Differential Sensory and Clinical Phenotypes of Patients With Chronic Widespread and Regional Musculoskeletal Pain

Marina M. Finnern; Dieter Kleinböhl; Herta Flor; Justus Benrath; Rupert Hölzl


Pain. 2020;162(1):56-70. 

In This Article



One hundred eighty-five patients with chronic primary pain of musculoskeletal origin participated in the study, which was part of a doctoral thesis.[13] The sample was the same as in our previous study on general components of pain chronicity[14] and was recruited from collaborating pain clinics. The mean pain duration of the sample was 48 months (SD = 29 months; range: 7–539 months). The majority (65.4%) had had pain for longer than 5 years. In addition, 41 sex- and age-matched pain-free controls (PFC) drawn from the volunteer pool of the Central Institute of Mental Health were assessed as reference for the sensory data of the patients. Supplemental Figure S1 (available at https://links.lww.com/PAIN/B131) presents details on the recruitment of patients, and Table S1 (available at https://links.lww.com/PAIN/B131) provides information on the inclusion and exclusion criteria. Briefly, patients were included if their major pain symptoms originated from the musculoskeletal system, lasted for at least 3 months, and were not related to acute trauma, inflammatory or neurologic disease, without radicular and neuropathic signs.[49,67] The study was approved by the Ethics Committee of the Medical Faculty Mannheim, University of Heidelberg.

Seventy-eight patients (73 women and 5 men) were diagnosed with fibromyalgia syndrome (FMS) according to the original ACR criteria of 1990 (pain in 4 quadrants and the axis plus 11 of 18 TPs).[74] One hundred seven patients (69 women and 38 men) fulfilled the criteria for CBP requiring that pain in the upper or lower back was the primary problem.

Age was comparable between the patient groups, and they did not significantly differ from PFC (mean: 48.9 ± 8.8). There were significantly more women than men in the FMS group[35] and the PFC (N = 30 women vs N = 11 men), whereas the CBP group had about an even sex distribution (Table 1). Although pain duration did not significantly differ, the Chronic Pain Grade (CPG)[43] was significantly higher in patients with FMS compared to patients with CBP (CPG III vs II; U test, P = 0.007). Daily pain medication use was also significantly higher in the FMS group (50%) than in the CBP group (22.4%, Ccorr, P < 0.001). Patients with FMS also had significantly more self-reported medical life-time diagnoses (Table 1).

Clinical Pain Assessment

The pain assessment of the patients involved the German versions of the West Haven-Yale Multidimensional Pain Inventory (MPI-D),[17] the German Pain Questionnaire,[48] the Fibromyalgia Impact Questionnaire,[8,51] and the Hannover Functional Ability Questionnaire (FFbH-R).[42] In addition, pain-related cognitions and coping with pain were assessed by the Pain-Related Self-Statements Scale (PRSS)[16] and fear avoidance beliefs with the Fear-Avoidance-Beliefs Questionnaire (FABQ).[53]

Pain loci were extracted from the German Pain Questionnaire. Reported pain localizations were aggregated into 9 regions according to Axis I of the IASP Taxonomy of chronic pain[34] and into 19 regions for the WPI using the regions of the Fibromyalgia Survey Questionnaire (FSQ),[28] the survey version of the FS scale.[71]

Sensory Testing

Pressure Pain Sensitivity. Sensitivity to pressure stimuli was assessed semiquantitatively by a standardized manual probe (≈10 N/cm2 per second up to ≈ 40 N/cm2 max) by trained medical personnel at 18 TPs and 14 control points (CPs) adapted from Refs. 52,74 (cf. Figure 2 below). Perceived pain intensity at the 40 N endpoint was rated on a visual analogue scale (VAS) from no to worst pain (transformed to values from 0 to 10). The lenient criterion of VAS ≥1 served to indicate a "sensitive point" to cover the complete sensitivity range all over the body. In addition, the spatial distribution of pressure sensitivity was evaluated by regionally aggregated measures. For this purpose, the 32 single ratings of pressure pain intensity in the manual probe were averaged for 5 body regions adapted from the Heidelberg pain drawing mask (head-cervical, thoracic, lumbar, upper and lower limb region; supplemental Table S2, available at https://links.lww.com/PAIN/B131).[23] In addition, pressure sensitivity was assessed quantitatively by a commercial algometer (Algometer; Wagner, Inc, Greenwich, CT) with an 8-mm diameter stimulation area. Phasic pressure pain thresholds ("just painful") were obtained bilaterally with the ascending method of limits at the ACR TP in the center of the trapezius and the control point on the thenar eminence (≈5 N/s ramps; mean of 3 trials). In addition, pain intensity was rated by a numerical rating scale (NRS) from no (0) to worst (10) pain for each completed series.

Heat Pain Sensitivity. Heat pain sensitivity was assessed at the trapezius (dominant side) and the thenar (nondominant hand) using a Peltier thermode (rectangular contact surface of 1.6 × 3.6 cm; PATH Tester MPI 100, Phywe, Göttingen, Germany; accuracy 0.05°C; 0.7°C/s heating and 2.0°C/s cooling rate).[19] Phasic heat pain thresholds were measured with the method of limits (means of 5 trials) and the method of adjustment, the latter in connection with the tonic threshold defined as the "just painful" temperature readjusted after 30 seconds (means of 3 trials). Suprathreshold sensitivity was assessed with the previously established tonic ("dual sensitization") protocol using 30-second constant stimuli at 9 temperatures from −1.0° to +1.0° with reference to individual self-adjusted thresholds.[38–40] The psychophysical function (Steven's coefficient) is determined from the intensity ratings at the end of the tonic stimulation on a numerically and verbally anchored VAS (0 = warm to 100 = very strong pain; "just painful" set at 40). The perceived change of temperature and heat pain over 30 seconds serves as marker of temporal summation or short-term sensitization reported behaviorally (ΔT°C readjustment) and by the VAS (ΔE rating with numerical temperature descriptors −0.55° to +0.55°C).

Assessment of Somatic Health, Comorbidity With Mental Disorders, and Psychosocial Factors

Somatic Symptoms and Health. Somatic symptoms were assessed by the SF-12, physical component (short form of the SF-36, Health-Related Quality of Life Questionnaire),[7] and the global symptom score of the Giessen Symptom Questionnaire (GBB-24).[6] Specific gastrointestinal disorders (functional dyspepsia/nonulcer dyspepsia and/or irritable bowel syndrome) were assessed by the questionnaire for functional gastrointestinal disorders (FGID) adapted from previous Rome criteria considering medically unexplained pain in the upper and/or lower abdomen.[12,31] Sleep quality was assessed by the Pittsburgh Sleep Quality Index.[9]

Comorbidity With Mental Disorders. General mental health was examined by the SF-12, mental health component, depression and anxiety by the German versions of the Center for Epidemiological Studies Depression Scale (ADS)[30] and the Trait Anxiety Scale of the State-Trait Anxiety Inventory.[44]

Psychosocial Factors. The Brief Stress Scale[15] was applied to determine self-reported stress burden in 4 areas of life (marital relationship, daily problems, social contacts, and trouble at work) sum score for overall strain. Physical load at work was assessed by the Dortmund questionnaire on workload (DFA).[41]

Data Analyses

For sensory and clinical data, a sequential multimethod strategy was used including correlational and contingency analysis with principal components (PCA) followed by latent class analysis (LCA) to identify indicators for stable sensory and clinical profiles to be combined in pain phenotyping. Details of the data reduction procedures and the multistep analyses are described in the supplemental material (available at https://links.lww.com/PAIN/B131).

This article reports the results of the final robust LCA of core pressure-sensitivity and clinical pain indicators defining coherent sensory-clinical phenotype clusters of patients across previous diagnostic groups. The identified sensory-clinical profile clusters were cross-validated by a stepwise linear discriminant function analysis applying Wilks' lambda for variable selection. These groups were compared with respect to the complete set of sensory and clinical pain markers not included in the profile definition (thermonociception, coping, and consequences of pain), comorbidity, and psychosocial factors.

Data were analyzed with the program packages IBM SPSS Statistics (version 25; Armonk, NY) for the PCAs and R program pvclust (version 3.4.1; The R Foundation for Statistical Computing, Vienna, Austria) and LatentGOLD program LC cluster (version 5.1; Statistical Innovations, Inc, Belmont, MA) for the LCA.[57,59,60] Cluster solutions at the parameter level were evaluated by arbitrary unbiased/bootstrap probability indices (AU/BP) with minimal AU ≥95% to protect against random errors.[57,59,60] The final sensory-clinical pain phenotype clusters were derived according to convergent fit indices (Log-Likelihood, Bayesian Information Criterion, and Akaike Information Criterion).[65]

Parametric t tests and nonparametric U tests for ordinal data were used for group and cluster comparisons confirmed by ex-post K-S 2-sample tests. Cross-classifications of diagnostic subgroups and clusters were evaluated by the corrected contingency coefficient (Ccorr). In view of current controversies about the validity of conventional significance levels,[1] exact probabilities are reported up to P = 0.001 where possible; else, classical significance was set at P < 0.05. Significance levels were familywise Bonferroni–Holm corrected within parameter domains. Effect sizes of group comparisons are reported as Cohen's d (interval scales) or suggested equivalents for ordinal scales (z values transformed to η2 [eta-squared]) or nominal scales (phi coefficient).[18]