This study employed a descriptive qualitative approach. Data were collected from patient-safety accident reports recorded in a Japanese database between January 2010 and December 2021.

Since 2004, the Japan Council for Quality Health Care has been conducting various activities to maintain public confidence in healthcare services as well as improve the quality of these services. These activities include evaluating medical services and the Project to Collect Medical Near-Miss/Adverse Event Information. As of December 31, 2021, 1,575 medical institutions (approximately 20% of the total number of hospitals in Japan) were registered with this project. Although participation in this project is mandatory for tertiary teaching hospitals (such as medical universities, national hospitals, and hospitals providing advanced treatments), other medical institutions have joined the project voluntarily. Overall, 273 hospitals (138,150 inpatient beds) were required to participate in this project from December 31, 2021. In total, 47,527 cases of medical adverse events were reported between October 2004 and December 2021, and 4,674 have been reported between January and December 2021. Information regarding medical adverse events is published quarterly and annually. Additionally, annual reports have been available to interested parties since 2010 on the project’s website (http://www.med-safe.jp/contents/english/index.html).

The following medical adverse events must be reported: (a) apparent errors in treatment or management that resulted in the patient’s mental/physical disability or death or those that required unexpected treatment, treatment to an unexpected extent, or other medical procedures; (b) unapparent errors in treatment or management that resulted in the patient’s mental/physical disability or death or those that required unexpected treatment, treatment to an unexpected extent, or other medical procedures (including events possibly associated with the treatment or management provided, limited to unexpected events); and (c) errors other than those described in (a) and (b) or any information conducive to preventing medical adverse events and their recurrence at medical institutions. Bleeding complications involving the brain and gastrointestinal tract may be relatively less reported in this system. Therefore, our study primarily evaluated adverse events related to “errors” in the ECMO procedures.

We used anonymized data from the database that were unlinked to individual patient information. The need for ethical approval and informed consent was waived because of the study’s retrospective nature and the anonymity of the analyzed data.

We searched the database for all ECMO-associated adverse events that were reported between January 2010 and December 2021; a free-text search was conducted to identify procedure-specific events. Because this database is only available in Japanese, the Japanese words for “ECMO,” “artificial cardiopulmonary device,” and “percutaneous cardiopulmonary support” were used for data extraction. In the event of a duplicate report, the data were integrated.

We collected the following data from the database: patient sex and age, location, clinical experience of the primary operator (in years), adverse events, causes of adverse events, and the reporter’s assessment of the possibility of residual disability. Based on the level of harm anticipated by the reporter, the accidents were then classified as follows: (a) event resulting in death, (b) event with a high potential for residual disability, (c) event with a low potential for residual disability, (d) event without a potential for residual disability, and (e) unknown. Two authors (HH and TK) independently reviewed the accident reports to determine their eligibility for inclusion and classified them by reviewing their free-text descriptions. Because of our focus on ECMO-associated adverse events, we excluded adverse events unrelated to ECMO and those related to mechanical cardiopulmonary support only during surgery.

We categorized ECMO-related adverse events into cannula malposition, accidental decannulation, bleeding, air in the circuit, thromboembolism, limb ischemia, gas supply issue, foreign body remnants, and others. Cannula malposition was defined as extravascular cannulation or cannulation of the wrong vessel (such as arterial-arterial and venous-venous cannulation). In contrast, decannulation was defined as accidental cannula removal that required reinsertion or position adjustment. Bleeding was defined as bleeding during or after managing patients undergoing ECMO. Therefore, bleeding due to extravascular cannulation was classified as cannula malposition, whereas bleeding or air in the circuit due to accidental decannulation was classified as decannulation.

First, we examined which adverse events were more likely to lead to death or had a high potential for residual disability. Second, we focused on cannula malposition and analyzed the patient’s age, types of ECMO, the vessel of cannulation, site of injury, physician’s department, physician’s experience (in years), location, and whether the cannulation was fluoroscopy- or ultrasound-guided. We differentiated extracorporeal cardiopulmonary resuscitation from the veno-arterial-ECMO since the cannulation techniques may differ. The ELSO guideline recommends a cut-down technique for cannula insertion as the first choice in patients with extracorporeal cardiopulmonary resuscitation because of no detection of femoral artery pulse [4]. In contrast, the percutaneous cannulation technique is recommended in veno-arterial-ECMO. Third, we focused on accidental decannulation and analyzed the level of harm, the type of cannula, the cause, and the considered factors. Fourth, we analyzed the bleeding site, causes of air in the circuit, and additional treatments administered for bleeding, the air in the circuit, and thromboembolism.