Assessment of pain in patients with dementia. Development of a staff-administered behavioural pain assessment tool

Abstract

Aims: The aim of this thesis is to develop a staff-administered behavioural pain assessment tool for older persons with dementia, to test the instrument with respect to reliability and validity, and to use it in the clinical setting of an entire nursing home (NH) population. Methods: In Paper I, the development of the Mobilization-Observation-Behaviour-Intensity-Dementia (MOBID) Pain Scale was described. In MOBID, the assessment of inferred pain intensity was based on the patient’s pain behaviour in connection with standardized, active guided movements of different body parts. The internal consistency and inter-rater reliability of pain behaviour indicators and pain intensity scores were tested through bedside investigation and video recordings of 26 patients with severe dementia. Face validity was discussed by a focus group. Different aspects of construct validity were investigated. Paper II documented the extended testing of the intra-rater and inter-rater reliability of the pain behaviour indicators and pain intensity scores of the MOBID Pain Scale by three external raters, using video recordings, concurrently and independently on days 1, 4 and 8. In order to also assess pain from internal organs, the head and skin, an extended instrument, the MOBID-2 Pain Scale, was presented in Paper III. It comprised the original MOBID, renamed MOBID-2 Part 1, and MOBID-2 Part 2, which registered pain behaviour related to internal organs, the head and skin. Monitored over time, caregivers’ observations were registered on pain drawings and inferred into pain intensity. Finally, overall pain intensity was assessed, including all observations registered in Parts 1 and 2. The internal consistency of the comprehensive MOBID-2 was examined for 77 patients. Furthermore, the inter-rater and test-retest reliability of pain behaviour indicators, pain drawings and pain intensity scores were tested. Arguments for face, construct and concurrent validity were added when pain scores from nurses using MOBID-2 were correlated with physicians’ clinical examinations and other pain variables. Paper IV was a cross-sectional study exploring the relationship between severity and diagnoses of dementia and the use of pain medication and other parameters of pain measured using pain intensity scores from MOBID-2 in 181 NH patients. Results: The results of Paper I suggested that registration of pain behaviour indicators during standardised movements, as measured by the MOBID, can be used reliably to disclose pain intensity inferred by nurses in elderly persons with dementia. Internal consistency of the MOBID items was found to be high ( =0.90). The inter-rater reliability of inferred pain intensity scores was high to excellent (ICC=0.70-0.96), but varied between poor to excellent for individual pain behaviour indicators ( =0.05-0.84). Arguments for construct validity were indicated, as the MOBID Pain Scale revealed significantly more pain than did pain scores during regular morning care. Video observation demonstrated higher pain intensity than bedside scoring. The pain intensity scores were highly correlated with the number of observed pain behaviour indicators. Finally, the overall pain intensity score was more associated with the highest pain score among the test items than with the mean score of all items. As demonstrated by Paper II, facial expression of pain was most commonly observed, followed by pain noises and defence. Using video recording, inter-rater reliability was highest for pain noises, followed by defence, and facial expression ( =0.44-0.92, =0.10- 0.76, and =0.05-0.76, respectively, on day 8). Of the movements, mobilisation of arms and legs was rated most painful. The intra-rater and inter-rater reliability of the overall pain intensity scores was very good, ICC(1,1) ranging from 0.92 to 0.97 and 0.94 to 0.96, respectively. As opposed to observed pain behaviour, the reliability of pain intensity scores tended to increase on repeated assessment. It was suggested that the overall pain score was based more on interpretation of the most pain provoking movement during assessment than on the total number of observed pain behaviour indicators. Using the MOBID-2 Pain Scale, the prevalence of any pain in patients with severe dementia was 81%, with predominance in the musculoskeletal system, as demonstrated in Paper III. Most frequent and painful was mobilising of the legs and arms (Part 1). Pain in the pelvis and/or genital organs was frequently observed in MOBID-2 Part 2. The internal consistency of the whole scale was highly satisfactory ( =0.82-0.84). Moderate to excellent inter-rater and test-retest reliability was demonstrated for pain behaviour indicators ( =0.44-0.90 and =0.41-0.83) and pain drawings ( =0.46-0.80 and =0.48-0.93). Moderate to excellent interrater and test-retest reliability (ICC=0.80-0.94 and ICC=0.60-0.94) was shown for pain intensity scores. The inter-rater and test-retest reliability for the overall pain intensity score was excellent (ICC=0.92 and ICC=0.94). Arguments for concurrent validity were indicated, as the overall pain intensity of MOBID-2 as observed by primary caregivers was correlated with physicians’ clinical examinations and pain variables. Indication of construct validity was provided, as both Part 1 and Part 2 were satisfactorily correlated with the overall pain score. Part 1 was more highly associated with the overall pain score, suggesting that pain behaviour occasioned by standardised movements may represent a more concrete pain concept than the observation of pain from internal organs, the head and skin, monitored over time. Paper IV indicated that patients with severe dementia have similar intensity, diagnoses and locations of pain to patients in other stages of dementia. Pain intensity measured by MOBID- 2 scoring did not differ between diagnostic groups of dementia. Patients with dementia who received opioids were more likely to demonstrate higher pain intensity scores than mentally healthy controls receiving opioids. It was suggested that these patients received less pain relief than they needed. The isolated increase of opioids may be limited by the high prevalence of ICD diagnoses and opioid side effects. The patients’ multi-morbidity and lack of communication require a comprehensive approach to pain assessment and treatment in a multidisciplinary perspective. Conclusions: The MOBID-2 Pain Scale is based on patients’ pain behaviour in connection with standardised active, guided movements of different body parts (Part 1), and pain behaviour related to internal organs, the head and skin (Part 2). Research evidence was provided that lent credibility to MOBID-2 as a reliable and valid nurse-administered assessment tool for inferred pain intensity. Using MOBID-2 in a cross-sectional study, it was suggested that patients with severe dementia and mixed dementia are at great risk of suffering from severe pain. Validity testing of a behavioural assessment tool is difficult, because the pain scores are indirectly observed and inferred by proxies (nurses). Future research should include extended testing of concurrent validity, comparing the MOBID-2 Pain Scale with other observational pain tools for patients with dementia. Future research should also explore the prevalence of pain in Norwegian NHs, as the findings presented in this thesis were based on data from only one NH. Implemented in a quality improvement programme, the use of the MOBID-2 Pain Scale may be an important contribution to improving pain assessment and treatment in NH patients.