Assessment of Chronic Pain: Domains, Methods, and Mechanisms

Abstract

Accurate classification of chronic pain conditions requires reliable and valid pain assessment. Moreover, pain assessment serves several additional functions, including documenting severity of the pain condition, tracking the longitudinal course of pain, and providing mechanistic information. Thorough pain assessment must address multiple domains of pain, including the sensory and affective qualities of pain, temporal dimensions of pain, and the location and bodily distribution of pain. Where possible, pain assessment should also incorporate methods to identify pathophysiological mechanisms underlying the pain. This article discusses assessment of chronic pain, including approaches available for assessing multiple pain domains and for addressing pathophysiological mechanisms. We conclude with recommendations for optimal pain assessment.

Accurate pain assessment is critical to the classification of chronic pain conditions. Indeed, the Core Diagnostic Criteria proposed in Dimension 1 of the ACTTION-APS Pain Taxonomy (AAPT) includes symptoms and signs of the pain disorder, and pain is, of course, the primary symptom for all chronic pain conditions ²⁶. Therefore, reliable and valid pain assessment is an essential component of the AAPT framework. In addition to its diagnostic importance, pain assessment serves several other valuable functions. First, pain assessment provides information regarding the severity of the condition. In addition to its diagnostic value, this information is critical for guiding treatment decisions. Also, pain assessment allows clinicians and scientists to monitor the longitudinal course of the pain disorder and to quantify treatment effects. Repeated pain assessment should inform pain treatment in much the same way as repeated blood pressure measurement informs treatment for hypertension. Finally, pain assessment can yield clues regarding the pathophysiological mechanisms underlying the pain condition, which can help guide treatment selection. The purpose of this article, included as part of a special supplement to the Journal of Pain, is to highlight important issues in pain assessment in the context of the AAPT ²⁶. While other reviews and book chapters have addressed pain assessment ^{8, 30, 38}, this manuscript presents a heuristic model for conceptualizing pain assessment in the context of evidence-based pain classification, and for conducting pain assessments that can ultimately provide information regarding pathophysiological mechanisms (see Figure 1). Assessment of the patient with chronic pain should also include assessment of other clinically important domains, such as psychological and physical functioning and quality of life. However, those issues will be addressed in separate articles in this Supplement (see Turk et al and Edwards et al), hence, this article focuses solely on assessing features related to pain and its underlying mechanisms. Specifically, we will discuss important domains of clinical pain that should be assessed and identify appropriate measurement tools. In addition, we will describe existing and emerging approaches to assessing pain mechanisms in clinical populations. The article will conclude with some recommendations for implementing pain assessment in order to enhance pain classification.

Figure 1.

Heuristic Model of Pain Assessment. This model depicts the two major goals of pain assessment: 1) assessment of pain burden, and 2) assessment of pain mechanisms. On the left side of the figure are depicted the domains of pain burden that should be measured. These measures primarily fulfill the goal of assessing pain burden (as indicated by the solid arrows), but some of these domains can also provide information regarding pain mechanisms (as indicated by the dashed arrows). The right side of the figure displays several common and emerging methods for assessing pain mechanisms (as indicated by the solid arrows). QST=quantitative sensory testing.

Domains of Clinical Pain to Assess

Sensory and Affective Qualities of Pain

Because pain is an internal, private experience, self-report remains the gold standard for its measurement. The most commonly assessed aspect of clinical pain is its sensory intensity. As summarized in Table 1, multiple approaches are available for assessing pain intensity, including categorical scales (e.g., Mild, Moderate, Severe), numerical rating scales (NRS), visual analog scales (VAS), and well-validated verbal descriptor scales that have excellent statistical properties (e.g. the Descriptor Differential Scale ³¹). The advantages and disadvantages of these different methods have been well described elsewhere.^{8, 30, 38} The NRS is the most commonly used method in clinical settings due to its ease of administration and scoring. A recent systematic review concluded that NRS showed higher compliance and ease of use than VAS.³⁷ These authors also reported a large variety of verbal anchors for upper end of NRS and VAS, the most frequently used including “Worst Possible Pain,” “Worst Pain Imaginable,” and “Most Intense Pain Imaginable.” Consistent with these findings, for most purposes we recommend using an 11-point or 101-point NRS, on which 0 represents “No Pain” and 10(0) represents either “the Worst Possible Pain” or “the Most Intense Pain Imaginable.” However, in young children or in populations with limited verbal abilities we recommend the Faces Pain Scale, which presents a series of pictures of facial expression depicting different levels of pain experience.⁵⁸ The time frame over which pain intensity is assessed deserves some mention. Because current pain may not accurately reflect a patient’s overall pain experience, instruments such as the Brief Pain Inventory (BPI) ^{14, 44} and the Graded Chronic Pain Scale ^{87, 88} ask patients to report their worst, least and average pain intensity over some period of time (e.g. the past 24 hours or the past week). This provides important information regarding the patient’s overall pain burden for a given period of time.

Table 1.

Approaches to Assessing Different Domains of Pain

Pain Domain	Measures	Comments
Sensory and Affective Qualities of Pain
Pain Intensity-The strength or “loudness” of the pain	Categorical Scales; Numerical Rating Scales (NRS); Visual Analog Scales (VAS); Faces Scale; Verbal Descriptor Scales; Brief Pain Inventory; Graded Chronic Pain Scale	0 (No Pain)-10 (Most Intense Pain Imaginable) NRS is recommended for most settings due to its ease of use and statistical properties.
Pain Affect-How unpleasant of disturbing the pain feels	Categorical Scales; Numerical Rating Scales (NRS); Visual Analog Scales (VAS); Faces Scale; Verbal Descriptor Scales	0 (Not at All Unpleasant)-10 (Most Unpleasant Feeling Imaginable) NRS is recommended for most settings due to its ease of use and statistical properties.
Perceptual Qualities of Pain-Description of sensory and other features of the pain, how the pain feels	McGill Pain Questionnaire (MPQ); PainDetect; Neuropathic Pain Scale; Neuropathic Pain Symptom Inventory; Leeds Assessment of Neuropathic Symptoms & Signs (LANSS); Dolour Neuropathique-4 Questions (DN4)	The MPQ yields sensory, affective & evaluative subscales. The other instruments are screening tool for identifying neuropathic pain features and for tracking outcomes from treatment for neuropathic pain.
Temporal Characteristics of Pain
Pain Duration-Time since onset of chronic pain in months or years	Retrospective self-report	Often difficult for patients to accurately report, especially with insidious onset of pain.
Pain Variability-The presence vs. absence of pain and fluctuations in pain intensity over time.	Patient report of the percentage of the waking day during which pain is present; Ecological Momentary Assessment (EMA)	EMA is more accurate but requires patient compliance, and electronic EMA requires specialized hardware and software. EMA can provide direct measures of pain variability, as well as other distributional measures.
Modifying Factors-Factors that exacerbate or ameliorate the pain	Retrospective self-report; EMA
Other Pain Features
Pain Location(s)-areas of the body in which patient experiences pain; bodily extent of pain	Pain Drawing (Paper-and-Pencil or Electronic)	Identifies specific areas of pain but also assesses how widespread the pain is.
Provocative Pain Measures-Measures collected via physical exam in order to provide diagnostic information	Straight Leg Raising; Digital Palpation	Straight leg raising is recommended for classifying low back pain in studies of invasive interventions; Digital palpation is part of the diagnostic exam for fibromyalgia and temporomandibular disorders.
Pain Behaviors-Overt behaviors that convey to the observer that the individual is experiencing pain.	Facial Expressions; Limping, Guarding, Bracing, etc…	Some bedside versions have been developed and validated

Pain intensity reflects the sensory component of pain, however, another important component of pain severity is pain affect, which refers to how unpleasant or disturbing the pain feels. Pain affect can be assessed using categorical scales, as well as NRS and VAS, where the scale endpoints are modified to range from “Not at All Unpleasant” to “Most Unpleasant Feeling Imaginable.” While in most instances, pain intensity and pain affect are highly correlated, under some circumstances these two pain dimensions can be independently modulated.^{32, 65} Therefore, assessing both dimensions of pain can provide valuable information.

While single item measures are most frequently used to assess pain intensity and pain affect, multiple item instruments can provide additional information regarding the sensory and affective qualities of pain. For example, a pain described as shooting and burning differs from a dull, aching pain, even though the two pains might be rated as equally intense on an NRS scale. One of the most widely used multiple item instruments for collecting information regarding pain qualities has been the McGill Pain Questionnaire (MPQ)⁴¹, which also has two validated short forms (SF-MPQ).^{22, 41} The MPQ presents 20 groups of words and the patients select all of the words that describe their pain. The MPQ yields several subscale scores, including sensory, affective, and evaluative scores. Both the original MPQ and the SF-MPQ-2 show high reliability and validity and some evidence suggests that these instruments can distinguish among different types of clinical pain.^{22, 41} A valuable aspect of these instruments is their ability to provide information regarding the perceptual qualities of the pain.

Several additional multi-item instruments have been developed as screening tools to specifically assess neuropathic qualities of pain, several of which are listed in Table 1. These instruments assess self-reported features such as dysesthesias, electric shock-like or shooting pain, numbness, pain in response to heat or cold, allodynia, etc., and some include responses to evoked stimuli. Most of these instruments have reasonable sensitivity and specificity for distinguishing neuropathic from non-neuropathic pain.^{11, 34} Furthermore, subgrouping of patients based on their potential mechanisms and assessment of treatment outcomes can also be performed using these types of questionnaires (e.g., painDETECT, Neuropathic Pain Symptom Inventory, NPSI).^{5, 27} Interestingly, even in many non-neuropathic pain conditions, a substantial proportion of patients endorse “neuropathic” pain qualities, which raises questions regarding the specificity of these signs and symptoms. Moreover, a recent systematic review found that many of these scales demonstrate inadequate measurement properties, emphasizing that these tools “should not replace thorough clinical assessment.” ⁵⁷

Temporal Characteristics of Pain

The temporal features of pain, while quite important, are less often systematically assessed. These include the duration and chronicity of pain, and the temporal pattern of the pain (e.g. episodic, chronic-recurrent, constant but fluctuating in intensity). Assessing the duration of pain (i.e. time since pain onset) is critical to classification of chronic pain. For pain conditions with an initiating event (e.g. trauma, surgery), patients are typically able to report duration accurately. However, for pain disorders with insidious onset, duration can be difficult to determine. For example, a patient with knee osteoarthritis may have intermittent mild pain for months or years before the pain becomes of sufficient severity or constancy to seek care. Thus, when attempting to determine the duration of chronic pain one should ask for how long the patient has experienced pain most of the time.

Other temporal features of the pain include the variability of the pain and its temporal patterns. This includes whether the pain is present all the time or whether the patient experiences pain-free episodes. One approach to assessing constancy is to ask during what percentage of his/her waking day a patient experiences pain. An important issue that impacts temporal variations in pain is factors that exacerbate or ameliorate pain. These factors should be assessed as they impact interpretation of temporal changes in pain and can have diagnostic and treatment implications. These temporal features of pain are usually ascertained historically via patient recall. Some neuropathic pain questionnaires (e.g., PainDetect) also include one or more items that query temporal aspects of pain (e.g., constant pain vs. sudden “pain attacks”).

A potential barrier to valid assessment of temporal features of pain is the inaccuracy of patients’ memory for pain. For example, the “peak-end phenomenon” has been well documented, such that when asked to report pain experienced over a recent period (e.g. the last week), patient recall is predominantly influenced by the worst pain experienced over that time (i.e. the peak) and the most recent pain they experienced (i.e. the end).^{66, 67, 78} One method designed to enhance patient recall is the Day Reconstruction Method (DRM), which asks individuals to recall episodes of time during the previous day in which they engaged in an activity, thereby “reconstructing” their day. Then, they are asked to report their pain level at that time or during that activity.³⁹ More optimally, one can use daily diary methodologies (retrospective, time, or episode-based) and may use paper-and-pencil or electronic prompting (e.g., interactive voice recording systems [IVRS], ecological momentary assessment [EMA]). EMA assesses the temporal dynamics of pain in real time, including modifying factors and is typically completed using an electronic device that prompts patients to report pain (and possibly other events or symptoms) at randomly determined intervals throughout the day. This approach that overcomes the recall and compliance issues that can undermine retrospective patient reports of pain, including end-of-day diaries.^{75, 80} Such high-resolution data can help reveal factors that exacerbate or ameliorate pain. Moreover, EMA can yield novel outcome measures beyond average pain intensity, including distributional measures (e.g. proportion of pain ratings above 50), measures of variability, and time-contingent measures (e.g. time of day when pain is highest), which may provide important information regarding treatment outcomes and pain mechanisms.⁷⁹ However, collection of EMA data is resource intensive and can be inconvenient for patients; therefore, these methods are more likely to be used in research than when a practitioner is trying to classify an individual patient.

Pain Location and Bodily Distribution

The location of pain can have obvious diagnostic implications, because current diagnostic systems, including AAPT, categorize pain conditions primarily by body site or organ system.²⁶ The Pain Drawing represents the most common method for assessing the location and bodily distribution of pain.³⁸ The pain drawing typically consists of front and back line drawings and patients are instructed to shade areas in which they experience pain, and these drawings can also obtain information about different features of pain (e.g. intensity, perceptual qualities) across different locations. Such drawings are incorporated into several pain instruments, including the MPQ, the BPI, the PainDETECT, and the Leeds Assessment of Neuropathic Symptoms and Signs (LANSS). Furthermore, it is important to ask the patients whether the pain radiates in different body areas. Also these areas should be captured in the pain drawing. Specific drawings for the head and face can be used for individuals with headache and orofacial pain. Scoring typically consists of tallying the number of body regions that experience pain. Some implementations include the option for patients to use symbols or colors to indicate different pain qualities (e.g. sharp/shooting pain vs. dull aching pain). Similarly, patients can be asked to provide separate intensity ratings for different pain locations. In recent years electronic pain drawings have been developed, using both computers and handheld devices.^{7, 21} Both paper and pencil pain drawings and their electronic counterparts have shown high reliability.^{21, 54, 55}

The location and bodily extent of pain have important diagnostic and treatment implications. Indeed, a patient presenting with low back pain who endorses widespread body pain presents a different diagnostic and treatment challenge than a patient with localized low back pain. Finally, some pain conditions such as fibromyalgia (FM) have diagnostic criteria that specify a threshold for the spatial distribution of pain.⁹²

Electronic Pain Measures

As noted above, electronic approaches are ideal for collecting daily pain ratings; however, computer and mobile device software is increasingly being used to assess multiple aspects of pain experience. For example, the Patient Reported Outcomes Measurement Information Systems (PROMIS) has implemented computer adaptive testing to optimize the efficiency of reliable and valid assessment of patient reported outcomes, including pain. PROMIS measures have recently been implemented in a chronic pain registry in order to collect research quality data in the context of clinical care ^{82, 83}. Also, numerous pain assessment apps are available for mobile devices; however, most of the scientifically validated apps are not available in app stores ¹⁶. One exception is the PainOmeter, which includes multiple pain scales and has been found to be user friendly in previous research ¹⁷. However, it seems inevitable that mobile apps for pain measurement will become routine in the clinical setting in the near future. Indeed, one would expect that smartphone apps tethered to activity tracking technology will allow clinicians and researchers to more systematically assess pain in the context of daily life, including the impact of pain on physical activity and sleep, and vice versa.

Provocative and Behavioral Pain Measures

In addition to patient reported measures, accurate pain classification sometimes requires provocative pain tests and/or behavioral pain measures, which are typically obtained in the context of a physical examination. For example, straight leg raising was recognized as valuable in classification of low back pain, particularly for studies of invasive interventions.¹⁸ Also, existing diagnostic criteria for temporomandibular disorders and FM require assessment of sensitivity to digital palpation.^{71, 93} In addition, examiner observation of pain behaviors can provide useful information. This includes facial expressions of pain and other overt expressions of pain (“pain behaviors”), such as limping, guarding, and bracing. The sophisticated systems for quantifying pain behaviors that have been developed for research purposes are unrealistic in the clinical setting.^{42, 43, 64} However, these have been adapted into bedside approaches, which have shown adequate reliability and validity.^{48, 68} Assessment of these nonverbal pain behaviors can be particularly helpful in patients with reduced communicative capacity, such as very young children and individuals with cognitive limitations.^{35, 58} Objective physiological responses can also be helpful as adjunctive pain measures in these populations. For example, skin blood flow has been successfully applied in neonates. ⁸⁶ Similarly, skin conductance has been used as a measure of response to nociceptive stimuli in anesthetized women. ⁸¹ While these types of autonomic measures can be useful component of pain assessment, they are nonspecific and should not supplant self-report measures when they can be obtained.

Assessment of Pain Mechanisms

The goals of diagnosis are to understand prevalence and to characterize subjects in clinical and research settings, in order to inform policy and regulatory decisions. But perhaps the primary goal is to guide treatment, and optimal treatment requires knowledge of underlying pain mechanisms. Therefore, assessment methods that provide information regarding the pathophysiological processes contributing to patients’ pain can be very helpful in promoting mechanism-based pain classification.^{12, 95} Given our limited understanding of the pathophysiological mechanisms responsible for most chronic pain disorders, it is not currently realistic to enact a completely mechanism-based approach to pain classification. However, in order to encourage the inclusion of mechanistic information in pain diagnoses, AAPT incorporated Dimension 5, which “includes putative neurobiological and psychosocial mechanisms contributing to the pain disorder, including potential risk factors and protective factors.”²⁶ Below, we discuss existing and emerging assessment methods that may provide important information regarding neurobiological pain mechanisms (see Baron et al ⁴ and Table 2). Approaches to assessing psychosocial mechanisms are discussed by Edwards and colleagues (this supplement).

Table 2.

Approaches for Assessing Pain Mechanisms in Clinical Populations

Approach	Objective	Comments
Quantitative Sensory Testing 12,69	To assess contributions of somatosensory and pain modulatory function to pain	The German Neuropathic Pain Network (DFNS) has developed a standardized protocol. Dynamic measures, such as temporal summation of pain and conditioned pain modulation assess pain facilitation and pain inhibition, respectively. Clinical use of these measures remains infrequent.
Skin Biopsies to Measure Cutaneous Nerve Fiber Density 1,40,61,72	To assess peripheral innervation or denervation that may contribute to pain	Decreased epidermal nerve fiber density (ENFD) has been observed in fibromyalgia, HIV-associated neuropathy and in peripheral small fiber neuropathy. Findings remain somewhat controversial and clinical use is rare.
Microneurography 73,74	To assess abnormalities of C fiber activity that may contribute to pain	Abnormal activity of C nociceptors has been observed in fibromyalgia and neuropathic pain. Not widely accepted for clinical use.
Functional & Structural Brain Imaging 15,60	To assess contributions of cerebral brain structure and function to pain	Reduced gray matter volume has been reported in several chronic pain conditions. Also, changes in structural and functional connectivity have been associated with chronic pain. Expense and lack of specificity limit current clinical utility.
Chemical Neuroimaging 13,60	To assess contributions of specific neurochemical systems to pain	Ligand-based imaging has demonstrated differences in opioidergic and dopaminergic systems in chronic pain. Magnetic resonance spectroscopy has shown abnormal glutamatergic function in fibromyalgia and in patients with pain after spinal cord injury. Expense and lack of specificity limit current clinical utility
Pharmacological Phenotyping 90	Uses pharmacologic probes to assess the contribution of specific neurochemical pathways to pain	Patients can be subgrouped based on their clinical response to drugs with known pharmacology, which provides information regarding mechanisms that contribute to their pain. Potentially promising, but limited empirical evidence to date.
Genotyping 19,59	To identify genetic markers of proteins or pathways that contribute to pain	Multiple genetic markers have been associated with both experimental and clinical pain, including the catechol-o- methyl-transferase gene, the mu-opioid receptor gene, the G-cyclohydrolase gene, and several sodium channel genes. Nonreplication of findings is common, and clinical utility remains low.

Quantitative Sensory Testing (QST)

QST refers to a set of methods in which patients’ perceptual responses to quantifiable sensory stimuli are assessed in order to characterize somatosensory function or dysfunction.¹² Multiple stimulus modalities can be used to elicit both painful and nonpainful percepts, most commonly including thermal (heat, cold) and mechanical (tactile, pressure, vibration) stimuli, but electrical, ischemic and chemical stimuli are also used. Stimulus modalities and parameters can be selected to preferentially engage different nerve endings, nerve fibers, and central nervous system pathways in order to systematically evaluate somatosensory transmission and pain processing. Moreover, dynamic QST approaches can provide valuable information regarding pain inhibitory and pain facilitatory function. Multiple authors have discussed the use of QST in the assessment and classification of pain in recent years.^{12, 63, 84} Below, we will briefly review the application of QST in the assessment of neuropathic pain, and the use of QST for determining pain modulatory function.

The clinical application of QST has been most well developed in the characterization of neuropathic pain.^{2, 34, 63} The most systematic approach has been developed by the German Neuropathic Pain Network (DFNS).⁶⁹ The DFNS approach obtains 13 different measures in response to thermal and mechanical stimuli, which reflect both gain of function (e.g. allodynia, hyperalgesia) and loss of function (e.g. insensitivity to cold or vibration) changes. The DFNS protocol has been shown to have excellent inter-rater and test-retest reliability ²⁸, and they have reported reference values from a group of pain-free controls.^{50, 70} Using this QST protocol, these investigators have identified multiple somatosensory profiles within major neuropathic pain diagnostic groups, suggesting that different mechanisms may be at play for patients with the same neuropathic pain diagnoses and that these subgroups respond differently to treatment.^{29, 51} Using the DFNS QST methodology, a previous double-blind randomized clinical trial in patients with peripheral neuropathic pain found that the the sodium channel blocker oxcarbazepine showed substantially greater efficacy in patients who showed sensory gain (i.e. hyperalgesia) compared to those who showed sensory loss. Importantly, these phenotype groups did not differ in their responses to placebo. Mainka et al. ⁵² treated patients with peripheral neuropathic pain with topical 8% capsaicin patches and analyzed treatment responders and non-responders retrospectively based on their baseline DFNS QST-profile. Capsaicin responders and non-responders could be distinguished based on the presence of cold- and pin-prick hyperalgesia but they did not differ regarding the other QST parameters. However, unlike the Demant et al study, this trial did not include a placebo condition; therefore, it is possible that these QST parameters may be predicting nonspecific treatment effects. Thus, using the DFNS protocol, QST has provided valuable information regarding somatosensory profiles in neuropathic pain, and this information has important mechanistic and treatment implications.

QST has been increasingly studied in non-neuropathic pain conditions. In this context, the general goal is to more globally characterize pain modulatory function in contrast to the greater focus on peripheral afferent function in assessing neuropathic pain. Yarnitsky and colleagues ⁹⁶ have proposed the concept of a pain modulation profile, which reflects an individual’s balance of pain facilitation versus pain inhibition. The most frequently used measure of pain facilitation is temporal summation, which refers to a transient form of central sensitization manifested by increased pain evoked by rapid repetitive stimulation at a fixed stimulus intensity. Pain inhibition is most commonly assessed via conditioned pain modulation (CPM), which refers to the decrease in pain evoked by one stimulus (the test stimulus) produced by contemporaneous application of a second pain stimulus at a different body site (the conditioning stimulus). Considerable evidence demonstrates that individuals with chronic pain often exhibit a pain modulatory imbalance, characterized by increased pain facilitation and diminished pain inhibition.^{12, 96} This pattern has been observed for multiple pain conditions, including FM, temporomandibular disorders, irritable bowel syndrome, and osteoarthritis.^{45, 46, 77, 84} Thus, using QST to assess pain modulatory profiles in patients with chronic pain conditions may provide mechanistically useful information with important treatment implications.

The QST approaches described above require specialized equipment and training and considerable time to complete; therefore, their integration into routine clinical assessment is unlikely. However, bedside methods are available, which are clinically feasible and can provide valuable information regarding somatosensory function. For example, von Frey monofilaments and tuning forks can be used to assess mechanical sensation. The monofilaments can also be used to assess punctate pain and mechanical temporal summation. Pressure algometry can assess pressure pain sensitivity, and metal rods that are heated or cooled can assess thermal sensitivity. Some evidence supports the reliability and validity of this approach ⁶²; therefore, bedside QST may provide an important next step in mechanism-based pain assessment.²⁷

Emerging Approaches in Mechanism-Based Assessment

Based on recent and ongoing research, several additional approaches to mechanism-based pain assessment may prove useful in the near future. Peripheral contributions to pain have been examined by quantifying innervation of cutaneous tissues using skin biopsies. For example, epidermal nerve fiber density (ENFD), which typically requires taking a punch biopsy of the skin, is a standard technique for diagnosing peripheral small fiber neuropathy; however, recent evidence shows ENFD to be unrelated to neuropathic pain.^{40, 72, 85} However, ENFD did correlate with clinical pain in patients with HIV-associated polyneuropathy.⁹⁷ Also, a substantially greater proportion of FM patients than controls showed decreased ENFD findings consistent with small fiber polyneuropathy.^{61, 9, 47} In contrast, FM patients showed increased peptidergic innervation of cutaneous arteriole-venule shunts compared to controls ¹. Another approach to identifying peripheral contributions to pain is microneurography, which permits direct recording of unmyelinated nerve via tungsten needles inserted into a peripheral nerve fascicle.²⁰ Serra and colleagues have demonstrated abnormal activity of C nociceptors among patients with fibromyalgia and patients with neuropathic pain.^{73, 74} Thus, skin biopsy and microneurography represent existing techniques that may be incorporated into future assessment protocols to provide information regarding peripheral mechanisms contributing to clinical pain.

Progress in brain imaging has been exponential in recent years, producing evidence of alterations in both brain structure and function among patients with chronic pain.¹⁵ For example, some investigators have reported reduced gray matter volume in patients with chronic pain compared to pain-free controls, and this was reversed by successful pain treatment in one study.^{6, 33, 76} Interestingly, reduced gray matter volume in several brain regions has also predicted visceral and somatic pain sensitivity in healthy individuals.^{23, 24} Diffusion tensor imaging has identified white matter abnormalities in several chronic pain conditions, suggesting altered structural connectivity between brain regions.^{3, 25, 49, 53} Also, resting state functional connectivity (i.e. covariations in activity among different brain regions when a person is at rest) differs for people with chronic pain versus healthy controls.^{3, 60} While the mechanisms driving these pain-related changes in brain structure and function remain largely unknown, the findings suggest that brain imaging may be a useful tool for mechanism-based pain assessment in the future. In addition, other brain imaging approaches can provide specific information regarding neurochemical alterations in chronic pain. Radio-labeled ligands can be used to identify altered function of specific brain neurotransmitter systems ^{13, 56, 94}, and magnetic resonance spectroscopy can reveal neurochemical alterations in specific brain regions.^{36, 60, 91} Of course, given their expense and the emerging nature of some of the findings, the above brain imaging methods are not yet ready for clinical implementation. Moreover, it is important to recognize that while brain imaging can provide valuable information regarding neural mechanisms that contribute to pain, brain imaging is not a substitute for patient self-report, which remains the gold standard for pain assessment.

Another approach to reveal pain mechanisms is to assess patient responses to pharmacologic interventions, or pharmacological phenotyping. That is, in a group of patients, some will respond positively to a pharmacologic probe targeting a specific mechanism, such as a serotonin-norepinephrine reuptake inhibitor, while others will respond to a drug targeting a different mechanism, such as a calcium channel alpha-2-delta agonist. Responses to these pharmacologic interventions provide information regarding the relevance of their respective targets to the pathophysiology of pain in that particular patient. This principle is illustrated in an intriguing case study of a patient with bilateral at-level spinal cord injury pain.⁹⁰ The clinical description of pain on both sides was identical; however, QST revealed evidence of central sensitization on the right and deafferentation on the left. Pregabalin was administered, which significantly reduced the pain on the right side (associated with central sensitization) but did not affect the left-side deafferentation pain. These findings reveal the mechanistic value of both QST and pharmacological phenotyping.

A final approach to mechanistic assessment is to examine genetic markers using patients’ biological samples. Using primarily candidate gene approaches, several genes have been associated with pain perception and with clinical pain across multiple studies, including the catechol-o-methyl-transferase gene, the mu-opioid receptor gene, the G-cyclohydrolase gene, and several sodium channel genes.^{10, 19, 59, 89} If polymorphisms of genes encoding proteins that are involved in nociceptive processing and pain modulation show different allele frequencies in patients versus controls, this potentially implicates that protein or pathway in the pain disorder. Hence, identifying subgroups of patients based on their genotype may help to stratify patients in a mechanistically meaningful way. At present, genetic testing is not routinely performed in the clinical setting, but its accessibility and potential to guide treatment makes genotyping an attractive future option for mechanistic pain classification.

Conclusions and Recommendations

Assessment of pain is a critical component of accurate classification of chronic pain conditions. Multiple domains of pain should be assessed, including pain intensity and pain affect, as well as other perceptual qualities of pain. Multiple reliable and valid approaches are available to assess these pain dimensions. In addition, temporal features of pain are important, including not only pain duration but also the temporal patterns of pain, which can provide important information to guide diagnosis and treatment. In the clinical setting, these features are most commonly assessed using retrospective recall; however, EMA provides much more accurate and high resolution data regarding temporal aspects of pain. An important goal of pain assessment is to elucidate the pathophysiological mechanisms underlying the pain. Several approaches that are presently relegated primarily to research may ultimately be integrated into clinical assessment in order to generate such mechanistic information. QST provides insight into somatosensory and pain modulatory function, and bedside QST methods are emerging that are clinically implementable. Also, skin biopsies and microneurography are producing clinically relevant findings related to studies of peripheral afferent innervation and function. Brain imaging is rapidly progressing, providing important information regarding brain structure and function and neurochemical systems that modulate clinical pain. Pharmacological phenotyping can likewise identify neurochemical systems that contribute to pain, and genotyping reveals biological systems and pathways that may contribute to pain.

Based on the above review of methods available for assessment of pain and its mechanisms, the following recommendations are offered to optimize pain assessment thereby promoting more accurate and mechanistically informative pain classification.

• -Assess four key components of pain in all patients: pain intensity, other perceptual qualities of pain, bodily distribution of pain, and temporal features of pain. This will enhance not only pain classification but also treatment planning and outcome tracking.
• -Consider incorporating mechanism-based approaches into clinical assessment protocols. This may include thorough assessment of perceptual qualities of pain, including screening tools for neuropathic pain. In addition, bedside QST approaches may be informative and feasible in many settings, whereas the full QST protocol has its place in the research setting and in phase II trials. Skin biopsies can also be used for certain patients, assuming the expertise for analysis in available.
• -Pain assessment needs to be combined with assessment of other important domains, including physical and psychosocial functioning. Approaches to assessment of these domains are discussed elsewhere in this supplement (Edwards, et al; Turk, et al).

Importantly, systematic pain assessment will improve pain diagnoses by reducing observer bias, which is more likely to emerge in the absence of carefully collected data regarding the patient’s pain. Therefore, thorough pain assessment should be consistently conducted, as the results represent the data that will inform research, diagnosis, clinical management, policy and decision making.

Perspective.

Pain assessment is a critical prerequisite for accurate pain classification. This article describes important features of pain that should be assessed and discusses methods that can be used to assess the features and to identify pathophysiological mechanisms contributing to pain.

• Pain assessment represents a critical component of chronic pain classification

• Multiple domains of pain should be assessed, including: pain severity, pain qualities, bodily distribution of pain, and temporal characteristics of pain

• Methods for assessment of pain mechanisms should also be considered, which include: quantitative sensory testing, brain imaging, epidermal nerve fiber density, microneurography, and pharmacological phenotyping