Programs for organ donation after death determination by circulatory criteria (DCD) have expanded through the establishment of national ethical, professional, and legal frameworks, aiding to alleviate the disparity between organ supply and demand.1 Most DCD donors are patients who have been admitted to the intensive care unit and who have died after removal of life sustaining measures (i.e., controlled DCD).2

Historically, the medical community has not needed to focus on the specifics of death determination by circulatory criteria. Nevertheless, as DCD increases, debates persist regarding how and when to determine death in this context.3,4 Specifically, there is an evolving concern about the time required for observation following circulatory arrest to rule out the possibility of an unassisted resumption of spontaneous circulation (autoresuscitation).5,6,7,8,9,10 Consequentially, when determining death in the context of organ donation, there is a struggle between two time-sensitive aspects. On one hand, there is the requirement to strictly adhere to the dead donor rule, which states that patients are required to be declared dead before the removal of vital organs for transplantation.11 The dead donor rule exists in many jurisdictions and is essential to maintain trust between the health professional community and the public as well as within the health professional community itself. Opposingly, to minimize organ ischemia and ensure both the functionality and sustainability of organ grafts, the period between the removal of life sustaining measures and the cessation of circulation and determination of death cannot exceed predetermined time limits. Conflicting pressures between ensuring permanent loss of circulation while also maintaining organ viability for transplantation renders critical the need to establish clear criteria for death determination.

Although the criteria required for determination of death have been standardized for determination of death by neurologic criteria,12 standards for death determination by circulatory criteria for organ donation are in question.13,14 Surveys in Canada and the Netherlands have shown variability in death determination by circulatory criteria practices by intensive care physicians, highlighting the need for standardization of practice.15,16

The primary objective of this study was to describe the practices of intensive care health care professionals for death determination by circulatory criteria. This study aims to answer the questions: 1) What are the most common diagnostic tests used to determine death after circulatory arrest reported in the intensive care unit? 2) Do these diagnostic tests vary by country or by assessor? 3) Do these diagnostic tests vary in the context of organ donation?

Methods

This study is a retrospective analysis of prospectively collected data from the Death Prediction and Physiology after Removal of Therapy (DePPaRT) study.17 The primary objective of the DePPaRT study was to describe the incidence and timing of resumption of cardiac electrical and pulsatile activity in critically ill adults who died after withdrawal of life-sustaining measures. Death determination variables were recorded by clinicians from 20 intensive care units in Canada (16 sites), the Czech Republic (three sites), and the Netherlands (one site). Patients with neurologic determination of death, a functioning cardiac pacemaker or without an arterial catheter were excluded. The protocol did not dictate how death determination was performed. As part of the DePPaRT study, individuals who were responsible for death determination after planned withdrawal of life sustaining measures were asked to complete a checklist questionnaire that instructed responders to indicate the time/date of declaration of death, which specific diagnostic tests were used for death determination after circulatory arrest, as well as their professional role (see Electronic Supplementary Material [ESM] eAppendix).

The research protocol was approved by the relevant institutional review board or ethics committee at each site (coordinating site: Children’s Hospital of Eastern Ontario Research Ethics Board, Ottawa, ON, Canada). The self-reported checklist included a statement of assumed implicit consent upon completion and submission.

Analysis

Data collected included country of origin, medical background of the respondent, and patient characteristics (male vs female patients; attempted and successful DCD vs attempted but not successful DCD vs not attempted DCD). Diagnostic tests used for death determination were collected using a checklist, which included options for absent heart sounds by auscultation, absent palpable pulse, absent pulse by audible Doppler, absent blood pressure by noninvasive monitoring, flat invasive arterial blood pressure (ABP) tracing, pulseless electrical activity (nonperfusing rhythm), flat electrocardiography (ECG) tracing, absent breath sounds by auscultation, absent pulse oximetry, unresponsiveness to painful stimulus, fixed and dilated pupils, and other diagnostic tests used (see ESM eAppendix). While strictly speaking, breathing is a neurologic sign, for the purposes of this analysis, it was considered a respiratory sign.

Statistical analysis

Characteristics of the study population were collected; continuous variables are summarized using mean and standard deviation (SD) and categorical variables as proportions. For each criterion, the proportion of missing values was also recorded. Descriptive summaries of subgroup populations defined by sex, countries, respondent type, and DCD status were compared using Pearson’s Chi square test or, when values in one of the cells of the contingency tables was less than 5, Fisher’s exact test.

We used hierarchical cluster analysis to describe the pattern in which the diagnostic criteria occurred together. A distance metric based on correlation between variables and then between clusters was used to define groupings of diagnostic criteria using a hierarchical clustering algorithm.18 Prevalence of each cluster was then calculated. Analyses were conducted using R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria). The clustering method for this study was chosen to accommodate the qualitative type of the outcome variable, i.e., determination of death. The aim of the analysis was to find a partition of a set of qualitative variables such that the variables within a cluster are strongly related to each other. We used an ascendant hierarchical clustering algorithm method for clustering variables using the decrease in homogeneity (i.e., dissimilarity measure) for the clusters being merged as our aggregation criteria. In this method, each object is assigned its own cluster initially and then the algorithm proceeds iteratively, joining the two clusters at each stage with the smallest dissimilarity measure, continuing until there is just a single cluster. The homogeneity of a cluster was defined as the sum of the correlation ratio between the variables and the center of the cluster. A center of a cluster for categorical variable is the first principal component calculated using mixture of ordinary principal component analysis and multiple correspondence analysis. The algorithm builds a hierarchy of the variables of interest represented as a dendrogram in the Figure. The dissimilarity measure is represented by the height variable on the x-axis of the dendrogram.

Figure
figure 1

Cluster analysis

Full size image

Results

A total of 631 patients were enrolled between 1 May 2014 and 1 May 2018 in the DePPaRT study.17 Forty-nine patients were removed because of incomplete data, and 582 death determination checklists were analyzed (Table 1). The mean (SD) age was 64 (15) yr and 226 (38.8%) patients were female. After withdrawal of life sustaining measures, 52 patients (8.9%) proceeded to DCD. There were 314 (54.0%) from 16 units in Canada, 230 (39.5%) from four units in the Czech Republic, and 38 (6.5%) from a single unit in the Netherlands.

Table 1 Study population characteristics
Full size table

The most commonly used diagnostic tests for the cohort as a whole included absent heart sounds by auscultation (81.8%), a flat invasive ABP tracing (77.0%), and flat ECG tracing (73.2%), followed by absent pulse by palpation (69.6%), absent breath sounds by auscultation (69.2%), and absent pulse oximetry (63.6%) (Table 1). Assessment of fixed and dilated pupils and unresponsiveness to pain were reported 57.4% and 51.5% of the time, respectively.

A hierarchical clustering analysis was conducted to determine which different diagnostic tests tended to be used together. Clinical signs such as absent heart sounds by auscultation, absent breath sounds by auscultation, and absent pulse by palpation were used together in 59% of the cases (Figure). Similarly, monitoring indicators, such as a flat arterial line, absent pulse oximetry, and flat ECG tracing, were used together in 53% of the cases. Neurologic indicators such as fixed and dilated pupils and unresponsiveness to painful stimulus were used together in 47% of cases and combined with clinical signs in 42% of cases.

There were differences in diagnostic tests used in different countries. In all three countries, absent heart sounds by auscultation were reported at least 80% of the time (79.9%–87%) and a flat invasive ABP tracing was reported at least 73% of the time (73.2%–95%, Table 2). Absent pulse by palpation and absent breath sounds by auscultation were reported approximately 74% of the time in Canada but less often in the Czech Republic and the Netherlands (ranging from 61.3% to 68%). In contrast, a flat ECG tracing was commonly reported in the Czech Republic and the Netherlands (87%), but only 61.5% of the time in Canada. Absent pulse oximetry was reported in 82% of the cases in the Netherlands, compared with 59.2% of cases in Canada and 66.5% of cases in the Czech Republic. Pulseless electrical activity was used 63% of the time in the Netherlands, 44.9% of the time in Canada, and 6.1% of the time in the Czech Republic.

Table 2 Diagnostic criteria by country
Full size table

Fixed and dilated pupils and unresponsiveness to painful stimulus were reportedly used in 51.0% and 43.9% of cases in Canada, respectively, compared with 63.5% and 62.6% in the Czech Republic. Fixed and dilated pupils were used 73.7% of the time in the Netherlands, while unresponsiveness to painful stimulus was used 47.4% of the time. Absent blood pressure by noninvasive monitoring and absent pulse by Doppler were rarely used in any country.

Diagnostic tests used for death determination varied by the clinician category. Residents and fellows most commonly used the clinical signs of absent heart sounds by auscultation (~90% of cases), absent breath sounds by auscultation (~80%), and absent pulse by palpation (~80%) (see Table 3). This can be contrasted with the most responsible physicians, who tended to use the monitoring indicators of the flat invasive ABP tracing (81.5%) and flat ECG tracing (81.5%), in addition to absent heart sounds by auscultation (78.0%). Fellows were more likely to report using these monitoring indicators than residents were (77.8% and 68.5% vs 61.5% and 58.0% for the flat arterial line and flat ECG tracing, respectively). The flat arterial line was the primary criterion reported by nurses (87.3%), followed by absent heart sounds by auscultation (77.8%). Nurses used flat ECG tracing, absent pulse by palpation, absent breath sounds by auscultation and absent pulse oximetry equally often (67%–73% of cases). Fixed and dilated pupils and unresponsiveness to painful stimulus were used more than half the time by residents, fellows, and the most responsible physician, but were used 12.7% of the time by nurses.

Table 3 Diagnostic criteria by respondent (N = 582)
Full size table

Diagnostic tests for death determination also varied significantly based on the patient’s donor status (see Table 4). Of the 182 (31.3%) patients who met eligibility criteria for DCD, the procedure was attempted in 75 patients and 52 (69%) of attempted DCD patients successfully donated organs. Patients in whom DCD was successful had death determined using a flat invasive ABP tracing 94% of the time (49/52), compared with patients in whom DCD was attempted but not successful (11/23, 48%) or non-DCD patients (388/508, 76.4%). Absent heart sounds by auscultation were used 52% of the time for successful DCD patients, compared with 91% for unsuccessful DCD patients and 84% for patients in whom DCD was not attempted. Absent breath sounds by auscultation were also less likely to be used for successful DCD patients (52%) than for unsuccessful (78%) and non-DCD patients (70.6%). In contrast, absent pulse oximetry and pulseless electrical activity were used to determine death in successful DCD patients 85% and 71% of the time, respectively, compared with unsuccessful DCD patients (44% and 35%) and patients in whom DCD was not attempted (62.3% and 26.4%, respectively).

Table 4 Diagnostic criteria by DCD status (N = 582)
Full size table

Within the Canadian patient cohort (N = 314), diagnostic tests used for death determination again varied by donor status. In Canadian patients who successfully donated organs by DCD (n = 45), death was most often determined using flat arterial line (96%), absent pulse oximetry (87%), and absent pulse by palpation (80%), whereas Canadians in whom DCD was not attempted (n = 247) had death determined most often using absent heart sounds by auscultation (84.6%), absent breath sounds by auscultation (79.4%), and absent pulse by palpation (72.9%). For those in whom DCD was attempted unsuccessfully (n = 22), the most frequently used criteria were absent heart sounds by auscultation (91%), absent breath sounds by auscultation (77%), and absent pulse by palpation (68%).

Discussion

Our study showed that the self-reported diagnostic tests used by respondents to determine death varied by country, profession (nurse, physician, physician seniority), and DCD status. The most common diagnostic tests reported for the group as a whole were absent heart sounds by auscultation (81.8%), flat invasive ABP tracing (77.0%) and flat ECG tracing (73.2%). In patients who successfully underwent DCD, death was determined most frequently using a flat invasive ABP tracing (94%), absent pulse oximetry (85%), and absent pulse by palpation (77%), as recommended by current Canadian guidelines. Death was determined using a flat arterial bood pressure tracing in 94% of successful DCD patients compared with 77% of non-DCD patients. For all patients, clinical cardiorespiratory examinations of absent breath sounds by auscultation, absent pulse by palpation, and absent heart sounds by auscultation were used together 59% of the time, while monitoring indicators such as a flat arterial line, absent pulse oximetry, and flat ECG tracing, were used together in 53% of the cases. In all three countries, absent heart sounds by auscultation and/or a flat arterial line were used at least 73% of the time, with much more variable use of other diagnostic tests. Fellows (critical care physicians in training) tended to use clinical signs more than most responsible physicians and nurses, who seemed to rely more on monitors.

Though most clinicians in Canadian sites seem to practice according to the standards within the "National recommendations for donation after cardiocirculatory death in Canada,”13 there was some variability. We hypothesize that over time, hospital and regional standards and practice may deviate because guidelines are often outdated or less accessible over time. New evidence may impact consensus recommendations and their validity. If true, guidelines need to be routinely updated and knowledge translation methods should be prioritized to ensure consistency in practice. Reassuringly, a flat arterial line was used almost uniformly, but heart and breath sounds were only used half the time, signaling reliance on technology in situations where monitors were available and timely assurance of death determination was required. Omitting clinical criteria might risk errors in death determination if monitors are inaccurate.

As DCD programs expand internationally, it becomes more important to develop clear, uniform, and consistent evidence-based guidelines to determine death that are updated at regular intervals. Without such guidelines we may not be able to adequately address the medical, ethical, and legal obligations to ensure public trust. A review of guidelines for death determination by circulatory criteria19 showed a lack of consistency in the criteria, the procedures/tests used to satisfy the criteria, and the wait period specified to show that the criteria have been met. Similarly, a study conducted by Wind et al.20 concluded that there was an expected relatively large variability in European DCD death determination protocols because of differences in legislation, expertise, experience, and organizational factors. Nevertheless, as medical education evolves and new guidelines are created, we are unable to say if these findings hold true.

We found that health care professionals tended to use multiple criteria to determine death, which are clustered together. Generally, these clusters could be categorized into neurologic criteria, cardiac/circulatory criteria, clinical examinations, and monitor data. This suggests that health care professionals may be using algorithms in which similar criteria are more frequently used together to determine death.

This study provides potential insight into the way experience shapes a given health care professional’s practice. We found that physicians in training appeared to rely more on clinical examinations whereas senior doctors appeared to rely more on clinical monitors for death determination. This may reflect a pattern of clinical evolution experienced by doctors as they gain knowledge and experience with regards to clinical assessment. Nevertheless, additional research is required to discover the origins of these differences in practice. Additionally, this finding speaks on the broader theme reflected originally in the survey conducted by Dhanani et al.15 that most physicians do not receive standardized formal training on determination of death, almost unanimously agreeing that further training, research, and written guidelines are needed.

This study had several limitations. Because the participating sites were self-nominated, the study findings are representative of a convenience sample of select sites in Canada, the Czech Republic, and the Netherlands, rather than objective, international evidence. While there were 16 sites included from Canada, there were fewer from the Czech Republic (three) and the Netherlands (one), allowing for the capture of more potential diversity in practice across the Canadian sites than the other two countries. Only eight DCD attempts (all successful) were included from the Czech Republic (four) and the Netherlands (three). As this study of death determination practice is a secondary analysis of prospectively collected data, we are unable to determine the cause of the variability. Questionnaire responses were self-reported as opposed to observational. The responses were also subject to variability in response because there was no explicit description regarding instructions on using the critical first criterion/criteria for death determination as opposed to including additional criteria observed later on. Additionally, the questionnaire prompt leaves room to interpret “criteria used” as any required criteria versus criteria that were actually done/measured. This may have led clinicians to be too inclusive of which criteria to include in their response. We recognize that the same responder may have participated in multiple death declarations at each study site. We were not able to account for this in our analysis as the questionnaire was completed in an anonymous fashion. This may have lead to bias, with over-representation of expert individuals. Though we were able to show distribution of sites, were not able to show distribution of individual respondents. We did not attempt to correlate questionnaire responses with their local standards/protocols.

Conclusion

This study provided insight into the practice of death determination by circulatory criteria related to country, profession, and DCD status. Though some variability exists, we are reassured that appropriate diagnostic tests are almost always used in the context of organ donation. In particular, the use of continuous ABP in DCD was consistent. Our findings emphasize the need for standardization of practice, especially within the context of DCD where there is both an ethical and a legal requirement to adhere to the dead donor rule, while minimizing time between death determination and organ procurement. Practice recommendations that are up to date, evidence-based, and properly disseminated are important to standardize practice and foster trust in deceased organ donation practice. Further research is required to investigate and describe the value of specific death determination criteria to move closer to the goal of developing such standardized protocols.