Cessation of Resuscitation in Pulseless Electrical Activity Out of Hospital Cardiac Arrest: A Mixed Methods Study of Senior Paramedic Decision Making

Pulseless electrical activity, a non-shockable cardiac arrest, is treated using advanced life support resuscitation. Typically, survival rates are poor and there is a paucity of evidenced-based guidelines for paramedics on when to cease resuscitation. This led to one UK Ambulance Service developing a local guideline to support senior paramedics when making cessation decisions. This study aimed to describe the patient characteristics, clinical features and system factors of pulseless electrical activity and explore the experiences of senior paramedics making autonomous cessation decisions. conicting and


Background
Pulseless electrical activity (PEA) (previously known as electromechanical dissociation), a non-shockable cardiac arrest rhythm, is where the heart generates electrical activity, but muscle contractions are absent or too weak to generate a pulse.(1) PEA is the rst documented rhythm in 31% of all out of hospital cardiac arrests in the UK, with 4.7% of patients surviving to hospital discharge.(2) Paramedics treat PEA with advanced life support (ALS), however, when treatment fails, patient survival is unlikely.(3) Given the low probability of survival, for some patients there may be a point where it is reasonable for the paramedic to cease resuscitation.
Prognosticating resuscitation outcomes was previously investigated to identify patients who bene t from continued resuscitation and those where resuscitation cannot succeed. (4,5) Early recognition of cardiac arrest, basic resuscitation, de brillation and post resuscitation care, (the chain of survival) when effective, can optimise patient survival. (3) In contrast, duration of cardiac arrest, co-morbidities, absence of a reversible cause and the return of spontaneous circulation (ROSC) should be considered when ceasing resuscitation.(3) Whilst senior paramedics may consider these patient characteristics and clinical features, UK guidelines state the evidence-base for PEA is limited. How senior paramedics decide to cease resuscitation for PEA has not previously been explored. (6) Methods

Study Aim
This study aimed to describe PEA patients where resuscitation was ceased and explore the experiences of senior paramedics who provide remote telephone support to making autonomous cessation of resuscitation decisions. Objectives 1. To describe the patient characteristics, clinical features and system factors of PEA patients where resuscitation was ceased.
2. To examine the associations between variables known to optimise survival. 3. To explore the perceptions and experiences of senior paramedics autonomously making cessation of resuscitation decisions.

Design and Setting
This study employed an explanatory sequential mixed methods design. It was conducted in a single Ambulance Service NHS Foundation Trust providing emergency and urgent care services across an area of 10,000 square miles and a population of 5.5 million.

Methods
Cardiac Registry data was retrieved for all adult patients suffering an Out of Hospital Cardiac Arrest. Cases subject to a coronial or police investigation were excluded. A consecutive sample of 57 patients reported PEA as the rst documented rhythm and senior paramedics ceased resuscitation in 50 patients which were reviewed in detail. Results from analysis of these data were used to inform semi-structured interviews. Interviews were conducted with senior paramedics who provide remote clinical support to on-scene paramedics.

Patient and public involvement
The Ambulance Service Patient and Public Involvement group, which included one cardiac arrest survivor supported this study. The term 'futile' was identi ed as meaning beyond all hope of patient recovery. The proposed research question and methodology was found appropriate. Group consensus was that paramedics should make cessation of resuscitation decisions based on individual circumstances. The group agreed they would not want resuscitation continued if their quality of life was expected to be poor.
Professional consultation with ve research paramedics supported the study protocol. They felt paramedics should be empowered to cease resuscitation for PEA, supported by clear guidelines. They highlighted some paramedics may autonomously stop resuscitation and this would reduce the sample size. Bias may be introduced due to the quality of the retrospective data. A variety of interview platforms were proposed suitable for shift work commitments and anonymity assured due to the sensitive nature of resuscitation decisions.

Data Collection and analysis
Patient data The Ambulance Service redacted identi able patient information. Retrospective consent was not required as data was fully anonymised. Senior paramedics completed a cessation of resuscitation checklist for all PEA patients when resuscitation was ceased. All checklist data from December 2015 to December 2018 was screened against the inclusion criteria; adult cardiac arrest patients with PEA. Patients below 17 years, no PEA or those subject to coroner's court or a police investigation were excluded. Checklist data was cross referenced for accuracy for against clinical records, the computer aided dispatch system and the cardiac arrest registry. Checklist variables identi ed for analysis are presented in Table 1. Descriptive statistics were applied to all variables identi ed in Table 1. We hypothesised there may be a difference in patients presenting with factors known to prognosticate survival (chain of survival: witnessed cardiac arrest, BLS and de brillation) compared to those patients without. Independent t-test was used to examine duration of ALS between de brillation/no de brillation groups and between ROSC and non-ROSC groups. Chi Square examined association between witnessed cardiac arrest, BLS and de brillation and ROSC. No effect size was required for this study and therefore no power calculation was applied. Statistical analysis was conducted using SPSS version 24. (7) Semi-structured interviews Six senior paramedics responsible for providing remote clinical support to on-scene paramedics were recruited as an expert sample during February and March 2019. The results from the quantitative phase informed the semi-structured questions for the qualitative phase. Interview questions were piloted with a critical care paramedic with one minor change required. Five interviews were by telephone, and one interview was conducted by email.
The study was underpinned by a pragmatic epistemology and critical realism ontology. Qualitative data was descriptive and applied content framework analysis. Existing theory was identi ed using deductive analysis and new theory identi ed using inductive analysis. Data was managed and constructed using NVIVO 11.4.2. (10) Re ective validation was applied to the dialogue between the interviewer and interviewee to reduce bias and increase the trustworthiness and credibility of the data.(11) Re exivity was supported using a diary to bracket assumptions and minimise the researchers in uence during data collection and analysis. Data was cross coded by two independent researchers. The consolidated criteria for reporting qualitative research checklist was applied. (12) Ethics Institutional ethics approval was provided by the Trust (ref 17-018) and the University (reference 1718424).

Patient data results
From the 1 st December 2015 to the 31 st December 2018, senior paramedics were contacted by the on-scene paramedic for clinical support for 57 PEA patients and resuscitation was ceased in 50.  and extended hospital transfer times were the most considered system factors (n=14, 8.5%). These results identi ed a number of PEA patients that presented with factors known to prognosticate survival, witnessed cardiac arrest, bystander resuscitation, de brillation, in addition to normal ETCO 2 values, and ROSC, albeit intermittent. These results informed the semi structured interview questions to examine why resuscitation was ceased (Appendix 1).

Interview ndings
The senior paramedics recruited were mixed gender, median age 40, ranging from 35-47 years. Average clinical experience was 14 years and 'on call' rota experience ranged from six months to nine years. Framework content analysis generated 349 codes, 12 sub-themes and four main themes: (1) De ning a futile resuscitation (2) The impact of ceasing resuscitation (3) perceived con ict between senior and onscene paramedics (4) supportive tools for ceasing resuscitation. Multifactorial decision-making informed by clinical judgement remained uid throughout the data. An overview of themes and sub-themes with illustrative quotes is presented at Table 5. "I may feel more con dent to say actually if it was an older person or more of the elder category then to consider that that might be their natural end of life". (SP3) The impact of ceasing resuscitation Con dence gained with experience and exposure "I am con dent in making a decision yes I've been making it for enough years so I feel con dent in making it…. is it always clear cut? The weight of responsibility when deciding futility "it did have its moments going about your daily business and you think bloody hell that is a bit hardcore so it's very easy writing out a word document on your form and lling it in but when you actually put someone's name in it and all their bits and pieces its quite oh actually someone just died and sometimes that was more stressful more poignant that being there". (SP4) Perceived con ict between senior and on-scene paramedics Con ict between senior and onscene paramedic "they are trying to sell me a situation to ful l their own agenda so their selling me a patient that's profoundly unwell when there actually a survivable aspect to it because it is frustrating". The 'checklist' as a safety net "I think the fallback option which is used is just follow a checklist in order for safe practice". (SP1) Checklist deviation and sound clinical reasoning "cessation of resuscitation checklist which we should use which helps govern our decision making……if a patient doesn't fall within that then we have to be pretty con dent and be very careful about calling a cessation of resuscitation attempt". (SP1) Checklist and moving forward 'I do think the checklist that we are using is almost double negative in terms of the questioning and it's still a little ambiguous in some areas". (SP3) Theme 1. De ning a futile resuscitation Senior paramedics applied a multifactorial approach in order to identify a futile resuscitation and subsequent cessation. Advanced age, co-morbidities, persistent PEA, low ETCO 2 and prolonged resuscitation duration were considered alongside a perceived poor life quality or the possibility of a natural end of life. All were balanced to provide a clinical view of the patient, enabling senior paramedics to apply their knowledge, experience and acumen to determine the resuscitation as futile: "what is our threshold and why are we resuscing people who've just reached the end of their life -that's not a reversible event". (SP5) Theme 2. The impact of ceasing resuscitation Repeated exposure to ceasing resuscitation increased con dence and the decision-making process became more comfortable over time; however, senior paramedics highlighted that the decision to cease was not easy or clear: "I got to a position of con dence sometimes I had to work at it I never felt it was a wrong decision to make or that I was second guessing myself once I've hung up but on occasion it took me a while to get there". (SP4) Theme 3. Perceived con ict between senior and on-scene paramedics Senior paramedics felt remote clinical decision-making provided a degree of separation from the scene of the cardiac arrest. At times senior paramedics felt con ict between themselves and the on-scene paramedic due to a perceived lack of accountability and responsibility when making cessation of resuscitation decisions: "I think that's an element of accountability paramedics in general aren't very comfortable with". (SP1)

Supportive tools cessation of resuscitation decisions
Senior paramedics highlighted that the PEA cessation checklist worked as a supportive decision tool. Deviation from the checklist was possible with careful considered clinical judgement. This introduced subjectivity between senior paramedics when ceasing resuscitation and as a consequence, variation to the decision-making process: "we've got the checklist we can still step outside the check list… but I think if you've got all the SCA's in a room and ran some scenarios I'm not entirely convinced they'd make the same consistent decision". (SP5)

Discussion
This mixed methods study was employed to describe PEA patients when resuscitation was ceased, the association between variables known to prognosticate survival and the autonomous experience of senior paramedics making cessation of resuscitation decisions. Synthesis of the quantitative and qualitative data summarised three categories 1) PEA within a local context, 2) senior paramedic experience of ceasing resuscitation 3) multifactorial decision-making.

PEA within a local context
In this study support from senior paramedics was obtained in only 6% of PEA patients who died at home: Patients may have transitioned from PEA to asystole and managed according to existing termination of resuscitation criteria, clinical advice was sought elsewhere or the on-scene paramedic made an autonomous decision to cease resuscitation. Paramedic-led decisions for cardiac arrest was previously explored in a qualitative study which found personal beliefs and values in uential, (14) however in this study, on-scene paramedic cessation of resuscitation decisions were not explored and therefore require further investigation. Previous studies found paramedics felt training to identify those patients with a chance of survival or when recovery is beyond all hope (futile resuscitation), inadequate. 8 This view was supported by the senior paramedics, who highlighted the need for improved paramedic education, speci c to PEA and resuscitation decisions. This nding may explain why some con ict was encountered between the senior and on-scene paramedic as the clinical reasoning as to why the resuscitation was ceased was different. This nding is similar to that of previous studies which found con ict between staff groups due to personal beliefs, moral practice, pressure and expectation. (8,16) Senior paramedics highlighted the challenges of remote clinical support. The on-scene paramedic was required to communicate patient information by telephone, adding to the pressures and time critical nature of resuscitation. The on-scene paramedic was able to build a patient history, including the environmental context, enabling a visual interpretation of the patients physical condition, including frailty. That said, the bene ts of remote clinical support were considered by the National Institute of Health Care Excellence, nding that unnecessary conveyence, speci cally in remote areas could be reduced. (17) The bene ts of remote clinical support in this study found that senior paramedics had time to balance the bene t versus harm of continuing or ceasing resuscitation, which in turn reduced human factors and bias.

Paramedic experiences of ceasing resuscitation
In this study, remote clinical decisions were supported using a locally derived checklist. Deviation from the checklist occurred when senior paramedics applied clinical judgement to cease the resuscitation, as patient recovery was felt beyond hope. Checklists to support clinical decisions in a previous study were found insu cient, as they omitted the multiple factors were required to build a holistic view of the patient. (19) This nding was congruent with the results of this study, as senior paramedics balanced multiple factors using clinical judgement, a vital process when deciding to continue or cease resuscitation.
In total senior paramedics documented 161 different rationales for ceasing resuscitation with multiple, yet different rationales for each patient. This result suggests senior paramedics introduced an element of subjectivity to the decision-making process. The decision to cease resuscitation, informed by their clinical judgement was found to be underpinned by their knowledge, experience and acumen, similar to that of a previous study. (9) Within the literature, this variability in practice was found to be acceptable, as resuscitation decisions were found to be unique to the circumstances of the cardiac arrest. (20) Senior paramedics felt patient survival was more likely when the cardiac arrest was witnessed, with bystander BLS, progression to a shockable rhythm, de brillation and ROSC, similar to factors which prognosticate in the chain of survival. Witnessed cardiac arrest and bystander BLS were statistically signi cant, similar to an observational study which found the recognition of cardiac arrest positively associated with bystander BLS.(10) Whereas cessation of resuscitation in a natural end of life, co-morbidities, advanced age, poor life quality, prolonged no ow time, no bystander resuscitation and persistant PEA, were more comfortable decisions as the patient had no realistic hope of survival.
Senior paramedics interpreted a number of clinical features in which to prognosticate the patients chance of survival. Duration of resuscitation was key, albeit not a statistically signi cant result. The majority of senior paramedics felt after 20-40 mins of ALS resuscitation, if signs of life were absent, the patient was beyond survival. Previous research was unable to identi ed an optimal duration of resuscitation for PEA, reporting anywhere between 10 to 47 mins. (21,22) In this study the mean ALS duration (54 minutes) was longer when compared to senior paramedic views and no assocation with de brillation and ROSC was found. The average ETCO 2 in this study was 2.3kPa. Senior paramedics did not base cessation decisions on ETCO 2 alone, however, a persistant low value was interpreted as a poor indicator. A systematic review applied ETCO 2 to prognosticate resuscitation futility and whilst associations between survival were found, a speci c cut off value could not be determined. 23 A slow heart rate with a wide QRS morphology, found in 92% of patients in this study was considered a terminal rhythm. These results were similar to one study which found a higher death rate in patients with a low, wide QRS, (24) however, incongruent with another study which reported ECGs as a prognostic marker inaccurate. (25) In this study, 44% of patients progressed to a shockable rhythm and 16% achieved ROSC, albeit intermittent. Previous studies found progression to a shockable rhythm (26,27) and ROSC (6,7) increased the chance of survivability. This result suggests senior paramedics ceased resuscitation in patients with known factors which prognosticate survival.
This nding requires further investigation as there may be a difference between patients presenting with known factor from the chain of survival, where resuscitation was not successful, compared to those patients beyond all hope of survival, known as futile. In addition, investigating a value or threshold of the clinical features would be helpful to determine the difference between those patients where resuscitation should be continued and in those patients where resuscitation should be ceased.

Multifactorial decision-making
In this study, the multifactorial nature of decision-making was not exclusive, similar to that of other studies. (15) No association between patient characteristics or system factors were found, however, senior paramedics felt the interpretation and balance of each was vital to the decision-making process. This approach was important for patients with a perceived poor quality of life, multiple co-morbidities or when establishing a natural end of life event. Senior paramedics acknowledged the subjectivity of interpreting the impact of comorbidities and life quality, however, if the chance of survival with a good functional recovery was low, the resuscitation was ceased. In a previous study, co-morbidities were found to negatively impact on patient survival. (13,28,29) Arguably, co-morbidities do not infer a poor quality of life, however, assessing this remotely, whilst providing time critical care was challenging. Senior paramedics did identify advanced age, frailty, terminal disease, informed by family wishes as a natural end of life. As this is not a reversible cause of cardiac arrest the decision to cease resuscitation was made with con dence. Senior paramedics were well supported within the literature to interpret these factors to enable a balanced decision. (28,29,30) Previously only clinical features have been investigated to prognosticate the outcome of resuscitation. (4,13) This approach was not supported by the senior paramedics. The clinical features previously described were interpreted, however, they were balanced along with no ow time, advanced age, co-morbidities, duration of resuscitation and quality of life. This multifactorial approach was felt necessary, to ensure a hollistic, best interest and individual decision, a view supported within the literature (13,19) Conclusion Senior paramedics interpreted and applied a multifactorial approach to ceasing resuscitation for patients in PEA. Patient characteristics, clinical features and system factors were balanced using clinical judgement, found to be vital to the decision-making process. This hollistic approach questions the the appropriateness of paramedics making cessation of resuscitation decisions based on clinical features alone. As prognostic factors for survival were present within the sample, further investigation is warranted to examine the difference between those patients unsuccesfully resuscitated and those patients identi ed as futile.

Limitations
The quantitative sample size was small due to the limited involvement of senior paramedics. This may have introduced a type II error, where results lacked the statistical power to detect a difference that may exist. No power calculation was applied. Missing data was identi ed for ETCO 2 (12%) and the effectiveness of bystander resuscitation or by whom it was conducted was not measured. One of the senior paramedic interviews was conducted by email, with no opportunity to probe the interviewee's responses.

Consent for publication
All data is fully anonymised. Written consent of was gained for each participant, including for publication.
Availability of data and materials