Pain, Opioid Analgesics, and Cognition: A Conceptual Framework in Older Adults

Abstract

Chronic pain is highly prevalent in older adults and is associated with poor functional outcomes. Furthermore, opioid analgesics are commonly utilized for the treatment of pain in older adults despite well-described adverse effects. Importantly, both chronic pain and opioid analgesics have been linked with impairments in cognitive function, though data are limited. In this manuscript we summarize the evidence and critical knowledge gaps regarding the relationships between pain, opioid analgesics, and cognition in older adults. Furthermore, we provide a conceptual framework to guide future research in the development, implementation, and evaluation of strategies to optimize analgesic outcomes in older adults while minimizing deleterious effects on cognition.

Introduction

In the United States, the number of persons aged 65 years and older is expected to double from 43 to 89 million by the year 2050 [1]. Among other implications, this increase will pose significant challenges to the healthcare system [2]. Chronic pain constitutes one key challenge, as the estimated prevalence of chronic pain in this population ranges from 45% to 80% [3–5], although there are limitations to these estimates. For example, pain symptoms may be viewed as a normal part of the aging process (e.g., musculoskeletal “wear-and-tear”) and may therefore be under-reported. Regardless of etiology, chronic pain is associated with numerous adverse outcomes, including increased healthcare resource utilization [6], decreased physical capacity [7, 8], inferior perception of health [9], and impaired quality of life [6, 8, 10]. Pain may also be associated with accelerated cognitive decline in older adults [11, 12]. Thus, chronic pain represents an increasingly important public health problem in this population.

Pain is often treated with analgesic medications. In particular, approximately 10% of older adults in the United States are receiving an opioid analgesic at any given time [13, 14], and approximately 10% of older adults who receive opioids proceed to long-term opioid use [15]. Apart from concerns that opioids may not be effective in managing chronic non-cancer pain, opioids may have significant deleterious effects in older adults. One issue is how opioids may affect cognitive function. While acute opioid administration may impact short-term cognitive performance [16], there is increasing concern that chronic use may have lasting cognitive effects [17]. However, it is difficult to distinguish potential effects of opioids themselves from the effects of pain, or the underlying conditions causing pain, on cognition. Thus, the relationships between pain, opioid use, and cognitive function are complex. Nonetheless, it is important to understand the mechanisms underlying these relationships to better accomplish the goal of adequately treating pain while maximizing preservation of cognitive function in older adults.

Herein we propose a conceptual framework to guide scientific investigations seeking to define the relationships between pain, opioid analgesics, and cognitive function in older adults. We present a brief summary of current evidence, propose potential confounders, mediators, and moderators of relationships, and identify significant gaps (Table 1) that remain in our understanding. This conceptual framework, while primarily designed for research applications, may ultimately assist in the development and implementation of analgesic strategies that improve functioning and minimize deleterious cognitive effects in older adults.

Table 1.

Significant gaps in our understanding of pain, opioids, and cognition relationships

Gap	Description and impact
Pain assessment in older adults with cognitive impairment	Incomplete understanding of the validity and applicability of pain assessment tools across varying levels of cognitive function; limited validation of multidimensional pain assessment tools in those with cognitive impairment; limited understanding of proxy reporting of pain; limited data on the relationships between behavioral symptoms and pain in those with advanced cognitive impairment
Longitudinal evaluation of cognitive function, pain, and/or opioid analgesic utilization	Few studies evaluate cognition longitudinally, which limits our ability to detect changes over time and establish temporal relationships; paucity of data evaluating longitudinal trends in pain features or opioid analgesic utilization, which limits our ability to assess relationships with cognitive function over time; failure to account for pain in studies of opioids and cognition (and vice versa)
Reliance on cognitive screening tests or clinical dementia diagnoses	Most studies rely on cognitive screening tests rather than formal neuropsychological testing, which limits our ability to understand mechanisms of observed relationships or identify early, subtle changes in domain-specific cognition; as cognitive function falls on a broad spectrum, reliance on cognitive screening tests or clinical diagnoses of dementia will miss patients with milder forms of cognitive impairment
Adjustment for key patient-related factors, such as medical and mental health conditions, sociodemographic features, medications, daily functioning and socialization, and genetic factors	Residual confounding is a ubiquitous concern in existing clinical literature, which may result in variable and conflicting results; many important factors that may confound, mediate, or moderate observed relationships between pain, opioids, and cognition are not included in existing models; the roles of concomitant medication use (e.g., non-opioid analgesics) and mental health features (e.g., depression) are unclear
Mechanistic features such as brain imaging or functional biomarkers	There is a paucity of evidence assessing changes in brain structure and/or function in older adults in relation to pain, opioid analgesic use, and cognitive function; biomarkers are notably absent; this impedes our ability to evaluate potential mechanisms or develop potential therapeutic targets

Methods

Conceptual framework development is generally based on identifying factors that are likely to be related to an outcome of interest (i.e., cognition in older adults), and then hypothesizing the potential relationships that exist between these factors and the outcome through the creation of a graphical framework. There is no standardized process or guidelines for conceptual framework development, although key steps include identification of relevant literature, expert discussion, and the creation of a consensus graphical depiction of proposed relationships [18]. While conceptual framework development differs from a systematic review where the goal is to summarize existing evidence to address a focused question, a standardized approach to identify relevant literature was utilized by entering the key words pain, opioids, cognition, cognitive function, and older adults into MEDLINE databases (original search date 6/1/2021, 51 articles in first screen), cross-referencing with review articles, and manual searching reference lists of identified studies.

Pain and Cognition

Pain is generally defined as an unpleasant sensory and emotional experience with substantial inter-individual variability [19]. It is often difficult to objectively assess, a challenge that may be magnified in the older adult, particularly in those with cognitive impairment. Previous studies have shown that pain is underreported in older adults, which is likely related to the widespread notion that pain is expected, and hence normalized, in this population [20–22]. Biophysical changes with aging may also contribute to different pain perceptions in older adults [23]. Specifically, aging has been associated with increased pain thresholds, reduced pain sensitivity, and impairment of endogenous pain inhibition systems [24].

Cognitive status may influence both the perception and reporting of pain [25, 26]. For example, several neurocognitive processes are known to influence the pain experience, including attention, memory (i.e., current pain experience is shaped by pain expectations from recalled experiences), and social/emotional processing [27]. These changes may be driven by degenerative changes to key neurological structures. Furthermore, anatomic disruptions to ascending or descending pain pathways (e.g., stroke, traumatic insult, atrophy) may impair normal pain signaling, perception, and reporting, along with emotional processing. These alterations may ultimately drive decisions regarding analgesic therapy, including opioids, which in turn may have reciprocal effects on underlying neurologic processes.

Most pain assessment tools in adults are dependent on preserved cognitive function. Therefore, obtaining information about a patient’s cognitive status prior to performing the pain assessment is important and may be performed utilizing standardized cognitive screening tools [28]. Examples of cognitive screening tools available include the Montreal Cognitive Assessment [29], Mini-Mental Status Examination [30], Mini-Cog [31], and General Practitioner Assessment of Cognition [32, 33]. A brief review of these screening tools, including strengths and limitations, is provided in Table 2. For patients with some degree of cognitive impairment that impairs accurate pain self-reporting, the use of standardized proxy-rating instruments may be helpful. However, such measures have not been broadly validated in this population [34–36]. This represents a critical gap when trying to optimize pain management in these individuals. Fortunately, pain assessment is possible even in the absence of proxy reporting. The applicability of the limited available tools to any individual patient largely depends on the severity of cognitive impairment. For example, for those with mild cognitive impairment who retain capability of pain self-reporting, many may adequately utilize unidimensional tools such as the numeric rating scale (NRS) [28]. Furthermore, multidimensional pain assessment tools such as the McGill Pain Questionnaire have been successfully employed in those with mild cognitive impairment [37, 38], although further work is clearly needed to validate multidimensional tools in patients or proxies across a spectrum of cognitive function. Those with more advanced cognitive impairment ultimately lose their ability to self-report pain. It is often difficult to determine if presumed or expressed signs and symptoms of discomfort are indicative of pain or manifestations of dementia-related behaviors. Pain assessment tools are available in those with advanced cognitive impairment, including the Pain Assessment in Advanced Dementia (PainAD) tool, which employs clinical observation to score 5 different behavioral items: breathing, negative vocalization, facial expression, body language, and consolability [39]. However, such pain assessment tools based solely on behavioral evaluation may lack specificity for pain complaints.

Table 2.

Summary of several commonly encountered cognitive screening tools available in clinical practice

	Features	Score Cut-offs	Limitations	Completion Time
MOCA [29]	Assesses domains of memory, visuospatial function, orientation, attention, language, and executive function; widely validated; high sensitivity for detecting mild cognitive impairment; available in multiple languages	<26 = abnormal	Longer completion time, mandatory training and certification, training cost but no cost to use tests clinically	10 minutes
MMSE [30]	Assesses domains of memory, visuospatial function, orientation, attention, language, and praxis; no training necessary; widely validated; available in multiple languages	<24 = abnormal	Outcomes vary by educational level; longer completion time; cost with purchasing tests	10–15 minutes
Mini-Cog [31]	Assesses domains of memory and visuospatial function; no training necessary; free to use; available in multiple languages	<3 = abnormal *Score <4 may be used as cut-off to detect milder forms of impairment	Limited assessment of various cognitive domains leading to lower sensitivity for detecting milder degrees of cognitive impairment	3 minutes
GPCOG [32, 84]	Contains a patient assessment and an informant questionnaire; No training necessary; free to use; available in multiple languages	>8 = cognitively intact <5 = cognitively impaired 5–8 = inconclusive (informant <3 = impairment)	Limited information on performance across cultures and languages	5 minutes

MOCA = Montreal Cognitive Assessment; MMSE = Mini-Mental Status Examination; GPCOG = General Practitioner Assessment of Cognition.

The influence of pain on long-term cognitive trajectories is not well understood, with only a few studies examining its association. In a longitudinal population-based study of approximately 10,000 adults greater than 62 years of age, individuals with persistent pain, defined by self-reporting of moderate-or-severe pain on two successive interviews performed 2 years apart, experienced greater cognitive dysfunction when compared to subjects without persistent pain [11]. However, this study was limited by its inability to address several important potential confounders, including opioid, psychotropic, and anticholinergic medication use, physical activity, and social participation. A more recent study utilizing participant data from the Framingham Heart Study found that the presence of widespread pain at a single assessment interval was associated with increased risk for dementia approximately 10 years later, although incomplete adjustment for confounding variables was similarly encountered [40]. Additional studies have noted that more than 50% of adults with chronic pain, not limited to older adults, experience impairments in everyday cognitive functioning, including in the domains of working memory and attentional functioning [41, 42].

Patients with severe dementia have also been reported to have a greater prevalence of pain symptoms than those with less severe impairment [43], although pain assessments have been limited by their dependence on non-standardized caregiver observations. The prevalence of chronic pain in patients with Alzheimer’s disease (AD) dementia has been estimated at approximately 45% [44]. Multiple studies have observed a positive correlation between pain intensity and dementia severity, albeit with imperfect pain assessment tools, suggesting concordance between the progression of pain and dementia symptoms [11, 45, 46]. Although worsening cognitive function may indeed be driven by persistent pain symptoms, it is also possible that underlying neurodegenerative changes may impact central pain signaling and processing, resulting in greater pain [47]. Furthermore, associations between chronic pain and cognitive changes in observational studies may be related to residual confounding, including factors such as analgesic medication use in those with chronic pain. These gaps in our understanding of the relationships between pain, analgesic medications utilized to treat pain, and cognitive function must be addressed through rigorous investigation, including detailed pain assessments. In addition, future studies must carefully consider that the relationships between cognition, pain, and analgesic use may be multi-directional.

It is also important to consider that dementia is a clinical syndrome in which daily functioning is impaired by cognition [48]. However, cognition, much like pain, falls along a spectrum, with many people developing milder forms of cognitive impairment. Mild impairment may eventually progress to dementia, which itself varies in severity. Furthermore, the underlying neuropathological processes leading to a clinical diagnosis of dementia are not uniform, resulting in numerous distinct dementia subtypes (e.g., Alzheimer’s disease, frontotemporal dementia, vascular dementia, Lewy Body dementia, Parkinson’s disease, and mixed dementia). Although it is important to define the dementia subtype, as this may affect prognosis and treatment options, the ability to reliably distinguish these may be limited in community settings and in those with more advanced disease. Further, it is increasingly recognized that these pathophysiological processes are often not mutually exclusive [49]. Investigations examining relationships between pain, analgesics, and cognitive function in older adults should carefully consider differential relationships that may be influenced by underlying neuropathological processes. For example, certain neurological conditions have an exceptionally high incidence of chronic pain (e.g., thalamic stroke) [26, 50]. Additionally, various neuropathologic processes are known to affect different brain regions, some of which may be more intricately linked to mood disturbances and pain signaling, processing, and modification. Thus, investigations based on indiscriminate diagnoses of dementia rather than detailed neurocognitive and neurological assessments may produce misleading results. There is a paucity of data assessing pain across dementia subtypes [51], and it is unknown how changes in brain structure in older adults may relate to pain control, and vice versa.

Finally, pain varies by inciting factors, clinical phenotypes (e.g., neuropathic, nociceptive, radicular), the presence or absence of tissue injury, chronicity, location, and intensity. Further, pain, like cognition, is not static over time but requires longitudinal evaluation congruent with neurocognitive evaluations. Studies of the relationships between pain and cognitive changes in older adults must carefully consider the contributions of each of these pain features, recognizing that there is likely to be substantial heterogeneity across individuals that may mask or exaggerate perceived associations between pain and cognition in population-based studies.

Opioid Analgesics and Cognition

Opioid analgesics are frequently utilized for the treatment of noncancer pain, despite recent efforts to reduce exposures to prescription opioids in the face of a national opioid crisis. In a recent meta-analysis of data from more than 26,000 patients (median age 58 years) with chronic noncancer pain enrolled in 96 randomized clinical trials, opioid analgesics were associated with statistically significant but small improvements in pain and physical functioning when compared to placebo [52]. These improvements were generally similar to those noted with the use of non-opioid analgesics, including non-steroidal anti-inflammatory drugs (NSAIDs), antidepressants, and anticonvulsant medications. In older adults specifically, the efficacy of opioids for chronic noncancer pain are uncertain, with limited observational data showing mixed results on daily functioning and social engagement [53]. Notwithstanding, analgesic decisions should be patient-centered with thorough evaluation of risks and benefits of therapy to the individual patient, as some patients may benefit from multimodal analgesic strategies that include prescription opioids [54, 55]. Regarding prescription opioids specifically, these medications are frequently prescribed for older adults and, in many circumstances, are clearly indicated (e.g., acute pain following major surgery or trauma, intractable malignancy-associated pain). In general, older adults have greater prescription opioid availability (i.e., the presence of an active opioid prescription) than their younger counterparts [56, 57]. However, concerns regarding appropriate medication management, the development of opioid dependence, and negative short- and long-term effects on cognition may lead some clinicians to refrain from prescribing opioids to their older adult patients and/or cause patients to underutilize them—although there are limited data exploring this possibility. Additionally, in older adults with underlying cognitive dysfunction, it is possible that prescription opioids are utilized for non-pain complaints, with recent studies highlighting increases in opioid utilization with parallel decreases in anti-psychotic medication use in those with dementia [58, 59]. There are at least two possible explanations for this: (1) clinicians may be increasingly treating non-pain dementia-related behaviors with opioids, perhaps as a way to minimize antipsychotic medication use; or (2) clinicians may be more frequently recognizing and targeting actual pain drivers of dementia-related behavioral disturbances with opioid analgesics. Indeed, there is compelling evidence that pain may indeed be an important driver of dementia-related behavioral issues [60]. The ramifications of either scenario are important; hence, work is clearly indicated to confirm and understand temporal changes in opioid analgesic utilization in this high-risk patient group. Most importantly, evidence-based analgesic strategies for the optimal treatment of pain in older patients with cognitive impairment remain to be defined.

It is important to recognize that the aging process leads to physiologic changes that may influence opioid pharmacodynamics, thereby affecting dosing and frequency of opioid administration [61]. Furthermore, the effects of concomitant medication use must be considered when evaluating the relationships between opioids and cognition. Multiple analgesics are often utilized simultaneously to target one or multiple underlying pain condition(s), including medications such as acetaminophen, NSAIDs, neuropathic agents, antidepressants, and opioids. Further, non-analgesic medication use is common in older adults, and polypharmacy has been associated with increased dementia risk [62]. Along with detailed assessments of the efficacy and short-term safety of concomitant medication use, assessments should extend to include long-term efficacy and safety data, most notably with information on sustained pain control, cognition, and other aspects of daily function (e.g., physical capacity, independence in activities of daily living).

Most initial studies comparing opioid analgesic use in older adults suggest a higher prevalence of use among those who are cognitively impaired [63]. In cross-sectional studies of more than 1 million older adults in Denmark, approximately 1/3 of those with dementia received an opioid medication in 2015 compared to less than 1/5 without dementia [58, 64]. In contrast, a cross sectional population-based survey in Finland noted lower analgesic use in those with dementia (33%) versus those without dementia (47%), though opioids were not distinguished from other analgesics [65]. In a recent US-based study focused on Medicare beneficiaries aged 65 years and older with a diagnosis of dementia (identified by ICD-9 diagnosis codes rather than formal clinical dementia assessments) [66], approximately 30% of patients (n = 737,839) were prescribed at least one opioid in 2014–2015 [67]. Similar proportions have been observed in long-stay US nursing home residents with moderate-to-severe cognitive impairment [68]. These findings suggest a high frequency of opioid analgesic availability among those with dementia. However, many questions remain. For example, the temporal patterns of opioid use and variations in use by geographic location are not well understood. In addition, definitions of how to quantify opioid utilization and characterize dementia have not been standardized between studies. Furthermore, opioids vary in their strength, routes of administration, duration of action, and metabolism, and it is critical to consider these features when assessing the safety and efficacy of prescription opioids in older adults. Finally, it is important to note that most US based studies predate the publication of the 2016 CDC (Centers for Disease Control and Prevention) guidelines on opioid prescribing. More research is needed to describe the impact of these guidelines on opioid use and prescribing trends at the patient and provider levels, respectively, in older adults with varying degrees of cognitive dysfunction.

Analgesics, including most notably opioids, may potentially impact long-term cognitive function, either positively or negatively. These relationships are likely relevant for patients across a broad spectrum of cognitive function. Although the mechanisms are not well understood, opioid-cognition relationships may be mediated by indirect or direct effects on circulating plasma cytokine concentrations, central pain signaling and psychosocial processing (i.e., depression, anxiety, self-efficacy beliefs), central cholinergic neurotransmission, and cortical and subcortical brain volumes [69–71]. To this end, brain structural and functional imaging are important components of mechanistic studies of brain aging, dementia, and chronic pain, though few studies have expanded the application of these methods to opioid analgesics. A limited body of evidence suggests that changes in brain morphometry (i.e., regional changes in gray matter volumes) may occur in those with long-term opioid use, including in regions involved in reward processing and cognition [72–74]. Most data are derived from the addiction medicine literature, assessing brain morphometry in young adults with opioid dependence, which is a distinct clinical group from older adults with chronic pain. A study of 11 adults (mean age 39 years) with chronic low back pain who were administered morphine daily for 1 month found that exposure to morphine was associated with decreased gray matter volume in several reward and pain-related regions, including the bilateral amygdalae, inferior orbitofrontal cortex, and pre-supplementary motor areas when compared to 10 chronic pain patients without morphine exposure [75]. Interestingly, these changes did not revert to baseline after cessation of opioid therapy. Similar changes in amygdala volume were also noted in 10 prescription opioid-dependent patients (mean age 30 years) [76]. There are no data regarding changes in brain morphometry with opioid use in older adults, although clearly this is an area that warrants further exploration.

Clinical data directly assessing long-term changes in cognition with prescription opioid use among older adults are limited [77], and available data are largely based on the utilization of cognitive screening tools rather than longitudinal assessment of global and domain-specific cognitive function. In the largest study to date (n = 3,434), the Cognitive Abilities Screening Instrument (CASI) was administered to participants at 2-year intervals to identify community dwellers at highest risk for dementia [78]. The hazard ratio (HR) for dementia was greatest in participants with the highest cumulative opioid exposure (HR 1.29, 95% confidence interval [CI] 1.02–1.62) compared to those with minimal or no opioid exposure. However, this study could not determine whether opioids were causally associated with risk of dementia or whether pain, which was not accounted for in the study, or other clinical or biochemical features were driving the observed relationships. To support the notion that pain could be a driver of observed relationships, participants with the greatest consumption of NSAIDs also had increased risk for dementia.

The Role of Other Patient-Related Factors

In addition to understanding the potential relationships between pain, opioids, and cognition, it is important to recognize other patient-related factors that may influence these relationships, including medical and mental health conditions, sociodemographic features, medication use, daily functioning and socialization, and genetic factors. Each of these features may confound, mediate, and/or moderate relationships between pain, opioids, or cognition. Additionally, they may serve as important covariates in models assessing cognitive outcomes (i.e., they may affect cognition but not necessarily pain or opioid use).

A simplified visual representation of these potential relationships is provided in a directed acyclic graph (Figure 1), in which the solid line between exposure (i.e., pain or opioids) and the outcome (i.e., cognition) represents a causal pathway. A confounder is a variable causally related to both the exposure and the outcome but does not fall on the causal pathway between exposure and outcome. Failure to adequately control for a confounder may result in the creation of an alternate path from the exposure to outcome, biasing the estimated relationship. As an example, depression may be associated with pain (or opioids) and may be causally linked to cognitive impairment. Without controlling for depression, investigators may find that pain is associated with cognitive impairment. However, after controlling for depression, this relationship may no longer be observed. A mediator is a variable that lies along the causal pathway between the exposure and outcome such that the relationship between exposure and outcome occurs, at least in part, through that variable. As an example, pain may lead to more frequent hospitalizations, and recurrent hospitalizations may lead to cognitive impairment. A moderator is a variable that may change the direction or magnitude of the relationship between exposure and outcome. For example, the relationship between the exposure of pain and the outcome of cognition may be different in patients who are married vs. those who are widowed or divorced. Finally, a prognostic covariate is related to the outcome but not the exposure. For example, genetic factors such as APOE e4 may confer risk for cognitive impairment but not pain. Adjustment for APOE e4 in this situation does not affect bias of the estimated relationship, but may improve precision.

Figure 1.

Schematic representation of the potential relationships linking pain to cognitive outcomes (see text for details).

Given a basic understanding of these potential relationships, we have provided an outline of some potential patient-related factors that may influence pain, opioids, and cognition relationships (Table 3). Briefly, medical conditions of interest include chronic pain, mental health diagnoses, and cardiovascular disease, amongst others. Sociodemographic features of interest include patient gender, education level, occupation, and marital status. Medications of interest include non-opioid analgesics, antidepressants, anticholinergics, and other centrally acting agents, which can often result in polypharmacy. Daily functioning and socialization features include physical function, sleep, family engagement, socialization, and independence. Finally, genetic factors include family history or genetic predisposition to chronic pain, substance use disorders, and cognitive decline, including dementia. Importantly, assessing all of these variables is unlikely to be feasible and/or may place substantial burden on study participants. As such, in the following conceptual framework (Figure 2), we highlight those variables that we deem to be of greatest potential importance in assessing the relationships between pain, opioids, and cognition, recognizing that other variables may also prove to be important.

Table 3.

Other patient-related factors which may influence pain, opioids, and cognition relationships

Category	Example of Potential Impact*
Medical conditions Chronic pain diagnoses Mental health conditions Cardiovascular disease Neurological disorders Diabetes mellitus Recurrent hospitalizations or frequent healthcare encounters	Mental health conditions such as depression may confound the relationships between pain and cognition, such that depression is associated with pain and is causally linked to cognitive impairment
Sociodemographic features Gender Education Geographic residence Occupation Marital status	Marital status may moderate the relationship between pain and cognition, such that the relationship is distinct when comparing those that are married vs. those that are divorced or widowed.
Medications Non-opioid analgesics Antidepressants Neuromodulating agents Anticholinergics Other centrally acting agents	Concomitant medications such as anticholinergics may lead to cognitive dysfunction, even if not associated with the exposure(s) of interest (e.g., opioids, pain)
Daily functioning and socialization Physical function Functional independence Socialization Family engagement Sleep	Physical impairments may mediate observed relationships between pain and cognition, such that those with pain may develop worsening physical function, which in turn may lead to impaired cognition.
Genetics Family history of chronic pain Family history of substance use Family history of cognitive impairment Known genetic predisposition to cognitive impairment (e.g., APOE e4)	Genetic factors, such as a family history of early onset dementia, may influence the onset and severity of cognitive dysfunction, even if not related to the exposure(s) of interest.

^*Provided examples are not comprehensive. Any given variable may play varying roles (e.g., confounder, mediator, moderator, covariate).

Figure 2.

Conceptual framework for improving analgesic outcomes in older adults, which focuses on the interplay between pain, cognition, and opioid analgesic treatment strategies. Assessments of the relationships between pain and cognition must include thorough consideration for opioid analgesic exposures, and vice versa. Similarly, the relationships between pain, opioids, and cognition may be confounded, mediated, or moderated by additional patient-related factors such as mental health, sleep, socialization, and physical function, among others.

Conceptual Framework to Optimize Analgesic Outcomes in Older Adults

Recognizing the interdependent themes of pain, opioids, and cognition, we propose a conceptual framework to guide future research (Figure 2). All scientific investigations of longitudinal changes in cognitive function in older adults must consider the potential contributions of both pain and opioid analgesics. For example, assessing the associations between chronic pain symptoms and changes in cognition while failing to account for potential confounding or mediation by opioid analgesics could lead to incomplete or inaccurate conclusions. Similarly, studies of opioid analgesics and cognition relationships that fail to account for pain may not provide an accurate representation of more complex relationships. It is therefore essential to include each of these factors in future investigations, which has been a recurrent gap in existing literature.

Key pain features that should be considered in future investigations include, but are not limited to, pain etiology, clinical phenotype, severity, inciting factors, chronicity, interference with daily functioning including physical function, social interaction, and mental health, and temporal changes in pain features in relation to the timing of cognitive assessment. While each of these features are important, we believe that pain etiology, chronicity, severity, and interference are likely the most important to incorporate and are most readily accessible with existing tools. Additionally, regarding temporal changes in pain features, it is possible that pain experienced at a discrete time point may drive longer-term cognitive changes. However, it is important to design longitudinal investigations with congruent assessments of pain and cognition over time. Concerning the exposure of opioid analgesics, key factors for consideration include the medication type, mechanism of action, dose, frequency, duration of therapy, and concomitant use of non-opioid analgesics, sedatives, and other centrally acting medications. Of these, we believe that dosing, duration of therapy, and concomitant medications are most crucial to incorporate. Finally, cognitive function must be assessed comprehensively. When possible, this should include the use of multi-dimensional neuropsychological testing to assess for potential differential effects across cognitive domains (i.e., memory, language, visuospatial skills, executive function) in addition to employing formal clinical diagnoses of mild cognitive impairment and dementia. Cognitive screening instruments, while widely available, are unlikely to provide the requisite discriminative ability to establish potential mechanistic relationships between pain, opioids, and cognitive function. Recognizing that formal neuropsychological testing may be time and resource intensive, future work should be devoted to the development of more accessible and less burdensome neuropsychological assessment tools that permit domain-specific assessments of cognition.

Beyond the interplay of these three factors, it is also critical to consider the contributions of additional features that may confound, mediate, or moderate exposure and outcome relationships (Table 3). Among these, we highlight those that we believe to be important for the relationships between pain, opioids, and cognition: mental health [79, 80], sleep [81], socialization [82], and physical function [83]. Each of these factors in isolation may be associated with pain, opioid analgesic use, and cognitive function, yet it remains unclear how these factors may influence pain-opioid-cognition relationships. It is also critical to recognize that these features may play various roles. For example, we previously shared how mental health conditions such as depression could confound pain and cognition relationships. Additionally, pain may exacerbate mental health symptoms, which in turn may mediate impairments in cognition (i.e., pain is associated with impairments in cognition, but the causal pathway between pain and cognition occurs, at least in part, through the relationship between pain and mental health symptoms; the latter of which are increased in those with pain). Conversely, mental health symptoms may exacerbate pain, which may lead to worsening in cognitive function either directly or indirectly (e.g., through increased exposure to opioid analgesics). Mental health symptoms may also moderate the relationship between pain and cognition, such that the relationship may differ in those with and without mental health symptoms. Careful consideration of the directionality of these relationships, recognizing potential bidirectionality, is essential in future investigations.

Failure to explicitly assess for these potential relationships through well-designed clinical research will undoubtedly lead to variable and conflicting results.

To fully under the relationships that exist between pain, opioids, and cognition in older adults, and truly advance the science forward, we must ensure that future studies employ a common lens in which to view this problem, of which this conceptual framework may serve as a guiding template. While this framework was developed for research applications, clinicians may use the synthesized data to recognize the potential relationships that exist between pain, opioids, and cognitive function. Finally, given the interdisciplinary nature of the identified factors that may influence these relationships, it is essential that future investigations be performed by multidisciplinary teams of investigators that possess complimentary expertise in these key areas (e.g., pain, cognition, aging, mental health, statistics, epidemiology).

Conclusion

Achieving acceptable analgesia, while minimizing medication-related adverse effects and preserving daily functioning, is critically important for older individuals in our aging population who suffer from painful conditions. In this review, we have summarized critical gaps in knowledge regarding the relationships between pain, cognition, and analgesic use in older adults. Despite numerous observational studies evaluating the relationships between pain and/or opioids with cognition, limited consideration of the potential interplay of these factors has led to variable and disparate results. Improved characterization of these relationships is essential for the development, implementation, and evaluation of strategies that maximize analgesia while minimizing deleterious effects on cognition and enhancing additional patient-centered outcomes such as functional status and quality of life. Our proposed conceptual framework may serve as a guide for future research in this important area.

 

Funding sources: Dr. N.S. Warner’s research is supported by a K23 Mentored Research Award (K23AG070113) through the National Institute of Aging (NIA) and by the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.

Conflicts of interest: There are no conflicts of interest to report,Prior presentations: None.