Cause and Effect or Conjecture? A Call for Consensus on Defining “Anesthesia-Related Mortality”

In this issue of the Journal, Dr. van der Griend and colleagues present a report entitled, “Postoperative Mortality in Children After 101,885 Anesthetics at a Tertiary Pediatric Hospital.”1 This is a very well written retrospective analysis of the anesthetic experience at a single tertiary care institution, the Royal Children's Hospital (RCH) in Melbourne, Australia. The objective of the study was to determine the incidence and nature of anesthesia-related mortality in pediatric practice at a large tertiary institution. The study population included all cases of children ≤18 years of age who had an anesthetic between January 1, 2003, and August 30, 2008, at the RCH in Melbourne, Australia. The investigators merged a database of anesthetics performed at the hospital with a regional mortality database. Cases of children dying within 30 days and 24 hours of an anesthetic were identified and the patient history and anesthetic record examined. “Anesthesia-related deaths were defined as those cases in which a panel of 3 senior anesthesiologists all agreed that anesthesia or factors under the control of the anesthesiologist more likely than not influenced the timing of death.”1 During the 68-month study period, 101,885 anesthetics were administered to 56,263 children. The authors report an overall 24-hour mortality from any cause after anesthesia as 13.4 per 10,000 anesthetics delivered; the 30-day mortality was 34.5 per 10,000 anesthetics delivered. The incidence of death was highest in children ≤30 days old. Patients undergoing cardiac surgery had a higher incidence of both 24-hour and 30-day mortality in comparison with noncardiac surgery. There were 10 anesthesia-related deaths, giving an incidence rate of 0.98 cases per 10,000 anesthetics performed (95% confidence interval [CI] 0.5 to 1.8). In all 10 cases, preexisting medical conditions were identified as being a significant factor in the patient's death. Five of these cases (50%) involved children with pulmonary hypertension. There were no anesthetic-related deaths in children who did not have major comorbidities. The authors conclude that anesthesia-related mortality is higher in children with heart disease and in particular those with pulmonary hypertension. The study results indicate an anesthesia-related mortality that is significantly higher than that reported previously. Prior studies have found an anesthesia-related mortality rate ranging from 0.36 to 2.9 per 10,000 anesthetics, depending on the definitions used.2,3 Increased perioperative mortality from any cause in newborns and in children with cardiac disease of all ages has been reported in other studies.4 This finding is important because it confirms the safety of pediatric anesthesia in healthy children. The present study includes other valuable information for the clinician. The data of the current study corroborate previous studies for children younger than 1 year5–7 and younger than 1 month8,9 undergoing anesthesia and surgery. The findings of increased risk for children younger than 1 year and especially younger than 1 month point to the need for greater caution when caring for these age groups. The study results reinforce the inherent risk in caring for children with heart disease. The data corroborate the previously published findings of investigators from the Mayo Clinic10 and the Perioperative Cardiac Arrest (POCA) Registry.4 Of note for the clinician is that more than half of all deaths occurred outside the operating room. While corroborating important findings of previous investigations, the current study presents a major challenge to the reader because of some methodologic issues. The study is focused only on mortality after anesthesia, excluding many interesting surrogate events (i.e., cardiac arrest) that are precursors. The publication by Odegard et al.11 from Boston Children's Hospital provided insight into 41 cardiac arrests in children with heart disease that did not result in a single death. The RCH paper has only 10 anesthesia-related deaths, insufficient to evaluate in a statistically meaningful fashion, or even to build a case control study. However, it would be quite useful for the clinician if the investigators had included a meaningful discussion of potential causes and preventability of these deaths. The present study provides a limited basis for comparison to prior results. Experienced clinicians caring for medically fragile or high-risk patients are already aware that anesthesia involvement may be coincidental to the death of the patient. The clinician's quest is to understand the mortality data in context of practice to learn what she or he can do to prevent or reduce the chance of perioperative mortality. The most interesting part of this study is the novel approach to defining anesthesia-related death. Dr. van der Griend and colleagues considered a death as anesthesia-related “when anesthesia or factors under the control of the anesthesiologist influences the timing of death,”1 as determined by a panel of 3 expert anesthesiologists. This is an interesting approach that has some merit for the reader; the data may point to cases of special concern in which the anesthesiologist may need to engage in further discussion with the surgeon and the parents of the patient about risk. The challenge for the reader is that the results of the study are difficult to generalize, because the method is not validated; it is uncertain whether a panel comprising 3 different expert anesthesiologists would reach the same conclusion as did the panel in this investigation. Furthermore, anesthesiologists often are faced with providing care for children who require emergent or urgent procedures or those who present for palliative care. Under such circumstances, induction of anesthesia, institution of mechanical ventilation, or administration of analgesics may initiate a physiologic cascade that conceivably could influence the timing of a patient's death. Whether such cases would or should be deemed to have been as a result of “anesthesia or factors under the control of the anesthesiologist,”1 needs to be validated through a rigorous assessment of interrater reliability. Previous studies have used generally accepted methods of classification such as a predefined interval between the induction of anesthesia and death or a consensus definition such as the one recommended by the Australian and New Zealand College of Anesthetists Mortality Committee.12 Numerous studies have examined perioperative mortality.13–27 In these reports, the definitions for deaths in which anesthesia was the primary cause or a contributing cause varied widely. Morray offered a comparison of the definitions used in different studies of “anesthesia-related” events.28 The definitions used by various investigators included the following: “not defined,” “contributed to in any way,” “totally or partially,” “attributed,” “solely,” “directly responsible/important contributor.” In addition to the various etiologies included in the syndrome of “anesthesia-related,” published reports vary widely in the period of observation associated from 24 hours, 2 days, or 30 days after emergence. So how do we make sense of all of this information? Although the findings of the current report substantiate much of what we already know and some things that should have been highlighted, the report engenders a call for action in developing a clearer definition of anesthesia-related death. There have been a number of calls over the years for development of a common set of definitions for critical events in anesthesia.29 Common definitions of terms allow the reader to understand and compare various published results with the possibility of making the results clinically relevant. The need for standardization is evident when reading the current paper by van der Griend et al.1 They speculate on why the rates are higher than those reported from the Mayo Clinic10 or the POCA Registry.3 Voluntary reporting in these other studies may play a role in explaining the differences. As the authors propose, “A better understanding of overall perioperative mortality rates, and in particular anesthesia-related mortality, may help anesthesiologists determine which patients are at higher risk and guide planning, resourcing, and expert staffing for these high-risk cases. Defining risk of anesthesia-related mortality is also important to help inform preoperative discussion with families, including informed consent for anesthesia.”1 The challenge of understanding “anesthesia-related/caused” mortality remains unanswered. Despite the fact that the findings of the current paper are not generalizable, the proposed “simpler [method; sic] to define anesthesia-related death when considering that anesthesia impacted on the timing of death rather than being the cause of death”1 warrants further discussion. Is it possible to develop a consensus within the specialty on a systematic process to define what constitutes “anesthesia or factors under the control of the anesthesiologist,” which contributed to adverse events or death? A well-defined process appropriately validated through formal consensus development (for example, see National Quality Foruma or American Heart Association30) may lead to a new, uniform, clinically useful definition of anesthesia-related death. DISCLOSURE Name: Jayant K. Deshpande, MD, MPH. Contribution: This author wrote the manuscript. Attestation: Jayant K. Deshpande approved the final manuscript.