Customize your JAMA Network experience by selecting one or more topics from the list below.
Medical research drives innovation and improves outcomes. In cardiology, advances in diagnosis, prevention, and treatment reduced the 30-day mortality from acute myocardial infarction from 20% to 12.4% between 1995 and 2014.1 Similarly, for women diagnosed with breast cancer, 5-year survival has increased from 74% to 88.5% over the past 4 decades2 in parallel with advances in diagnosis and targeted therapy, including chemotherapeutic regimens such as cyclophosphamide, methotrexate, and fluorouracil (CMF), trastuzumab, and aromatase inhibitors.3
However, patient outcomes also improve in the absence of specific technological advances. For example, it is estimated that between 1990 and 2017, the global number of deaths from sepsis declined from 15.7 million (95% uncertainty interval [UI], 14.7-16.7 million) to 11.0 million (95% UI, 10.0-12.1 million).4 This improvement has not occurred because of new drugs and new procedures but rather from improved approaches to resuscitation and physiologic support and from understanding the effects of often subtle variability in management strategies that translate into better patient outcomes. A recurring theme of these insights is that although physiologic support is good, restraint in providing that support is often better, for example, controlling blood glucose levels but not too rigidly using insulin5; administering transfusions but accepting a lower hemoglobin threshold at which to make the decision to transfuse6; or providing ventilator support but limiting distention of the lungs.7 Underlying this theme is an emerging recognition that in acute critical illness, what is normal is not necessarily optimal and that interventions bring both benefits and harms.
In this issue of JAMA, Lamontagne and colleagues8 report the results of the 65 trial, an evaluation of blood pressure targets to guide vasopressor therapy. The emergency hemodynamic management of septic shock includes administration of fluids to restore a relative intravascular volume deficit, and the use of vasopressor agents such as norepinephrine to increase blood pressure when fluid alone is insufficient to do so. Increased blood pressure is thought to result in augmented tissue perfusion, although the optimal threshold is unknown. Previous guidelines of the Surviving Sepsis Campaign had recommended that a mean arterial pressure (MAP) threshold of 65 mm Hg should be targeted,9 based on observational data that tissue perfusion is maintained at a MAP as low as 65 mm Hg.10 Implicit in that recommendation is an assumption that harm is only a consequence of a MAP that is too low, and not one that is too high. Prior studies have suggested that a higher blood pressure may be harmful11,12 and that targeting a MAP of 65 mm Hg results in MAP levels of 75 mm Hg or higher because infusions are increased when the pressure is lower than the target but not necessarily reduced when it is higher than the target blood pressure level.12 Moreover, the physiologic objective is not pressure but flow, and for a given pressure, flow is increased when resistance is reduced, as it is in septic shock.
Building on the findings of an earlier pilot trial,13 the 65 trial investigators tested the hypothesis that targeting a MAP of 60 to 65 mm Hg in older patients would be more effective than usual care as reflected in reduced all-cause mortality at 90 days. Patients older than 65 years with vasodilatory hypotension were randomized to receive vasopressors guided either by MAP target (60-65 mm Hg, permissive hypotension) (n = 1291) or to receive usual care (at the discretion of the treating clinician) (n = 1307). The researchers showed satisfactory adherence with the intervention, as reflected in an 11.3% occurrence of nonadherence and in the separation of the mean blood pressures achieved (although MAPs for the permissive hypotension group appeared to average around 65 mm Hg).
At 90 days, 500 of 1221 (41.0%) patients in the permissive hypotension group had died compared with 544 of 1242 (43.8%) in the usual care group (absolute risk difference, −2.85%; 95% CI, −6.75 to 1.05; P = .15), which favored the permissive hypotension approach but failed to meet prespecified criteria for superiority (an absolute risk reduction of 6%). The duration of vasopressor usage and the amount of drug administered were reduced in the permissive hypotension group, with a difference in mean duration of vasopressors of −9.9 hours (95% CI, 62; –14.3 to −5.5 hours) and a difference in median dose of vasopressors of (8.7 mg, 95% CI, −12.8 to −7.6 mg, norepinephrine equivalent). Both a prespecified analysis adjusted for imbalances in key baseline variables and a post hoc subgroup analysis of patients with chronic hypertension at baseline yielded odds ratios (ORs) suggesting the possibility of better survival associated with permissive hypotension, with an adjusted OR of 0.82 (95% CI, 0.68 to 0.98), and an adjusted relative risk of 0.84 (95% CI, 0.71 to 0.99), respectively. As importantly, there was no evidence that a lower blood pressure target was associated with later ischemic sequelae such as acute kidney injury or cognitive impairment in survivors.
The 65 trial is an example of what has been disparagingly called a “negative” or “null” trial,14 yet it is also an important contribution to knowledge that will likely change practice and may help to reduce mortality. How can this be possible?
The concept that a trial is either positive or negative derives from trials of novel drugs or technologies. The investigators posit a minimally important treatment effect that either is or is not achieved when the intervention is applied; the result is a categorical decision about efficacy. Such a dichotomous decision is appropriate when the intervention is untested: in that case, not only is the intervention new and a departure from standard care, it carries unknown risks and costs, so the threshold for adoption is high.
In contrast, the 65 trial explored the merits of an intervention that is in widespread clinical use—vasopressor therapy—but for which optimal titration is uncertain. This uncertainty reflects inherent variability in clinical practice and changing clinical perspectives on the validity of conventional wisdom. Declining rates of acute gastrointestinal stress bleeding, for example, have led investigators to ask whether the harms of prophylaxis may outweigh the benefits15; concerns that the oxygen-carrying capacity of red blood cells might be jeopardized by prolonged storage led to trials of fresh vs conventionally stored red blood cells,16 and speculation that oxygen may be harmful during acute inflammation has prompted analyses of reduced oxygen targets in critical illness.17 The outcomes of these trials were far from predictable, and practice variability reflects uncertainty in understanding optimal therapy in the context of profoundly disrupted homeostasis. This variability is what makes clinical trials in critical care so important but also so potentially controversial.18
There are 2 common approaches to the design of a clinical trial that evaluates practice variability. Investigators might first seek to measure the spectrum of that variability through preliminary observational studies or studies of clinician attitudes and then define study groups that represent 2 plausible but separate approaches along a spectrum of practice.6 Alternatively, the investigators might opt to evaluate a specific interventional strategy and compare this with a “usual care” control, as was done in the 65 trial. Both approaches have shortcomings. Definition of 2 different treatment groups based on 2 discrete points on a spectrum of practice possibilities (for example, a liberal vs a conservative approach) potentially opens the trialists to accusations that one or other intervention is widely divergent from usual care and thus unethical or that the optimal approach is in the middle.19 On the other hand, the use of a usual care control assumes that standard care is optimal care and risks contamination of the experiment because of secular changes in practice over time.20 In many cases, usual care is often far from standardized care, as the substantial variability around MAP readings in the usual care group of the 65 trial attests.
So how should the results of the 65 trial be incorporated into clinical practice? From a scientific perspective, the trial was indeterminate. It failed to support its primary hypothesis, although it generated findings that were consistent with that hypothesis and suggested that further work might well be informative. But for the clinician caring for a patient with vasodilatory shock, the message is different. Nothing in the results suggests that artificially raising the blood pressure by administering more vasopressors provides benefit for patients; in fact, the signal suggests that this deliberate intervention may be harmful, particularly among patients for whom expert opinion had previously recommended such an approach,9 those with preexisting hypertension. Clinicians could be reassured in the finding that assiduously targeting an arbitrary blood pressure is not helpful; scientists could ask whether interventions should be titrated to pressure at all but and instead should focus on physiologic variables that reflect flow, such as capillary refill.21 Clearly, caution must be exercised in avoiding a blood pressure that is too low to enable perfusion, but the 65 trial raises the question, “Should an upper limit be placed on a MAP target?”
Randomized clinical trials yield insight, not instructions. The message of the 65 trial by Lamontagne et al is not that all patients should be treated the same, rather that treatment decisions can be individualized around a somewhat lower mean MAP comfort point. A MAP target of 60 to 65 mm Hg appears to be more than adequate for most older patients with vasodilatory shock, and, contrary to conventional wisdom, it is possible that a lower target threshold may be more beneficial in older patients with preexisting hypertension. Some “negative” trials can contribute to changes in clinical practice.
Corresponding Author: John C. Marshall, MD, St Michael’s Hospital, 30 Bond St, Fourth Floor Bond Wing, Room 4-007, Toronto, ON M5B 1W8, Canada (firstname.lastname@example.org).
Published Online: February 12, 2020. doi:10.1001/jama.2019.22526
Conflict of Interest Disclosures: Dr Marshall reported receiving personal fees from AKPA Pharma, Baxter, and Critical Care Medicine, for which he serves an associate editor; receiving nonfinancial support from Adrenomed; and serving as the chair of the International Forum for Acute Care Trialists.
Marshall JC. Choosing the Best Blood Pressure Target for Vasopressor Therapy. JAMA. Published online February 12, 2020. doi:10.1001/jama.2019.22526
Create a personal account or sign in to: