Are Clinical Prediction Rules Used in Spinal Cord Injury Care? A Survey of Practice

Abstract

Background:

Accurate outcome prediction is desirable post spinal cord injury (SCI), reducing uncertainty for patients and supporting personalized treatments. Numerous attempts have been made to create clinical prediction rules that identify patients who are likely to recover function. It is unknown to what extent these rules are routinely used in clinical practice.

Objectives:

To better understand knowledge of, and attitudes toward, clinical prediction rules amongst SCI clinicians in the United Kingdom.

Methods:

An online survey was distributed via mailing lists of clinical special interest groups and relevant National Health Service Trusts. Respondents answered questions about their knowledge of existing clinical prediction rules and their general attitudes to using them. They also provided information about their level of experience with SCI patients.

Results:

One hundred SCI clinicians completed the survey. The majority (71%) were unaware of clinical prediction rules for SCI; only 8% reported using them in clinical practice. Less experienced clinicians were less likely to be aware. Lack of familiarity with prediction rules was reported as being a barrier to their use. The importance of clinical expertise when making prognostic decisions was emphasized. All respondents reported interest in using clinical prediction rules in the future.

Conclusion:

The results show widespread lack of awareness of clinical prediction rules amongst SCI clinicians in the United Kingdom. However, clinicians were positive about the potential for clinical prediction rules to support decision-making. More focus should be directed toward refining current rules and improving dissemination within the SCI community.

Introduction

Spinal cord injury (SCI) is a debilitating condition with significant impact on quality of life.¹ Post injury, it is vital for patients and their families to understand likely prognosis. Patients want to know their potential for mobility, upper limb function, and independence.² Clinicians also need accurate predictions for improved communication, treatment planning, and multidisciplinary decision-making.³

Several tools stratify patients based on their impairments to determine injury severity and enhance clinical communication. The International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI), produced by the American Spinal Injury Association (ASIA), is the standard and involves clinical assessment of motor and sensory function to define the level and extent of injury.⁴ ISNCSCI and other assessments are sometimes used to make predictions about recovery potential. Research shows, for example, that patients graded A on the ASIA Impairment Scale (AIS) component of the ISNCSCI rarely regain ambulatory function, whereas for grade D, more than half may regain the ability to walk.⁵ Predictions are often less accurate for grades B and C.⁶ However, AIS is not a sensitive or precise prognostic tool.⁷

Due to the heterogeneity of patient presentations and challenges in making precise functional predictions, there is strong scientific and clinical interest in reliable prognostic tools for determination of recovery potential.⁵ Predictive factors investigated include clinical measures such as motor scores or AIS^8–11; radiological factors, such as MRI^5,12; cerebrospinal fluid biomarkers¹³; and neurophysiological measures, such as motor^14,15 and somatosensory evoked potentials.¹⁶ The primary functional outcome is typically ambulation, often measured through the Spinal Cord Independence Measure (SCIM)^8,17 or Functional Independence Measure (FIM).¹⁸ Whilst many use regression techniques to understand factors relevant to prediction of outcomes, there is increasing interest in machine learning techniques.^19–21 A subset of research has focused on developing clinical prediction rules (CPR), a grouping of clinical indicators used to predict outcomes or inform clinical decisions.^22,23

A key CPR for SCI prognosis was developed by van Middendorp et al.⁸ This predicts the likelihood of independent ambulation based on five factors: age, motor power in L3 and S1 myotomes, and light touch of L3 and S1 dermatomes. This rule has been independently validated in large data sets.^24,25 Development of the model by Hicks et al.¹⁰ found that prediction accuracy was still good for a three-item rather than a five-item model. However, accuracy of the model is reduced for AIS B and C, suggesting that accuracy is overdetermined by AIS categories A and D, where outcomes are easier to predict.²⁶ Attempts to validate the model for clinical use also identified several problems,¹⁸ and accuracy of the model was found to be no better than that of expert clinicians.²⁷ Other CPRs exist, but most are not well validated and emphasize walking over other outcomes.^17,28 Although more recently, an upper limb CPR has been produced.²⁹ See Table 1 for examples of specific CPRs used in SCI.

Table 1.

Specific clinical prediction rules (CPR)

Study	Outcome of interest	Statistical model	Key predictors identified	CPR accuracy
Van Middendorp et al., 2011	Independent indoor walking ability Assessed by SCIM	Multivariate logistic regression	L3 and S1 motor score; L3 and S1 light touch sensory score; age (<65 vs. >65 years)	AUC: 0.956, 95% CI: 0.936-0.976, p<.0001
Hicks et al., 2017	Independent walking ability Assessed by FIM locomotor score	Multivariate logistic regression	L3 motor score; S1 light touch sensory score; age (<65 vs. >65 years)	AUC: 0.866, 95% CI: 0.816-0.916, p<.01
Jean et al., 2021	Independent outdoor walking ability Assessed by SCIM	Multivariate linear regression	Highest motor score of L3 and L5; preservation of light touch at S1	Classification accuracy: 84.91%; sensitivity: 84.21%; specificity of 85.54%
Hori et al., 2022	Independent living (custom 5-point scale)	Classification and regression tree	FIM total score; AIS grade; FIM bed to chair score; FIM eating score; residual function level	AUC: 0.813 (95% CI 0.787-0.839); classification accuracy: 78.6% (76.380.7); sensitivity: 80.7% (78.9 -82.4); specificity: 75.1% (72.0-78.0)

Note: Clinicians were asked about their knowledge of four specific prediction rules for spinal cord injury (SCI) from the literature. These were CPR reported to be straightforward to use in clinical practice. See reference list for more details. Measures of accuracy were those reported in the relevant article. Respondents could also select “other” and suggest CPR not in this list. AIS = American Spinal Injury Association Impairment Scale; AUC = area under the curve; CI = confidence interval; FIM = Functional Independence Measure, SCIM = Spinal Cord Independence Measure.

Accurate prognostic tools are urgently needed in SCI to reduce uncertainty for patients and enable personalized treatment approaches.³⁰ Given the lifetime specialist management required for SCI³¹ and limited resources of spinal services,³² tools that inform early decisions are vital. Despite attempts to develop prediction rules, their impact on SCI care remains uncertain. This study aimed to address this knowledge gap by examining the use of existing CPRs in clinical practice and exploring the methods clinicians currently use for predicting SCI patient outcomes. The objective was to understand the knowledge of, and attitudes toward, CPR for SCI and general attitudes to using CPR in clinical practice. Clinicians were asked about what other information they use to evaluate likely outcomes in SCI populations. Through understanding current CPR use and clinicians' perspectives, this study may inform the development and implementation of more effective prognostic tools and optimization of SCI care.

Methods

Ethics statement

The study was evaluated using decision tools from the Imperial College London Governance and Integrity Team and was classified as a “service evaluation.” It did not require formal ethical approval. There were no issues of confidentiality because all responses were anonymous. To ensure quality and integrity of the project, the protocol was presented to the Imperial College Healthcare National Health Service Trust clinical governance committee and subsequently registered with the clinical audit team (registration number 816). Before participating, respondents could review an information sheet, with participation indicating implied consent.

Recruitment

A convenience, cross-sectional sample of UK clinical staff involved in SCI care was recruited for this exploratory survey of practice. Clinicians were approached via professional and special interest groups: the Multidisciplinary Association of Spinal Cord Injury Professionals (MACSIP), the Association of Chartered Physiotherapists in Neurology (ACPIN), and the Royal College of Occupational Therapists specialist section in neurological practice (RCOT SSNP). The survey was distributed via email to the respective mailing lists of these groups, with permission. The survey was also shared via contacts in several UK spinal injury units and acute National Health Service (NHS) Trusts treating patients with SCI.

Sample size

Population size was estimated based on membership data from special interest groups previously mentioned. Membership of MASCIP was reported to be 1956 in 2022 with approximately 40% of those being either therapists or doctors and therefore likely to be involved in making functional predictions. The ACPIN mailing list had 1800 members, 148 of whom report working with SCI. The RCOT mailing list had 1462 members but no data on numbers working with SCI. If we assume the proportion of occupational therapists working with SCI is similar to physiotherapists, this represents a population of approximately 1100, acknowledging potential overlap between MASCIP and the other groups. Increasing this to account for therapists working in acute settings treating SCI and those who have worked with SCI in the past, we estimated a population of 2000.

Using a 95% confidence level and a 10% margin of error, the estimated ideal sample size was therefore 92, as per the Qualtrics sample size calculator. A 10% margin of error is slightly narrower than commonly used in survey-based research,³³ because this survey aimed to produce a preliminary description of trends.³⁴ Response rates cannot be accurately calculated as the survey was also distributed through spinal units and acute NHS Trusts so true exposure is unknown.

Survey design

The survey was created by the research team to explore knowledge of and attitudes to CPR in general and SCI specifically. It comprised 32 potential questions covering the following: respondent demographic characteristics (e.g., profession, level of experience with SCI patients), attitudes toward CPRs, awareness and use of existing CPRs (identified from a review of the SCI literature), reasons for using or not using CPRs, other predictors used in clinical practice (e.g., AIS grade, neurological level, etc.), and perceived accuracy of these. Questions were largely multiple choice or Likert scales, but some allowed for free text, enabling respondents to give more detailed responses. To reduce missing data, responses were mandatory for all questions except for the final free-text comment section. However, questions were presented according to survey logic, so respondents would only answer questions relevant to them. For example, those who reported not being aware of any existing CPR were not asked whether they used CPR in their practice. The survey was available from January 2023 until late May 2023; respondents followed an anonymous link; once started, respondents had 7 days to complete the survey. All responses were examined with a custom Python script to check for duplicates by comparing demographic information; none was identified. The possibility of spurious responses cannot be completely ruled out, but as the distribution was via professional networks, we feel this is unlikely.

The survey was hosted on the Qualtrics platform, which allows fully anonymous responses. Details were collected regarding respondents' profession, grade, UK region, type of work setting, total experience working with SCI patients, and proportion of time spent with SCI patients (see eAppendix for an example of the survey document).

The survey was piloted with a smaller group of clinicians and a user experience researcher to ensure clarity and relevance of question items and was updated accordingly. The survey underwent further rounds of revisions in response to clinical governance committee suggestions and after input from health care professional reviewers from the ACPIN special interest group.

Data analysis

Quantitative analysis

Descriptive statistics were generated using the Qualtrics platform and Microsoft Excel; custom Python scripts were used for advanced statistical tests. To explore relationships between demographic factors and attitudes toward, or use of, clinical prediction rules, answers to demographic questions were cross-tabulated with answers about clinical prediction. Apparent associations led to further statistical testing. The Shapiro-Wilk test assessed normality for continuous variables; t tests or Mann-Whitney U tests were subsequently used to examine group differences. Categorical variables were analyzed using chi-square tests. The significance level was set at α = 0.05.

Qualitative analysis

Thematic analysis was conducted according to the key steps as described by Ritchie et al.³⁵ These are familiarization, construction of an initial thematic framework, indexing and sorting, reviewing data extracts, and analysis and interpretation. Initially, the researchers familiarized themselves with the free-text responses and identified preliminary themes for each question. An analytical framework was collaboratively developed with other authors. The data were then coded according to this framework, and a final descriptive narrative was agreed upon for the free-text responses, where appropriate.³⁵ Due to the limited volume of free-text data, multiple iterations of analysis were not required.

Results

There were 151 responses to the survey, of which 93 were complete. The remaining 58 were incomplete: 51 of these only answered demographic questions, and 12 out of those 51 only answered the initial question: “Have you worked with SCI?” These 51 responses were not included in the main analysis. The remaining seven incomplete responses did answer some questions about clinical prediction rules and were therefore included alongside the complete responses, making a total of 100 analyzed responses.

Of the incomplete responses that were included (seven total), four provided their opinions on CPR but did not provide information on other predictors used in clinical practice, two reached the last mandatory section on other predictors but did not rate the predictors they had selected, and one reached the penultimate mandatory section, answering the question, “What would convince you to use a new CPR?” Where a question was not answered by all eligible respondents, this is noted in the results section.

There were some demographic differences between those who completed only the demographic section and those who progressed beyond this, discussed briefly below. All respondents had experience in working with SCI.

Demographics

Demographic features of respondents are described in Table 2. Various professions, levels of experience, and regions were represented. The majority of respondents were senior physiotherapists based in London and the southeast of England. Ninety-two percent of respondents reported working primarily in the NHS. Regarding incomplete responses (n = 39), there were a greater number of “other allied health professionals” amongst people who only completed the demographic section (4/39, 12%) relative to those who went on to answer questions about CPR (3/100, 3%). There was also a greater proportion of people with less than 5 years of experience in SCI (18/39, 46% vs. 25/100, 25%) and a smaller proportion of senior (band 7 and 8) clinicians (14/39, 36% vs. 72/100, 72%).

Table 2.

Respondent demographic information

Demographics	Total
Profession
Physiotherapist	84
Occupational therapist	9
Doctor	1
Nurse	2
Other AHP	3
Other	1
Experience with SCI, years
0-2 years	6
2-5 years	19
5-10 years	24
10-20 years	30
20-45 years	21
Location
Scotland	4
Wales	8
Northern Ireland	1
England - northeast	0
England - northwest	9
England - Yorkshire and the Humber	5
England - East Midlands	3
England - West Midlands	6
England - East of England	3
England - London	39
England - southeast	14
England - southwest	8
Professional grade
Band 8	28
Band 7	44
Band 6	24
Band 5	1
Band 4	1
Other	2
Work setting
Acute inpatient	57
Inpatient rehab	32
Outpatient neurology	23
Community	17
Other	17
Proportion of work with SCI
Less than 25%	46
25%-50%	19
51%-75%	3
More than 75%	32
Sector
NHS	92
Private	8

Note: Numbers indicate totals for each category. As total number of respondents was 100, numbers also represent percentages. Professional grades indicate Agenda for Change (National Health Service [NHS]) bandings; higher numbers indicate more senior grades. Respondents could select multiple options for work setting. AHP = allied health professional.

Attitudes toward clinical prediction rules

Attitudes toward CPR were broadly positive (see Figure 1). At least 80% of respondents agreed with the statements, “I would be interested in using a CPR” (81%), “A CPR can help with goal setting” (80%), and “A CPR can help plan rehab strategies” (81%).

Figure 1.

Level of agreement with different statements about clinical prediction rules (CPR).

Awareness of existing clinical prediction rules

Most respondents were not aware of any clinical prediction rules for SCI (71%). For those who were, most were aware of Van Middendorp⁸ (17 of 29; Figure 2). People selecting “other” stated the ISNCSCI and “presence of proprioception.” Note that these are prognostic factors rather than prediction rules per se.

Figure 2.

Awareness of clinical prediction rules (CPR). Left subplot shows proportion of respondents (100 total) who were either aware or unaware of CPR for spinal cord injury (SCI). Subplot on right side shows number of people aware of each named CPR (out of 29 total respondents). Respondents could select multiple options, hence numbers total more than 29. Named CPR: Van Middendorp et al., 2011,⁸ Hicks et al., 2017,¹⁰ Jean et al., 2021,¹⁷ Hori et al., 2022.²⁸

Using CPR in practice

For those who reported being aware of CPR, only eight people reported using them in clinical practice, which was 8% of the total respondents. Of those who reported using CPR in clinical practice, the majority reported using Van Middendorp⁸ (4 “yes” responses), followed by Hicks¹⁰ (2) and then Jean¹⁷ (1). People who selected “other” reported using ISNCSCI or AIS. All respondents who reported using a CPR were in broad agreement with statements used to describe them, namely: “The predictions are accurate,” “The evidence base is strong,” “It is easy to communicate to patients,” “It is valid for a range of presentations.”

Reasons for not using CPR

Of those who reported being aware of CPR but not using them in practice (21 total), respondents reported that this was because they thought predictions from CPR are not accurate (6/21), that the evidence is weak (6/21), and that they are not useful for patients (5/21). Fewer people thought that they are not useful for their practice (3/21) and that they don't have time to use them (3/21). Only one person thought that they were difficult to use.

Analysis of text entries: reasons for not using CPR

Nine respondents offered more detail regarding their reasons for not using CPR (via free text). Thematic analysis of their responses revealed the following key themes.

Familiarity with CPR

Respondents mainly reported that lack of familiarity with CPR was the reason for not using them; CPR are not ingrained in their clinical practice. Other peoples' lack of awareness was reported as a barrier, making it difficult to communicate CPR within the multidisciplinary team. Respondents reported that they would like to include them in their practice, and it would be useful to know more about them.

“I'm not as familiar as I would like to be with them”.

Relevance

One respondent reported that CPR are less relevant in later stages of the patient journey and that predictions are more useful in acute settings.

Expectation management

One respondent reported that CPR should be used with caution due to the difficulty around managing expectations and the risks of introducing or eliminating hope.

“Working in the acute setting, you have to be very careful about communication around expectations and likelihood and hope with patients…I think any CPR should be used with caution.”

Deciding to use a new CPR

Respondents were asked what would convince them to use a new CPR. Suggested features were “proven predictive accuracy,” “simple to use,” “strong research base,” “helps to plan rehab,” and “uses imaging or neurophysiological data.” The first four were selected by most respondents (78%-86%). Least likely to be selected was the inclusion of imaging or neurophysiological data (35%). No respondents selected “I'm not interested in using a CPR.” This question was answered by 96 respondents.

Other predictors used in practice

Respondents were asked to select other factors used in predicting outcomes. All potential predictors were selected as useful by more than 20% of respondents (Figure 3). AIS grade and neurological level of injury (NLI) were selected as important predictors by almost every respondent (97% and 98%, respectively). Other important predictors were functional status and mood/cognition, as well as presence of complications or spasticity. Most predictors were felt to be only moderately predictive, with responses skewing toward “not accurate” rather than “extremely accurate.” This question was answered by 95 respondents, and 93 provided ratings of predictors.

Figure 3.

Ratings of accuracy for various prognostic indicators of function in spinal cord injury (SCI). Respondents only rated predictors that they reported using. Total number of responses for each predictor indicates how many respondents reported using it. AD = autonomic dysreflexia; AIS = ASIA Impairment Scale; FIM = Functional Independence Measure; SCIM = spinal cord independence measure; UTI = urinary tract infection.

Analysis of text entries: other predictors used

Respondents who answered “other” suggested several other predictors they believed to be important: Factors relating to patient history were mentioned multiple times, for example, age, comorbidities, and previous level of function. Patient personal factors were also identified as important in recovery, with motivation, understanding and acceptance of the condition, mental health, adherence to rehabilitation, and access to compensation all mentioned. Specific outcome measures were suggested such as the SCIFERTO tool, which describes the best possible functional outcome for each level of injury,³⁶ and the EQ5DL, a measure of health status and quality of life.³⁷ Other factors that did not fall into specific categories were laryngeal pathology, time to surgery, and etiology of injury (traumatic vs. acquired).

Associations between demographics and attitudes to CPR

It was hypothesized that attitudes toward, and use of, CPR might vary according to demographic features of respondents. Cross-tabulations and plots were examined for potential relationships between demographic features and responses to questions about clinical prediction rules. Apparent differences in “years of experience” and “professional grade” among those who were aware versus unaware of CPR led to further statistical testing.

For “years of experience,” the Shapiro-Wilk test revealed that data were not normally distributed (W statistic = 0.91, p < .001); nonparametric tests were subsequently used. A Mann-Whitney U test revealed a significant difference between the aware and unaware groups (U statistic: 1332.00, p = 0.022, n_Aware = 29, n_Unaware = 71). The aware group had a median of 15 years of experience compared with 8.5 for the unaware group, indicating that more experienced clinicians were more likely to be aware of CPR (Figure 4).

Figure 4.

Boxplots comparing years of experience between respondents who were aware vs unaware of existing clinical prediction rules. Red lines indicate median values; boxes represent interquartile range (IQR); whiskers extend to 1.5 x IQR; circles denote outliers. Asterisk indicates a significant difference (p = .022).

For “professional grade,” a chi-square test of difference indicated a significant difference between those who were aware versus unaware of CPR (χ² = 12.12, p = .03, df = 5, n = 100). More band 8 clinicians were observed in the aware group (15) than would be expected by chance (8), and fewer band 6s (4 vs. 7) and 7s (9 vs. 13) were observed.

No significant associations were found for other demographic features, as shown in Table 1.

General views on CPR

At the end of the survey, respondents could add comments via free text, and 18 people did so. Analysis of themes was carried out to identify key issues. The themes are framed by the question “What are your views on clinical prediction in spinal cord injury?” and are listed and described below.

Prediction is complex

Clinicians emphasised the difficulty in making accurate predictions due to the complex nature of SCI and the large number of interacting factors.

“Predicting outcome can be complex and inaccurate due to the huge number of personal and environmental variables.”

Clinical experience is key

Several made the point that there is no substitute for clinical experience and use of CPR should be in a specialist context where there is access to people with extensive relevant experience.

“Whilst a predictive tool may have its uses, it should only be used alongside well-informed and experienced clinical reasoning in a specialist setting.”

Prediction is harder for those with incomplete injuries

Respondents mentioned that predictions are more difficult for patients whose injuries are classified as AIS incomplete.

“For incomplete injuries, presentations can differ so vastly…it can be difficult to predict their outcomes.”

Impact of setting

Respondents reported that health care setting impacts the ability to make predictions, that it is harder in the intensive care unit, and that predictions are rarely made in acute settings.

Attitudes to CPR

Positive. Several respondents reported that having a validated CPR would be useful and that they would like to learn more.

“A prediction tool would be useful if accurate.”

Negative. Respondents expressed concerns around the risks of CPR stating the obvious, oversimplifying, and potentially eliminating hope.

Implementation and use of CPR

Respondents suggested ways in which CPR could be implemented, including identifying specific timescales and standardization for their use in the UK.

Discussion

The survey offers a glimpse into how SCI clinicians in the UK perceive CPR. Most respondents were experienced senior physiotherapists, mainly from London and the southeast of England, working in acute inpatient or neurological rehabilitation settings. The predominance of physiotherapy responses in part reflects the distribution of professions working in SCI rehab, for example, the membership of MASCIP has approximately 2.5 times as many physiotherapists as occupational therapists. It may also reflect the fact that existing clinical prediction rules are focused on ambulation, which tends to be a physiotherapy rehab activity. Higher representation of senior clinicians might be attributed to their active participation in special interest groups and more extensive experience with SCI.

Although there was general positivity toward CPR, with many clinicians open to utilizing them, a striking 71% were not familiar with any SCI-specific CPR. Less experienced clinicians, who might benefit most from such tools, were often unaware of them. This suggests a potential gap in training or exposure to these prediction rules. Research suggests that junior clinicians may benefit most from CPR.^38,39

Very few clinicians who were familiar with CPR used them, crediting their validity and accuracy. Those aware but not using them had reservations about their reliability and the supporting evidence, reflecting the state of research into CPR for SCI. The most extensively validated clinical prediction rule⁸ and the most recognized in our survey by Van Middendorp et al.⁸ has not demonstrated good clinical utility,²⁷ and doubts have been raised about its relevance for the broad spectrum of SCI presentations.^30,26 Respondents offered further reasons for not using CPR, most commonly a lack of familiarity, suggesting a disconnect between research and clinical practice. Ethical considerations, such as managing patient expectations and hope, emerged as another barrier.

Despite most clinicians reporting not being aware of CPR, or choosing not to use them, all respondents expressed interest in using CPR in the future, dependent on them having proven utility and strong scientific support. The value of neurophysiological and imaging data was less recognized, which may relate to the fact that the evidence underpinning these prognostic factors is yet to be established.⁴⁰ Note however that preliminary research⁴⁰ and evidence from stroke⁴¹ suggest combining neurophysiological measures with clinical measures may improve predictive power and overcome weaknesses of CPR that have previously only used clinical data.

Although participants reported using a range of potential predictors in their practice, the most common were AIS grade and NLI. These convey important information about the extent of injury, likely functional status, and probable endpoints.⁴² However, they are not strictly predictive and limitations with the ISNCSCI (of which AIS and NLI are components) are a significant motivator for the development of better prognostic tools for SCI.¹⁶ It is notable that despite respondents reporting that they use a wide range of predictors in their practice, the level of confidence in these was relatively low, with most predictors being rated as only moderately or slightly accurate. Respondents also emphasized the importance of personal patient factors in recovery. These factors have been largely unexplored in the SCI literature around prognosis, although research on patient outcomes in other conditions has highlighted the importance of self-efficacy in recovery.^43,44 It may therefore be useful for future prognostic models to incorporate these factors.

At the end of the survey, respondents were given the opportunity to express their views on CPR in free text. Clinicians most frequently highlighted the difficulty in making predictions in this complex and heterogenous patient group. They emphasized the importance of clinical experience in making prognostic judgments and the potential risks of using prediction rules that may oversimplify or eliminate hope for patients. Despite this, respondents also reported that an accurate prediction rule would be a useful clinical tool that they would be interested in using.

Conclusion

Clinicians in the UK show considerable interest in CPR for SCI, but awareness remains limited. This highlights the necessity for intuitive, validated tools that are able to be implemented in clinical practice and can effectively predict functional outcomes in SCI. Given the limited awareness and utilization of existing tools, especially among less experienced clinicians, dissemination strategies targeting these groups may be beneficial.

Future research should prioritize developing and validating such tools and ensuring their findings are conveyed effectively to the SCI community. Regarding existing tools, there would be benefit in qualitative research examining how these tools can best be integrated into current clinical pathways, as well as implementation studies specifically examining how introduction of these tools can impact clinical practice.

Given the concerns about the accuracy of current CPR, further refinement is required, possibly through integration of neurophysiological assessments or use of machine learning techniques. SCI practice may be able to follow recent developments in stroke care as better CPR are developed. The most compelling case for the widespread adoption of CPR in SCI will come from their proven clinical effectiveness.

Limitations

This survey was based on a relatively small sample size and predominantly included physiotherapists from London and the southeast of England. Caution should be exercised in extrapolating these findings to other professional groups and geographical regions. Despite these limitations, the survey detected a difference in awareness of CPR based on clinical experience, while other demographic features did not significantly affect responses.

We also note that less experienced clinicians were more likely to answer only demographic questions, possibly indicating a lack of confidence in discussing CPR. This suggests that our survey may have overestimated awareness, given the predominance of responses from senior clinicians.

Therefore, even though the survey reveals useful insights into CPR awareness and attitudes among SCI clinicians, the results may not be fully representative of the broader professional community.

Funding Statement

Financial Support The work was funded as part of a research fellowship grant to Rowan Boyles from The Imperial Health Charity and NIHR Imperial Biomedical Research Centre 2022-2023: NeuroMap - Evaluation of cortical mapping for people with incomplete spinal cord injury. This work was supported by Imperial Health Charity and the NIHR Imperial Biomedical Research Centre (BRC). Infrastructure support for this research was provided by the NIHR Imperial BRC.