Revised chronic widespread pain criteria: development from and integration with fibromyalgia criteria

1 Introduction

The idea of chronic widespread pain (CWP) had its beginning in the American College of Rheumatology (ACR) 1990 criteria for fibromyalgia [1]. The criteria authors sought a pain variable that would identify potential fibromyalgia patients who would then undergo follow-up tender point examinations. They identified what they called “widespread pain,” defined as pain above and below the waist, on both sides of the body and in the axial region, and with no exclusions as to the source of the pain (WP1990). WP1990 was an ad hoc criterion, developed for the sole purpose of aiding in the diagnosis of fibromyalgia and not for classifying persons with respect to a disease process or as a separate, freestanding disorder.

In the years that followed, however, CWP as a diagnosis was ubiquitous. By 2015 more than 1,500 citations to CWP were identified [2], and as a freestanding entity it was the subject of many comprehensive reviews [3], [4], [5]. CWP is scheduled to be added as a diagnostic category in ICD-11 [6], [7], [8]. In a comprehensive 2016 review of 100 studies, more than 80% used the 1990 (WP1990) criteria or some modification of it, and ICD-11 considers fibromyalgia to be a form of CWP [7]. The ACR definition (WP1990) is somewhat ambiguous and open to different interpretations [2]. It can be satisfied by pain in as few as three sites and can identify as CWP individuals with limited body pain as well as those with extensive pain. Deficiencies in the WP1990 definition led to alternative definitions [9], [10]. In 2016, in a modification of fibromyalgia criteria, CWP was redefined [10]. This new definition (WP2016) required pain in at least four of five pain regions, but, like WP1990, did not address the fundamental question of how many pain sites were required. That is, it addressed geographic spread but not the extensiveness of pain. WP2016, like WP1990, was defined to function as a criterion for fibromyalgia rather than as a definition that could be used outside of the context of fibromyalgia criteria.

In this report we studied CWP assessments in population-based, community-based and specialty clinic settings to analyze and characterize CWP definitions used with fibromyalgia criteria as well as freestanding categories as to symmetry and geographic spread, pain density or diffuseness, and associations with clinical severity measures. To evaluate clinical severity of CWP definitions, we measured clinical severity characteristics of patients satisfying each set of CWP criteria and used WP1990 criteria as a goal post of least clinical severity and 2016 fibromyalgia criteria as the measure of greatest clinical severity. To characterize symmetry and geographic spread, we evaluated a matrix of widespread pain index (WPI) pain sites and WP2016 pain regions. Based on our findings related to symmetry and geographic spread, pain density and clinical severity, we propose modifications of the CWP definition to provide a simple, unambiguous definition that is suitable for clinical and research use.

2 Methods

3 Results

Table 1 describes the 15 pain sites in four settings before and after adding the WP2016 criterion. There is considerable pain site involvement in all settings, with the extent of pain involvement increasing (columns 2–5) from the limited pain in the German general population (GP), through the primary care setting (PC), the NDB survey of rheumatoid arthritis patients (RA), and the NDB survey of a group with fibromyalgia, osteoarthritis and other non-inflammatory disorders (NonInfl). The most common pain site in all groups was low back, occurring in 25.3%, 41.9%, 58.3% and 76.1% in GP, PC, RA and NonInfl, respectively. The sites with the greatest pain scores were the axial region, shoulders and hips, in a consistent pattern across all study groups. This pattern of increasing severity is reflected in mean WPI scores (1.3, 2.4, 5.8 and 7.7) and in the proportions meeting the WP2016 criterion (5.0%, 10.0%, 38.0% and 51.2%), the proportion with fibromyalgia (FM 2016) (1.4%, 5.5%, 24.6% and 32.0%), as well as in mean PSD scores (3.0, 6.1, 10.7 and 12.4). When the WP2016 criterion is applied (columns 6–9), all groups demonstrate substantially increased levels of pain-related variables (WPI, SSS and PSD).

In Table 2 we examined 4 definitions of CWP using the combined NDB population. The WP2016 criterion requires 4–5 pain regions and by definition has a minimum WPI of 4 (light and darker shaded boxes in the 4–5 grouping). The WP2019 criterion requires 4–5 pain regions and a minimum WPI of 7 (lightly shaded boxes in 4–5 grouping). The WP≥7 criterion (lightly shaded boxes in regions 2–5) includes all patients with WPI≥7, including those with fewer than 4 regions, and thus is the only definition of CWP without at least 4 quadrant pain. The WP1990 criterion is not represented in this table because it can count some pain sites twice. In general WP1990 can be thought of as being similar to WP2016, but with a minimum WPI and region count of 3. Thus, the WP≥7 differs from the WP2019 by allowing fewer than 4 regions and the WP2016 differs by allowing a WPI<7.

Table 2:

Range of widespread pain index (WPI) scores according to painful region classifications in the combined NDB population (n=35 239).

1. WP2016: Shaded boxes in regions 4–5 from WPI 4–19. WP2019: lightly shaded boxes in regions 4–5 from WPI 7–19. WP≥7: shaded boxes in WPI 7–19.

The 4 CWP definitions are compared in Table 3.

From the easiest to the most difficult to satisfy CWP definition, the percent satisfying the definition is WP1990 (51.3%), WP2016 (41.7%), WPI≥7 (37.6%) and WP2019 (33.9%). For those CWP positive for the 4 groups, the percent satisfying FM 2016 criteria are 51.1%, 63.6%, 69.0% and 76.6%, respectively. Kappas for the agreement of CWP methods and FM 2016 criteria are 0.490, 0.681, 0.737 and 0.801. We also compared clinical characteristic of the 4 definitions with each other and with FM 2106 positive patients. With each stricter CWP definition, clinical scores worsen. A dramatic example is the percent disabled in the NDB which increases from 15.9% in WP1990 to 28.1% in WP2019 and to 32% in FM 2016.

Table 4 shows that WP2019 positive and negative subjects in the NDB had very large mean differences in clinical variables.

To get further insight into these differences we graphed the distribution of clinical variable scores according to WP2016 status for 4 variables. As shown in Fig. 1, although the mean scores and distributions of the variables differed considerably between WP2016 and non-WP2016 groups in Table 4, there was considerable overlap in variable scores. For example, subjects with fatigue scores of 3 and 8 can be found in both groups and subjects with limited as well as substantial functional disability (HAQ) can be found in both groups. Thus, membership in WP2016 does not guarantee only severely affected patients for important clinical variables.

Fig. 1:

Graphs of scores for HAQ functional disability, PHQ-15 categories, symptom severity score (SSS) and VAS fatigue according to WP2019 (−) and WP2019 (+) status in the combined NDB datasets. Values on x-axis are percent values for bars and sum to 100%.

4 Discussion

The data of this study show that different definitions of CWP identify different sets of patients, though with overlapping membership. By far, the old WP1990 definition is easiest to satisfy, and 51.5% of NDB patients met that definition. At the other end of the spectrum WP2019 classified just 33.9% of patients as CWP. Comparing these 2 definitions (WP1990 and WP2019) by kappa statistics with FM 2016, kappas of 0.490 and 0.801 were noted. That is, WP2019 is in substantial to almost perfect agreement with FM 2016 while WP1990 reaches the moderate agreement level [21]. Moreover, as shown in Table 3, WP2019 contains patients with the most abnormal clinical scores compared with all other definitions. And with population data the WP1990 definition identified 2.8 times more subjects than WP2019 (8.8% vs. 3.1%). In addition to severity differences, Table 2 shows that WPI≥7 identifies persons with 3 or fewer pain sites, as does WP1990. We believe it may be time to retire the WP1990 definition and replace it with WP2019, as we discuss below.

WP1990 was never intended to define CWP in the way it is now being used. It was simply an ad hoc criterion to aid with fibromyalgia diagnoses in epidemiologic studies [1], [2]. Recently, Dean suggested that “widespread pain” is not different from “multi-site pain,” and that counting the number of sites could provide an equivalent measure of widespread pain [22]. The Dean study used a comprehensive pain manikin of 35 pain sites and selected the cut point of ≥8 as the multisite equivalent of CWP based on comparisons with the ACR 1990 CWP criteria – the least effective of our definitions. Subsequently, Arnold and coworkers published criteria for fibromyalgia [23] that utilized the multisite method of Dean et al. [22]. They proposed a definition of multisite pain that required 6 of 9 body areas with pain that was stated to yield a prevalence of CWP consistent with WP1990. The number of pain sites is strongly correlated with somatic and psychological symptom burden in the general population [22], [24], [25], [26]. In addition to the WPI, the 2011–2016 fibromyalgia criteria provide a polysymptomatic distress scale that is even more strongly correlated with outcomes than a count of painful sites [13].

In the current study we created WPI≥7, a measure that is simply a count of 7 sites without respect to the number of regions included. In general, we found that with this definition 10.0% of pain sites were in regions 2–3 (Table 2), 37.6% of patients satisfied the WPI≥7 criteria, and its kappa with FM 2016 was 0.737. Thus, it performed similarly to our recommended WP2019 criteria, differing primarily in diffuseness, allowing 10.0% of patients with 3 or few regions of pain to be diagnosed with CWP. In making our recommendation for WP2019 over WPI≥7 we opted for a definition that did not allow persons with (mostly) non-symmetrical and limited regions of pain. WP2019, for example, will not be positive in patients with regional pain syndromes, as multisite pain can be [27], [28]. Importantly, WP2019 represents the logical intersection of WPI≥7 and WP2016, preventing selection of subjects with fewer than 4 regions (WP2016) and fewer than 7 pain sites (WPI≥7).

What exactly is CWP? As originally intended, it was bounded by the minimum number of symmetrical body pain sites required to diagnose fibromyalgia. When sufficient symptoms (or tender points in 1990) were added, fibromyalgia could be diagnosed. The ACR 2010 and the 2011 fibromyalgia criteria eliminated the CWP criterion and relied for the most part on the “WPI≥7” criterion to effectively act like the CWP criterion. In 2016, with WP2016, the widespread pain criterion was reinstalled to prevent persons with low (2–3) region counts from being diagnosed as fibromyalgia. The multisite definition of CWP [22], the WPI≥7 and the WP2019 are all similar. Standardization of the multisite definition to readily available WPI≥7 or WP2019 data would advance the science of CWP and move toward a common definition, as it generally recognized that the WP1990 definition has substantial deficiencies.

The definition of CWP that we have used in this paper, and that has been used in hundreds of other studies [2], refers only to pain. Recently, new definitions related to pain have been proposed and have become part of ICD-11. In these definitions, CWP is considered to also be a disease [8]. According to ICD-11, CWP (MG30.01) is a form of chronic primary pain (MG30.0) and is present when A. Chronic pain (persistent or recurrent for longer than 3 months) is present in at least three body quadrants plus the axial skeleton (4 of 5 regions) AND B: the pain is associated with at least one of the following: B.1 Emotional distress due to pain is present or B.2: The pain interferes with daily life activities and social participation; AND C: the pain is not better accounted for by another chronic pain condition. This definition appears to define a CWP as a disease or disorder [8].

Our proposed criteria, as well as the standard WP1990 criteria, does not consider items B and C of the ICD-11 definition. Rather, it addresses part A. However, Tables 3 and 4 provide partial measures of B.1 and B.2 with respect to the 4 criteria evaluated in this study as well as to FM 2016. Changes in PSD and disability across the criteria (Table 3) suggest that B.1 and B.2 are more likely to be satisfied with WP2019 then WP1990. The ICD-11 widespread pain disorder definition also requires “C.” Our definitions deliberately do not consider “C,” but can be further subset to just identify “primary” pain, or they can be used in the face of mixed primary and secondary pain after further study. ICD-11 defines CWP as “diffuse pain.” Considering ICD-11, our paper can be considered to show how best to identify diffuse pain. Within the context 4 quadrant pain, WP2016 and WP2019 are the best ways to do this (Tables 2–4).

Neither FM nor CWP are manifestly distinct entities. Instead their boundaries are man-made. Fibromyalgia with its criterion of WPI≥7 could as easily have been WPI≥8. CWP as defined by White [9] and by WP1990, WPI≥7, WP2016 and WP2019 (Table 3) all identify different patients. Practically, if fibromyalgia is to be diagnosed for clinical and research purposes it would seem to make sense to use a common, agreed upon definition, such as FM 2016. WP2019 effectively is the CWP run-up to FM 2016. The two diagnoses could provide substantial standardization for fibromyalgia and CWP. We recommend the use of instruments that can be easily administered and which assess CWP at a glance, such as the WPI or a pain diagram or the Michigan body map [29], and the use of the PSD scale provides a clear estimation of disease severity without the problems of uncertain dichotomization.

A major limitation to our study is that it requires use of the ACR 2010/2011 fibromyalgia criteria pain site map and WPI variable. It seems clear that somewhat different maps and variables could provide similar results. The advantage of the methods assessed in this paper is that they work easily with current definitions and variables for diagnosis and assessment of fibromyalgia and that it is possible to use more than one definition of CWP once a pain assessment is complete. This could allow for different definitions in different settings and support future research into the best definition. Finally, use of the WPI allows for classifications similar to WPI≥7 or multisite approach [23].

How should we choose a WSP definition? Symmetry and geographic spread, diffuseness of pain and clinical severity are important. If, as in agreement with ICD-11 definitions [7], CWP should be “diffuse,” then WP2019 and WPI≥7 best fit that designation, as they identify the most pain sites (Table 2). Additionally, as shown in Table 2 and the results section, the WP1990 criterion and WPI≥7 can admit patients with asymmetric pain and low region scores (2–3), and WP1990 can include those with as few as 3 pain sites. For such reasons, we would recommend discontinuing WP1990 use. WP2016 and WP2019 are similar, and WP2019 is actually WP2016 with the requirement of a WPI score of at least 7. WP2019 has the advantage that it identifies most patients with fibromyalgia. The overall agreement with FM 2016 was 91.5%, kappa 0.801, and thus it could act as a reasonable surrogate for fibromyalgia, allowing estimation of fibromyalgia without the necessity of collecting the symptoms variables of the somatic symptom scale. Our choice would be to recommend the WP2019 definition of CWP.

In summary, data from our analyses suggest that the ACR 1990 CWP is not an effective tool to identify CWP and should be replaced. CWP (WP2019) can be defined by 4-region pain and a total pain site score ≥7. This definition provides a simple, unambiguous measure that is suitable for clinical and research use as a standalone diagnosis that is integrated with fibromyalgia 2016 definitions.