PainChek Solving the silence of pain in dementia Mustafa Atee Research Fellow Faculty of Health Sciences, School of Pharmacy & Biomedical Sciences Curtin University, WA
Conflict of Interest Research Fellow, Curtin University Research Scientist, PainChek Ltd Shareholder in PainChek Ltd
Objectives of Presentation Highlight the challenges associated with assessing pain in people with dementia Provide an overview of PainChek Discuss the evidence behind PainChek
Overview Background Pain and dementia Pain behaviours Pain assessment tools PainChek Conceptualisation Development Description and design Clinical Studies Current Insights Impact on Care
Background: Pain and Dementia
Pain Subjective (i.e. internal experience) Biopsychosocial phenomenon Acute vs chronic (persistent) Source: http://www.nd.edu.au/ data/assets/pdf_file/0010/139492
Pain Behaviours/Expressions/Displays Actions or postural displays that are enacted during the experience of pain Observable cues of pain e.g. Verbal (e.g. pain noise/words) vs non-verbal (e.g. body posture, grimacing) Communicative (e.g. facial expressions) vs protective (e.g. guarding) Subtle (e.g. changes in ADL) vs obvious (e.g. limping) Convey information about patient s pain severity Difficult to suppress non-verbal expressions of pain Martel et al, 2012; Steinkopf, 2016; Hadjistavropoulos & Craig, 2002
Pain Behaviours Communicative Discernible for observers but subject to interpretation More pronounced than protective behaviours as a signal of pain Examples Facial expressions (e.g. grimacing, wincing) Vocalisation/verbalisation/pa ra-verbal pain expressions (e.g. moaning, groaning, sighing, pain noises/words) Protective Defensive mechanism or reflex permitting escape from source of pain (Griensvan et al, 2014) Greater influence on observer judgement of pain related limitations (Martel et al, 2012) Examples Limping Holding Guarding Touching Reduction in ADLs Social activities Recreational activities Sleep problems Uncooperative behaviour Subtle Non-specific behaviours Difficult to interpret as pain Examples Freezing Defence Agitation Guarded behaviour Upset Sleep problems Uncooperative behaviour Verbal abuse Resistance to care Hoarding Inappropriate sexual behaviour Changes in ADLs Sullivan M et al,2006 Husebo et al 2007, 2009
Pain-related Behaviours American Geriatric Society (AGS) American Geriatric Society Panel on Persistent Pain in Older Persons The management of persistent pain in older persons. J Am Geriatr Soc. 2002;50:S205 24
Big problem. Up to 85% have pain Emotional and behavioural discomfort + epat Poor QoL
Pain in Dementia: Alzheimer s Disease Relatively strong associations have been shown between pain and depression, as well as unspecified behavioural problems. Associations of pain with agitation, aggression, delusions, wandering, and resistance to care have also been established, although the link is less consistent. Distressing behaviours aggression, phobia, anxiety strongly associated with pain Guarding is better indicator of pain than rubbing, bracing, and restlessness in older adults with dementia who are unable to communicate their pain experiences. Ford B et al. 2015;16(5):692-700
Facial Responses to Pain Amongst the three categories of non-verbal behavioural responses to pain, namely, facial expressions, vocalisations and body movements, the facial expression of pain has been studied most extensively. The reason why research on pain behaviour has mostly focused on the facial expressions of pain is that facial expressions are readily accessible, are highly plastic, and are believed to be the most specific, encodable form of pain behaviour in humans (Williams 2002 ). Miriam Kunz, 2015
Science behind the epat Facial Action Coding System (FACS) Catalogue of facial expressions Used to describe changes, contraction, or relaxation of facial muscles Each facial muscle movement is described as Action Unit (AU) Combination of AUs produces facial expressions e.g. Sadness = AU 1 + AU 4 + AU 15 Limitations Difficult Time consuming Laborious Bartlett et al. Curr Biol 2014;24
Facial Action Unit (AU) Codes of Pain
Facial Features of Pain Pain intensity & nature How pain affects someone Miriam Kunz, 2015
Patterns of Pain Facial Expressions Pattern I: tightening of the muscles surrounding the eyes with furrowed brows and wrinkled nose (a + b + c) Pattern II: furrowed brows with tightening of the muscles surrounding the eyes (a + b) Pattern III: opened mouth with tightening of the muscles surrounding the eyes (b or d + e). Miriam Kunz, 2015
What We Know Pain is detected by different facial expressions Facial expressions of pain is one of the strongest indicators of pain in older persons with dementia and are particularly useful and sensitive to assess pain in dementia (Kunz et al, 2007; Sheu et al, 2011; Lints-Martindale et al, 2012) Facial expression of pain is more frequent and intense in people with dementia (Kunz et al, 2007) Image source: https://i.pinimg.com/736x/a4/fc/4f/a4fc4f2ce5374d4e cd5dd3b6f42cdc5f--drawing-tips-drawing-tutorials.jpg
What do we know from the literature? People with dementia have lost their learnt pain behaviours (Kunz et al 2007) Observational scales with objective facial measures have better psychometrics than those containing abstract or vague facial descriptors (Sheu et al, 2011; Beach et al,2016)
Pain Facial expressions Pain Behaviours Restlessness Freezing Moaning Guarding Altered Behaviours Agitation Resistance to care Aggression Sighing Distress Limping Groaning Fidgeting
Pain Assessment in People with Dementia
Observational Pain Scales Curtin University is a trademark of Curtin University of Technology CRICOS Provider Code 00301J
Current observational tools Sub-optimal Lack consistency Subject to bias Under-utilised epat
Pain assessment for people with dementia 28 tools reviewed Conclusion Limited evidence of validity, reliability and clinical utility epat
What we know from the literature There is evidence that pain is under detected and undertreated for people with communication difficulties, including people with dementia (Scott et al, 2011). Pain tools exist but there is evidence that they are not used in everyday practice (Manias, 2012). Observational tools improve pain recognition by >25% above chance (Lukas et al, 2013). Tools containing common pain related facial action units have higher sensitivity, reliability and validity (Sheu et al, 2011)
PainChek World s first pain assessment & monitoring medical device for people with dementia
Our Solution
PainChek : World s first pain assessment and monitoring medical device for dementia State-of-art technology Smartphone technology Point-of-care app Secure cloud-based data management Automated facial recognition Real time automated facial analysis Evidence-based solution Clinically validated tool 4 areas of the face 9 objective facial descriptors (facial microexpressions) Communicative, protective and subtle behaviours
PainChek : Medical Device TGA registration and CE Mark July 2017 Class 1 medical device
PainChek : Design Point-of-care App on a smart device Portability Automation and digitisation Hybrid model Automated facial analysis Binary (Yes/No) guided checklist
PainChek Continuous monitoring Pain Chart Longitudinal assessment Temporal patterns and trends Patient s pain profiling
PainChek : Description 6 domains Domain 1: Face Domain - Ekman s Facial Action Code System (FACS), 1978 Domain 2-6: Voice, Movement, Behaviour, Activity, Body - American Geriatric Society Indicators of Persistent Pain (AGS, 2002) 42 items Comprehensive - Communicative, protective and subtle pain features
PainChek : Administration Binary scoring (Yes/No) Domain 1: Face Domain - Automated facial video analysis - Computer vision algorithm Domains 2-6 - Digital record of non-facial pain cues Automatic calculation of final pain score and severity
PainChek vs Abbey Pain Scale
Domain 1: Facial Analysis
Domain 2: The Voice
Domain 3: The Movement
Domain 4: The Behaviour
Domain 5: The Activity
Domain 6: The Body
Automated Scoring Final score= total domain scores Numerical value indicates pain intensity - No pain: 0-6 - Mild pain: 7-11 - Moderate pain: 12-15 - Severe pain: 16-42 Pain intensity groups are comparable to Abbey Pain Scale scoring
PainChek Clinical Studies
PainChek : Evidence-Based Solution Study 1 Objectives To determine the relationship between PainChek and Abbey Pain Scale (APS) pain scores To investigate psychometric properties of PainChek when compared to the APS
Study 1: Methods Setting 3 metropolitan RACF in Perth Subjects Residents with moderate to severe dementia (MMSE < 19) Purposive convenience sampling technique Consent Resident s power of attorney Ethics HREC Curtin University Participating aged care groups 44
Study 1: Protocol RA Carers Blinded assessments Part of routine care: At rest and with movement Care workers assisted with answering some questions about residents behaviours e.g. sleeping/eating pattern 45
Study 1: Results Residents N = 40 70% females Mean age: 80.0 ± 9.1 years Age Range: 60-98 years 39 Caucasians, 1 Asian Various pain conditions 46
Study 1: Results Atee et al. Journal of Alzheimer's Disease, vol. 60, no. 1, pp. 137-150, 2017
Study 1: Results Atee et al. Journal of Alzheimer's Disease, vol. 60, no. 1, pp. 137-150, 2017
Validity & Reliability Findings Validity: the ability of the tool to measure what is intended to measure Reliability: the degree to which an assessment tool produces stable and consistent results 49
Concurrent Validity Activity Correlation Coefficient (95% Confidence Intervals) Overall (epat Score vs. Abbey Pain Scale Score) 0.88 (0.86-0.90) Rest 0.88 (0.84 0.91) Movement 0.89 (0.86 0.92) Guide to Correlation Coefficient: ± 0.70-1.00 Strong ; ± 0.30-0.69 Moderate; ± 0-0.29 None (0) to weak
Discriminant Validity Pain scores higher with movement versus at rest when assessed using either the APS or PainChek. Association between PainChek and APS scores was not influenced by the timing (at rest or with movement; p=0.795). 51
Reliability Data Guide to Cohen s Kappa: κ<0.60 = inadequate; κ =0.61-0.80 = good; κ=0.81-1.00 = very good Atee et al. Journal of Alzheimer's Disease, vol. 60, no. 1, pp. 137-150, 2017
Study 1: Conclusions PainChek has demonstrated excellent performance against the APS regardless of involved activity when assessing clinical pain in patients with moderate to severe dementia. 53
PainChek : Evidence-Based Solution Study 2 Objectives To investigate psychometric properties of PainChek when compared to the APS To confirm previous findings
Study 2: Results Dementia and Geriatric Cognitive Disorders, vol. 44, no. 5-6, pp. 256 267, 2017
Study 3: Clinical Implementation Clinical trial (Trial ID: ACTRN12616001003460) Primary aim To evaluate the feasibility and clinical effectiveness of the epat (PainChek ) in assessing pain among residents with moderate-to-severe dementia. Secondary aim To determine the clinical impact of the use of epat (PainChek ) pain management and associated pain-dependent behavioural outcomes such as agitation and aggression in residents with moderate-to-severe dementia and their management. 56
Implementation Studies Protocol 57
Stage 1. Pilot Implementation Study: Inter-rater Reliability This was a sub-study of the Pilot Implementation Study in which data obtained for the Familiarisation Period was used to assess the reproducibility of epat pain scores when independent raters assessed pain in the same subjects Protocol Single aged care site in WA Residents subsample: n=10, 40% male Raters subsample: n=11 Clinical Nurse (n=1), Registered Nurses (n=4), Enrolled Nurses (n=5), and Care Worker (n=1) Results 76 assessments (Rest =19 pairs, Movement =19 pairs) over 2 weeks Pain category agreement Overall [combined (Rest + Movement)]: 84% (i.e. excellent agreement) During Rest: 100% (i.e. perfect agreement)
Stage 1. Pilot Implementation Study Inter-rater Reliability Activity Weighted Kappa (κw) 95% CI Rest n=38 Movement n= 38 0.72 0.58-0.86 0.66 0.48-0.84 Guide to Cohen s Kappa: κ<0.60 = inadequate; κ =0.61-0.80 = good; κ=0.81-1.00 = very good
Pilot Implementation Study Impact of Episode and Occasion on Pain Scores N (%) Total pain scores (Mean ± SD) Automated facial score (Mean ± SD) Episode No pain 50 (65.8) 3.6 ± 1.7 * 1.9 ± 0.8 Pain 26 (34.2) 9.3 ± 2.9 * 2.5 ± 0.6 Occasion Rest 38 (50%) 4.0 ± 2.2 1.7 ± 0.7 Movement 38 (50%) 7.3 ± 3.7 2.5 ± 0.6 * The two-tailed P value is less than 0.0001 The two-tailed P value equals 0.0012 The two-tailed P value is less than 0.0001
PainChek Current Insights
PainChek Impact on Care
Impact on Care Delivery Simple to use, point of care App Improve patient outcomes Designed to be used by anyone (healthcare professionals and carers) who cares for people for dementia It will empowers carers, enhancing their roles Facilitate workforce transformation within aged care
PainChek Website www.painchek.com
Q & A session
Acknowledgments
PainChek Solving the silence of pain in dementia Mustafa Atee Research Fellow Faculty of Health Sciences, School of Pharmacy & Biomedical Sciences Curtin University, WA