The new goal in goal-directed sepsis therapy: Pink pinkies
ANDROMEDA-2 uses capillary refill time to drive resuscitation for sepsis
There’s a new goal in goal-directed therapy for sepsis, and it’s right at your fingertips. It works as well as any other, which is to say, not very well.
Capillary refill time is a fast and easy test to assess peripheral perfusion. It’s now been tested as a primary endpoint of resuscitation for septic shock in two multicenter international trials.
The ANDROMEDA Trials
ANDROMEDA-SHOCK
The first multicenter ANDROMEDA trial found that among 424 patients with septic shock, those randomized to a care strategy targeting capillary refill time ≤ 3 seconds had numerically lower mortality (35% vs 43%) and less organ failure, and received less IV fluid, compared to usual care. A post hoc bayesian analysis concluded it was 90% likely that the strategy saved lives.
ANDROMEDA-2
In an impressive accomplishment, at 86 ICUs in 19 countries, including many Latin American nations with limited health system resources, 1,501 patients with a diagnosis of septic shock (requiring norepinephrine at a median dose of 0.22 μg/kg/min, with 20% patients on a second vasopressor) were randomized to either usual care or protocolized resuscitation to meet a goal of capillary refill time ≤ 3 seconds.
Capillary refill time was measured in a standardized way, pressing the fingertip with a microscope slide for 10 seconds after blanching, then releasing and timing color return (to a subjective “normal”) with a stopwatch.
Patients who had cap refill > 3 seconds entered the algorithm, which was complex.
It primarily involved checking for fluid responsiveness and individualizing volume resuscitation, utilizing a variety of techniques, including pulse pressure variation, passive leg raises, IVC collapse on ultrasound, and the end-expiratory occlusion test (read more here).
They also prescribed unorthodox physiology-driven moves to boost cap refill, such as:
Titrating norepinephrine to maintain diastolic blood pressure > 50 mmHg;
Giving dobutamine to patients with low ejection fraction on bedside echocardiography (shout out to Rivers NEJM 2001!);
Titrating norepinephrine to MAP >80 mmHg or giving dobutamine infusions to maintain cap refill in patients with chronic hypertension
The algorithm guided their resuscitation for the first 6 hours.
Patients with normal cap refill times were rechecked at the bedside every hour.
In addition to the care teams, trained study personnel were at the bedside helping facilitate the protocol. At some sites, those trained study personnel were helping manage patients in both arms.
Results
After 28 days, mortality was identical in both groups (26.5% vs 26.6%).
Cap refill-resuscitated patients had a median one fewer day of organ support (mechanical ventilation, vasopressors, or renal replacement therapy), compared to the usual care group (3 vs. 4 days, statistically significant).
In their 6 hours of protocolized treatment, intervention patients received:
Less fluid (595 mL vs 847 mL, a 252 mL difference)
More dobutamine (12% vs 5%) and norepinephrine (94.7% vs 91.4%);
and more frequently achieved cap refill < 3 seconds (86% vs 62%), but had no difference in:
Lactate levels (3.2 vs 3.5 mmol/L) or
Central venous oxygen saturation (74% vs 72%).
Central venous pressure (9.1 vs 9.8 mmHg).
We’re politely ignoring the decision to use a win ratio as the primary outcome.
So, how could ~700 patients have less organ failure but identical mortality? One possibility is that using cap refill significantly improved care. Other explanations are possible, though.
Unmeasured Care Differences (and Similarities?) Between Groups
Trained trial personnel assisted the primary teams at the bedside in the first six hours to help implement the trial protocol on the intervention patients. Those patients got more attention and care. This extra care could be responsible for the difference in outcomes, rather than the protocol itself.
This was not a cluster-randomized trial. So a patient in one bed was on the protocol, while a nearby patient (possibly cared for by the same physician) was not. Further, at some centers, trained trial personnel assisted in caring for patients in both the intervention and control arms. This is a recipe for unconscious bias seeping into care decisions.
Decisions to stop organ support are inherently subjective and vulnerable to bias. As an example, we know that just adding a second spontaneous breathing trial to the day can reduce clinicians’ willingness to extubate patients. This trial went further, introducing multiple behavioral and psychological variables into the equation.
Outcomes were widely divergent between centers, which is partly due to chance—there were 86, and most of them (49) enrolled fewer than 10 patients. One center in Mexico was the largest single recruiting center, enrolling 16% of the patients (n=229), and found a strong benefit, producing much of the positive signal. Results at the ~68 centers enrolling 1 to 19 patients each are not individually shown in eFigure 5 in the supplemental appendix, but together they found no benefit.
It’s entirely possible that protocolized resuscitation driven by capillary refill times resulted in reduced need for organ support. But the potential influence of the above aspects of the ANDROMEDA-2 trial, and the very small differences in aggregate treatments between groups, together reduce the confidence that the intervention itself produced faster recovery.
Is Cap Refill the New Lactate?
The good news is that ANDROMEDA-2’s results offer something for everyone.
Do you enjoy the intellectual and technical challenge of carefully analyzing and tailoring each patient’s treatment to their physiology as you understand it? If so, using cap refill can help guide your decisions, and might even reduce your septic patients’ duration of organ support. Plus, although the mortality figures were virtually identical, the confidence interval included the possibility of lives saved (or lost).
Do you instead believe that sepsis care can be somewhat standardized, with appropriate tweaks for each patient that do not require the tenacity employed in this trial? You can be reasonably well-assured that your patients will survive as often as they would with a more intense and analytical approach.
Like lactate, capillary refill most of all gives the clinician something desperately desired in a high-stakes and uncertain situation: information to direct action in a seemingly rational way.
But like lactate, the information provided is often misleading. Less than two-thirds of patients in the control arm achieved cap refill < 3 seconds (an absolute 24% fewer than in the intervention arm), yet they stubbornly survived anyway, at a numerically identical rate.
The off-label use of norepinephrine and dobutamine to boost diastolic pressure and cardiac function in sepsis aren’t likely to spread widely in the U.S., as there is no direct evidence here for their benefits as components of the bundle.
More fundamentally, there is a meaningful debate among serious people about whether permissive hypotension might actually be superior to an obsessive approach to “optimizing” hemodynamics in sepsis.
When a question as basic as that remains unanswered, it’s not surprising that targeting care to keep pinkies pink didn’t save lives in septic shock.
ANDROMEDA-2 is an impressive achievement by hundreds of dedicated investigators across the globe. Using capillary refill time makes logical sense and might direct sepsis care positively more often than not. Or, it might be another dead-end trend and generator of unhelpful busywork.
The equivocal results mean that you get to decide what it means to your practice (at least, until the sepsis regulators get hold of it). Whatever you choose, godspeed.