## Introduction This study was born with the idea of a randomized clinical trial in cardiopathic children \< 5 years old who have an operation with hypothermia and extracorporeal circulation (ECM), to verify the use of protective ventilation during ECM. The standard treatment does not involve ventilation and the lung is left to collapse and then re-expanded at the end of the operation, since oxygenation is still guaranteed by the ECC. The outcome is being on mechanical ventilation (post-operative, in intensive care) for more than 48h. ## Population There were 188 children recruited (patients from an old database were also used as controls), of whom 48 were excluded because of severe pulmonary problems or other reasons (eg, postoperative ECMO). Partial ventilation problem Of these 140, n = 15 were partially ventilated i.e., the surgeon started protective ventilation in ECMO but then decided for various complications to stop ventilation (the expanding lung disturbs the operative field). ## Design of the analysis Given the difficulty of a serious randomization, the experimental design has turned to an a posteriori evaluation by balancing the variables of the two groups through propensity score (inverse probability of treatment weighting, IPTW). In fact, at a first glance at the descriptives it was quite clear that despite a design with a pseudo-randomization (one patient ventilated and two non-ventilated, depending on the type of heart disease, so as to have balanced controls and treated not only for absolute number but also for incidence of the various heart diseases), the surgeon tended to ventilate the less severe patients, who therefore, in any case, would have been less in intensive care even without treatment. Given also the problem of the partially ventilated, I did the same analyses both with n = 125 (ventilated and nonventilated only), and with n =140, considering the 15 partially ventilated as "intention to treat"