Colored Pain Drawing as a Clinical Tool in Differentiating Neuropathic Pain from Non-Neuropathic Pain

Abstract

Objectives

This is a prospective, blinded, case–control study of patients with chronic pain using body diagrams and colored markers to show the distribution and quality of pain and sensory symptoms (aching, burning, tingling, numbness, and sensitivity to touch) experienced in affected body parts.

Methods

Two pain physicians, blinded to patients’ clinical diagnoses, independently reviewed and classified each colored pain drawing (CPD) for presence of neuropathic pain (NeuP) vs. non-neuropathic pain (NoP). A clinical diagnosis (gold standard) of NeuP was made in 151 of 213 (70.9%) enrolled patients.

Results

CPD assessment at “first glance” by both examiners resulted in correctly categorizing 137 (64.3% by examiner 1) and 156 (73.2% by examiner 2) CPDs. Next, classification of CPDs by both physicians, using predefined criteria of spatial distribution and quality of pain-sensory symptoms, improved concordance to 212 of 213 CPDs (Kappa = 0.99). The diagnostic ability to correctly identify NeuP and NoP by both examiners increased to 171 (80.2%) CPDs, with 80.1% sensitivity and 80.6% specificity (Kappa = 0.56 [95% confidence interval: 0.44–0.68]). The severity scores for pain and sensory symptoms (burning, tingling, numbness, and sensitivity to touch) on the Neuropathic Pain Questionnaire were significantly elevated in NeuP vs. NoP (P < 0.001).

Conclusions

This study demonstrates good performance characteristics of CPDs in identifying patients with NeuP through the use of a simple and easy-to-apply classification scheme. We suggest use of CPDs as a bedside screening tool and as a method for phenotypic profiling of patients by the quality and distribution of pain and sensory symptoms.

Introduction

Pain drawings (PDs) are pictorial representations of pain in the form of body diagrams. These diagrams are drawings of the complete human body, front and back, upon which patients draw or shade the locations and qualities of pain sensations by means of symbols, shading, or cross-hatching. PDs have been used for many years in pain clinics as part of a standard evaluation of patients with chronic pain [1–3]. PDs facilitate physician communication with patients and provide information to physicians about the distribution of pain at a glance. Traditionally, no specific instructions are provided to patients, other than how to shade the diagram, and there are no standards for systematic interpretation of the characteristics of PDs. PDs were first described by Palmer in 1949 [4] and were originally introduced to determine whether pain was “real” or a manifestation of psychological disorders. Three different scoring methods were described, primarily to predict psychological function [2, 5], and several studies reported on the utility of PDs as a tool for detecting psychological distress in pain patients [1, 2, 6–10]. The value of PDs as an indicator of pain psychopathology has been inconclusively debated, and several others have questioned the performance characteristics of PDs in detecting psychopathology [11–22].

Bertilson and colleagues found that PDs are useful in distinguishing neuropathic pain (NeuP) from musculoskeletal pain in the neck and shoulder regions [23] and later reported that a simple PD combined with a physical examination is a more sensitive diagnostic method for detecting spinal nerve involvement than is magnetic resonance imaging [24]. The reliability in using PDs may be limited by clinical knowledge of the patient’s history [25].

However, the utility of PDs—specifically colored pain drawings (CPDs), in which each color is assigned a specific somatic sensation, as is often reported in NeuP—has not been explored until now. An individual patient’s CPD is of important and informative value to the clinician because it provides an immediate and comprehensive visual overview of the critical characteristics of the patient’s pain. Such characteristics include the quality of pain and sensory symptoms, as expressed by colors specifically assigned to each sensation and their location on the body. Consequently, sensory symptoms frequently associated with NeuP, such as burning, aching, numbness, and tingling, are used to present both the quality of these symptoms and their distribution in one picture. This property of CPDs can be used in clinical and research settings to identify individuals with NeuP by its distribution and qualities. CPDs could also be useful in differentiating NeuP from other types of pain.

The purpose of this study was to determine how well the information presented in the form of CPDs (i.e., the distribution and characteristics of pain and sensory symptoms as represented by different colors) performs when clinicians use CPDs to distinguish NeuP from nociceptive pain (NoP).

Methods

This is a prospective case–control study conducted at a single academic medical institution.

Participants

Patients with chronic pain attending the pain clinic and the spine clinic at the University of Wisconsin Hospital and Clinics were recruited. Inclusion criteria were age ≥18 years, pain duration of >3 months, established pain diagnoses, receiving treatment for pain at the University of Wisconsin pain clinic or spine clinic, and ability to read and understand instructions in English. Patients with cancer pain, headache disorders, or visual impairment and those who declined to participate were excluded.

Variables, Data Sources, and Measurements

Each patient completed a CPD and three questionnaires: a demographic survey, the Neuropathic Pain Questionnaire (NPQ) [26], and the State-Trait Anxiety Inventory (STAI Form Y-1) [27].

The demographic survey collected information on age, gender, marital and work status, race and ethnicity, and education level. Each patient also rated pain intensity on a numeric pain rating 0–10 scale, where 0 is no pain and 10 is the worst pain imaginable.

The CPD form consisted of an anterior and posterior outline of the human figure on which patients colored the distribution and quality of pain and pain-associated sensory symptoms with colored markers. There were seven color choices to code for seven qualities of pain and associated sensory symptoms: yellow= aching pain, blue= burning pain, red= stabbing pain, black= numbness, green= tingling/pins and needles, orange= hurts to touch, and purple= any other pain quality or symptom in the affected body part (Figure 1).

Figure 1.

Anterior and posterior depictions of asexual human form with color coding key for pain and sensory symptoms quality.

The NPQ is a 12-item, validated, self-report questionnaire used in screening for NeuP. This questionnaire is based on qualities of pain inferred from pain descriptors and provides a quantitative measure (on a 0 to 100 scale) for each of the pain descriptors. It has a 66.6% sensitivity and 74.4% specificity in distinguishing patients with NeuP from patients with NoP in the original validation sample [26].

The STAI [27, 28] is a reliable and validated measure of anxiety in diverse populations and has been used extensively in research and clinical practice. It consists of separate self-reported scales for measuring state anxiety, which is an outcome of outside influences, and trait anxiety, which is reflection of inherited and developed tendencies. The S-Anxiety scale (STAI Form Y-1) consists of 20 statements that evaluate how respondents feel “right now, at this moment.” The T-Anxiety scale (STAI Form Y-2) consists of 20 statements that assess how people generally feel. For the present study, we used the STAI S-Anxiety scale. This scale evaluates feelings of apprehension, tension, nervousness, and worry, in addition to assessing how people currently feel. Scores on the S-Anxiety scale increase in response to psychological stress and physical danger and decrease as a result of relaxation training.

The Physician Clinical Diagnosis Survey was developed specifically for this study and included medical and pain diagnoses (Supplementary Data). The pain diagnoses were reached by experienced pain physicians (treating physicians) on the basis of all clinical information available, including comprehensive history, physical, and neurological examination and laboratory, imaging, and electrophysiological studies.

Ethics Approval

The Health Sciences Institutional Review Board of the University of Wisconsin, Madison, approved all procedures described in this report. Informed consent was obtained from all study subjects. All study procedures were in accordance and compliance with the Health Insurance Portability and Accountability Act (HIPAA) regulations and the institution’s policies and guidelines for protection of human subjects. No material inducements were provided to subjects for participation in the study. None of the project personnel received any incentives for recruiting subjects for the study or for any other purpose.

Procedure

Participants with established pain diagnoses were identified by the clinic nurse and invited to participate in the study during their routine follow-up clinic visits. Those who consented to participate were given a sealed packet of coded survey questionnaires to be completed at the conclusion of the clinic visit. Participants were instructed to place their completed pain diagram and questionnaires in sealed envelopes and return the sealed envelope to the clinic staff.

Participants were instructed to color areas corresponding to their pain on both the anterior and posterior depictions of a human figure, using different colors for each specific quality of pain they were experiencing. In addition, subjects completed the NPQ and STAI Form Y-1.

All sealed envelopes were sent to the research office for data entry and analysis. The coded diagnosis surveys completed by the treating physicians were returned separately in sealed envelopes to the research office. Treating physicians in the study were not privy to patient questionnaires and pain diagrams.

Review of CPDs

Pain diagrams were scanned by research staff and forwarded to two study pain physicians (MB and NS), a neurologist and a physiatrist, who were blinded to each participant’s medical and pain diagnoses. Each of the study physicians independently reviewed each CPD and reported the presence of NeuP or NoP in two steps: 1) In Step One, each PD was classified as NeuP or NoP on the basis of a diagnostic impression (i.e., “Is neuropathic pain present: yes or no?”). 2) In Step Two, each CPD was classified through the use of a predefined criterion based on the neuroanatomic pain distribution pattern (such as distal sensory length-dependent neuropathy, mononeuropathy, radiculopathy, plexopathy, etc.) and the presence of four specific sensory qualities (numbness, tingling, burning pain, hurts to touch) (Table 5). During Step Two, classification was ranked as: 1) definite neuropathic—recognizable neuroanatomic distribution plus 3 or 4 out of 4 prespecified pain qualities; (Figure 2, diagrams A to D, F, H, K, and L); 2) probably neuropathic—(2a) recognizable pain distribution plus 1 or 2 pain qualities or (2b) ≥2 pain qualities and no recognizable distribution (Figure 2, diagrams E, G, I, and J); 3) indeterminate—multifocal pain, only recognizable distribution or only 1 pain quality; or 4) not neuropathic—lack of recognizable pain distribution and quality or pain consistent with NoP, such as localized pain in joints, as is seen in arthritis (Figure 2, diagrams M to P). The process was iterative, and the first round of analysis was based solely on the uninstructed study physician’s impression mimicking a clinical use of pain diagrams. The second round was conducted after a session during which both participating study physicians developed and agreed on a scoring process. Results presented here are those of the analysis after the second round.

Table 5.

Criteria used to classify colored pain diagrams

A. Is there a recognizable NeuP pattern (plausible neuroanatomic or pathophysiological explanation):

Yes

No

B. If yes, determine body zone and distribution of NeuP pattern:

Body Zone

1: Head and face

2: Right upper body quadrant and upper limb

3: Left upper body quadrant and upper limb

4: Right lower body quadrant and lower limb

5: Left lower body quadrant and lower limb

6: Trunk only (neck, chest, back, abdomen)

Distribution

D1. Radicular

D2. Stocking or stocking and glove

D3. Peripheral nerve (ulnar, intercostal, intercostobrachial, sciatic, peroneal, etc.)

D4. Cranial nerve: trigeminal

D5. Hemibody (upper and lower limb ± trunk on same side, e.g., in stroke)

D6. Half body or all four limbs (both lower limbs or all four limbs ± trunk as in Spinal cord injury)

D7. Regional (whole upper or lower limb)

D8: None of the above

C. What features of NeuP are present?

C1. Definite: distribution + >2 pain qualities (burning, tingling, numbness, hurts to touch)

C2. Probable:

C2a: Distribution + ≤2 pain qualities (burning, tingling, numbness, hurts to touch)

C2b: No recognizable distribution but 2 or more pain qualities (burning, tingling, numbness, hurts to touch)

C3. Indeterminate: all-over body pain, multifocal pain, does not meet C1 and C2 criteria

C4: No features of NeuP

Figure 2.

Representative examples of NeuP and NoP seen in this study. (A) Mononeuropathy, right meralgia paresthetica: Definite NeuP, plausible distribution, and 3 qualities of NeuP (burning, tingling, hurts to touch). (B) Right lumbosacral radiculopathy: Definite NeuP, plausible distribution, and 3 qualities of NeuP (burning, tingling, numbness). (C) Right cervical radiculopathy: Definite NeuP, plausible distribution, and 3 qualities of NeuP (burning, hurts to touch, numbness). (D) Mononeuritis multiplex (right ulnar, saphenous): Definite NeuP, plausible distributions, and 4 qualities of NeuP (burning, tingling, hurts to touch, numbness). (E) Lumbosacral radiculopathy: Probable NeuP, plausible distribution, and 2 qualities of NeuP (tingling, numbness). (F) Cervical radiculopathy: Definite NeuP, plausible distributions, and 4 qualities of NeuP (burning, tingling, hurts to touch, numbness). (G) Hemibody pain (thalamic pain syndrome): Probable NeuP, plausible distributions, and 2 qualities of NeuP (burning, numbness). (H) Spinal cord injury (Brown-Sequard syndrome): Definite NeuP in lower half of the body, plausible distributions, and 3 qualities of NeuP (burning, tingling, numbness); also with probable NeuP in left upper limb and neck. (I) Distal length-dependent sensory polyneuropathy in both lower limbs: Probable NeuP, plausible distributions, and 2 qualities of NeuP (tingling, numbness). (J) Hemibody pain (thalamic pain syndrome): Probable NeuP, plausible distributions, and 1 quality of NeuP (burning). (K) Spinal cord injury: Definite NeuP, plausible distributions, and 4 qualities of NeuP (burning, tingling, hurts to touch, numbness); also has non-neuropathic neck and shoulder pain. (L) Distal length-dependent asymmetric sensory polyneuropathy: Definite NeuP, plausible distributions, and 3 qualities of NeuP (burning, tingling, numbness). (M) Polyarthralgia and myofascial pain: Non-neuropathic pain; distribution and pain qualities do not match with NeuP. (N) Ankle pain with referred leg pain: Non-neuropathic pain; distribution and pain qualities do not match with NeuP. (O) Polyarthralgia: Non-neuropathic pain; distribution and pain qualities do not match with NeuP. (P) Buttock pain with referred thigh pain: Non-neuropathic pain; distribution and pain qualities do not match with NeuP. (Q)  Axial low back pain with referral to thighs: Non-neuropathic pain; distribution and pain qualities do not match with NeuP. (R) Neck pain and forehead pain, cervicogenic headache: Non-neuropathic pain; distribution and pain qualities do not match with NeuP.

Statistical Analysis

Data were summarized as mean (SD) and N (%). NPQ outcomes were examined for appropriateness of the parametric t test as a sensitivity analysis. We examined the results of both the t test and the Wilcoxon rank-sum test, and all tests resulted in the same conclusion. Therefore, we present t test P values. Chi-squared tests were used for comparisons between groups for categorical variables. Kappa statistics and 95% confidence intervals (CIs) were used to assess inter-rater agreement and concordance with the clinical diagnosis. All analyses were conducted at a 5% significance level via R for Statistical Computing [29]. Sensitivity and specificity were calculated in terms of correctly classifying neuropathic (sensitivity) or non-neuropathic (specificity) clinical diagnosis by pain diagram diagnosis.

Results

Two hundred and twenty-one participants were recruited, and 8 participants were excluded from analysis because of incomplete or missing data. Of the 213 included participants, 151 (70.9%) were diagnosed on clinical grounds with NeuP and 62 with NoP, which reflects the referral pattern to these clinics. The NeuP group included 106 participants who had a singular pain diagnosis of NeuP (e.g., peripheral painful neuropathy as the only pain diagnosis), and 45 participants with a mix of NeuP and NoP (e.g., a participant who had NeuP in the form of radiculopathy as the predominant pain and NoP in the form of axial mechanical and myofascial low back pain). The NoP group included 62 participants with osteoarthritis, joint pain, mechanical back pain, fibromyalgia, and chronic pain due to indeterminate origin or type. (Table 1).

Table 1.

Total number and numbers with NeuP, mixed pain, and NoP

Diagnoses	Number of participants
Total enrolled	221
Missing data	8
NeuP	151
NeuP only	106
Mixed pain (NeuP + NoP)	45
NoP only	62

For analysis, all subjects with mixed pain were analyzed with the NeuP group.

Descriptive participant data are provided in Table 2, and they show that age was the only statistically significant difference between the clinical diagnostic groups.

Table 2.

Demographic characteristics by color pain diagnosis and by clinical diagnosis

	Clinical Pain Diagnosis
Characteristic	NeuP (n = 151)	NoP (n = 62)	P
Age, years	49.5 (13.1)	45.1 (12.7)	0.025
Sex, female	85 (56.3%)	37 (59.7%)	0.763
Caucasian	139 (93.3%)	59 (96.7%)	0.515
Average pain rating (0–10)	5.8 (1.9)	5.6 (2.3)	0.749
Worst pain rating (0–10)	8.3 (1.8)	8.1 (1.8)	0.356
Duration of pain			0.427
<1 year	27 (18.1%)	13 (21.0%)
1 to 5 years	65 (43.6%)	29 (46.8%)
5 to 10 years	31 (20.8%)	7 (11.3%)
>10 years	26 (17.4%)	13 (21.0%)

Reported as mean (SD) or N (%).

The NeuP group included peripheral and central NeuP disorders, as seen in Table 3. A majority of the patients had peripheral NeuP in the form of radiculopathy, followed by post-traumatic neuralgia and idiopathic painful polyneuropathies. The radiculopathy group included subjects with cervical and lumbar radiculopathy resulting from herniated nucleus pulposus or spinal stenosis or occurring after spinal surgery (discectomy, laminectomy, or fusion). Although the majority of participants had a single NeuP diagnosis, 14 participants had more than one type of NeuP (e.g., peripheral neuropathy and radiculopathy).

Table 3.

NeuP diagnoses

Diagnoses	No. Diagnoses
Central post-stroke pain	4
Multiple sclerosis with myelopathy	2
Spinal cord injury pain	4
Phantom limb pain	2
Radiculopathy	47
Plexopathy	6
Idiopathic polyneuropathy	18
Painful diabetic polyneuropathy	10
Trigeminal neuralgia/neuropathy	4
Post-herpetic neuralgia	5
Peripheral nerve entrapment	5
Post-traumatic and postsurgical neuropathy	32
Complex regional pain syndrome	15
More than 1 NeuP diagnosis	14

During Step One of the CPD analysis, which relied on the first impression only, examiner 1 correctly classified 137 (64.3%) of all CPDs. Examiner 2 correctly classified 156 (73.2%) of all CPDs. Agreement between the two examiners on first impression was 184 (86.4%) of 213, showing strong agreement (Kappa: 0.73 [95% CI: 0.63–0.82]).

During the Step Two analysis, the examiners’ agreement increased to 212 out of 213 (99.5%) (Kappa: 0.99 [95% CI: 0.97–1]). The consensus diagnosis based on CPDs after Step Two correctly classified 171 (80.3%) of participants with a sensitivity of 80.1% [121/151; 95% CI: 72.7–86.0] and a specificity of 80.6% [50/62; 95% CI: 68.3–89.2].

A few examples of CPDs from patients in this study with representative NeuP and NoP disorders are provided (in Figure 2, A through L are examples of NeuP and M to R are examples of NoP). These drawings illustrate the information that can be gleaned from CPDs.

The NPQ score differentiated NeuP from NoP on clinical grounds and when CPDs were used. A number of items on the NPQ, such as burning pain, numbness, sensitive to touch, electrical, freezing, tingling, and increased pain due to touch, significantly differentiated NeuP from NoP in both the CPD consensus and the clinical diagnosis groups (P < 0.05; Table 4).

Table 4.

NPQ and Spielberger STAI by NeuP diagnosis

	Color Pain Diagram Consensus			Clinical Pain Diagnosis
Pain and Anxiety	NeuP (n = 133)	NoP (n = 80)	P	NeuP (n = 151)	NoP (n = 62)	P
NPQ score	0.7 (1.1)	−0.4 (0.8)	<0.001	0.6 (1.1)	−0.4 (0.9)	<0.001
NPQ ≥0	87 (72.5%)	17 (23.9%)	<0.001	89 (65.9%)	15 (26.8%)	<0.001
Burning pain	49.1 (31.6)	22.7 (28.5)	<0.001	46.7 (32.0)	20.8 (28.0)	<0.001
Sensitive to touch	49.8 (35.5)	22.8 (29.0)	<0.001	46.2 (36.1)	23.9 (29.1)	<0.001
Shooting pain	52.1 (34.8)	45.2 (33.3)	0.154	52.1 (34.5)	43.0 (33.3)	0.077
Numbness	48.1 (34.6)	18.3 (24.7)	<0.001	44.7 (34.6)	17.9 (25.5)	<0.001
Electric pain	36.9 (37.5)	17.8 (28.6)	<0.001	35.3 (36.9)	16.6 (28.2)	<0.001
Tingling pain	40.4 (29.6)	16.9 (24.2)	<0.001	37.7 (29.6)	16.3 (25.1)	<0.001
Squeezing pain	27.7 (33.8)	20.0 (30.0)	0.095	25.1 (32.8)	24.3 (32.3)	0.882
Freezing pain	19.6 (31.9)	3.2 (13.7)	<0.001	17.2 (30.3)	4.7 (17.8)	<0.001
Unpleasant	68.7 (24.4)	59.4 (25.2)	0.009	67.2 (24.1)	60.2 (26.9)	0.077
Overwhelming pain	67.3 (26.2)	58.5 (28.2)	0.025	65.1 (26.6)	61.2 (28.8)	0.360
Increased pain with touch	51.4 (37.0)	28.3 (32.7)	<0.001	47.8 (38.0)	30.3 (31.6)	0.001
Increased pain by weather	46.5 (39.5)	27.0 (30.6)	<0.001	43.0 (38.1)	30.1 (34.9)	0.020
STAI Anxiety total	41.7 (13.8)	40.1 (12.3)	0.359	41.5 (13.9)	40.1 (11.7)	0.453

Reported as mean (SD) and N (%).

There was no statistical difference in anxiety between patients with NeuP and NoP according to both diagnosis definitions, with mean (SD) total scores being 41.5 (13.9) and 40.1 (11.7), for clinical diagnosis, respectively (P = 0.453; Table 4).

Discussion

This is the first study to provide prospectively obtained information about the performance characteristics of CPDs in differentiating NeuP from NoP in a clinical setting. These results demonstrate a high rate of accuracy of CPDs in identifying NeuP in a routine clinical setting, as well as a high degree of concordance between two study physicians. Concordance with clinical diagnoses (the gold standard) and between study physicians was increased by applying predetermined grading criteria to the CPD analysis. These characteristics are within the range of other clinical tools used as aids to medical diagnosis [30, 31].

Previous research documented similarly high inter-rater reliability between examiners for quantifying the clinical data found in PDs [2, 32], with a test–retest reliability of r = 0.85 [3, 33, 34] and responsiveness to change with treatment [5, 35]. PDs have been useful in differentiating the types and prevalence of musculoskeletal pain conditions [36]. Distinctive and identifiable patterns in PDs have been reported for cervical facet joint pain [37], discogenic pain [38, 39], and herniated discs identified by myelography [39]. Ohnmeiss described the utility of PDs in assessing patients’ pain response to computed tomography / discography and reported 78% accuracy of PDs in differentiating patients with false positive pain responses [38]. The pattern and type of pain on PDs correlated with pain provocation on discography, and 82.7% with pain on computed tomography / discography had PDs classified as indicative of disc pathology [39, 40]. Additionally, the location of pain on PDs showed significant correlation with level of painful disc disruption on computed tomography / discography [39, 40]. PDs can also aid in diagnosing the level and degree of disc herniation [7, 41] and distinguishing among different types of lower back pain [42–44]. In one study, eight expert physicians with experience in the use of PDs in patient evaluations were able to classify with 51% accuracy patient-completed PDs from five lumbar spine disorder categories: benign disorder, herniated disc, spinal stenosis, underling disorder, or psychogenic disorder [42]. Comparison of PDs classified by lower back pain experts with classification using discriminant analysis and computerized artificial neural networks showed associations between the predicted pain patterns for each diagnostic group made by an expert and the computer-generated pain patterns [43].

Research on the role of PDs in adult patients with pain has largely been confined to spine pain. Very few studies are available on the use of PDs for other types of pain conditions, such as chronic upper-extremity pain [45], orofacial pain [46], or hip osteoarthritis [47]. Poulsen reported that the most common locations and distributions of pain in hip osteoarthritis are the greater trochanter, groin, thigh, and buttock [47]. A recent report described the use of PDs in identifying central sensitization in patients with hip osteoarthritis. A greater pain extent on PDs in hip osteoarthritis was significantly associated with measures of central sensitization, i.e., higher pain scores on widespread pain index, Pain DETECT, and pain catastrophizing and lower pressure pain thresholds on quantitative sensory testing [48]. Conclusions about symptom distribution from PDs in the case of musculoskeletal pain may be improved by the concomitant assessment of pain catastrophizing [49].

Making a diagnosis of NeuP can be complex and time consuming and can involve multiple steps. It begins with obtaining information on the distribution and quality of pain (i.e., symptoms), followed by a detailed and meticulous neuro-sensory exam to demonstrate painful and sensory signs corresponding to the symptoms identified in the patient’s history. Further workup with imaging, electrophysiological, or histopathological studies is often indicated to confirm a diagnosis of NeuP [50]. The diagnosis of NeuP can be missed or delayed because of failure to appreciate and correctly interpret the symptoms and signs of NeuP. A CPD captures characteristics of pain quality and distribution in a single graphic presentation in a time-efficient manner. CPDs can signal the presence or absence of NeuP and lead to a directed history and physical examination, thereby enhancing diagnostic accuracy. On the basis of information from symptoms and signs, the diagnostic process can be further refined, by combining the information from the CPD with information on other aspects of pain, such as pain interference with functioning from the Brief Pain Inventory. The present study is the first study to establish CPD as a useful clinical tool in demonstrating the presence of NeuP in patients with chronic pain. Results from this study provide support that information from the NPQ further delineates and confirms specific symptoms in support of a NeuP diagnosis.

Interpretation of CPDs for the purposes of making a NeuP diagnosis requires a good knowledge of neuroanatomy. Certainly, a first step is to determine plausibility—for example, that no coloring is done outside the body diagram and that the distribution of colored areas represents projections of plausible anatomic structures. A lack of familiarity with specific knowledge and nuances of neurological localizations related to peripheral and central sensory pathways can be overcome by referring to published anatomic charts of innervation or figures available through mobile medical applications.

The promising results of this study were likely possible because both study physicians are pain physicians with extensive knowledge of the peripheral and central nervous anatomy and innervation, as well as the distribution of sensory abnormalities in NeuP resulting from injury and diseases of the peripheral and central nervous system. Their knowledge and experience equipped them to classify pain as NeuP or NoP. The participation of experienced pain physicians was essential for this study. Pain physicians are familiar with the diagnosis and treatment of NeuP and are well equipped to diagnose and treat NeuP on the basis of the information provided in the study. Another aspect of this study is that these two physicians developed agreement criteria for Step Two, which harmonized their efforts.

Completion of CPDs by patients is intuitive and efficient with respect to resources and time; it requires only a set of coloring pens and paper, and the drawing takes a few minutes to complete. The information from CPDs described here could also be captured on electronic devices such as tablets with appropriate applications, though it remains to be confirmed whether digital body schemas increase the accuracy and precision relative to paper-and-pen PDs [51–53]. Most NeuP disorders have readily recognizable signs and symptoms. This serves as the basis of neurological localization and diagnosis for disorders such as peripheral neuropathy, mononeuropathy, and radiculopathy. CPDs have the potential to localize pain and associated symptoms to the central nervous system, spinal segment, or peripheral nerve distribution by referencing to known anatomic charts and body maps [54]. In our present study, there is a strong agreement between experienced examiners (inter-rater reliability) for identifying NeuP patterns by using criteria that combine pain distribution and quality. In routine clinical practice, clinicians can recognize that NeuP has a characteristic pattern and qualities, which are distinct from those of NoP. However, careful testing of this assumption has not been systematically carried out.

Although several manual scoring systems have been described previously, these have been used to screen and psychologically classify patients with pain [19, 55, 56]. A statistical model of body pain areas colored in PDs and negative emotions has been shown to predict a large component of the variance of pain intensity in patients with fibromyalgia [57]. Broadbent et al. reported an association between PDs and illness perceptions, mood, and health outcomes [58]. No study has previously assessed the role of CPDs in screening for the presence or absence of NeuP.

Potential limitations of CPDs include imprecise coloring of CPDs by participants. This could lead to errors in pain description, such as coloring outside the body part, wrong color choice, and confusion of sides (i.e., right and left on the front and back). Use of color as the primary mode of communication, as in CPDs, is of limited usefulness for patients and clinicians who are color-blind. Other limitations include failure to confirm that the body area described clinically is the same as the area depicted in the CPD. This may have lowered Kappa values for concordance with clinical diagnosis. Another possible limitation is that almost one third of the patients had lumbar radiculopathy. This may make it easier to use a PD, because radiculopathy is usually more easily anatomically defined (dermatomes) than some other neuropathic conditions (like post-laminectomy pain syndrome) that can be more variable in distribution or “vague” when described by patients. Therefore, with a large proportion of patients having radiculopathy, this may have made it “easier” for pain physicians to interpret the CPD. Finally, we have not evaluated the performance characteristics of CPDs in clinicians with limited clinical experience and training, such as residents and pain medicine fellows.

Future studies are needed to reproduce the results of this study and determine whether CPDs enhance a clinician’s ability to recognize the presence of NeuP.

Conclusions

In conclusion, the CPD is a useful clinical tool to screen for the presence or absence of NeuP. There is strong agreement between examiners indicating inter-rater reliability for identifying NeuP if strict prespecified criteria for interpretation are used to classify CPDs. Future research will further identify and describe distinctive and recognizable patterns of NeuP based on both the distribution and qualities of pain.

 

Funding sources: The project was supported by the Clinical and Translational Science Award (CTSA) program through the National Institutes of Health National Center for Advancing Translational Sciences (NCATS; grant UL1TR002373). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of interest: There are no conflicts of interest to report.

Scott Hetzel, MS, Researcher/Institute for Clinical and Translational Research Biostatistcs & Epidemiology Research Design (ICTR BERD) Core Manager, conducted the statistical analysis. Study sponsorship.