Pain is an almost universally experienced phenomenon (Wall et al, 2006). As a physiological sensation, it serves to protect an organism from harm and limit damage. The influence of psychological factors on the perception of pain is well recognised (Darnall et al, 2016). Acute post-surgical pain can persist well beyond 3 months (Richebé et al, 2018). This chronic post-surgical pain can affect up to one in 10 surgical patients (Glare et al, 2019), with one in 100 experiencing intolerable pain (Fletcher et al, 2015).
While there are many risk factors for developing chronic post-surgical pain including mood, pre-existing pain and the use of pre-operative analgesia, the trajectory of the acute pain course in the days following surgery as well as the amount of time spent in severe pain seem to be important factors (Lavand'homme, 2017). This provides an opportunity for careful in-hospital monitor ing, which could potentially impact outcomes.
Understanding a patient's response to pain and its management is vital (Faculty of Pain Medicine, 2021) for:
- Assessing the effectiveness of treatment
- Prompting an increase or reduction of analgesia
- Identifying neuropathic pain, warranting adjunct analgesic regimens
- Addressing psychological risk factors
- Detecting adverse effects of an analgesic regimen.
During pain assessments, some nurses may collect only simple pain scores, whereas others may capture detailed, highly specific data around pain characteristics, interventions and prognosis (Radnovich et al, 2014). Data collection pathways should be standardised, and always be centred on effective clinical management and monitoring in line with national and local strategic needs (Mac Neela et al, 2006).
A pain assessment should aim to effectively assess all of the above and be dynamic enough to discern new and evolving pains that occur while the patient is in hospital. There is detailed literature on what pain characteristics, interventions and additional metrics constitute a thorough pain assessment (Schug et al, 2015). Standardisation of the means in which these components are documented and stored is essential to ensure a wide range of characteristics, behaviours, emotions and interventions can be identified, captured and monitored quickly and inexpensively (Dansie and Turk, 2013). The purpose of this effort is to promote evidence-based practice and data democratisation, within a cultural wrapper of continual improvement.
Many NHS providers are still using paper-based pain assessments. This is associated with numerous clinical and operational risks (Lichtner et al, 2016), which are amplified when patients require multiple assessments while in hospital. Digitised pain assessments can be described as best practice, as they are easily accessible and cannot be misplaced, and because output data can be stored and analysed quickly (Boochever, 2004).
Structured interfaces
Structured interfaces within an electronic health record, for example forms and templates, provide a method for rapid and consistent data capture by front-end users and fast mapping and manipulation by back-end users such as analysts. Well-designed structured interfaces balance volume and adaptability with the need for consistent and rapid data collection, interpretation and manipulation (Cimino et al, 2001; Van Vleck et al, 2008).
Structured data (pre-defined data), such as those that come from tick boxes, are straightforward to store and analyse. As such, essential data items for monitoring or improvement should be structured wherever possible.
The method in which pain assessments are completed and documented lends itself well to a structured interface. Key data entry fields, such as pain characteristics and manifestations plus associated conditions can be tied to radio buttons, tick boxes, smart search boxes and free-text entry boxes, embedded with a terminology package such as SNOMED CT (SCT) (SNOMED International, 2022a).
Clinical terminologies
Clinical terminologies are collections of medical terms that can be processed by computers. They can be used to capture and transform clinical records into structured data for intelligence generation, providing insights that can lead to action being taken.
Compared to classifications such as ICD-10 and OPCS, a clinical terminology's scope is significantly wider and provides a paradigm for a clinical snapshot of time (Bowman, 2005). Classifications provide a transformed and aggregated viewpoint of a concluded episode of care, making them unsuitable for many clinical purposes where specificity and expressibility are important (SNOMED International, 2022b).
SCT is a clinical terminology mandated for use in the UK (meeting information standard SCCI0034) and, through its design and framework of concepts (clinical entities) and relationships, every term has a computer processable definition of its meaning (National Information Board and Department of Health and Social Care, 2014; NHS Digital, 2020). SCT can be embedded throughout medical records and front-end interfaces, for example in nursing assessments and care plans, to support rapid, accurate and consistent data entry, storage, retrieval and communication (SNOMED International, 2022c).
Aims
The pilot study aimed to assess the impact and significance of a newly structured digitised form from a quality, safety and efficiency standpoint.
Methods
Background: record-keeping and standards
Nursing records should be clear, concise and accurate (Nursing and Midwifery Council, 2018). However, literature has shown that documentation of relevant pain features and descriptors can be poor during pain assessments (Lichtner et al, 2014).
There are core standards for acute and chronic pain management, including the contents and regularity of a pain assessment (Faculty of Pain Medicine, 2021). The successful adoption and application of these standards must be governed by a robust clinical audit process designed to continually assess operational and clinical performance against key targets (National Pain Audit, 2011).
Study framework
The study was designed to inform the need for more comprehensive research into nursing assessment interface design. Two acute pain specialist nurses and a consultant anaesthetist compiled a list of factors that were essential for a thorough pain assessment, based on current best practice principles (National Pain Audit, 2011; Faculty of Pain Medicine, 2021). This resulted in 18 categories covering aspects of pain characteristics, patient impacts and management (Table 1).
Table 1. Critical factors for adult pain assessments
Factor | Description | Data type* | Example value* |
---|---|---|---|
SedationS,Q | Assessment of alertness | SNOMED CT | 248234008Mentally alert (finding) |
Physiological site(s)S,Q | Anatomical site(s) of pain | SNOMED CT | 771367001Amputation stump of left lower limb (body structure) |
DistributionQ | Spread of pain over defined area | SNOMED CT | 9972008Radiating pain (finding)| |
Pain at restS,Q | Rest scores | SNOMED CT | 40196000Mild pain (finding)| |
Pain on movementS,Q | Movement scores | SNOMED CT | 50415004Moderate pain (finding)| |
Assessment toolQ | Assessment type | SNOMED CT | Known omission—authoring in progress. Placeholder concepts used |
Pain descriptor(s)Q | Characteristic feature(s) of pain | SNOMED CT | 49575005Shooting pain (finding)| |
NatureQ | Acute or chronic | SNOMED CT | 274663001Acute pain (finding)| |
OnsetQ | Onset of pain | Date | 01/01/2021 |
PeriodicityQ | Continuous or intermittent | SNOMED CT | 314642004Intermittent pain (finding) |
Known/suspected causeQ | Cause of pain | SNOMED CT | 107401000119105 Acute postoperative pain (finding) |
Aggravating factorsQ | Non-pharmacological factors increasing pain perception | Free text | Free text |
Alleviating factorsQ | Non-pharmacological factors decreasing pain perception | Free text | Free text |
Impact on sleepQ | Assessment of sleep impact | SNOMED CT | 301345002Difficulty sleeping (finding)| |
Impact on activities of daily livingQ | Assessment of activities of daily living | SNOMED CT | 284977008Difficulty dressing (finding) |
AnxietyQ | Assessment of anxiety | SNOMED CT | 48694002Anxiety (finding) |
Low moodQ | Assessment of mood | SNOMED CT | 366979004Depressed mood (finding) |
Non-pharmacological intervention(s)S,Q | Started/evaluated | SNOMED CT | 229559001Transcutaneous electrical nerve stimulation (regime/therapy) |
Q: factor used for quality scoring
S: factor used for safety scoring
*Data types and values in new pain assessment (intervention)
Note: SNOMED CT content was taken from the latest UK release at the time of writing in 2021 (NHS Digital, 2021)
A digital form was designed and created with the purpose of supporting rapid documentation of the outlined factors. The form consisted of >50 dynamic fields, with SCT embedded throughout to ensure compliance with the national mandate (NHS Digital, 2020).
Data collection
The study was split across three sequential phases: baseline; intervention; and re-evaluation of baseline.
These stages were called: baseline—pain assessment (BPA); intervention phase—new structured pain assessment (NSPA); and re-evaluation of baseline—pain assessment (RBPA). The original pain assessment template was used during the BPA and RBPA phases and the new SNOMED CT-enabled template during the NSPA phase.
These stages ran sequentially in 2-week periods starting from 24 January 2021. Samples of pain assessments were generated from every second assessment completed in each phase by four acute pain team nurses who were unaware of the purpose of the study. Duplicate patient records were excluded to ensure all observations were independent—one patient, one pain assessment.
BPA covered normal documentation practice using the original pain assessment form. NSPA consisted of a 30-minute education session and the new digitised structured form was used. RBPA covered the return to normal practice, using the pre-existing form, and was designed to assess whether benefits remained once the structured form had been removed.
In all assessments, across all stages, visual analogue scale (VAS) scores were used to rate pain intensity.
Sample
A power analysis was performed to determine the minimum sample sizes required for the large desired effect size (1.2) (Sawilowsky, 2009) at the conventional statistical power of 0.8 (Serdar et al, 2021).
The analysis showed the total number of subjects required was 12 in each group, and this was used as a minimum sample parameter for study. The sample size (BPA: 14; NSPA: 15; RBPA: 16), was aligned with comparable NHS-based pilot studies (Crossley, 2016; Velazquez Cardona et al, 2019; Barker et al, 2022).
Data analysis
Across the three stages, the content of the assessments was scored, aggregated and analysed by two senior clinical analysts. Points were awarded where each individual factor listed in Table 1 had been recorded in a given assessment.
Any statement against a variable, whether determinant or non-determinant, was considered a positive result; for example, ‘patient has neuropathic pain’ and ‘patient does not have neuropathic pain’ were considered a positive for descriptors of pain as the characteristics of the pain had been considered). Each variable was allocated a binary flag (1 or 0); the value of 1 was given if content relevant to a variable was documented, regardless of frequency.
The time taken for the analysts to extract and interpret the content of each assessment was recorded in seconds.
Statistical analysis
The mean count of the recorded factors, the mean count of the five safety-related variables and the mean time taken to extract the data at each of the stages were compared.
To determine whether differences between the means of these stages were statistically significant, Welch's t-test was used. The use of this test is justified because each form represents an individual sample from the distribution of its stage and those distributions have unknown variance (Liu, 2015). To assess whether the documenting of active pain-related sleep conditions had improved, the portion of patients who had an active pain-related sleep condition recorded was compared across each phase of the study.
To take account of the small sample size, Fisher's exact tests were used to determine whether these differences were statistically significant.
The null hypothesis was that the samples from each stage were drawn from distributions with the same mean. Differences were considered to be statistically significant where P<0.05.
Ethics
The study was approved by the local panel for clinical audit and followed all local policies. All clinical data used were anonymised and patients gave consent via site governance procedures.
Results
Quality
The overall quality of pre-existing assessments was suboptimal across all variables (BPA/RBPA), with a low consolidated average score of 4.7 (25.9%). When NSPA was compared to established practice (BPA/RBPA), there was a large improvement in scoring with the average score tripling to 15.9 (88.3%) (differences: versus BPA: 11.9; 95% CI (10.7–13.1); P<0.001; versus RBPA: 10.6; 95% CI (9.3–11.9); P<0.001). There was no significant difference in quality scores between the RBPA and BPA (difference: 1.3; 95% CI (–0.3, 2.9); P=0.12) (Table 2 and Figure 1).
Table 2. Statistical comparisons using Welch's t-test
Metric | Comparison | Means | Difference | 95% confidence interval | P |
---|---|---|---|---|---|
Quality (score) | NSPA/BPA | 15.9/4.0 | 11.9 | (10.7–13.1) | <0.001 |
NSPA/RBPA | 15.9/5.3 | 10.6 | (9.3–11.9) | <0.001 | |
RBPA/BPA | 5.3/4.0 | 0.7 | (–0.3, 2.9) | 0.12 | |
Safety (score) | NSPA/BPA | 4.9/2.7 | 2.2 | (1.5–2.9) | <0.001 |
NSPA/RBPA | 4.9/3.0 | 1.9 | (1.0–2.8) | <0.001 | |
RBPA/BPA | 3.0/2.7 | 0.3 | (–0.7, 1.30) | 0.55 | |
Efficiency (seconds) | NSPA/BPA | 36.3/122.9 | –86.6 | (–117.6, –55.6) | <0.001 |
NSPA/RBPA | 36.3/116.7 | –80.4 | (–102.9, –57.9) | <0.001 | |
RBPA/BPA | 116.7/122.9 | –6.2 | (–44.3, 30.9) | 0.73 |
BPA=baseline—pre-existing assessments; NSPA=intervention—new structured assessments; RBPA= re-evaluation of baseline—pre-existing assessments
Across all homogeneous variables, there were variations in documentation between the stages, most notably with the capture of pain-related sleep difficulties. Use of the new form resulted in a significant increase in the identification of patients experiencing difficulties in going to and/or maintaining sleep because of pain (Table 3). More than two-thirds of patients had an active sleep difficulty documented when the new form was used (NSPA); with the pre-existing form (BPA/RBPA), the highest proportion was less than 20% (RBPA). This finding highlights a significant morbidity factor that was not routinely collected in native data sets (Table 3).
Table 3. Significant results from analysis of pain-related sleep difficulties
Comparators | Sample sizes | Sleep evaluations | Active difficulties | P value |
---|---|---|---|---|
NSPA/BPA | 15/14 | 15/1 | 10/1 | <0.002 |
NSPA/RBPA | 15/16 | 15/5 | 10/3 | <0.012 |
Significance calculated using Fisher's exact test
BPA=baseline—pre-existing assessments; NSPA=intervention—new structured assessments; RBPA=re-evaluation of baseline—pre-existing assessments
Safety
The new assessment form significantly outperformed the pre-existing assessments (BPA/RBPA), with an average consolidated safety score of 4.9 (97.3%) (differences: versus BPA: 2.2; 95% CI (1.5–2.9); P<0.001; versus RBPA: 1.9; 95% CI (1.0–2.8); P<0.001).
No significant differences in the recording of safety features were found when comparing existing practices before the study and after the structured form was removed (difference: 0.3; 95% CI (-0.7, 1.30); P=0.55) (Figure 2).
Efficiency
Data extraction and interpretation averaged 122.9 seconds during BPA, 36.3 seconds at NSPA and 116.7 seconds at RBPA. NSPA significantly outperformed both BPA and RBPA (differences: versus BPA: -86.6; 95% CI (–117.6, –55.6); P<0.001; versus RBPA: -80.4; 95% CI (–102.9, –57.9); P<0.001), more than halving the time on average to complete the interpretation and scoring for each assessment.
There was no significant difference in data extraction and interpretation time between pre-existing practice before the new structured form was introduced and after it was removed (differences: RBPA versus BPA: –6.2; 95% CI (–44.3, 30.9); P=0.73) (Figure 3).
Discussion
The new form delivered improvements across all measured metrics of quality, safety and efficiency. This carries notable clinical and operational benefits.
Impact on service
The new form significantly improved the quality of the recorded information for each assessment and, simultaneously, reduced the time taken for data extraction and interpretation. This shows it can facilitate a faster stream of more consistent, accurate data flows to support secondary functions such as reporting, clinical audit and research. Grounding the form in evidence-based practice (Faculty of Pain Medicine, 2021) provides a simple means for ongoing evaluation and service improvement.
Sleep and mental health are negatively affected by pain, and impaired functionality in these areas also exacerbates the pain experience (Chouchou et al, 2014; Zarean et al, 2021). Data on sleep and mental health were not readily available within the study organisation. This makes it difficult to assess the size of the problems, monitor interventions and calculate resource requirements. Regular recording of sleep disturbed by pain and its psychological impacts will provide data on the size and types of negative experiences caused by acute pain, and provide insights into potentially useful interventions. For example, mindfulness and guided imagery have low side-effect profiles and have been shown to be effective in reducing opiate dependence (Mehl-Madrona et al, 2016).
Impact on patients
Despite heightened awareness of and investment in pain management, significant post-surgical pain remains problematic (Gan et al, 2014). Temporal pain profiling, as made possible with structured data capture as demonstrated in the new form (NSPA), may be part of the solution as it makes it easier to identify when and what treatment regimen changes are appropriate.
For example, temporal pain profiling may reinforce the rationale for weaning and discontinuation of opioids, and supports stewardship of these medications for patients (Hah et al, 2017). It may also suggest where non-pharmacological interventions (e.g. pulsed electromagnetic fields, mindfulness and massage) may be suitable alternative or adjunctive tools (Tan et al, 2007). This is important given the lower potential side-effect profile and proven efficacy of these treatments (Schug et al, 2015).
Linking patient pain profiles to drug algorithms and data from other clinical areas can reduce information redundancy and allows personalised care to be provided (Pickering et al, 2016).
Profiling also helps to detect individual differences and achieve equality across diverse patient groups, an example being those with learning disabilities (Faculty of Pain Medicine, 2021).
Impact on nurses
It is important that the pain assessment acts as a component of an integrated, learning health record where information is linked with data from other areas almost instantaneously. Assimilation with other assessments, results and interfaces could allow nurses to readily assess and monitor the impact of pain on function. This could include compliance with services, such as physiotherapy, and the impact of physiological activity, such as deep breathing, on pain experience.
The pre-existing form was free-text heavy, supporting high volumes of unstructured data entry. The content was greatly influenced by individual differences between nurses' documentation styles. The results show the new form provided a clear, rapid mechanism for standardising the information captured and reduced undesirable variation.
The new form also showed a greater sensitivity in identifying factors of morbidity that negatively affect patient experience and outcomes, notably sleep difficulties. Sleep deprivation reliably correlates with new incidents and exacerbations of pain (Finan et al, 2013), so swift and continual management is important. The results suggest this is overlooked in established practice, and highlight a potential area of oversight that should be further investigated.
Making all relevant psychological and clinical factors available, in conjunction with nursing diagnoses, supports nurses to better understand specific patient behaviour and responses, and therefore respond more appropriately with targeted remedial action(s) (Correia and Duran, 2017; NANDA International, 2010). All components that affect overall pain experience should be documented for assimilation in decision support tools; the greater the volume of relevant content, the better tools will be at supporting the workforce.
It is worth noting that a low safety score did not mean that the patient did not receive appropriate care nor that any safety protocol was breached. However, from a medicolegal perspective, all nurses should consider: if it is not documented, it did not happen.
Moreover, using SCT as the core terminology supports local and national interoperability, as multiple domains use the terminology as their standardised clinical vocabulary (NHS Digital, 2020). This is particularly important in the sphere of pain, as effective management requires collaborative working in multidisciplinary team pathways.
Impact on analytics
The time taken for data extraction and interpretation was reduced, on average, by 69.6% when using the new form. Intuitively defining and structuring the data capture mechanisms in the form resulted in the elimination of time-consuming free text scanning. The time saved could be used to provide and validate more frontline analytics initiatives.
A key feature of the new form was the autofeed of documentation from one assessment to another. This allows temporal data touchpoints to be created and facilitates the tracking of changes in key characteristics or management, which are not possible where data points are not consistently mapped. Identifying subtle changes in individual pain profiles are essential to optimising care and reducing clinical risk.
Limitations
The study was limited somewhat by the COVID-19 pandemic. Reduced capacity resulted in a small sample size (45 patients), although this sample size was deemed suitable for results. A follow-up study with a larger sample size would further solidify and validate the findings of this study, although the results are unlikely to change given the large differences found during significance testing.
As elective surgeries were paused during the sampling period, the sample consisted of predominantly emergency patients with complex histories. Although unlikely to result in any material changes to the outcomes of the study, expansion of the sample to include elective surgeries could be insightful and further support the conclusions.
Future improvements
Given the reduction in data extraction and interpretation times, it can be concluded that clinician data recording times would also improve in parallel. This aspect could be an interesting area of further development.
Pain assessments were scored based on two senior analysts' interpretation of intuitively clear information. Scoring of pain assessments by clinical staff may be different and knowing this would also enhance the results.
Patient involvement and satisfaction with their pain management plan are paramount to compliance and successful outcomes (Stinson et al, 2013). As part of an interoperable health record, interfaces can be designed to incorporate both practitioner and patient input, which may encourage shared planning and goal-setting. Integration of patient satisfaction ratings and self-defined descriptors could be a useful enhancement for future study.
Conclusion
Intelligent, structured forms are highly effective for documenting pain assessments and provide notable benefits from quality, safety, efficiency and interoperability standpoints. Nurses and analysts should collaborate to improve data capture and downstream flows to optimise care, support the local workforce and continue to push the frontier in evidence-based practice.
KEY POINTS
- Assessments lend themselves well to structured forms with a design focused on rapid, accurate and consistent data entry, storage, retrieval and communication
- A structured form can significantly improve the quality of information recorded for each assessment, providing assurance of safe and evidence-based practice
- Intuitively defining and structuring data capture mechanisms through a structured form can eliminate free text scanning, significantly reducing the time spent on extraction and interpretation
CPD reflective questions
- What aspects of pain should be captured in a pain assessment form?
- Do your pain assessments focus purely on the physical manifestations of pain, or do they also monitor hidden impacts such as sleep disturbance and anxiety?
- How many structured and free-text interfaces do you use to capture patient data? If they were aligned into one structured interface, would it improve quality, safety and efficiency?