation in medicine.
Shashikanth Manikappa is director of Anaesthesia & Perioperative Medicine, Casey Hospital, Melbourne and honorary faculty to Monash University and The University of Melbourne, Australia, and a visiting professor to Rajarajeshwari Medical College and Hospital, Bangalore, India. Dr Manikappa is an instructor for the Ultrasound Education Group, The University of Melbourne, Australia ,and course director & instructor for the Bedside Echo Scan Training (BEST) workshop, in India.