How to do clinical trials in the ICU

Site-initiated clinical trials are challenging, particularly with vulnerable populations like ICU patients. But they can be useful if you can answer “Yes” to these questions: (1) Could the answer change practice? (2) Does my ICU see enough of these patients to answer the question? (3) Can I treat both the investigational and control group patients in a fair manner? Other approaches to ICU clinical research, such as joining another multicenter clinical trial and observational (prospective or retrospective) studies whose answers have explanatory power may also be valuable. Regarding changing practice: The trial's answer should be important enough that any intensivist treating such a patient would care about it. Examples of questions that probably aren't important enough are: (1) Is there a difference in ARDS outcome if the tidal volume is 4, 6, or 8 mL/kg?, (2) Is knee-high pneumatic compression inferior to thigh-high pneumatic compression for thromboembolism prophylaxis in anticoagulant-contraindicated patients?, and (3) Do piperacillin-tazobactam and cefepime have equivalent efficacy for empiric coverage of hospital-acquired infections? The first question probably has a patient-specific, non-generalizable answer; the second, with small “dose” differences has high risk of being a negative study; the third answer likely depends on local organisms, and equivalence studies require very large patient numbers. Studies whose answers probably could change practice include: (1) Does adding Extracorporeal CO2 Removal (ECCO2R) to intubated ventilation in combined respiratory and metabolic acidosis improve survival/shorten ventilator time?, and (2) Does extracorporeal CPR (ECPR) for witnessed out-of-hospital cardiac arrest with a shockable rhythm save lives? Regarding the ECCO2R proposal, current practice involves hyperventilating high-dead-space patients, pressors, bicarbonate infusion, maybe dialysis, and outcomes are poor. ECCO2R can reduce ventilatory and alkalinizing/dialysis requirements1. Regarding the ECPR proposal, current survival rates reported after 60 min of CPR without ECPR for possibly salvageable patients are 9%; with ECPR, 22%. Moreover, there are no clinical trials published2. The second criterion is important because if your site doesn't see enough patients, coordination with multiple centers and delay can imperil completion. Estimating sample size early for feasibility is important. For the ECCO2R proposal, an estimate for 30-day survival of such patients is 25%. To double survival to 50%, 2-sided p = 0.05, with 80% power, sample size is 65/group, 130 patients altogether. For the ECPR proposal, an estimate without ECPR is 5% survival after an hour of CPR. To raise survival to 25%, 2-sided p = 0.05, with 80% power, sample size is 58/group, 116 patients. These numbers may be practically achieved in some centers after 2-3 years, when the answers will still be pertinent. The third criterion addresses whether the question is practical: Can a protocol fair to patients be devised and executed? For both ECCO2R and ECPR proposals, the protocols must (1) Establish criteria to assure no prolongation of life just to favor one unblinded group or another, and (2) Establish a process to assure uniform patient entrance and procedures among participating physicians and centers. Careful thought and collegial consultation at the start of study planning is important for helping ICU patients with your clinical research. Preparing the final study report similarly requires close attention and consultation3.