How to leverage pain assessment data to inform clinical decision-making
INTRODUCTION
Over 3,000,000 pain assessments have been completed using PainChek® to date from residential aged care facilities in Australia, New Zealand, the UK, Canada, the U.S., and beyond. In addition to empowering residential aged care facilities with more accurate pain assessment and management, this data reveals crucial information that can be harnessed by residential aged care facilities to improve care for residents and streamline operational efficiency for carers.
Beyond empowering residential aged care facilities with more accurate pain assessment and management, PainChek data can guide and support healthcare leaders as we all strive to innovate, improve outcomes, and enhance best practice pain assessment.
In this report, we’ll share how to analyse pain assessment data, and explore how residential aged care providers can harness this information for clinical decision-making.
About PainChek: How It Works
Before delving into the data, it’s critical to grasp exactly what PainChek does, and how it works.
PainChek’s mission is to give a voice to those who cannot reliably verbalise their pain, such as those living with dementia or cognitive impairments.
Data reveals that 54% of those living in residential aged care have a formal diagnosis of dementia, and around 60% to 80% of nursing home residents with dementia regularly experience pain. However, many of these residents are unable to verbally report or articulate their pain. This may lead to individuals being mislabelled as difficult or aggressive before the cause of the behaviour has been uncovered, or to the inappropriate prescription of antipsychotic medications.
A core component of PainChek® technology is the ability to use artificial intelligence (AI) to detect micro-facial expressions that indicate pain. Facial expressions are an encodable form of pain behaviour, which provides carers and clinical staff with a wealth of information to determine whether pain is present. Facial analysis is also particularly suited for those with dementia, who typically have a reduced ability to filter their facial expressions.
Once the facial analysis is done, a clinician then completes their assessment across a further five domains, which include the resident’s voice, movement, behaviour, activity, and the body. Based on these indicators, PainChek® will return either a No Pain (0-6 indicators), Mild (6-12 indicators), Moderate (12-15 indicators), or Severe (16-42 indicators) result.
In total, there are 42 total indicators in a PainChek® assessment
When starting a pain assessment, clinicians must first determine the assessment timing. This is when they choose if the resident is at rest, or post-movement. Next, the clinician chooses whether to use the AI technology to assess pain, or whether the assessment situation requires a manual assessment. These are situations where it’s perhaps not possible to conduct an auto-analysis, such as due to poor lighting or if the resident is displaying high levels of agitation.
Interpreting pain assessment data using PainChek Portal
All providers have access to the PainChek Portal as part of their PainChek license. This portal allows care providers and clinicians to access pain assessment at a company, facility, and individual user level. Clinicians can log in at any time, anywhere, and see pain levels down to the facility, and resident history.
In addition to the PainChek Portal, care providers also receive a monthly management dashboard report as an added service. This report provides detailed information on how the tool is being used and how often it is being used, and includes information such as assessment history, key assessment indicators, and chronological pain scores. Providers can also view their top-performing facilities and identify power users in the report.
How to leverage pain assessment data for clinical decision-making
Pain assessment data offers valuable information around when assessments are being conducted, how assessments are being conducted, which facilities and carers are performing more assessments, and which facilities are lagging behind.
When providers harness this data at a company, facility and individual level, it provides insights into pain levels and trends in residents, as well as guidance on areas of improvement for training, onboarding, and usage. In turn, this information can be leveraged to ensure facilities and carers deliver best practice pain assessment for residents.
Below, we explore some of the most common ways that providers can leverage the data in the PainChek Portal to extract insights and guide clinical decision-making.
When pain assessments occur
Mapping the time and number of pain assessments against the pain scores of residents provides macro-level insights into pain levels and trends throughout the day. In this example above, we can see that most assessments at this facility are conducted during the day, with the highest number of assessments occurring between 9AM and 10AM.
The other notable aspect of this graph is the percentage of pain assessments that returned a moderate or severe pain rating. This is demonstrated by the colour: the darker the red, the more moderate to severe pain assessments occurred in that hour.
What is evident here is that while far fewer pain assessments occur during the night, those assessments typically identify higher levels of pain. A provider reviewing this data could draw the conclusion that pain is an underlying factor for restlessness or disrupted sleep patterns, and that appropriate pain treatment methods need to be administered in the evening.
Going even further to the data at a facility level, clinicians can see when pain assessments occur during shifts, or even at an individual resident level. Based upon this information, a carer can identify any trends of pain in a resident over any given day and tailor the treatments accordingly.
Point of care assessment
Another valuable insight for clinicians is the point-of-care assessment. This data reveals insights into how pain assessments are being conducted, either when a resident is at rest, or post-movement.
In this chart from an example facility above, we can see that more than 80% of pain assessments returning no pain were conducted when the resident was at rest. As we progress through the pain categories, there is a clear correlation between conducting the pain assessment post-movement and identifying higher levels of pain. This is consistent with the research, which shows pain is more common post-movement.
From this data, a facility manager may advise carers to implement post-movement assessments where possible, to support the accurate identification of pain. This ratio is also valuable to help clinicians ensure they are maintaining best practice when it comes to pain assessment.
In the monthly reporting dashboard, providers can view a number of key assessment indicators to determine how many assessments are conducted post-movement, and how many pain assessments are recording a zero pain score.
In this chart, the provider can see that only 22% of pain assessments were conducted post-movement and 19% of assessments are returning a zero pain score. This is a key indicator that more training needs to be done to ensure clinicians are following best practice and conducting assessments post-movement.
Based upon this data, the provider could dig deeper to see which facilities are conducting more assessments at rest. PainChek’s monthly management report offers insights into which facilities have the highest percentage of post-movement analyses, as well as those that may be lagging behind. In this case, a workshop or training event may be valuable to allow carers from top-performing facilities to train those that have less assessments conducted post-movement.
Evaluating data related to pain domains to rapidly identify multidimensional pain indicators
PainChek® assesses 42 indicators of pain grouped into six domains: face, voice, movement, behaviour, activity, and body. The graph from an example provider shows which domain has the highest impact for a particular pain rating — in other words, which domain contributed most to an overall rating score.
Here, the provider can see that In assessments returning a pain rating of none, the face and activity domains contributed the most indicators. The activity domain covers indicators such as resisting care or altered sleep cycles, which appear to be the most prominent in the ‘none’ pain rating. In addition, indicators in the voice domain, which covers indicators such as moaning, crying, and sighing, are less frequently detected in lower pain assessments but far more prominent in moderate and severe pain.
Going into further detail at a common pain indicator level, these graphs from an example provider show the common pain indicators for each pain category. From this data, we can see that the ‘Confused’ indicator is commonly present for lower levels of pain, and accounts for around 30% of all indicators that are present.
When we look at distressed or restlessness behaviours, these become more strong features in the moderate to severe pain assessments. However, pain is a highly subjective experience and we are looking at high level trends. The data actually supports this, in that it’s dynamic, and best practice pain assessment requires staff to be both confident and competent in identifying the range of indicators that make up the multidimensional nature of pain.
From this data, a facility manager could extract key insights into which behaviours are most prevalent in residents with moderate and severe levels of pain. This information would allow carers to rapidly identify indicators of pain in different residents, and ensure that treatment is delivered in a timely fashion.
For example, based on these indicators, a facility could focus on the identification of subtle pain indicators in training, such as confusion or restlessness, to help carers understand when pain may be present and a PainChek pain assessment needs to be conducted.
Regularity of pain assessments
Another key metric is the regularity of pain assessments and, in particular, how much time passes between a pain assessment and a follow-up assessment. Providers can see when follow-ups are conducted in residents at a facility level or a resident level, or even zoom in on when follow-up assessments are conducted for those with none to mild pain, compared to those returned a moderate or severe rating.
Tracking pain assessment follow-ups is a really powerful metric, particularly when we analyse the data at a facility or individual level. This data enables clinicians and facility managers to track assessments, in order to ensure they’re being followed up in compliance to the local clinical policies at hand. Clinicians can also couple this data with other sources, such as medication or treatment information, to generate even more sophisticated reporting.
At an individual level, let’s take an example resident. The chart above shows PainChek® assessments completed on a resident over a 2.5 month time period. Each box represents an assessment, and the colour of the boxes reveals what the assessment pain rating is. The coloured dots running along the top of the boxes tell the clinician whether the assessment was conducted at rest, or post-movement. The detail along the bottom is our follow-up time intervals.
In mapping this data, the clinician can see that on the first of May 2020, that assessment returned a ‘none’ pain rating that was conducted at rest. It then took 18 days and 12 hours for the next follow-up assessment to occur. It then returned a ‘mild’ pain rating. It then only took 1 hour for a follow up assessment to occur, which again was taken post movement and this time the pain rating went back to a none.
Monitoring facility and user adoption to improve uptake and training
Providers can also use pain assessment data for the purposes of change management. Implementing new clinical practices in a facility requires a significant amount of effort, such as communication, training and support. A provider or facility manager can therefore use pain assessment data to monitor uptake and ongoing utility of PainChek® at an organisational, facility, and user level.
At the facility view, this example above is based on a provider with 5 sites, who implemented PainChek approximately 18 weeks ago. This mapped data shows how many pain assessments each facility is undertaking as a percentage of all assessments.
In the example above, it’s quite clear that there are some sites that are very strong users compared to others. A provider can use utilisation data to identify which sites may need additional support and training. It also gives clinical leaders at each site a comparison point for their own pain assessments, particularly if they want to share data between facilities.
Similar to the facility level graph, PainChek clients can track usage at an individual carer level. This monitoring usage data is a way to recognise champions, and also tailor retraining and mentoring opportunities for others.
Finally, this graph offers a combination of utility and how PainChek is being used in an example facility with 50 staff using PainChek®. Together, these clinicians have amassed a total of 10,000 PainChek® pain assessments. Each box on this graph represents an individual PainChek user at this facility.
The deeper the shade of the box, the more assessments that user has conducted — essentially, the dark purple boxes are essentially power users. However, it’s also important to note where the box sits on the X axis, which represents the percentage of auto facial assessments vs. manual assessments.
In this example, a clinician can track usage levels, as well as an indicator of how assessments are being conducted. Those users that are on the right hand side of X axis predominantly use AI when conducting their pain assessments. Moving towards the centre of the graph, the boxes or users are more typically half-half when it comes to automatic facial assessments compared to manual pain assessments.
It’s great to have high usage, but power users also need to be following best practice. A clinical leader viewing this data can subsequently identify retraining opportunities to strengthen best practice pain assessment across the facility.
Conclusion
Digital health technologies and data supports objectivity to a subjective and prevalent condition such as pain.
Pain data can help clinical leaders and carers continuously improve how pain assessment is undertaken, particularly in those who cannot verbalise their pain, and is critical for:
- Identifying trends in pain assessment delivery
- Tracking pain levels over time
- Flagging pain outcomes and risks
Real-time health data also supports the embedding of new clinical practices and technologies.
You don’t need to be a data scientist to leverage data for clinical outcomes. At PainChek, all of our clients can access their information using the PainChek Portal, and receive a monthly reporting dashboard containing this information. For those that want to conduct a more sophisticated analysis, we also provide access to data via an API.
To learn more about how PainChek® could enable best-practice pain management within your organisation, book a one-on-one session with a member of our team.
Sources:
