In the beginning, there was physiology. Intensive care originated with the provision of invasive support for failing organs — and several decades of critical care practice were characterized by the use of increasingly sophisticated means to achieving physiologic ends. Dobutamine and transfusion to improve systemic oxygen delivery in shock. Mechanical ventilation to normalize PaO2 and PaCO2. Physiology-guided critical care interventions were inherently logical and personalized. Many improved patients’ appearance, silenced the alarming monitors, and stabilized the labsheet. Unfortunately, a number of physiology-guided critical care interventions also killed patients1–3. The arrival of randomized clinical trials to critical care in the mid-1990s4 heralded a new era in which the interventions provided in the ICU were placed back under the microscope with a new standard: improving patient outcomes.
The arrival of large-scale randomized trials in critical care, and the immediate recognition large trials could identify interventions capable of saving lives in the ICU1, gave rise to a novel challenge in critical care, one has persisted over the following two decades. While physiology-based interventions apply naturally to a single patient or a small group of patients, trials require populations. The results of any trial, therefore, require translation back from the population studied to individual patient in the ICU.
So how should one determine whether the results of a given clinical trial are likely to apply to a specific patient? A common approach is to consider whether the patient in front of you would have been ‘eligible’ for the trial. If the patient seems to meet the inclusion and exclusion criteria spelled out in the study methods, the intervention probably applies to them. If not, laissez-faire. Such an approach is encapsulated in an article in CCM this month by Dr. Ivie et al5. In to “assess the generalizability of the most highly cited RCTs relevant to intensive care” the authors screened 93 ICU patients at two academic centers using the inclusion and exclusion criteria of 15 highly cited critical care trials. Consistent with the author’s hypothesis, less than half of patients met criteria of any of the 15 landmark trials, largely due to not fulfilling inclusion criteria. The authors conclude “Our work…suggests eligibility criteria used in [critical care] RCTs impose limitations on the generalizability of their findings.”
This approach misses the subtle but fundamental distinction between ‘study population’ and ‘domain’ when assessing to whom the results of a clinical trial apply. The study population is the group of patients who are enrolled in a clinical trial – a product of the inclusion criteria (usually defining the disease of interest), the exclusion criteria (often focused on safety and targeting patients at the right amount of risk for the outcome), and the rate of ‘eligible-but-not-enrolled’ patients (patients for whom the patient, surrogates, or physician refuses participation)6. The domain is the population of patients to whom the results of the trial apply. These are not the same. The study population is easy to determine…just read the methods and table 1. The domain requires the reader to understand the study intervention, the population evaluated, and determine whether there is some physiologic reason why the same results seen in the trial would not apply in the patient in front of them.
A simple example is prisoners. Prisoners are excluded from essentially every critical care trial for regulatory reasons. Prisoners are therefore almost never in the study population. But if a prisoner comes in with florid ARDS and septic shock, would the results of major ARDS and sepsis trials still apply to them? Of course they would. There is no physiologic difference between prisoners and non-prisoners with the same disease – and therefore, despite not being in the study population, they are in the domain…and the trial results apply.
This subtle difference between ‘study population’ and ‘domain’ has been highly problematic for the translation of critical care evidence into practice. In 2001, the landmark ARMA trial showed that ventilation with lower tidal volume improved mortality among patients with ARDS1. Only ARDS patients were in the study population, but a reasonable physiologic argument could be made that lung-protective ventilation would likely have the same benefit among other types of ventilated critically ill adults who didn’t meet strict ARDS criteria. Some centers took the ‘domain’ approach and applied the ARDSNet protocol to all ventilated patients. Others took the ‘study population’ approach and waited 15 years while the evidence was gradually expanded to support the benefit of protective ventilation among surgical patients7 and ventilated ICU patients generally8.
Of the 93 patients in the study by Ivie et al, only 4% had ARDS…but 66% were mechanically ventilated. I would argue that the ‘domain’ of the ARMA trial alone means the results apply to at least two-thirds of the study patients…to say nothing of the other 14 trials. There are legitimate arguments to be made that clinical trials should better deal with the patient heterogeneity inside9,10 and outside6 of the study. But given the tremendous advances in ICU care the last 20 years have offered, I think the onus is on physicians to use the knowledge of physiology that was the original basis for critical care to determine whether there is any physiologic reason the intervention studied would have different results in your patient. I would not discard high-quality RCT evidence based on the minutia of the inclusion and exclusion criteria. After all, the study by Ivie et al only considered patients eligible who “were at least 18 years old…and were admitted to a participating ICU in November 2010 or July 2011”. None of the patients in my ICU would have qualified.
- Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. The Acute Respiratory Distress Syndrome Network. N Engl J Med. 2000;342(18):1301-1308. doi:10.1056/NEJM200005043421801.
- Perner A, Haase N, Guttormsen AB, et al. Hydroxyethyl starch 130/0.42 versus Ringer’s acetate in severe sepsis. N Engl J Med. 2012;367(2):124-134. doi:10.1056/NEJMoa1204242.
- Ferguson ND, Cook DJ, Guyatt GH, et al. High-frequency oscillation in early acute respiratory distress syndrome. N Engl J Med. 2013;368(9):795-805. doi:10.1056/NEJMoa1215554.
- Bernard GR, Wheeler AP, Russell JA, et al. The effects of ibuprofen on the physiology and survival of patients with sepsis. The Ibuprofen in Sepsis Study Group. N Engl J Med. 1997;336(13):912-918. doi:10.1056/NEJM199703273361303.
- Ivie RMJ, Vail EA, Wunsch H, Goldklang MP, Fowler R, Moitra VK. Patient Eligibility for Randomized Controlled Trials in Critical Care Medicine: An International Two-Center Observational Study. Crit Care Med. October 2016. doi:10.1097/CCM.0000000000002061.
- Arabi YM, Cook DJ, Zhou Q, et al. Characteristics and Outcomes of Eligible Nonenrolled Patients in a Mechanical Ventilation Trial of Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med. 2015;192(11):1306-1313. doi:10.1164/rccm.201501-0172OC.
- Futier E, Constantin J-M, Paugam-Burtz C, et al. A trial of intraoperative low-tidal-volume ventilation in abdominal surgery. N Engl J Med. 2013;369(5):428-437. doi:10.1056/NEJMoa1301082.
- Serpa Neto A, Cardoso SO, Manetta JA, et al. Association between use of lung-protective ventilation with lower tidal volumes and clinical outcomes among patients without acute respiratory distress syndrome: a meta-analysis. JAMA J Am Med Assoc. 2012;308(16):1651-1659. doi:10.1001/jama.2012.13730.
- Iwashyna TJ, Burke JF, Sussman JB, Prescott HC, Hayward RA, Angus DC. Implications of Heterogeneity of Treatment Effect for Reporting and Analysis of Randomized Trials in Critical Care. Am J Respir Crit Care Med. 2015;192(9):1045-1051. doi:10.1164/rccm.201411-2125CP.
- Prescott HC, Calfee CS, Thompson BT, Angus DC, Liu VX. Toward Smarter Lumping and Smarter Splitting: Rethinking Strategies for Sepsis and Acute Respiratory Distress Syndrome Clinical Trial Design. Am J Respir Crit Care Med. 2016;194(2):147-155. doi:10.1164/rccm.201512-2544CP.