Multi-Center Validation of Pain Assessment in Advanced Dementia (PAINAD) Scale in Malaysia

Abstract

The detection of pain in persons with advanced dementia is challenging due to their inability to verbally articulate the pain they are experiencing. Pain Assessment in Advanced Dementia (PAINAD) is an observer-rated pain assessment tool developed based on non-verbal expressions of pain for persons with severe dementia. This study aimed to perform construct validation of PAINAD for pain assessment in persons with severe dementia in Malaysia. This was a prospective cross-sectional study conducted from 27 April 2022 to 28 October 2022 in eight public hospitals in Malaysia. The PAINAD scale was the index test, and the Discomfort Scale—Dementia of the Alzheimer Type (DS-DAT) and Nurse-Reported Pain Scale (NRPS) were the reference tests for construct and concurrent validity assessment. Pain assessment for the study subjects was performed by two raters concurrently at rest and during activity. The PAINAD score was determined by the first rater, whereas the DS-DAT and NRPS were assessed by the second rater, and they were blinded to each other’s findings to prevent bias. PAINAD showed good positive correlations ranging from 0.325 to 0.715 with DS-DAT and NRPS at rest and during activity, with a p-value of <0.05. It also demonstrated statistically significant differences when comparing pain scores at rest and during activity, pre- and post-intervention. In conclusion, the PAINAD scale is a reliable observer-rated pain assessment tool for persons with severe dementia in Malaysia. It is also sensitive to changes in the pain level during activity and at rest, pre- and post-intervention.

1. Introduction

Pain is defined as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage or described in terms of such damage” [1]. The sensation of physical discomfort could be due to tissue damage giving rise to nociceptive pain or nerve damage causing neuropathic pain [2]. Pain can be accentuated by psychological factors such as depression or anxiety. Psychogenic or functional pain can arise in the absence of any tissue or nerve damage and is known to be an emotional expression of an underlying stress, psychosocial disorder, or unexpressed emotional conflict [3].

Pain has detrimental effects on the well-being of older adults. It can precipitate delirium, hinder activities of daily living, and cause instability and immobility. It may also disrupt the physiological functions of an already frail older person by increasing the sympathetic drive, causing increased heart rate and cardiac output that could precipitate myocardial ischemia; reducing the breathing or cough effort, causing an increased risk of hypoxemia and pneumonia; and also precipitating ileus or constipation [4].

Dementia is a general term used to describe progressive cognitive decline, which impairs activities of daily living. There are many subtypes of dementia; Alzheimer’s disease is the most common cause of dementia, followed by vascular dementia, mixed dementia, dementia with Lewy bodies, frontotemporal dementia, etc. [5]. Pain is common among people with dementia, and more than half of them suffer daily pain [6]. A study showed that 72% of people living with dementia in nursing homes experienced nociceptive pain, and 25% experienced mixed nociceptive and neuropathic pain [7]. Another study showed that around 5.8–14.0% of people living with dementia in nursing homes experienced acute pain during the study period, 14.8–22.3% suffered from chronic pain, and 3.4–10.1% had both acute and chronic pain [8].

Generally, people with dementia can still express their feelings and describe their pain during the early and moderate stages of dementia. As the disease progresses, managing pain in advanced dementia becomes very challenging due to their inability to verbally articulate the pain being experienced; pain is always underrecognized in this group of vulnerable patients. The Ministry of Health Malaysia Guideline on Pain Management in Adult Patients, 3rd Edition, suggests that the FLACC (Face, Legs, Activity, Cry, Consolability) scale can be used for cognitively impaired elderly patients [4]. However, the evidence for FLACC in cognitively impaired elderly is scarce. FLACC was originally created for use in young children [9], not for cognitively impaired elderly; thus, these two groups of patients might behave differently during pain.

A systematic review of pain assessment tools in older adults with dementia showed that about 24 pain assessment tools are available to date [10]. In this study, it is recommended to use a self-rated tool whenever possible in persons with dementia; as the disease progresses and patients become unable to rate themselves, then an observer-rated scale is recommended. Among all the pain assessment scales for use in those with dementia, the Pain Assessment in Advanced Dementia (PAINAD) scale has been gaining popularity in recent years because it is user-friendly and does not require extensive training [11,12]. It is practical for use in daily clinical settings, with moderate to good consistency for detecting pain in persons with advanced dementia. It is also useful for gauging improvements following interventions for pain.

The PAINAD scale was first introduced in 2003 [13] and several validation studies have subsequently been performed, demonstrating good reliability and consistency in pain assessment in persons with advanced dementia [14,15,16,17,18,19]. It consists of five domains: breathing, negative vocalization, facial expression, body language, and consolability. Each domain is rated on a 3-point scale according to the specific descriptions provided for different levels of pain. A group in Singapore reported concordance of the PAINAD and Nurse-Reported Pain Scale (NRPS) in detecting pain among nursing home residents with moderate to severe dementia [11]. Discomfort Scale—Dementia of the Alzheimer Type (DS-DAT) and PAINAD were significantly correlated in a small cohort of persons with severe dementia with the absence of verbal communication [13]. PAINAD was validated in reference to the Disability Distress Assessment Tool (Dis-DAT) in the United Kingdom but significant false positives were found among patients with psychological distress [14].

Pain assessment in people with advanced dementia heavily depends on non-verbal pain behavior manifestation. However, people living with dementia from different ethnic and cultural backgrounds may express different non-verbal pain behaviors [20]. PAINAD has not been validated as a pain assessment tool among persons with severe dementia in Malaysia. Hence, it is essential to run a construct validation study before we promote the use of this pain assessment tool in our daily clinical practice. The objective of this study was to assess the construct validity of the PAINAD scale in local settings by comparing the performance of the PAINAD scale with that of DS-DAT and NRPS in identifying different pain levels among persons with severe dementia.

2. Methods

2.1. Study Design, Setting, and Duration

This prospective cross-sectional study was conducted from 27 April 2022 to 28 October 2022 in 8 public hospitals in Malaysia. It was approved by the Medical Research and Ethics Committee, Ministry of Health Malaysia on 27 April 2022 with the project identification code of NMRR ID-22-00274-0C3 (IIR). Six of the study centers are located in Peninsular Malaysia—Hospital Kuala Lumpur, Hospital Tengku Ampuan Rahimah, Hospital Taiping, Hospital Tuanku Ja’afar, Hospital Sultan Ismail, and Hospital Sultanah Nora Ismail—while Sarawak Heart Centre and Hospital Queen Elizabeth are in East Malaysia. Every study center has in-house geriatrician(s) who functioned as the principal investigators at their sites, and the recruitment of study participants was conducted in either the General Medical Wards or Geriatric Wards of the respective centers.

All persons with severe dementia admitted to General Medical Wards or Geriatric Wards during the study period were recruited after consent was obtained from their next of kin or legally acceptable representative (LAR), regardless of the reason for admission or pain status on admission. The diagnosis of severe dementia was made based on the criterion of major neurocognitive disorder, at the severe stage in DSM-5 [21]. Patients who were ventilated and/or sedated were excluded from this study.

The targeted sample size was set as 100 participants because this number can provide a good power of estimation in a validation study, and we set the minimum sample size at 25 by the rule of a sample size at least five times the number of items [22,23].

2.2. Study Procedures and Data Collection

Demographic data, including age, gender, ethnicity, type of dementia and year of diagnosis, reason for current admission, and prescribed medications at the time of the study, were obtained once each participant was enrolled in the study. The pain assessment was conducted by the co-investigators at any time after enrolment. Patients who were discharged and readmitted within the study period were reassessed as new study participants. Training for all the co-investigators was conducted before the recruitment period to ensure standardization of the pain scales’ administration. The pain scales used in this study were PAINAD (Appendix A), DS-DAT (Appendix B), and NRPS (Appendix C). PAINAD was the index scale, whereas DS-DAT and NRPS were the reference scales for concurrent validation. These two scales were chosen as reference scales based on the previous validation performed in the original PAINAD validation study and Singapore validation study [13,19].

The pain assessment was conducted by two raters administering the different pain scales simultaneously. The first rater assessed the pain using the PAINAD scale, while the second rater assessed the pain using DS-DAT and NRPS. They observed the study participant at rest and during activity (bathing/sponging/transfer/therapy session) for 5 min before they administered the pain scale(s). The two raters were blinded to each other’s findings. If the PAINAD scale score was 3 or above, the participant was considered to be in pain. According to the study protocol, the rater then alerted the treating clinician regarding the finding as soon as possible, so that proper pain management could be carried out according to the standard practice in the local setting. The same pain assessment procedure was performed again 24–48 h after the initial assessment in a similar manner for those participants who scored 3 or more on the PAINAD scale. Any new medication or intervention started by the treating clinician was recorded.

2.3. Statistical Analysis

The data analysis was performed using IBM SPSS version 21. Demographic data are presented in a descriptive manner. The Kolmogorov–Smirnov test and kurtosis value were used to assess the normality of data; the difference between pain scores at rest and during activity was evaluated by using the paired t-test or Wilcoxon signed rank test, depending on the normality of the data distribution. The same statistical analysis method was also used to assess the difference in pain scores pre- and post-intervention. Examining the PAINAD scale’s ability to differentiate pain levels was part of the construct validity test. At the same time, concurrent validity among the PAINAD scale, NPRS, and DS-DAT was analyzed by using the Pearson or Spearman correlation test, depending on the normality of the data. All statistical analyses were two-tailed, and a p-value of <0.05 was set as a statistically significant point.

3. Results

A total of 34 participants were recruited in this study, as shown in

Table 1. Demographic data.

Variables (N = 34)	n (%)
Age (years, mean ± SD)	79.76 ± 7.85
Duration of illness (years, mean ± SD)	4.88 ± 3.63
Gender
Male	9 (26.5)
Female	25 (73.5)
Ethic
Malay	11 (32.4)
Chinese	18 (52.8)
Indian	2 (5.9)
Aborigine	3 (8.8)
Subtype of dementia
Alzheimer’s disease	14 (41.2)
Vascular dementia	7 (20.6)
Mixed AD and VaD	8 (23.5)
Other	2 (5.9)
Unsure	3 (8.8)
Initial PAINAD score ≥ 3	14 (41.2)

; none of the participants were readmitted. Their mean age was 79.76 years old (SD ± 7.85), predominantly female (73.5%). In terms of ethnicity, Chinese comprised 52.8% of the study population, followed by Malay, 32.4%; Aborigine, 8.8%; and Indian, 5.9%. Concerning the subtype of dementia, the majority were diagnosed with Alzheimer’s disease (41.2%), followed by mixed Alzheimer’s disease and vascular dementia (23.5%) and by vascular dementia alone (20.6%). The rest were other types of dementia or unsure of the exact subtype of dementia diagnosis. On average, the participants of this study had been diagnosed with dementia 4.88 years (SD ± 3.63) before recruitment, but the duration of illness varied widely, ranging from 1 year to 18 years.

The mean PAINAD score upon recruitment was 0.85 (±1.58) at rest and 2.79 (±2.27) during activity. There were 14 patients who scored 3 or more on the PAINAD scale during the initial assessment, and the findings were highlighted to the treating physicians for further pain management. Six of them were prescribed analgesia for pain management, five received both pharmacotherapy and non-pharmacotherapy, one patient received only non-pharmacotherapy, and two patients were put under observation, with no specific intervention performed in the first 24–48 h after the initial assessment.

The paired t-test was used to compare pain scores at rest and during activity; all pain scales demonstrated significantly higher scores during activity than at rest, as shown in

Table 2. Comparison of pain scores at rest and during activity.

	Mean Difference	Std Deviation	Std Error Mean	95% Confidence Interval of the Difference		p-Value
				Lower	Upper
PAINAD activity–PAINAD rest	2.213	1.922	0.208	1.649	2.777	˂0.001
DSDAT activity–DSDAT rest	5.021	6.208	0.906	3.198	6.844	˂0.001
NRPS activity–NRPS rest	0.804	0.866	0.126	0.639	1.148	˂0.001

. The PAINAD score was, on average, 2.213 (SD ± 1.922) higher during activity than at rest; the mean differences in the DSDAT and NRPS scores were 5.021 (SD ± 6.208) and 0.804 (SD ± 0.866), respectively. All these differences were statistically significant with p-values of <0.001. Additionally, the PAINAD scale showed positive correlations in the Pearson correlation test with DSDAT and NRPS, both at rest and during activity; all p-values were less than 0.05 (

Table 3. Correlation matrix between PAINAD, DSDAT, and NRPS at rest and during activity.

Pain Scales	At Rest		During Activity
	DSDAT	NRPS	DSDAT	NRPS
PAINAD	0.680 **	0.325 *	0.715 **	0.527 **
DSDAT		0.506 **		0.552 **

* p < 0.05, ** p < 0.01.

). At rest, positive correlations were shown between PAINAD and DSDAT (0.680, p < 0.01), between PAINAD and NRPS (0.325, p < 0.05), and between DSDAT and NRPS (0.506, p < 0.01). During activity, positive correlations were also demonstrated between PAINAD and DSDAT (0.715, p < 0.01), between PAINAD and NRPS (0.527, p < 0.01), and between DSDAT and NRPS (0.552, p < 0.01).

However, in further data analysis of those participants who scored 3 or more during the initial PAINAD assessment, only the PAINAD scale demonstrated a statistically significant difference (mean difference 1.462, SD ± 1.050) when comparing pre- and post-intervention pain scores during activity. There were no significant differences between the first and second assessed scores for the DSDAT and NRPS (

Table 4. Comparison between pain scores before and after pain intervention.

		Mean Difference Pre- and Post-Intervention	Std Deviation	Std Error Mean	95% Confidence Interval of the Difference		p-Value
					Lower	Upper
PAINAD	At rest	0.923	1.754	0.487	−0.137	1.983	0.082
	During activity	1.462	1.050	0.291	0.827	2.096	<0.001
DSDAT	At rest	−0.385	4.234	1.174	−2.943	2.174	0.749
	During activity	2.000	5.508	1.528	−1.328	5.328	0.215
NRPS	At rest	0.077	0.277	0.077	−0.091	0.245	0.337
	During activity	0.000	1.080	0.300	−0.653	0.653	1.000

). The pre- and post-intervention PAINAD score at rest also showed no significant difference.

4. Discussion

The majority of persons with advanced dementia (PWAD) have lost their ability to communicate effectively, but they may express themselves via body language, facial expression, and vocalization. Over the past two decades, many observer-rated scales for discomfort in PWAD have been developed to assist healthcare workers in providing better quality of life during end-of-life care [6]. DS-DAT was the earliest observer-rated objective assessment tool, developed in 1992 to assess discomfort in PWAD [24]. It was chosen as one of the reference scales for our PAINAD validation study because it is a comprehensive assessment tool consisting of nine domains with a score range of 0–27; it can provide a more solid construct and content validity compared to other pain scales. However, the scoring system is complex and time-consuming, and substantial training is required for healthcare workers to master the skill, which makes it less practical for routine clinical use in busy wards or clinic settings [25].

PAINAD, which was developed later, offers a more practical alternative [13] because it is easier to administer and only minimal training is required. In our construct validation study, PAINAD showed a good positive correlation with DS-DAT and NRPS across multiple inpatient healthcare facilities in Malaysia. PAINAD demonstrated consistent performance in other countries’ validation studies [14,15,16,17,19], showing that it can be used in multi-ethnic populations without significant cultural bias. In addition, PAINAD also has good performance in differentiating the different pain levels experienced by PWAD during activity and at rest.

Both PAINAD and DS-DAT scores were higher during activity in our study, suggesting that the discomfort may be due to somatic pain. Comparing the scores of the three pain scales after pain intervention (Table 4), only PAINAD during activity showed a significant difference in our cohort. This highlights its potential role in monitoring clinical responses in patients with dementia. On the other hand, it may suggest DS-DAT to be overly complex or NRPS to be insensitive in detecting pain level changes after intervention. A study in the U.K. demonstrated a significant decrease in PAINAD scores following pain treatment, which included medications, positioning of the body, dental treatment, and others. However, the same study also noted that about one-third of the elevated PAINAD scores among the study participants were false positives, suggesting that the scale may detect other psychosocial distress [14]. Furthermore, assessment should always be accompanied by appropriate monitoring and management. Bettina et al. demonstrated that agitation was significantly reduced with individual daily pain treatment according to a stepwise protocol [26].

NRPS is a proxy-rated pain scale for PWAD, derived from the Verbal Descriptor Scale (VDS) or Verbal Rating Scale (VRS) [27]. It consists of four categories: no pain, mild pain, moderate pain, and severe pain. The nurse in charge of the participants completed the NRPS based on their observation and experience. We chose this scale as one of the reference tests for validation because it is easy to administer and has good agreement with PAINAD in inpatient and nursing home settings [26,28]. We observed a moderate correlation between these two assessment tools, but NRPS did not show a significant difference after pain intervention. The nursing shift-working system and rapid turnover may hinder accurate pain estimation by nurses in local healthcare facilities. NRPS may still be a valuable method for nurses to monitor pain due to its simplicity, but the implementation of PAINAD assessment among nursing staff will confer additional value.

The mediation of pain is complex. It involves several neuronal systems mediating cognitive-evaluative, sensory-discriminative, and behavioral aspects [29]. The neurodegenerative process of dementia may affect how pain is mediated and its severity. Hence, the rationale for adopting behavioral observation scales such as PAINAD is not merely due to the lack of communicative ability but also because the manifestation of pain can be very different in PWAD. Under-recognition of pain or discomfort will lead to inadequate intervention and exacerbate neuropsychiatric symptoms. In order to enhance pain management in dementia, better education and effective facilitation of pain assessment within the organization have been proposed [29,30].

Our main study limitation was the small sample size; the number of participants recruited during the study period was only one-third of our target recruitment. We believe that this could be due to limited patients with advanced dementia being managed as inpatients or due to under-recognition of dementia, especially in general medical wards. The COVID-19 pandemic and the blooming of teleconsultation services might be one of the contributing factors to the low admission rate of PWAD during the study period. We also recognize the potential bias as assessors were not blinded to intervention during the second assessment. This was minimized by preventing the involvement of the assessors in managing pain.

5. Conclusions

In conclusion, the PAINAD scale is a reliable observer-rated pain assessment tool for persons with advanced dementia in Malaysia, as demonstrated in this study. PAINAD showed good correlations with DSDAT and NRPS during the assessment at rest and during activity. It was also sensitive to pain level changes during activity and at rest, pre- and post-intervention.