Multisensory sensitivity differentiates between multiple chronic pain conditions and pain-free individuals.

Individuals with chronic pain and pain-free individuals were recruited to participate in an online survey study using the Research Electronic Data Capture (REDCap) interface to collect all data. The primary pain cohorts that were targeted included people with FM, migraine, or chronic LBP. These three conditions were chosen to reflect different degrees of central mechanism involvement. For example, while chronic LBP can involve central mechanisms [ 3 , 50 ], LBP patients report primarily localized pain and sensitivity as opposed to the widespread pain and symptomology typical in FM [ 1 , 16 , 21 , 35 ]. Migraine shares characteristics of both LBP and FM, as it typically involves relatively localized pain yet can involve a myriad of symptomology [ 24 ]. All participants were required to be at least 18 years old and fluent in English to participate in this study. Additional inclusion criteria for the pain-free group included: no current pain and no history of chronic pain in the past year. Recruitment involved multiple Institutional Review Board (IRB) approved methods, including university-wide mass email, Facebook, and Twitter. We collaborated with the National Fibromyalgia Association to recruit subjects with FM and the American Chronic Pain Association to recruit individuals with all types of chronic pain through email and website notices. Due to initial low recruitment of chronic LBP patients and pain-free adults aged over 30 years, we also specifically recruited these populations through Amazon Mechanical Turk, an online marketplace for recruiting individuals to complete small tasks for monetary compensation. Those recruited through Mechanical Turk were additionally required to have completed 50 tasks on Mechanical Turk with an approval rate of 95% or higher to encourage high quality responses. Interested participants were asked to complete a set of 5 surveys, encompassing demographic information, their pain condition(s), multisensory sensitivity (MSS), general pain-rating schema, and several psychological traits, including anxiety, depression, negative affect, and positive affect. The surveys took approximately 10–15 minutes to complete. Respondents who completed the surveys were entered into a random drawing (1 out of every 10) for a $20 Amazon e-gift card as an incentive if they provided their contact information. The survey study was approved by the University of Iowa Biomedical IRB with exempt consent in accordance with the Declaration of the World Medical Association. A total of 1262 individuals submitted at least a portion of all surveys out of 1563 respondents who initially signed on (80.7% completion rate, Figure 1 ). If less than 10% of the total number of survey items were missing, missing values were imputed as the mean of the remaining items. If more than 10% of items were missing, the survey was considered incomplete, and subjects were excluded from these analyses (n=41). Additionally, participants who did not meet their respective self-identified category: chronic pain (i.e., did not claim any chronic pain conditions) or the pain-free cohort (i.e., reported 1 or more COPCs) criteria, were excluded (n=16). Thus, a total of 1205 individuals were included in the final dataset: 785 individuals with chronic pain and 420 pain-free adults. However, 28 participants reported 1 or more COPC that did not include LBP, migraine, or FM; thus, for the diagnosis-based analyses the sample size was only 1177. These individuals were included only in the second aim evaluating COPC-based groupings. The sample sizes for each chronic pain subgroup (diagnosis-based and COPC-based) are provided in Figure 2 . Demographic information collected for all subjects included age, sex at birth, height, weight, race, and zip code. Subjects were asked to indicate if they regularly took prescribed or over-the-counter medications to prevent or treat any of chronic pain conditions (including depression and anxiety medications as often these are prescribed both for pain and psychological benefit) and whether they had used any opiate pain relivers (e.g., OxyContin, Vicodin, Percocet, etc.) or marijuana products (e.g., CBD oil, weed, gummies, etc.) during the past four weeks (yes/no responses). The presence of multiple comorbid pain conditions lasting at least 3 months in the past year was assessed using a list of common chronic overlapping pain conditions (COPCs) proposed by the Institute of Medicine Consensus Report [ 44 ], including upper and lower extremity musculoskeletal pain conditions ( Table 1 ). Current pain, average pain and worst pain in the last 7 days were assessed using a 0–10 pain rating scale, where 0 = no pain and 10 = the maximum pain you have ever felt or can imagine feeling, allowing responses in whole or half numbers. Pain-rating schema, or the general strategy used to conceptualize numerical pain ratings, was assessed using a brief version of the Pain Schema Inventory (PSI) [ 18 ]. Six items from the original 18-item instrument measured self-perceived pain intensity corresponding to hypothetical mild, moderate, and severe joint pain and headache situations. The mean of the six items provides a pain-rating schema index as an indicator of how on average the 0–10 scale may be conceptualized by different individuals. The original PSI demonstrated good validity with excellent internal consistency (Cronbach’s alpha = 0.96) and reliability (ICC = 0.98) [ 18 ]. MSS was assessed using the multisensory amplification scale (MSAS), a validated measure of MSS [ 58 ] modified from the Somatosensory Amplification Scale [ 10 ] to incorporate additional sensory domains. The MSAS consists of 12 items that assess sensory sensitivity (high and low) to normal everyday occurrences involving visual, auditory, tactile/food texture, olfactory and internal somatic sensations. Subjects were asked to rate how true each statement was to them (i.e., “Strong perfumes or colognes really bother me”) using a Likert-type scale ranging from 1 to 5 (1 = not at all true, 2 = a little bit true, 3 = moderately true, 4 = quite a bit true, 5 = extremely true). The sum of the 12 items is used as the MSAS score (range 12 – 60). Higher total scores indicate higher MSS. The MSAS has demonstrated good construct validity, internal consistency (Cronbach’s alpha = 0.82), and excellent test-retest reliability (ICC 3,1 = 0.90) [ 58 ]. In addition to the continuous MSAS score, each individual was assigned to a categorical MSS level: high, high-average, low-average, and low based on previously reported sex-specific quartile cut-offs (see Table 2 ) [ 58 ]. Respondents (N=647) in the previous study included both pain-free individuals and those reporting a range of pain intensities, recruited regardless of health conditions [ 58 ]. Thus, these cut-offs better represent the normal range of MSS in a more generalized population than if we had used quartiles based solely on this targeted sample. Low MSAS cut-offs represents the 1 st quartile responses from this prior study, low-average represents the 2 nd quartile, high-average represents the 3 rd quartile, and high MSAS is based on the 4 th quartile responses. Thus, the range of normal variation in MSS responses from very low sensory sensitivity to very high sensory sensitivity is represented. These cut-offs are sex-specific as women demonstrated greater MSS than men on average [ 58 ]. Several common psychological traits were assessed for use as covariates in this study as they have been shown to influence pain. The Patient-Reported Outcomes Measurement Information System (PROMIS) Anxiety and Depression 4-item short forms were used to assess anxiety and depression constructs, respectively. For these, participants were instructed to rate how often they felt each statement in the past seven days (i.e., “I felt worthless”) on a 1 to 5 scale: never, rarely, sometimes, often, and always. Thus, higher scores reflect greater anxiety and depressive symptoms. The PROMIS-Anxiety and Depression scales have been approved as reliable measures with good construct validity and internal consistency (Cronbach’s alpha = 0.89 and 0.93, respectively) [ 28 ]. Positive and negative affect was assessed using the Positive and Negative Affect Schedule (PANAS), a 20-item instrument that includes a list of 10 positive and 10 negative feelings and emotions [ 56 ]. Subjects rate how much each word applies to them in general using a 5-point Likert scale, with 1 being very slightly or not at all and 5 being extremely. The sum of the corresponding item ratings was used to quantify negative and positive affect, where higher scores reflect higher degrees of each construct. The PANAS displays good internal consistency with Cronbach’s alpha ≥ 0.84 for each scale and good convergent and discriminant validity [ 56 ]. Due to known correlations between anxiety, depression and negative affect, a single negative factor was computed for use as a covariate in the statistical analyses. A single composite score was created by summing z-scores for the anxiety, depression, and negative affect scales and transforming them to T scores (range 0 to 100) for ease of interpretation [ 53 , 59 ]. That is, higher scores indicate stronger negative emotionality and psychological distress. Positive affect remained separate as positive and negative affect are independent constructs [ 56 ]. Only participants who completed all key surveys were included in the analyses, assessing basic demographic information including pain and pain conditions, MSS, negative emotionality surveys and pain-rating schema. Descriptive analyses were performed on all variables, reporting mean ± standard error of mean (SEM) or proportion (%) as appropriate. Chronic pain participants were first classified into 5 subgroups considering the three primary pain conditions targeted and their overlap as follows: LBP only, LBP + migraine, migraine only, FM + migraine, and FM only ( Figure 2a ). Because FM by definition involves some form of neck or back pain [ 62 ], we did not separately evaluate a group of FM + LBP as we cannot be confident that those reporting “FM only” do not also experience LBP, but simply consider it a part of their condition. To assess differences in MSS between these five diagnosis-based patient subgroups and the pain-free cohort, general linear model (GLM) analyses were performed where the continuous MSAS score was the dependent variable and subgroup allocation was the primary independent variable. Thus, a total of 6 subgroups were considered: 1 pain-free and 5 chronic pain groups (e.g., LBP, migraine, or FM only, as well as LBP + migraine and FM + migraine). To evaluate the inclusion of multiple covariates in the GLM model, a series of hierarchical models were considered with 5 combinations of 4 potential covariates: 1) demographic variables, 2) pain-rating schema index, 3) composite negative factor, and 4) positive affect. At each level of the hierarchical progression, we first assessed for a significant sex interaction with diagnosis subgroup. If there were no significant sex interactions, the GLMs were repeated without the interaction term to simplify model interpretation and optimize study power. Model residuals were tested for normality using the Kolmogorov-Smirnov test. Model fit tests were assessed for each level to identify the model that best represented the data, based on the Loglikelihood Ratio Test and Akaike Information Criterion (AIC). Optimal model fits are reflected by lower AIC values, coupled with significant Ratio tests to indicate significant improvement over other models [ 15 ]. To adjust for consideration of multiple models, significance between models was assessed as p < 0.01. Using the optimal model, when significant between-group differences in MSS were found, post-hoc tests were performed using false discovery rate (FDR) adjustment for multiple comparisons (15 pairwise subgroup combinations possible) [ 11 ]. Standardized effect sizes (Cohen’s d with 95 th confidence intervals, CIs) were computed for the mean differences in MSAS scores between each subgroup pairing using adjusted least square (LS) means to account for age, sex and BMI, and standard deviations (SD) with adjustment for unequal sample sizes [ 25 ]. Effect sizes were operationally defined using the following cut-off values: small (d ≥ 0.2), medium (d ≥ 0.5), and large (d ≥ 0.8) [ 14 ]. To address our secondary aim, the individuals with chronic pain were sub-grouped largely by the number of pain co-morbidities more broadly ( Figure 2b ). Because FM inherently involves pain in multiple anatomical regions, which respondents may or may not conceptualize as additional pain conditions (e.g. back, arm, or leg pain), we considered FM separately when categorizing number of COPCs. Thus, we divided participants into four categories with least to most pain comorbidities as follows: isolated pain conditions only (i.e., reported only 1 pain condition that was not FM); ≥ 2 pain comorbidities but not FM; FM with zero or one other comorbidity; and FM with ≥ 2 other pain comorbidities. Further, when identifying number of pain comorbidities in people with FM, we did not include spine or musculoskeletal extremity pain when counting additional comorbidities, as they are typically part of the diagnostic symptoms of FM [ 62 ]. Thus, the additional co-morbidities identified for FM patients included all other options listed in the first row of Table 1 . Further, to reduce the likelihood of over-estimating comorbidities that may in fact represent a common underlying condition, similar or regional pain conditions were considered as a single pain condition for these analyses. For example, LBP and upper back/neck pain responses were counted as ‘spine pain;’ chronic migraine or tension type headaches were counted as ‘headache;’ all noted upper extremity (UE) pain conditions were counted as ‘UE pain’; and all lower extremity (LE) pain conditions were counted ‘LE pain.’ This promoted a conservative estimate of different pain co-morbidities for use in the subgrouping classification. The natural order of the COPC-based subgroups along the chronic pain continuum, potentially representing increasing degrees of central sensitivity involvement from pain-free individuals through those with FM with 2 or more COPCs ( Figure 2b ). To assess the likelihood of having more COPC involvement as a function of MSS, we computed odds ratios (ORs) between the COPC-based subgroups and degree of MSS based on the normative sex-specific MSS quartile allocations ( Table 2 ). Multinomial logistic regression analyses were used to compute the ORs of being in each successive COPC-based subgroup relative to being the pain-free group for both high and low MSS levels using the low-average MSS level (normative 2 nd quartile level) as the reference. This reference enabled assessment of both vulnerability and resilience to more COPCs (i.e., indirect marker of centrally mediated pain) based on MSS. The same series of hierarchical models to identify optimal model covariates were considered for these logistic regression analyses. Similar goodness of fit metrics as described above were used. To further characterize MSS, we assessed the proportion of each MSS level within each pain subgroup and assessed inter correlations between MSS and the psychological traits with the adjustment of age, sex, and BMI (Pearson correlation coefficients). Because of the difference in proportion of men and women across the three pain cohorts, we also performed an exploratory analysis repeating the between group differences in MSS with only women. Because LBP patients were specifically recruited through both social media and through Amazon Mechanical Turk, whereas the other cohorts were largely recruited through social media and Association communications, we conducted a secondary analysis to determine if the LBP participant demographic characteristics differed between recruitment sources. Finally, to determine if pain intensity differed with MSS, we compared current pain and worst 7-day pain across MSS categories within each pain subgroup, with adjustment for age, BMI, sex, and pain-rating schema. All statistical tests were conducted using SAS, version 9.4, USA. Significance level was set at P ≤ 0.05 unless otherwise noted (e.g., Goodness of Fit tests and FDR adjustment).

Of the 1205 included in the final analysis, 95.7% of subjects were from the United States; a total of 47 states were represented, with the largest proportion (47.8%) coming from Iowa ( Supplemental Materials , Figure S1 ). Just under 4% of the study sample came from 12 additional countries (See Supplemental Materials , Table S1 ). The demographic information across the total study population, considering both chronic pain subgrouping strategies and the pain-free cohort are presented in Tables 3 and 4 . All demographic variables were significantly different between at least one and often more of the chronic pain subgroups and the pain-free group (p < 0.0001). That is, age, BMI, and proportion of females were generally higher in the FM and multiple co-morbidity groups than in the pain-free, migraine, or LBP subgroups. Age did not differ between pain-free individuals and those with 1 or multiple COPCs, but those with FM and any number of COPCs were older. Further, the highest proportion of males was in the LBP subgroup (50.4%) compared to all other subgroups including pain-free individuals (2.5 – 26.7%). The pain-rating schema index showed pain-free individuals tended to rate mild, moderate, and severe hypothetical pain more than 1 point lower on average (0 – 10 scale) than those with FM or multiple COPCs, with increases spread incrementally across pain subgroupings ( Tables 3 and 4 ). FM patients reported higher current pain and worst 7-day pain than non-FM patients even after adjusting for age, sex, BMI, and pain-rating schema (see also Supplemental Materials , Figure S2 ). Medication use and/or use of marijuana-derived products were significantly greater in all chronic pain subgroups than in pain-free individuals and generally increased with greater number of COPCs ( Table 4 ). When considering pain diagnosis subgroups, those with only migraine reported the lowest use of opiates or marijuana compared to the other 4 pain subgroups (6.7% vs 31 – 46%), but similar prescription or over the counter pain medication use (92% vs 83 – 98%, Table 3 ). Negative psychological traits were higher and positive affect lower across all pain subgroups compared to the pain-free cohort ( Supplemental Figure S3 , Tables S2 and S3 ), particularly for those with FM. LBP participants recruited from Amazon Mechanical Turk had more males, were more racially and ethnically diverse, and reported higher use of opiates or marijuana (p ≤ 0.03 for all), but no differences in age, BMI, other pain medications, MSAS, pain rating schema, or any of other psychological traits were observed compared to those recruited through social media (p > 0.05, Supplemental Table S4 ). The comparisons of continuous MSAS scores across diagnosis-based subgroups and across COPC-based subgroups both remained significant across all five hierarchical models tested (group, p ≤ 0.0001). The model residuals were normally distributed (p > 0.15). Based on goodness of fit tests amongst the five models, adjustment for age, sex, BMI, pain-rating schema, and negative emotionality, but not positive affect was chosen (model 4, Table S5 ). MSAS scores were significantly higher in those with any pain diagnosis than in pain-free individuals, even after adjusting for age, sex, BMI, negative emotionality, and pain-rating schema (adjusted p ≤ 0.0007 for all, Figure 3a ). Effect sizes relative to the pain-free cohort ranged from moderate to large (Cohen’s d = 0.32– 1.20, Table 5 ). Among the pain groups, MSS generally increased as hypothesized: LBP only < migraine only < FM only (adjusted p ≤ 0.02, Cohen’s d = 0.38 – 0.71, Table 5 ). The difference in MSS between those reporting both LBP and migraine were not significantly different than those reporting either condition without the other, whereas those reporting FM and migraine had significantly higher MSS than those with FM but not LBP or migraine. However, this difference was small, Cohen’s d = 0.19. Note also that these classifications only considered presence of these three targeted diagnoses and no other comorbidities. When classifying groups by considering number of pain comorbidities, pain groups showed significantly higher MSS than the pain-free cohort (adjusted p ≤ 0.0001 for all, Cohen’s d = 0.57– 1.13, Figure 3b and Table 5 ), except for the group of individuals with only isolated pain conditions (adjusted p = 0.18). When comparing across pain groups, the MSS score was increasingly higher across subgroups (adjusted p ≤ 0.004, Cohen’s d = 0.46 – 0.97). It is worth noting that although FM with ≥ 2 pain comorbidities group showed statistically significant higher MSS than the FM with one or no pain comorbidity group (adjusted p = 0.02), the effect size (Cohen’s d = 0.18) was very small and may not be clinically meaningful. The likelihood of being in pain subgroups with more COPCs between MSS quartiles were significant across all five models (group, p ≤ 0.0001) with normally distributed model residuals (p > 0.15). Based on goodness of fit tests amongst the five models, the same covariates as for the GLM analyses were again chosen as the most appropriate to include in the model (model 4, Table S6 ) as the further improvement with model 5 was marginal. Thus, the multinomial logistic model included age, sex, BMI, pain-rating schema, and negative emotionality as covariates. The ORs of being in one of the COPC-based subgroups versus being in the pain-free cohort for those with high, high-average, or low MSS relative to low-average MSS (2 nd quartile) are shown in Figure 4 . Those with the highest MSS levels had 2 – 5.5 times higher odds of having 2 or more pain co-morbidities: OR [95% CI] = 1.98 [1.15, 3.42] to 5.56 [2.74, 11.28] (p ≤ 0.0001) across subgroupings. Further, those with the lowest MSS levels were 55%–87% less likely to have 2 or more pain comorbidities than the low-average MSS (OR [95%CI] = 0.45 [0.22, 0.88] to 0.13 [0.05, 0.39]; p ≤ 0.0001). However, MSS levels did not alter the odds of having an isolated pain condition versus no pain condition ( Figure 4 , Supplemental Table S7 ). Further, there was no difference in odds of having any number of pain comorbidities between those with high-average versus low-average MSS levels. Our exploratory analyses assessing these relationships in female participants only generally showed similar findings ( Supplemental Tables S8 and S9 ). The proportions of MSS levels within each pain subgroup are presented in Figure 5A and 5B . MSS levels were approximately equally distributed in the pain-free group representing normative quartiles. However, the distribution of the highest MSS level became increasingly disproportionate for LBP, migraine, and FM diagnoses (44.5% – 85.5%) and with increasing COPC classification (41.8% – 84.1%). The comparisons of current pain or worst pain in last 7 days across MSS levels within each subgroup for both classifications were also conducted. Within each pain subgroup, there were generally no significant differences in current pain, or worst last 7-day pain observed across MSS levels (p > 0.1, data not shown). The intercorrelations between MSS scores and psychological constructs are provided in Table 6 . The negative emotionality factors, depression, anxiety, and negative affect were all highly intercorrelated (R ≥ 0.73, p < 0.0001), demonstrating 53 – 56% shared variance. While significant, only weak to moderate correlations were observed between MSS and the psychological constructs (R ≤ 0.41, p <0.0001), corresponding to 4 – 16% shared variance.

As hypothesized, MSS varies with type and number of chronic pain conditions thought to involve a spectrum of central sensitivity. FM patients reported higher MSS than migraine patients, followed by LBP patients, with moderate to large effect sizes. Moreover, those with high or low MSS, based on normative data cut-points, showed increased risk or resilience, respectively, to having multiple COPCs, also considered to be an indicator of centrally mediated pain. Thus, self-reported MSS may serve as an additional indirect marker of generalized CNS processing of sensory input that is relevant to pain. Our findings suggest variation in self-reported MSS, assessed with a brief survey, may be a clinically feasible tool for potential use in phenotyping clinical pain populations. Neuroimaging studies have elucidated the overlap in regions involved in processing painful and non-painful sensations. The “pain matrix,” brain regions commonly associated with pain processing [ 2 ], are also activated by non-painful stimuli [ 17 ]. More precisely, functional magnetic resonance imaging (fMRI) studies demonstrate substantial overlap in brain region activation with noxious and non-noxious sensory stimuli [ 33 , 45 , 54 ]. For example, common ascending and descending connectivity patterns are seen with both laser-induced pain, selectively activating nociceptive Aδ and C fibers, and electrically-induced nonpainful tactile sensations [ 33 ]. Yet, unique neuroimaging patterns are also seen with painful stimulation [ 33 ]. Similarly, in children and adults with sensory modulation disorders (SMD), neurophysiological EEG and psychophysical studies show shared mechanisms between pain and non-painful multisensory processing [ 4 – 8 , 49 ]. Furthermore, individuals with SMD exhibit normal peripheral sensory processing but sensitized central pain pathways [ 4 , 6 , 7 ]. Several central mechanisms have been proposed to explain the relationship between SMD and pain sensitization, such as altered cortical excitability, altered endogenous sensory modulation, and imbalances in inhibitory and excitatory systems [ 5 ]. This body of literature examining altered pain processing in those with SMD compliments the current findings that self-reported MSS is related to type and number of chronic pain conditions. The current results further indicate this relationship is not limited to those with clear SMD. Our results suggest individual differences in MSS could serve as an additional marker to differentiate a range of pain subtypes. Compared to QST and psychological factors, MSS has drawn far less attention in the pain literature. While a few studies have investigated non-painful sensory sensitivity in select pain conditions, even less have considered overlapping pain conditions [ 39 ]. One study found that generalized sensory sensitivity was associated with the presence of COPCs in a pelvic pain cohort, a mixed-pain cohort, and healthy adults [ 51 ]. Our findings support and extend this by demonstrating that MSS varies across multiple common chronic pain types driven by different degrees of central sensitivity. We further show the magnitude of MSS scores increase across the ordered pain subgroups, with markedly higher odds of having 2 or more COPCs and/or FM in those with high MSAS scores, i.e., greater than 36/60 in men and 40/60 in women. We uniquely show that those with low MSS scores (i.e., < 24/60 in men, < 29/60 in women) had significantly lower risk of, or increased resiliency to, having multiple COPCs or FM. The significant differences in risk attributed to MSS occurred despite adjustment for a composite negative emotionality factor and propensity to conceptualize pain ratings differently across subgroups, i.e., pain-rating schema. This suggests that MSS is a conceptually distinct construct that is not simply reflecting bias in symptom report and/or negative affect. This was supported further by the notably weaker correlations between MSS and multiple negative emotionality factors than observed within those same emotionality factors. While adjusting for negative emotionality constructs is relatively common in pain studies, the adjustment for pain-rating schema is novel and provides additional evidence of a robust finding. Similarly, prior work using principle factor analysis revealed somatosensory amplification, a similar construct but representing fewer sensory domains, is distinct from catastrophizing or pain-related fear [ 30 , 31 ]. Indeed, somatosensory amplification displayed the strongest correlation with acute cold pain quality compared to other psychological factors [ 31 ]. Lastly, the Central Sensitivity Inventory (CSI) is a screening tool to identify patients with central sensitivity syndromes. Along with typical symptomology, the CSI includes 2 items assessing light and smell sensitivity [ 42 ]. Although not assessing MSS, this further supports the premise that non-painful sensory sensitivity is relevant to centrally mediated pain conditions. The idea of stratifying patients according to pain-related symptom profiles has been suggested to prescribe more efficient treatments that target the underlying mechanisms or to forecast treatment responders. For example, in multiple neuropathic pain patients, the Pain Detect Questionnaire was used to identify subgroups based on somatic sensation and pain quality, to improve targeted treatment [ 9 , 36 ]. Similarly, a review study divided FM patients into four subtypes based on patient characteristic profiles, suggesting a different treatment emphasis for each [ 46 ]. One of the subtypes identified was FM patients with “somatization,” suggesting that psychotherapy should be a primary intervention approach. It is not necessarily surprising that psychotherapy is cited as the first line of treatment for pain patients with amplified somatosensation, as some have regarded somatization as a psychological dysfunctional process [ 10 , 37 , 47 ]. Interestingly, the review also reported that patients were reluctant to undergo psychological intervention as they perceived their heightened somatic sensitivity complaints were more than psychological [ 46 ]. Our current findings may be more consistent with these patient perceptions than the authors’ original suggestion, if one assumes that somatization is similar to MSS. That is, MSS may indicate altered physiological CNS processing distinct from psychological distress. This in turn may further imply that alternative therapeutic approaches targeting MSS specifically, such as desensitization interventions commonly used with autism spectrum disorder [ 29 ], may be beneficial for this subset of chronic pain patients. Accordingly, future studies are needed to better identify optimal first-line treatments for those with chronic pain and elevated MSS. The concept of resilience to pain has largely been studied in the psychological field with several plausible resilience factors cited such as optimism, social support, secure attachment, trait resilience and active coping [ 22 , 55 ]. Whereas few physiologic factors have been identified to explain the phenomenon of resilience to pain. We propose, however, the protection or resilience to COPCs based on MSS is independent of optimism or positive affect. Indeed, we found the weakest association between MSS and positive affect of the psychological constructs evaluated (R = −0.19, Table 6 ), aligning with prior findings [ 31 ]. Further, the inclusion of positive affect as a covariate did not substantially improve model fits (model 5), thus was not ultimately used in the final analyses. Collectively we conclude that low levels of MSS may reflect physiological protection mechanisms involving CNS processing rather than simply a psychological-based resiliency factor. Longitudinal studies are warranted to determine whether MSS levels are a pre-existing vulnerability or protective factor for the development of central pain conditions including COPCs or whether elevated MSS can develop because of altered CNS processing. Interventions focusing on normalizing elevated sensory sensitivity have long been successfully practiced in treating autism spectrum disorders, where SMD is a primary concern [ 29 ]. It is less clear if these approaches will reduce pain in those with elevated MSS. However, emerging evidence suggests sensory interventions, such as light therapy, have promise. Nonspecific LBP patients reported lower pain and depression with white light exposure [ 32 ]. Similarly, green light improved pain and quality of life in fibromyalgia patients [ 40 ] and decreased headache frequency by 58% – 70% in migraine patients [ 41 ]. Several limitations need to be considered. First, this study primarily targeted patients with FM, migraine, or LBP and thus the evaluation of MSS may not generalize to all phenotypes of pain. Secondly, age and sex distribution differed across some groups. FM patients were above 53 years old on average and were mostly women; non-FM pain patients and pain-free individuals were in middle age; and LBP patients included more men. However, the age and sex differences may be inherently associated with the nature of the specific diseases. Further we previously found MSS did not significantly differ with age [ 58 ] and when evaluating only female participants, the MSS and pain relationships were largely consistent with the results including men. Despite the relatively small sample size in some subgroups, we were able to identify significant differences between groups partly due to the large effect sizes. MSS differentiates pain phenotypes thought to be characterized by increasing degrees of central sensitivity, indicating MSS may serve as an additional indirect marker of CNS sensory processing. Furthermore, our findings suggest MSS is distinct from psychological constructs, thus providing unique information beyond positive or negative emotionality. The current study highlights the potential application of the MSAS as a brief tool that can be used to characterize or possibly infer high and low risk for central contributions across pain conditions in clinical or research settings. Future studies are needed to better ascertain whether desensitization interventions may be effective regarding pain-related outcomes or when applied proactively could reduce the risk of developing persistent pain conditions.

Chronic pain continues to be challenging to treat; however, mechanism-based interventions may improve pain management [ 13 ]. As chronic pain can involve both peripheral and central mechanisms [ 52 ], efforts to discern the degree of central involvement have been frequently investigated. Markers such as quantitative sensory testing (QST) [ 61 ], psychological factors [ 34 ], and brain connectivity networks [ 48 ] have been used to characterize pain profiles and infer the role of centrally-mediated pain in humans in an effort to better identify underlying mechanisms. Recently, it has been suggested that heightened generalized or multisensory sensitivity (MSS) may be associated with a sensitized central nervous system (CNS) [ 8 , 16 , 60 ] and thus represent a vulnerability or risk-factor for centrally-mediated pain. A handful of studies have investigated MSS in one or more pain conditions with varying degrees of central involvement. For example, patients with fibromyalgia (FM), often cited as involving central mechanisms [ 1 , 21 , 43 , 48 , 52 , 60 ], report higher MSS than those with rheumatoid arthritis [ 60 ], complex regional pain syndrome [ 12 ], or osteoarthritis [ 16 ] using various measures of MSS. Similarly, migraine patients show a high incidence of sensory over-responsiveness in relation to pain attack aura [ 23 ]; however most studies involving migraine assess only unimodal sensory processing [ 24 , 38 ]. Further, several studies link pain with sensory over-responsiveness in those with clinical manifestations of sensory modulation disorder [ 4 – 8 ]. This emerging evidence supports elevated MSS in individuals with pain conditions thought to involve central mechanisms, such as FM and migraine [ 20 , 60 ], yet is not observed universally across all pain conditions [ 16 , 43 , 60 ]. It remains unclear, however, how well MSS differentiates between multiple pain conditions typically associated with varying levels of central mechanism involvement. While patients often identify with a primary pain diagnosis, many have additional co-morbid pain conditions. For instance, 45%–80% of fibromyalgia patients report migraine [ 20 , 57 ] and 49%–72% report low back pain (LBP) [ 19 , 26 ]. Chronic overlapping pain conditions (COPCs) are generally thought to involve more central mechanisms than isolated pain conditions [ 1 , 39 , 51 , 63 ], hence sometimes referred to as Central Sensitivity Syndromes [ 27 ]. So while studies show various forms of sensory hypersensitivity in patients with FM or migraine, [ 24 , 60 ]; few also evaluate for the presence of other pain comorbidities. Indeed, one report identified higher probabilities of having comorbid COPCs based on higher scores on the Generalized Sensory Sensitivity scale (GSS) [ 51 ]. Collectively, the growing evidence supports that normal variation in MSS could serve as an additional marker to help phenotype chronic pain patients but is constrained by the lack of clear comparisons between multiple conditions using a single instrument, and the lack of consideration of additional co-morbid pain conditions. Thus, to address these issues, our primary aim was to determine MSS differences between three chronic pain conditions (i.e., FM, migraine, and LBP) in isolation and combined, compared to individuals without chronic pain. Our secondary aim was to assess how well MSS predicts the likelihood of having multiple co-morbid and widespread pain conditions. This knowledge may help better understand the role of MSS as a marker for profiling chronic pain conditions and aid in inferring associated underlying mechanisms.