OUP user menu

Discriminative Value of Tender Points in Fibromyalgia Syndrome

Nurettin Tastekin MD, Kaan Uzunca MD, Necdet Sut PhD, Murat Birtane MD, Oznur Berke Mercimek MD
DOI: http://dx.doi.org/10.1111/j.1526-4637.2009.00784.x 466-471 First published online: 1 March 2010


Objective. The aim of this study is to assess the discriminative value of all tender points, alone and in combination, that are designated as criteria for fibromyalgia diagnosis by the American College of Rheumatology (ACR), by investigating the appropriate pressure magnitude that should be applied during tenderness examination.

Design. Cross-sectional.

Patients. This study was performed on 66 patients with fibromyalgia diagnosed according to ACR classification criteria and 50 control subjects.

Setting. The outpatient rheumatology clinic of a PM&R department of a university hospital.

Intervention. Pressure pain threshold values were measured by a dolorimeter on nine specific point pairs in both groups and a cutoff value for discriminating positivity and negativity was calculated for each. Then the most valuable tender point pairs were assessed for discrimination of fibromyalgia syndrome using ACR criteria set as a reference standard.

Results. All tender points with determined pressure cutoff values were found out to significantly discriminate fibromyalgia syndrome and their area under curve values ranged from 0.779 to 0.934. Univariate logistic regression analysis revealed that lateral epicondyle and supraspinatus point pairs had the most powerful discriminative ability (odds ratio = 113.6 and 45.0, respectively). Multiple logistic regression analysis with backward stepwise method showed that lateral epicondyle and second rib point pairs were most discriminative with sensitivity and specificity rates of 87.9–94.0% and 77.3–84.0%, respectively.

Conclusions. Fibromyalgia syndrome can have potential to be recognized simply by pressing fewer tender point areas but with various pressure cutoff levels identified for each tender point areas.

  • Fibromyalgia
  • Chronic Pain
  • Tender Points
  • Dolorimeter
  • Pain Assessment and Methods


Fibromyalgia patients must have widespread pain in all four quadrants of their body for a minimum of 3 months and at least 11 of the 18 specific tender points, according to the classification criteria for fibromyalgia syndrome (FMS) published by the American College of Rheumatology (ACR) in 1990 [1]. Increased number of tender points (TPs) is one of the main determinants of FMS.

If the patient senses tenderness when the physician gradually applies digital pressure up to approximately 4 kg/m2 on one of the 18 specific points, then this point is counted as a TP. When the sum of these tender points exceeds 11, it is significant for classification according to ACR criteria especially in research settings. The ACR Committee study for FMS diagnosis [1] reported that the combination of widespread body pain and the determination of more than 11 TPs had sensitivity of 88.4% and specificity of 81.1% for FMS classification. The sensitivity value increased to 90.1% and specificity value decreased to 77.7% with an accuracy value of 84.2%, when TPs were examined alone.

It is well known that pain threshold values on specific pressure fibromyalgia points are lower in FMS patients than in healthy people [2,3]. On the other hand it has been reported in population-based studies that significant tenderness can be achieved on specific points also in healthy individuals [4,5]. Therefore, a more convenient evaluation is needed to discriminate tenderness between healthy individuals and patients with FMS.

ACR classification criteria forced a strict “at least 11 tender points” presence. Although the criteria are mainly recommended for research purposes, they have been used in some clinical settings as well to diagnose FMS for years. Some authors criticized this issue in recent years and asked what should be done for patients with very clear FMS symptoms but with 9 or 10 tender points examined [6–8]. In this regard, there seems to be a problem that some patients with classical symptoms may be excluded because they exhibit fewer than 11 TPs or pain threshold of more than 4 kg/cm2 on some specific points. This argument grew interest on methods and interpretation of TPs evaluation and consequently revealed some questions in practitioners' minds, whether it is a strict obligation to find out more than 11 TPs for FMS diagnosis and how will the patient exhibiting typical FMS symptoms with widespread pain lasting for more than 3 months, but having 8–10 TPs, be handled in clinical evaluation. As an 18-point evaluation is a time-consuming process during a busy office work, it would be very practical to make the correct FMS diagnosis with less TPs count. The anatomical localization and surrounding tissues of nine pairs of TPs show significant variability as some lay on a thick fat tissue, some on a tendon origo or on the middle portion of a muscle, and some could be on the skin surface just on a bone tissue. So a standard pressure magnitude of 4 kg/cm2 might not be logically capable enough to reveal tenderness in all these anatomical areas showing major anatomical differences. Hypothetically, the structural and anatomical variations of the points might make some specific pairs of tender points more valuable for FMS diagnosis. If this hypothesis comes true, then the diagnosis will be achieved with less TP with the same sensitivity, specificity, and accuracy as much as ACR criteria. As the studies on TP evaluation is lacking and pressure pain thresholds show marked differences among different TP areas [6,9–11], studies that questioned the place of TP evaluation in FMS diagnosis are needed.

The aim of this study is to assess the discriminative value of each of TPs one by one or in combination, by investigating the appropriate pressure magnitude that should be applied during tenderness examination.

Materials and Methods

Study Participants

Seventy-two female patients with FMS classified according to ACR criteria [1] with digital TP evaluation in our outpatient rheumatology department were invited to the study. Patients with lumbar discopathy [2], with cervical discopathy [2], with cardiac disease [1], and with inflammatory joint disease [1] were excluded from the study. Thus, 66 patients with no concomitant musculoskeletal disorder were evaluated. None of the patients used any analgesic drugs nor did they exercise within the previous 24 hours in any way that could interfere with the tender point evaluation. Control subjects were healthy individuals between 35–60 years of age similar to the age interval of the patients, with standard daily living activities selected from hospital staff and patient relatives. The exclusion criteria for controls included spinal deformity, other systemic diseases causing chronic pain, and use of medications for any musculoskeletal, neurological, or cardiovascular diseases. The study participants were informed about the study process and they all gave informed consent.

Tender Point Areas and Digital Evaluation According to ACR Criteria

An investigator with 10 years experience performed digital palpation, using a digital TP counting method according to standard ACR definition [1], to identify the patient group by ACR criteria, and also performed the same procedure in control subjects. Nine symmetric TP areas known to be specific for FMS were processed. Digital counting was performed by pressing the tender areas with approximately 4 kg/cm2 pressure, estimated by when the pulp of the thumb whitened. The nine TP areas defined by ACR were as follows: 1) Occiput, at the suboccipital cervical muscle insertions; 2) low cervical, at the anterior aspects of the intertransverse spaces at C5–C7; 3) trapezius, at the midpoint of the upper border; 4) supraspinatus, at origins, above the scapula spine near the medial border; 5) second rib, at the second costocondral junctions, just lateral to the junctions on upper surfaces; 6) lateral epicondyle, 2 cm distal to the epicondyles; 7) gluteal, in upper outer quadrants of buttocks in anterior fold of muscle; 8) greater trochanter, posterior to the trochanteric prominence; and 9) knee, at the medial fat pad proximal to the joint line[1].

Dolorimetric Tender Point Counting Method

One hour after digital palpation, two dolorimetric evaluations were performed with a 10-minute interval by second and third experienced investigators who were blinded to the digital examination in both groups, to establish pressure pain threshold for TPs. The pressure pain threshold is defined as the minimum force applied that induces pain. Mechanical pressure, determined as kilogram (kg) per 1 cm2 skin region on the sensitive areas was applied at a 90° vertical angle by a Fisher's hand dolorimeter (Pain Diagnostic & Treatment [PDT, Italy]). Previous studies have shown pressure threshold measures obtained using algometers with a 1-cm2 contact area to have acceptable inter-rater and intra-rater reliability of pressure scores over time [12]. The examiners increased the pressure at a rate of approximately 1 kg/cm2 per second. Subjects were instructed to say “yes” when the sensation of pressure changed to pain, and then the pain pressure threshold was recorded [1]. Two threshold measurements with an interval of 10 minutes were performed at each area and the arithmetical mean of two values were taken as the final threshold value. This method has demonstrated reliability [13]. After finding the mean pressure values of each of the specified 18 points, the arithmetic pain threshold mean of the symmetrical points were taken; thus, nine values, one for each pair of specific points, were calculated.

TPs count determined by digital palpation and dolorimeter was performed also in the control groups. If the pain threshold remained under 4 kg/cm2 by dolorimeter on a specific point, then this point was identified as tender.

Statistical Analysis

The age and body mass index (BMI) differences between groups were compared using the Student's t-test for normal distributed data. Mean pain threshold values between groups were compared by using Mann–Whitney U tests. In FMS group, hierarchical cluster analysis was used to gather dolorimetric measurements into meaningful clusters. Squared Euclidian distances served as the similarity measure and Ward's method was used to cluster the participants. The best cluster solution was selected using a dendrogram plot. The independent effect of the dolorimetric measurements on each of the points to discriminate the presence of FMS was assessed using univariate logistic regression analysis, and overall effect was assessed using multivariate backward stepwise logistic regression analysis. Diagnostic power of the dolorimetric measurements was assessed using the receiver operating characteristics (ROC) curve analysis and optimal thresholds for each measurement were determined. Area under the curve (AUC) of the ROC curve was used to assess the predictive power of measurements. A plot of true positive rate against false positive rate was made and the AUC was measured. The sensitivity and specificity rates of the scores for FMS were estimated by cutoff points. The AUC is a measure of the overall discriminatory power of the prognostic variable. A value of 1.0 indicates perfect discrimination, a value of 0.5 indicates random prediction, and a value of lower than 0.5 indicates no discriminative power. A P value of <0.05 was considered statistically significant.


No significant difference (P > 0.05) was found in age between the 66 patients with FMS (43.9 ± 6.5) and 50 controls (41.5 ± 6.9), nor in BMI values for FMS (29.01 ± 5.48) and controls (27.30 ± 5.40).

Mean number of TPs counted by digital palpation was found to be 15.68 ± 2.58 in FMS, and 4.48 ± 3.54 in controls. The dolorimetric evaluation of TPs revealed significantly lower pain thresholds in every examined point, when compared with those of control subjects (P < 0.001). The pain threshold values of specific points are shown in Table 1.

View this table:
Table 1

Pain threshold pressure values by dolorimetry in patients with fibromyalgia and control subjects

Tender PointsRightLeft
FMS (n = 66)Control (n = 50)FMS (n = 66)Control (n = 50)
Occiput3.3 ± 1.24.5 ± 1.63.3 ± 1.24.8 ± 1.8
Low cervical3.2 ± 1.05.3 ± 1.93.1 ± 1.05.2 ± 1.7
Trapezius4.1 ± 1.37.0 ± 2.14.2 ± 1.57.1 ± 2.0
Supraspinatus4.2 ± 1.57.4 ± 1.94.3 ± 1.58.0 ± 1.8
Second rib3.1 ± 0.95.4 ± 2.13.1 ± 0.95.5 ± 2.2
Lateral epicondyle3.7 ± 1.67.0 ± 2.13.8 ± 1.77.2 ± 1.9
Gluteal5.1 ± 1.88.3 ± 1.95.3 ± 1.88.3 ± 2.0
Greater Trochanter5.1 ± 2.38.2 ± 1.95.2 ± 2.17.8 ± 2.1
Medial knee3.8 ± 1.55.9 ± 2.13.8 ± 1.56.4 ± 2.2
  • All of the data shown are mean (±) SD.

  • All values are mean pressure levels that reveal pain in all examined tender points and the difference between patients and conrol subjects were all significant (P < 0.001).

  • FMS = fibromyalgia syndrome.

To define an optimal decision threshold of several total scores, ROC curve analysis was performed. Cutoff values, predictive accuracies, and AUCs are shown in Table 2. The diagnostic values of each point when the cutoff value was taken 4 kg/cm2 as expressed in ACR, criteria are also shown in Table 2. All of the parameters were significantly discriminated FMS from controls, and their AUCs ranged from 0.779 to 0.934. Among them, lateral epicondyle and supraspinatus were the two most powerful discriminating sites to determine FMS (+) case status. Optimal cutoff point for lateral epicondyle measurement was less than or equal to 4.8. At this cutoff point, the sensitivity for correct prediction of FMS diagnosis was 87.9%, and the specificity rate was 94.0%.

View this table:
Table 2

Calculated AUC values of point pairs and sensitivity-specificity rates for both the study cutoff points and cutoff points recommended by the ACR

AUCStd. Error of AUCPStudy Cutoff pointsSensitivity (%)Specificity (%)Cutoff points ACRSensitivity (%)Specificity (%)
Occiput0.7790.042<0.001≤3.875.870.0≤4.078.8 64.0
Low cervical0.8610.034<0.001≤4.593.964.0≤4.081.8 70.0
Trapezius0.9020.029<0.001≤5.4586.482.0≤4.051.5 96.0
Supraspinatus0.9310.022<0.001≤5.5584.890.0≤4.053.0 98.0
Second rib0.8780.032<0.001≤3.877.384.0≤4.086.4 70.0
Lat. epicondyle0.9340.024<0.001≤4.887.994.0≤4.071.2 98.0
Gluteal0.8880.031<0.001≤7.487.978.0≤4.024.2 98.0
Gr. Trochanter0.8430.036<0.001≤7.081.872.0≤4.034.8100.0
Medial knee0.8630.033<0.001≤4.578.882.0≤4.063.6 84.0
  • ACR = American College of Rheumatology; AUC = area under the curve.

Results of the univariate logistic regression models for predicting FMS are shown as forest plot in Figure 1. Each variable significantly (P < 0.001) affected the outcome. When we investigated dolorimetric measurements, lateral epicondyle and supraspinatus were the two most significant variables on outcome. The FMS risk in lateral epicondyle (+) and supraspinatus (+) patients were higher than lateral epicondyle (−) and supraspinatus (−) subjects (odds ratio [OR] = 113.6 and 45.0-fold, respectively).

Figure 1

Odds ratios (95% confidence ıntervals) per measurement relative to the reference category of cutoff points. (1: Occiput [≤3.8], 2: Low cervical [≤4.5], 3: Trapezius [≤5.45], 4: Supraspinatus [≤5.55], 5: Second rib [≤3.8], 6: Lateral epicondyle [≤4.8], 7: Gluteal [≤7.4], 8: Greater trochanter [≤7.0], 9: Medial knee [≤4.5].)

Hierarchical cluster analysis was performed on the tender points to examine whether they could be grouped in meaningful clusters. Dendrogram plot was drawn in Figure 2. As shown in the figure, two meaningful clusters were obtained.

Figure 2

Dendrogram plot of dolorimetric measurements. (1: Occiput, 2: Low cervical, 3: Trapezius, 4: Supraspinatus, 5: Second rib, 6: Lateral epicondyle, 7: Gluteal, 8: Greater trochanter, 9: Medial knee.) Points 3, 4, 6, 7, and 8 were clustered at point of lateral epicondyle and named as cluster 1. Points 1, 2, 5, and 9 were clustered at point of second rib and named as cluster 2.

Results of the multiple logistic regressions for predicting FMS with backward stepwise method are shown in Table 3. Of the nine variables (occiput, low cervical, etc.) that were entered into the logistic regression model, two variables (second rib and lateral epicondyle) were included to the model. Other variables were excluded from the model due to the nonsignificant effect. Univariate logistic regression analysis revealed that supraspinatus had the second higher OR value (Figure 1); however, it was clustered with lateral epicondyle due to interactions (Figure 2). Therefore, supraspinatus could not pass into the last step of the multivariate logistic regression model. Using these two points (second rib and lateral epicondyle), classification accuracy was found to be equivalent to the use of nine points. Overall accuracy of the model was 90.5%, with sensitivity and specificity rates of 87.9% and 94.0%, respectively.

View this table:
Table 3

Results of the multiple logistic regression analysis with backward stepwise method according to the optimal cutoff points for predicting fibromyalgia syndrome

POR (95% CI)
Second rib 0.006 6.5 (1.7–24.6)
Lateral epicondyle<0.00164.7 (15.5–270.3)
  • Reference categories: second rib ≤3.8, lateral epicondyle ≤4.8.


In clinical practice, reliable application of a standard pressure of 4 kg/cm2 on tender soft tissue areas with different underlying anatomic properties requires considerable experience, generally raising concerns about the validity of fibromyalgia diagnosis. Moreover, the obligation of requiring positive tender points in 11 out of 18 areas to make the ACR a criteria-based FMS diagnosis is also debated [6,7,13,14].

Our study investigated whether we could reliably diagnose fibromyalgia with less than 11 tender points and also identify more discriminating tender points. While assessing the diagnostic values of the points, we tried to determine the magnitude of pressure the clinician must apply on each point for better discrimination of tender and nontender points.

Prior studies revealed significantly lower pressure pain threshold levels in patients with FMS when compared with those of healthy subjects [3,6]. Our study confirmed the results of these studies and we found lower pressure pain threshold levels in FMS patients than controls on every nine pairs of TPs as expected. This finding emphasizes the validity of the selected 18 points on the body for FMS diagnosis, as no tender point area had higher pain threshold levels in FMS patients than controls.

After detailed analysis, we observed that every point had a different cutoff value for discrimination of positivity and negativity. This finding tested our study hypothesis that questioned the application of a standard pressure on all tender point areas in FMS diagnosis. We found that every point reacted to different pressure levels, most likely due to differences in underlying fat tissue thickness, soft tissue surfaces, and closeness to neighboring bone muscle and nerve tissues. Marques et al. [9], in their study that assessed pain threshold levels and life quality in FMS patients, showed that the second rib, low cervical, and lateral epicondyle points had higher sensitivity when compared with the others; whereas the gluteal area and trochanter major points were observed to have lower sensitivity. McVeigh et al. [10] confirmed the Marque et al. findings and reported high pressure pain thresholds on these two points. Our findings are similar in that we observed cutoff points of pressure lower than 4 kg/cm2 on occipital and second rib points. Low cervical, lateral epicondyle, and medial knee points had cutoff points much closer to 4 kg/cm2. However, these values were higher than 5.5 kg/cm2 on trapezius and supraspinatus points and very high up to 7 kg/cm2 on the gluteal and greater trochanter points. Generally, tender points with thicker underlying soft tissue had higher pain threshold values (e.g., low sensitivity to pressure). These results raise the question as to which points are more valuable for diagnosis; those with lower thresholds or those with higher thresholds. Our analysis results suggest that gluteal and greater trochanter points with thicker underlying soft tissue had lower sensitivity values for FMS diagnosis, while superficial points like occiput and second rib were more sensitive to the diagnosis. However, these two deeper points had extremely high specificity for FMS diagnosis. It means that all FMS patients need not have gluteal and trochanter positivity; however, if these points are positive at 4 kg/cm2 pressure, patients are very likely to have FMS.

We observed different diagnostic value levels among various tender points when we compared specific cutoff points with standard cutoff points of 4 kg/cm2 proposed by ACR. In this situation, we performed a multiple regression analyses and tried to understand which points or point combinations were more valuable for diagnosis FMS and whether fewer than 11 tender points would be as successful as the full 11 points. This analysis revealed that the combination of positive lateral epicondyle and second rib tenderness was more valuable using study cutoff pressure levels.

Katz et al. [7] suggested that patients whose number of tender points was 5–11 were frequently classified as having FMS and 11 tender points was not the optimum cutoff point in clinical practice. They concluded that ≥6 positivity would efficiently discriminate diagnosis and nondiagnosis. Our findings seem to be more optimistic than Katz et al. [7], by demonstrating that the positivity of only two point pairs (lateral epicondyle and second rib) reliably discriminates FMS diagnosis. This suggests that a more efficient office procedure may adequately identify patients with FMS who will respond to treatment. However, the application of our findings to clinical practice remains to be tested. For example, will patients with FMS diagnosed by the more efficient method respond similarly to established clinical treatments?


A major aim of studies evaluating TPs in FMS is to enhance the quality of the diagnostic procedure. This study indicates the importance of specific tender points and that it is not appropriate to apply a standard magnitude pressure to all TPs because differences between the pressure pain threshold values of TPs have been observed using dolorimeters. This study suggests that FMS diagnosis can be achieved by pressing fewer TP areas using various pressures cutoff levels identified for each tender point areas. Whether a more efficient procedure for clinical practice, as suggested here, will effectively identify FMS patients who respond to appropriate treatment, remains to be tested.


  • No grants or support provided for the study. No financial relationship with any company whose products may be related to the topic of this manuscript for all authors.


View Abstract