Assessing the quality of ReSPECT documentation using an accountability for reasonableness framework.

BACKGROUND: The Recommended Summary Plan for Emergency Care and Treatment (ReSPECT) form, which supports the ReSPECT process, is designed to prompt clinicians to discuss wider emergency treatment options with patients and to structure the documentation of decision-making for greater transparency.
METHODS: Following an accountability for reasonableness framework (AFR), we analysed 141 completed ReSPECT forms (versions 1.0 and 2.0), collected from six National Health Service (NHS) hospitals in England during the early adoption of ReSPECT. Structured through an evaluation tool developed for this study, the analysis assessed the extent to which the records reflected consistency, transparency, and ethical justification of decision-making.
RESULTS: Recommendations relating to CPR were consistently recorded on all forms and were contextualised within other treatment recommendations in most forms. The level of detail provided about treatment recommendations varied widely and reasons for treatment recommendations were rarely documented. Patient capacity, patient priorities and preferences, and the involvement of patients/relatives in ReSPECT conversations were recorded in some, but not all, forms. Clinicians almost never documented their weighing of potential burdens and benefits of treatments on the ReSPECT forms.
CONCLUSION: In most ReSPECT forms, CPR recommendations were captured alongside other treatment recommendations. However, ReSPECT form design and associated training should be modified to address inconsistencies in form completion. These modifications should emphasise the recording of patient values and preferences, assessment of patient capacity, and clinical reasoning processes, thereby putting patient/family involvement at the core of good clinical practice. Version 3.0 of ReSPECT responds to these issues.

Introduction

The Recommended Summary Plan for Emergency Care and Treatment (ReSPECT) is a new emergency care and treatment planning (ECTP) process, introduced into more than 150 National Health Service (NHS) hospitals in the UK since late 2016. ReSPECT was developed to overcome the harms associated with do not attempt cardiopulmonary resuscitation (DNACPR) processes. Audits and evaluations of DNACPR forms have shown that they are completed and interpreted inconsistently by medical staff, thereby carrying unintended consequences for patients, such as the denial of other types of treatment.2, 3, 4, 5 These forms are often not accompanied by transparent documentation of decision-making processes, and do not contextualise the DNACPR decision within the patient’s preferences and wider clinical care.6, 7, 8 Additionally, because DNACPR forms tend to be institutionally specific, they cannot be transferred across medical settings.8, 9 Finally, a key challenge in using DNACPR forms is how to prompt clinicians to include patients or their families in decision-making discussions.4, 10, 11

By contrast, the ReSPECT form (Fig. 1), which supports the ReSPECT process, was designed to establish a shared understanding with patients and/or their relatives of the patient’s condition(s), the patient’s values, and what treatments will realistically accomplish the outcomes patients prioritise, while avoiding those they fear. The form, which is carried by patients across medical settings, prompts clinicians to discuss wider emergency treatment options with patients and structure the documentation of decision-making for greater transparency. The form prompts recording of diagnostic information, clinical reasoning and patient preferences, and brings these together in the documenting of clinical recommendations. These include a CPR recommendation (with CPR attempts either recommended or not recommended), preceded by an open text box that prompts clinicians to “provide clinical guidance on specific treatments that may or may not be wanted or clinically appropriate”, thereby emphasizing the wider treatment context in which CPR should be embedded. The reverse side of the form prompts clinicians to assess the patient’s capacity, and to document the appropriate involvement of patients with capacity, or of the relatives or advocates of patients without capacity, in the ReSPECT discussion. Taken together, the sections of the ReSPECT form are aimed at promoting the consistency, transparency, and ethical justifiability of clinical decision-making.

Fig. 1

ReSPECT form, versions 1.0 and 2.0.

The present study, performed as part of the larger ReSPECT Evaluation Study, aims to assess the extent to which ReSPECT forms completed during acute hospital admissions accomplish the ethical aims of the form’s design. The analysis is guided by three key research questions:

Are recorded decision-making processes in ReSPECT forms consistent across patients and clinicians?

Is the recorded decision-making process in ReSPECT forms transparent?

Is the reasoning process recorded in ReSPECT forms ethically justifiable?

Methods

Six hospital sites in England, providing specialist services on a national level, were recruited. The sites were purposefully sampled for diversity in the following categories: approach to the implementation of ReSPECT, Care Quality Commission (CQC) banding (‘needing improvement’ or ‘good’, indicating the hospital’s performance against the independent regulator’s standards for care), admissions volume, hospital type, location, and the ethnicity and socioeconomic status of patient populations served. NHS ethics (reference 17/WM/0134) and Confidentiality Advisory Group (CAG) (reference 17/CAG/0060) approvals were obtained. Participating sites’ research and development departments provided research governance approvals. Hospital characteristics are described in Table 1.

Table 1

Hospital characteristics.

Characteristic	Hospitals (n = 6)
Hospital type	Teaching: n = 4
	District general: n = 2
Urban/rural populations served	Urban populations: n = 3
	Urban and rural populations: n = 2
	Rural populations: n = 1
Affluent/deprived populations served	More deprived than the average in England: n = 4
	More affluent than the average in England: n = 2
Geographical region in England	Midlands: n = 3
	North: n = 1
	South: n = 1
	East: n = 1

The study team developed a sampling frame based on a pilot sample collected in the first study site (n = 20), with the aim of identifying diverse records. Wards were selected by the research team, following discussion with participating hospital research teams, to ensure adequate coverage of a range of wards and clinical specialities. On a designated day, records were collected from all adult in-patients on these wards, unless they opted out. Using these records, in each study site, an NHS researcher (e.g., a research nurse) purposively sampled the records of patients with a ReSPECT form on file through a sampling frame, with the following categories: ReSPECT decision, patient age, type of ward, and emergency or elective admission. Each category had a target of patient records to be sampled (e.g., 4–6 patients age 20–50, 6–14 patients with emergency admissions). Where targets could not be met (e.g., if there were fewer than four patients under 50 with a ReSPECT form), the NHS researcher negotiated a suitable alternative with the research team.

The data collected included both the completed ReSPECT form and sections of the patient’s notes related to the ReSPECT discussion and/or decision, if available; if ReSPECT was not mentioned in a patient’s notes, the NHS researcher submitted the ReSPECT form only. The NHS researchers redacted patient and clinician identifiers (e.g., patient name, address and NHS number, clinician name) from all records, and securely sent these pseudo-anonymous records to the research team via an encrypted NHS platform. Data were collected from July 2017 to January 2020, within the first two years of ReSPECT implementation in each of the participating hospitals. Version 1.0 and Version 2.0 forms were included. Because the sections and items in both versions are similar and directly comparable, this did not affect the analysis (Fig. 1).

We used an accountability for reasonableness (AFR) framework to develop a quality standard for ReSPECT form completion. AFR is an ethical decision-making framework that emphasizes fair decision-making processes rather than particular moral theories. It focuses on transparency of decision-making, with decisions based on reasons that stakeholders can agree are relevant. Using AFR, we developed an evaluation tool that enabled a structured qualitative analysis. Based on the design of a similar tool developed for a previous study on ethical decision-making in critical care admissions, the evaluation tool assigned scores to each section of the ReSPECT form considered relevant to assessing the quality of the recorded decisions. While the tool assigned numerical values to items on the ReSPECT form (e.g., a score of ‘2’ for a detailed response, ‘1’ for a basic response, and ‘0’ for no response), it was aimed at supporting a focused qualitative analysis of a large sample of forms. In constructing the tool, emphasis was placed on the extent to which the records demonstrated consistency, transparency, and ethical justifiability (Table 2).

Table 2

Criteria for assessing completed ReSPECT forms and associated patient records.

Criterion	Definition	Assessed sections of the completed ReSPECT forms and patient records
Consistency	Whether clinicians completed the form, including treatment recommendations, to the level of detail requested on the ReSPECT form and associated guidelines,18 across patient records.	All sections
Transparency	The extent to which information was provided about diagnosis, patients’ wishes and preferences, reasons for recommendations, and who took part in the discussion.	Section 2: summary of relevant information
		Section 4: clinical recommendations for emergency care and treatment
		Section 6: involvement in the plan
		Relevant sections of patient notes
Ethical justifiability	(1) Whether the reasons given for the recommendations explicitly took into account the patient’s current clinical condition and ability to benefit from future interventions, as well as the patient’s wishes and preferences (either directly sought from patients with capacity or indirectly through discussion with a legal proxy, family or friends).(2) Whether, in taking these factors into account, the clinician completing the form had documented the weighing of potential burdens and benefits of treatments for this particular patient.	Section 2: summary of relevant information
		Section 3: personal preferences to guide this plan
		Section 4: clinical recommendations for emergency care and treatment
		Section 5: capacity and representation at time of completion
		Section 6: involvement in the plan
		Relevant sections of patient notes

The tool was developed and refined over several meetings of the study team. The first version was developed by Author 8 and Author 9, general practitioners with expertise in medical ethics and medical sociology, respectively; Author 2, a health services researcher with a background in critical care nursing; Author 3, an acute medicine consultant; and Author 1, a medical anthropologist. Each version was tested through the team’s inter-scoring of the pilot records. To finalize the evaluation tool, Authors 1, 2, 8 and 9 met for two analysis sessions. Working with a sub-sample of 30 completed forms, each form was scored by two of the co-authors. Through this process of continuous trial and refinement, the team agreed on clear guidelines for assessing the quality of each section of the ReSPECT form (Table 3).

Table 3

Examples of guidelines for the evaluation of the completed ReSPECT form.

Item	Evaluation guidelines
Section 2, “reasons for the preferences and recommendations recorded”	Reasons should be specific (e.g. ‘because the patient is unlikely to benefit from invasive ventilation, this treatment is not recommended’) for the item to score a point. If reasons are non-specific (e.g. ‘treatment likely to be futile’), the item does not score a point.
Section 4, “Now provide clinical guidance on specific interventions that may or may not be wanted or clinically appropriate, including being taken or admitted to hospital ± receiving life support”	For the item to score two points, specific interventions (such as antibiotics, dialysis, intravenous fluids, non-invasive ventilation) should be mentioned. If only general instructions (e.g., not for ITU, for ward based care) are mentioned, the item scores one point.

Using the final version of the evaluation tool (Supplementary Table 1), Author 1 analysed all forms, while Authors 2, 8 and 9 provided inter-scoring, each analysing about a third of the forms. Rather than test for inter-rater reliability, inter-scoring was undertaken to inform and enhance the discussion of the findings. Additionally, each co-author took notes about findings within the completed forms that required further discussion. The team then met for comparative analysis and discussion of findings, and achieved consensus on key findings in the three areas of interest.

Findings

A total of 162 completed forms were collected. Twenty-one were excluded: two forms were duplicates, and all 19 forms from one site incorporated ReSPECT into an existing record-keeping system, and could not be directly compared to the other ReSPECT forms in the sample. Thus, 141 forms from five NHS hospitals were included (Table 4).

Table 4

Sample characteristics.

ReSPECT form version (n = 141)		Format (n = 141)		Patient notes (n = 141)
Version 1.0	Version 2.0	Paper	Digitized	Available	Missing
n = 87 (61.7%)	n = 54 (38.3%)	n = 121 (85.8%)	n = 20 (14.2%)	n = 113 (80.1%)	n = 28 (19.9%)

Consistency of recorded decision-making processes

Recommendations to provide or withhold CPR were consistently recorded on all forms. Recommendations to focus on life sustaining treatment or symptom control (section 4) were completed (signed on paper forms or selected on digitized forms) on 90 forms (63.8%) and free text recommendations on other treatments were recorded in most cases (84.4%). However, recommendations for specific interventions substantially varied in detail and clarity. Just under half of the forms (n = 65, 46.1%) mentioned specific treatments, (e.g., antibiotics, inotropes, non-invasive ventilation) while other forms (n = 54, 38.3%) provided only general instructions (e.g. “not for ITU”, “for ward based care”). Thus, the quality of documentation was inconsistent.

Inconsistencies were observed in the documentation of who took part in the ReSPECT discussion (Section 6). Most forms included some information on who was involved in making the plan (n = 97, 68.8%). However, interpretations varied across the forms. For example, in forms that indicated the patient lacked capacity, just over half (n = 31, 54.3%) included the names or roles (e.g., son, daughter, wife) of the relatives who participated in the discussion, while others (n = 26, 45.6%) included only the names of the doctors involved in the discussion, or no information at all.

Transparency of recorded decision-making processes

Most forms (n = 121, 85.8%) included lists of chronic conditions and comorbidities; acute diagnosis and reason for the current admission were included in less than half (n = 57, 40.4%) (section 2). Reasons for treatment recommendations were rarely articulated (section 2), with only 13 forms (9.2%) explicitly stating why a particular treatment was recommended or not. In section 3, where patients’ preferences were stated, this was often brief (e.g., “comfort”, “keep comfortable”). In section 4, forms rarely stated directly that patient preferences informed treatment recommendations, although a few notable exceptions were identified. For example, one form stated: “[Patient] + family do not wish for resuscitation, escalation of care to ITU or invasive ventilation. [Patient] would like to prioritise comfort and symptom control”.

Of the 121 forms that featured a senior responsible clinician signature line (the digitized forms did not have this signature line), 48 (39.7%) included an appropriate senior responsible clinician signature in the correct place. In many cases, the forms were signed by senior clinicians on the clinician signature line but not the senior responsible clinician line, indicating this might be a form design issue (section 7).

Few forms (n = 11, 7.8%) indicated where records of ReSPECT discussions could be found in the main clinical record (section 6), although most records (n = 90, 63.8%) included a mention or description of the ReSPECT discussion in the patient notes. Some forms (n = 44, 31.2%) included no record of who was involved in the discussion. At times, the information included in the notes did not match the information included in the form; for example, notes might have indicated detailed instructions about escalation of care or discussions of palliation which were not included in the form. In some cases, the patient notes recorded who participated in the discussion, while the form did not.

In the available notes for patients who were identified as not having capacity, no formal capacity assessments were recorded. Indirect references to the patient’s capacity (e.g., “delirium”, “unable to discuss with pt given dementia/state of confusion”) were noted in 17 of 38 records (44.7%). Patient notes sometimes indicated a discussion with family members was held after the form was completed, but there was no indication these patients’ ReSPECT forms had been updated to note discussions with relatives were held (section 3).

Ethical justifiability of the recorded reasoning process

Only one form directly referenced the balance of benefits and burdens of treatment, with the reasons given for the recommendations explicitly taking into account the patient’s current clinical condition and ability to benefit from future interventions: “Given good premorbid function, recommend full escalation of treatment”. More broadly, the reason for an emergency care treatment recommendation was explicitly stated in a small minority (n = 13, 9.2%) of records (section 2). Of the forms that provided reasons, most stated that a particular treatment would be futile, for example, “Should patient require CPR – it will prove futile”, or that CPR would not be in the patient’s best interest, e.g., “CPR likely futile + not in patient’s best interests”.

In most forms, the question about mental capacity was completed (n = 122, 86.5%) (section 5). This number included 65 patients recorded as having capacity (53.3%), and 57 patients recorded as not having capacity (46.7%). Among the patients who were recorded as having capacity, the patients’ preferences and wishes were recorded in some cases (n = 34, 52.3% on the scale, and n = 17, 26.2% in the free text box) (section 3). Among the patients who were recorded as not having capacity, the involvement of someone close to the patient (a relative or friend) in the discussion was recorded in just over half of cases (n = 31, 54.4%) (section 6).

Discussion

Through comparing 141 completed ReSPECT forms and associated patient notes, we found that all forms recorded recommendations related to CPR and most included some information about other treatment recommendations, indicating alignment with the stated aim of the ReSPECT process. However, inconsistencies were observed in the level of detail, with some forms listing specific interventions (e.g., “for NIV” [non-invasive ventilation]) and others providing generalised instructions (e.g., “not for ITU”). Shortcomings were observed in the transparency of recommendations, and ethical justifiability was difficult to assess because many forms lacked documentation of reasons for treatment recommendations. Although most forms recorded patients’ capacity, the preferences and wishes of patients with capacity and the involvement of relatives of patients without capacity were recorded in just over half of cases. Evidence of clinicians weighing the balance of benefits and burdens to inform a treatment recommendation was rarely captured.

The ReSPECT process was designed to situate discussions and decisions about cardiopulmonary resuscitation in the wider context of patient treatment planning, to facilitate shared decision-making and to provide clear recommendations for future decision-making. Our review of completed ReSPECT forms suggests that ReSPECT prompts consideration of wider treatment options to some extent. However, our analysis of ReSPECT documentation did not provide evidence that doctors fully engaged in the decision-making logic that ReSPECT promotes. In the great majority of forms, no reasoning was provided for treatment recommendations, thereby reducing the transparency of decision-making. A lack of explicit reasoning in recorded decision-making has also been observed in critical care contexts, suggesting that this phenomenon is not unique to the ReSPECT or ECTP processes. Moreover, the forms’ transparency was compromised by inconsistencies in recording who took part in ReSPECT conversations, often leading to ambiguous and incomplete accounts of whose voices were heard in the discussion, and which clinicians ultimately participated in making the recommendations.

The inconsistencies observed in this sample of completed ReSPECT forms carry implications for their ethical justifiability. The lack of explicit documented reasoning about treatment recommendations entailed an interpretive leap from diagnoses to treatment recommendations, implying that the latter were inevitable rather than subject to a careful weighing of burdens and benefits. Inconsistencies in recording patient values and preferences and family participation in ReSPECT conversations obscured the extent to which patients and relatives were included in the decision-making processes, suggesting that patients’ wishes might not always be included in treatment recommendations. Related to this were inconsistencies in the recording of patients’ capacity. Although patients’ relatives were involved in most cases where patients lacked capacity, the apparent lack of inclusion of relatives in some cases aligned with practices observed in previous research on DNACPR processes.10, 11 This is concerning given the legal requirement to consult with someone close to the patient unless impracticable or inappropriate.

The findings suggest a gap between the ReSPECT form’s design and use. Although the ReSPECT form outlines a decision-making logic through the wording and ordering of the items, with each section calling for the documentation of key steps (e.g., clinical reasoning, patient wishes) leading up to treatment recommendations, hospital doctors do not record their reasoning on the form. Rather, ReSPECT forms are used to record treatment recommendations about escalation of treatment, often abbreviated and using medical acronyms, with reasoning processes left unarticulated. This suggests a tension between doctors’ need to communicate clinical information quickly, definitively and succinctly on an emergency form, and the ReSPECT process’ emphasis on the careful documentation of patient and family involvement in decision-making, wider treatment contexts, and the balance of burdens and benefits entailed in treatment recommendations.

Strengths and limitations

A particular strength is the study’s focus on assessing ReSPECT documentation, which has both legal and patient care ramifications, and may influence the development of ReSPECT and its ongoing implementation. The novel contribution of this tool is the AFR approach and the focus this places on the importance of reasoning and justification of recommendations, informed by considerations of harm, benefit and patient autonomy. Compared to more standard audit tools, this provides a key additional element which better reflects the aims of the ReSPECT process, and will be relevant for future audits of ReSPECT and ECTPs more generally. Another strength is the study’s inclusion of five NHS hospitals, representing different implementation timelines, localities and procedures. However, because data were collected during the first two years of ReSPECT implementation, findings might reflect documentation practices during early adoption. An additional limitation relates to patient notes, which were missing from one-fifth of records. In some cases, the NHS researcher indicated no relevant notes were available, but in other cases no reason was indicated. This limited our ability to assess the extent to which ReSPECT conversations had been documented in the patient notes across the sample. Because we defined relevant sections of patient notes as those that included mentions of ReSPECT, it is also possible that capacity assessments were mentioned in other pages of the patient notes, such that our findings concerning doctors’ documentation of capacity may be incomplete. Finally, because paper forms constituted the great majority of the sample, we could not determine whether digitized forms might carry different implications, though our analysis found that issues related to consistency, transparency and ethical justifiability were similar across the sample, regardless of the medium for form completion.

Implications

The findings suggest that improvements could be made in both form design and ReSPECT training. The design of the form could be reconfigured to compel clinicians to follow the logic of the form; this could be achieved through rewording certain sections (e.g., removing the word ‘optional’ from section 3, on patient values and preferences), reordering certain items (e.g., moving the reasoning item from section 2, where diagnoses are listed, to section 4, where treatment recommendations are recorded), and clearly delineating instructions (e.g., clarifying that the names of all clinicians, family members and advocates who took part in the discussion should be included in section 6). This would clarify each section’s purpose and how the sections build together toward treatment recommendations. Training in the ReSPECT process could be improved by emphasizing the recording of reasoning processes, assessment of patient capacity, patient values and preferences, and patient/family involvement as critical to good clinical practice, and as key to the accurate completion of the ReSPECT form. These recommendations have been fed back to the ReSPECT team as part of their iterative and consultative process of development. Version 3.0 of the form includes changes that reflect these recommendations.16, 17

Conclusion

The analysis of completed ReSPECT forms has shown that the introduction of ReSPECT has had some positive effect, moving the focus away from CPR to include wider treatment recommendations. This is an important step to reduce unintended harms previously associated with DNACPR forms. However, the quality of form completion suggests that clinicians only partially document the decision-making logic ReSPECT promotes. Wide variability in clarity and specificity was observed for treatment recommendations, most forms did not articulate clinical reasoning for the treatment recommendations made, and patient preferences were inconsistently recorded. Our recommendations have contributed to Version 3.0 of the ReSPECT form. The future implementation of ReSPECT should continue to address these issues through form design and training. This is of particular importance against the backdrop of the Covid-19 pandemic, during which concerns about how CPR discussions and decisions are conducted have been raised, highlighting the urgent need for improved guidelines and training in newly-adopting medical practices.

Credit author statement

All authors have made substantial contributions to: (1) the conception and design of the study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, (3) final approval of the version to be submitted.

Conflicts of interest

GDP and CAH are members of the ReSPECT national working group. ZF chairs the ReSPECT subcommittee for the Resuscitation Council (UK). AMS, FG, CAH, GDP and ZF received grants from the UK National Institute for Health Research during the study. KE, CJH, AW and JG have no conflicts of interest to declare.

Ethical information

National Health Service (NHS) ethics (reference 17/WM/0134) and Confidentiality Advisory Group (CAG) (reference 17/CAG/0060) approvals were obtained.