“IT'S not like ten rats and a t test” was Fred Cheney's reassuring comment as we struggled through yet another rewrite of the Pediatric Perioperative Cardiac Arrest (POCA) manuscript.1Indeed, outcomes research is not easy. The investigator attempts to objectify a data set that remains stubbornly nuanced, subjective, and incomplete. It is likely that Randall Flick, M.D., and his coauthors from the Mayo Clinic would agree. In this issue of Anesthesiology, Flick et al. 2report the incidence of and outcomes from perioperative cardiac arrest (PCA) in anesthetized children at the Mayo Clinic during the past 17 yr. Mayo Clinic investigators have been pioneers in the use of the electronic medical record for the creation and maintenance of a single-institution outcomes database. In the current report, PCA occurred most often in children with congenital heart disease as a result of factors not related to anesthesia. While cardiac surgery accounted for only 5% of all procedures, 87.5% of all arrests occurred in patients with congenital heart disease, usually during cardiac surgery as a result of failure to wean from cardiopulmonary bypass. Anesthesia factors were related in only 7.5% of all arrests, with an incidence of 0.65 per 10,000 anesthetics. Only six anesthesia-related arrests occurred in noncardiac cases during the 17-yr study period.
Interpretation of outcomes data is also not easy for the reader, given the absence of standardization among the myriad studies published during the past five decades. The reader must closely examine the fine print: What were the demographic characteristics of the patient population (e.g. , age, American Society of Anesthesiologists physical status, surgical category, and emergency status)? How was “anesthesia related” defined (e.g. , “preventable,”“associated,”“causative,”“human error”)? What time frame was included in the term “perioperative” (postanesthesia care unit discharge to 30 days)? Was reporting voluntary or mandatory? Only by appreciating these definitions and other details of methodology can the reader interpret the Mayo Clinic findings and put them in the context of previous reports of cardiac arrest in anesthetized children.
As a single tertiary referral center, the Mayo Clinic is probably not representative of the population-at-large of patients, physicians, or surgical procedures. Nonetheless, the data presented by the Mayo researchers are of high quality. Numerator and denominator data are reliable, given that reporting was mandatory and occurred within a single institution. Patient demographics and total caseload were known with a high degree of reliability. Underreporting was possible but unlikely.
The authors compare their results to those of the POCA Registry.1However, the Mayo Clinic and POCA Registry data banks are very different, each with strengths and weaknesses. In the POCA Registry, more than 60 institutions contributed PCAs to a central data bank maintained by the American Society of Anesthesiologists Closed Claims Project staff at the University of Washington. Underreporting or biased reporting (e.g. , withholding sensitive cases) occurred at an unknown rate. Participating institutions contributed their annual patient demographics and total caseload with a variable degree of reliability and accuracy. Therefore, incidence calculations were probably less reliable than those from the Mayo Clinic series, given the likely inaccuracy of both numerator and denominator.
On the other hand, the POCA Registry has accumulated many PCAs from multiple institutions, allowing analysis of cause of arrest and factors related to arrest. An inclusive definition of “anesthesia related” facilitated this process. Cases were deemed anesthesia related if anesthesia personnel or the anesthesia process played at least some role (ranging from minor to total) in the genesis of cardiac arrest. Flick et al. applied a more restrictive definition of anesthesia related. For example, arrest from massive trauma, embolic events, uncontrolled hemorrhage, and the metabolic consequences of massive transfusion (including hyperkalemia) were defined as “non-anesthesia attributed.” Patients “in extremis” upon arrival to the operating room were also excluded, even if the anesthetic might have contributed to cardiac arrest. Inclusion of such cases would have increased the number of anesthesia-related cardiac arrests in the Mayo Clinic series from 6 (23%) to 14 (53%) of the 26 arrests occurring in noncardiac patients. Exclusion of such cases may preclude identification of problems of interest to anesthesiologists. For example, when an anesthesiologist unknowingly administers blood with a high potassium concentration to a patient who subsequently suffers a hyperkalemic cardiac arrest, that arrest is anesthesia related and is potentially preventable.
Whether because of a restrictive definition of anesthesia-related cardiac arrest or because of other reasons (including high-quality care), the Mayo Clinic reports that an anesthesia-related arrest in a noncardiac patient occurred on average only once every 3 yr! Meaningful analysis of cause of arrest or of factors related to arrest was not possible because of these small numbers. Likewise, a multivariate analysis of factors relating to survival from cardiac arrest could not be performed. It is interesting that the Mayo Clinic group did not find the same decline in the incidence of PCA in children during the 17 yr of the study as they did in their adult population.3Perhaps this lack of change resulted from small numbers of arrests reported in children, although other factors (e.g. , increasing patient acuity) are also possible.
Regardless of these concerns over definitions and methodology, Flick et al. deserve our thanks and congratulations for their contributions. Their data complement the findings of the POCA Registry and other series of pediatric PCAs from around the world. As noted in their report, the absence of standardization of definitions and methodology remains a serious problem. It has even been suggested that improved outcomes for anesthetized patients during the past 50 yr could be an artifact of the heterogeneity of definitions and methodology.4A coordinated effort to eliminate this heterogeneity is required. The Mayo Clinic group, under the auspices of our national organizations, should play a leadership role in an effort to standardize definitions and coordinate data acquisition and analysis. By creating national and international data pools, we can firm up what remains a soft science.
* Perioperative Services, Phoenix Children's Hospital, Valley Anesthesiology Consultants, Ltd., Phoenix, Arizona. jmorray@cox.net. † Department of Anesthesiology, University of Washington School of Medicine, Seattle, Washington.