Studies show that people with advanced dementia are at high risk for under-identification and under treatment of pain. To date, no valid, sensitive and specific pain intensity measure for persons with dementia exists.
The goal of this study is to develop a pain intensity measurement strategy for persons with dementia and limited ability to self-report pain.
Aim 1: Develop an efficient and specific measure comprised of behavior observations that best discriminate pain from other constructs.
Aim 2: Evaluate the validity of PIM-D scores in the cross-sectional sample of NH residents.
Aim 3: Evaluate the responsiveness of the PIM-D to analgesic therapy.
Aim 4: To evaluate the incremental validity of PIM-D scores compared to existing pain behavior measures.
Item reduction was achieved using a modified Delphi method followed by a statistical analysis of the data to identify items that best predict the Expert Clinician Evaluation of Pain. The sample is comprised of 207 residents with moderate or severe dementia from four Community Living Centers and 12 community nursing homes in Alabama and Southeastern Pennsylvania. Study measures included evidence-based observer-recorded indicators of pain (pain-related behaviors, surrogate pain ratings), and known correlates of pain (sleep, agitation, and depression). All participants were evaluated by an expert clinician who made summary assessments of pain intensity, which were used as the "gold standard" measure of pain.
Aim 1: Based on findings from the Delphi panel, we created a 39-item PIMD. We then applied three statistical approaches: 1) LASSO; 2) forward stepwise variable selection and 3) linear regression using items selected by expert opinion to build multivariate predictive models for expert clinician pain evaluation ("gold standard" measure). As a result, we reduced the PIMD to 7 items with each item scored on a 0 to 3 (0= "absent" to 3 = "severe").
Aim 2: Preliminary analyses demonstrated small to moderate, statistically significant correlations with variables (i.e., depression, agitation, sleep disturbances, painful diagnoses) that are known to be associated with pain. However for two variables, painful diagnoses and sleep disturbances, the correlations were negative, which was not expected. Further analyses are necessary to understand these findings.
Aim 3: As reported previously, we were unable to evaluate the responsiveness of the PIMD to analgesic therapy, due to insurmountable challenges in data collection. As an alternative, we examined differences in PIMD scores according to activity level, hypothesizing that scores during physical activities (e.g., transferring) would be significantly higher than scores for persons at rest. Our analyses upheld this hypothesis, although scores while ambulating were surprisingly low. We suspect this is because residents who ambulate probably experience less pain to begin with.
Aim 4: To evaluate the incremental validity of the PIMD, we examined the correlation between the PIMD and another widely used behavioral pain measure, the MOBID (R2 = 0.57, p < 0.001). This finding supports the validity of the PIMD, but additional work is necessary to examine if the PIMD outperforms the MOBID.
This project influences VHA patient care because: (a) VHA devotes significant resources to long term care (b) the prevalence of dementia in the veteran population is high and is increasing, and (c) the VHA has identified pain control as a priority clinical and research area. In addition to disseminating project findings via the typical academic channels, we also are working closely with the GEC leaders to communicate our findings to CLC leadership, clinicians and staff.
- Ersek M, Neradilek MB, Herr K, Hilgeman MM, Nash P, Polissar N, Nelson FX. Psychometric Evaluation of a Pain Intensity Measure for Persons with Dementia. Pain medicine (Malden, Mass.). 2018 Sep 10.
- Ersek MT. Palliative care in nursing homes: challenges and opportunities. Paper presented at: Supportive Care Coalition National Palliative Care Congress; 2013 Jun 3; Anaheim, CA.