Pain Evaluation and Management in Patients With Limited Ability to Communicate Because of Dementia

Key Points

• Pain is challenging to assess in older adults with moderate to severe dementia because of limitations in ability to communicate pain because of cognitive and linguistic impairments.

• While older adults with mild to moderate dementia may be able to report on their pain using unidimensional scales, validated systematic observational approaches such as the PACSLAC-II or the PAINAD are needed to monitor pain levels in those with moderate to advanced dementia.

• Consensus guidelines for assessing pain in people with dementia are available.

• Research on nonpharmacologic pain management interventions is limited but does suggest that physical therapy, exercise therapy, sensory stimulation, and various comfort modalities show promise.

• Pharmacologic pain management research specifically focusing on older adults with dementia has been limited, and clinicians often extrapolate from studies based on younger persons without dementia.

• The pharmacologic modalities should be chosen based on goals of therapy, as discussed with the older adult and caregiver, and also tailored and monitored according to other medications used and concurrent comorbidities managed.

• A multimodal interdisciplinary approach of a combination of nonpharmacologic and pharmacologic modalities for pain may be particularly appropriate in older adults with dementia and should be tailored and monitored for safety according to goals of therapy, and other medications used to managed comorbidities.

Pain Assessment Using Self-Report

Although dementia compromises linguistic and cognitive abilities in general, self-report tends to remain a valid indicator of pain in early to moderate stages of the dementia process. A recent study of residents with cognitive impairment showed that less than half could self-report pain using a simple zero to ten numeric rating scale. Weiner et al. found that the mean Mini-Mental Status Examination (MMSE) scores in two samples of older adults with cognitive impairments, who were able to self-report pain, ranged from 18.6 to 22. Those unable to self-report pain had MMSE scores with a range of 12 to 13.1, classified as having severe dementia. Chibnall and Tait studied older adults with mean MMSE scores of 18 and found good support for the psychometric properties of simple self-report tools.

Interestingly, Ferrell et al. demonstrated that 83% of research participants with an average MMSE score of 12 could complete at least one of four simple unidimensional pain assessment tools (i.e. The Present Pain Intensity Scale [PPI] of the McGill Pain Questionnaire [MPQ], a verbal rating scale [VRS; involving four to six verbal descriptors indicative of different pain intensities], a 100 mm visual analog scale [VAS], the Rand Coop Chart and the Memorial Pain Assessment Card subscale ; the two latter tools are not used commonly in this population), with the Pain Intensity Scale (PPI) of the MPQ having the highest completion rates. The PPI is a zero to five scale that uses both numeric and verbal descriptors of pain intensity. The findings of Ferrell et al. imply that it may be beneficial to attempt several different unidimensional self-report tools (e.g. a zero to ten pain intensity numeric rating scale [NRS] anchored by polar opposites such as “no pain” and “most pain,” VRS, PPI) in older adults who have moderate dementia to determine which tool is the best suited for the individual patient.

Among the most commonly used pain self-report tools among younger adults is the horizontal visual analog scale (i.e. a 10 cm line anchored by the polar opposites “No Pain” and “Extreme Pain”). This scale, however, is not recommended for older adults because it has been associated with frequent unscorable responses in this population. Nonetheless, better results have been demonstrated through the use of vertical analog scales.

The Colored Analog Scale (CAS) has also been used with some success in samples of persons with dementia. The CAS is usually presented on a rectangular piece of plastic that looks like a ruler. On one side, there is a shape that starts thin at the bottom and gradually increases in width, and deepens in color (from light pink to red) as one moves toward the top. There are “no pain” to “most pain” anchors at the top and bottom of the scale. The patient is asked to move a plastic guide to indicate the location on the scale that best describes his or her pain. The CAS was originally developed for young children with marginal self-report skills but has been used successfully in samples of older adults with mild to moderate dementia. However, older adults with severe dementia are often unable to complete the tool. Scherder and Bouma suggested an approach that can be used to evaluate the patient’s ability to comprehend the scale (e.g. asking the patient to point to the scale, point to the plastic glide, point to where the glide should be placed when there is no pain).

Although no test can determine with certainty the extent to which a patient can self-report pain effectively, a rule of thumb has been suggested based on the literature mentioned above. Specifically, it would be reasonable to expect that an MMSE score of 18 or higher would indicate that a senior would likely be able to use unidimensional pain measures such as the zero to ten pain intensity NRS and the CAS. In contrast, those with MMSE scores below 12 would be more likely to have difficulty with the tools. Nonetheless, self-report should be attempted in all cases as some older adults with low MMSE scores may still be able to provide valid self-report. Based on the literature and our direct experience in clinical settings, we recommend that the NRS, PPI, and/or CAS be attempted. Where scales are presented visually (e.g. the CAS or the VRS), it is important to ensure that large fonts are used to compensate for visual impairments that become more common in older age groups. While multidimensional pain assessment scales (e.g. the short version of the Geriatric Pain Measure [GPM], ^, which assesses disengagement because of pain and pain with ambulation in addition to pain intensity), have been shown to be useful with older adults who are cognitively intact, research is needed on the development of suitable multidimensional pain assessment tools for older adults with moderate to advanced dementia.

Pain Assessment Using Observational Assessment Tools

Given the limited ability of many older adults with moderate to severe dementia to effectively self-report pain, there has been a proliferation of observational measures designed to assess pain in this population (see Hadjistavropoulos et al. for a review of various available tools). Many of these measures have limited validity evidence, but some have been studied more and have a substantial evidence base behind them. Among the most studied measures are the Pain Assessment in Advanced Dementia (PAINAD) and the Pain Assessment Checklist for Seniors with Limited Ability to Communicate (PACSLAC and PACSLAC-II) scales. ^,

The PAINAD consists of five items (Breathing, Negative Vocalization, Facial Expression, Body Language and Consolability), with each rated on a zero to two scale based on specific criteria. For example, Negative Vocalization is rated as follows: 0 = Normal; 1 = Occasional moan or groan, low-level speech with a negative or disapproving quality; 2 = Repeated troubled calling out, loud moaning or groaning, crying. The PAINAD has demonstrated good psychometric properties, including the ability to differentiate painful situations from non-pain related states, ^, and been translated in several languages.

Comprehensively covering all pain assessment domains recommended by the American Geriatrics Society, ^, the original PACSLAC scale consisted of 60 items organized into the following categories (Facial Expression, Activity Body Movement, Social Mood Personality, and Other, which includes physiologic changes, eating/sleeping changes and vocal behaviors). Each item is scored as present or absent, and the scale usually takes less than 5 min to complete. The PACSLAC has been shown to have good internal consistency and to differentiate pain related from non-pain related experiences. When used regularly in an LTC environment, the PACSLAC was shown to result in reduced pain levels (as assessed by LTC staff), increased use of pro re nata analgesic medication, and reduced nurse stress. In a study of nursing staff who used the PACSLAC, the PAINAD, and the Doloplus ^, (which is a validated tool that has been used widely in France with English translations available from the developers’ website), Zwakhalen et al. demonstrated that, while all three tools had good psychometric properties, nursing staff expressed a preference for the PACSLAC. Notably, the PACSLAC has been found to account for more variance in differentiating pain from non-pain related states than other commonly used observational pain assessment tools, even when several other tools were used in combination. The PACSLAC has been translated into several languages.

Despite the strong psychometric properties of the PACSLAC ^{, ,} , a newer version, the PACSLAC-II, was developed. The 31 item PACSLAC-II was intended to be a briefer scale that would maintain comprehensive coverage of the pain assessment domains recommended by the American Geriatrics Society, would exclude many items that frequently overlap with delirium, and would emphasize items informed by previous investigations, including laboratory research (e.g. specific facial responses that are frequently associated with pain). The psychometric properties of the PACSLAC-II have been supported. ^{, ,} The PACSLAC-II has been found to account for more variance in differentiating pain related from non-pain related states than other commonly used tools, including the original PACSLAC, even when several other tools were used in combination. Moreover, regular use of the PACSLAC-II in an LTC environment, combined with physician involvement, was shown to lead to reductions in benzodiazepines compared to a control group of patients whose caregivers did not use the PACSLAC-II regularly. The reductions in benzodiazepines may have been because of a better understanding of agitated behaviors that could have resulted from pain instead of another source of distress. Other research has suggested that, under professional guidance and direction, laypersons (e.g. informal caregivers) should be able to complete the PACSLAC-II and the PAINAD in a valid fashion allowing for the provision of important pain assessment information by family members to health professionals. More recently, an app version of the PACSLAC-II, with the capability of graphing patient scores over time (allowing LTC staff to observe deviations from patients’ regular pattern of scores), was developed and evaluated with good results. The PACSLAC-II has been translated into French, Farsi, and other languages. ^, A copy of the PACSLAC-II has been included as an Appendix of this chapter.

Recently, Kunz et al. developed a 15 item “meta-tool” for pain assessment in “impaired cognition, especially dementia,” the Pain Assessment in Impaired Cognition-15 (PAIC15) using items from preexisting scales such as the PACSLAC and the PAINAD. Tools that were published after 2012 (e.g. the PACSLAC-II) were not considered. The selection of items and tools was based on expert consensus and studies that evaluated some of the psychometric properties of candidate items. Although promising, a limitation of this work is that a direct comparison of the validity of the PAIC15 with that of other well validated tools in assessing pain in dementia was not reported. Pending such a direct comparison, the clinical value added because of the PAIC15 introduction remains to be determined through additional research.

On occasion, cutoff scores for observational pain assessment tools have been proposed in the literature. However, we have explicitly argued against such practice on the following grounds: a score on an observational assessment tool would vary as a function of situational and personal factors such as duration of observation (e.g. longer observation periods would likely result in higher scores than short observation periods) and the nature of the dementing illness (e.g. dementias primarily affecting the frontal lobes would like result in more disinhibited pain responses compared to dementias primarily affecting the posterior part of the brain). Other factors, such as facial paralysis or mobility limitations, would also affect the repertoire of pain behaviors being displayed. The wide variety of factors, besides pain, that can affect scores on observational tools implies that the use of universally applicable cutoff scores would be problematic. As an alternative to the use of cutoff scores, it has been suggested that an idiographic approach be adopted wherein each patient is assessed regularly over time, and clinicians look for fluctuations in the patient’s typical pattern of scores. Unexpected increases or decreases from such a pattern of scores would imply changes in status.

Other Dimensions of the Pain Experience

The clinician needs to be cognizant that persistent pain affects a wide variety of quality of life dimensions, including mood and level of agitation. As such, a thorough assessment of the persistent pain would also involve consideration of such dimensions. Useful psychometric evaluation tools for mood, agitation, and quality of life, specifically designed for people with moderate to severe dementia, rely on input from knowledgeable informants and include the Cornell Depression in Dementia, Alzheimer Disease Related Quality of Life-Revised tool, and the Cohen-Mansfield Agitation Inventory.

It is also important to acknowledge that multiple medical and psychological (particularly depressive) conditions and syndromes contribute to the pain experience (e.g. multiple pain diagnoses, sleep disturbance, musculoskeletal pain, chronic wounds, neuropathy), delirium ^, and that the pain itself is a result of a multifactorial pathway. Assessing and addressing multimorbidity, often through careful review of the medical record, will be important for effective and safe pain management.

A Clinical Approach to Assessment of the Older Adult With Limited Ability to Communicate Because of Dementia

Several guidelines and consensus groups have described clinical approaches designed to optimize and facilitate the assessment of patients with dementia. ^, These approaches have much in common. A general approach to the pain assessment of older adults with cognitive impairments, reflecting key points presented in various guidelines, is outlined in Table 79.1 .

One of the most recent and up-to-date guidelines for pain assessment in the patient unable to self-report is the 2019 Position Statement of the American Society for Pain Management Nursing. The statement emphasizes the following general recommendations: 1) be aware of potential causes of pain; 2) attempt self-report; 3) observe patient behaviors; 4) solicit proxy reporting of pain behaviors and activity changes; and 5) attempt an analgesic trial. The position statement recommends using validated assessment tools as appropriate (minimizing the emphasis on vital signs for pain assessment, regularity of pain assessment), and follow up assessments following the intervention. When soliciting proxy reports of pain, clinicians should note that proxy and patient reports of pain do not always correspond. Proxy reports should be considered in the context of all other sources of information about the patient, including results of the physical examination. Moreover, in many instances, nonpharmacologic interventions may be appropriate prior to attempting an analgesic trial.

Because of concerns about limited personnel and other resources that characterize many LTC facilities, a group of geriatric pain and public policy experts recommended a feasible model of pain assessment for LTC facilities. Specifically, they outlined the following recommendations:

• Every LTC resident should be assessed for pain on admission and at least once a week thereafter using a validated tool. Assessments would need to be more frequent when pain is suspected and/or if symptoms appear to escalate.

• A treatment plan should be documented in the chart within 24 h for patients with suspected moderate to severe pain. Depending on the circumstances, the treatment plan can be simple (e.g. administer acetaminophen) or more thorough following a medical examination.

• 24 h following treatment plan implementation, side effects of treatment should be evaluated and addressed.

• Pain should be re-evaluated within 24 h of treatment plan implementation.

Research has demonstrated that stakeholders (e.g. front line health professionals, LTC administrators) assessed these guidelines as feasible, desirable, and appropriate. Moreover, the feasibility of implementation of the aforementioned four clinical guidelines, with no significant added resources, has been demonstrated ^, although the sustainability of such guidelines over time would require ongoing management support. The impact of such guideline implementation on patient outcomes deserves further study.