The study involved 16 women: eight had been diagnosed as having fibromyalgia on the basis of the American College of Rheumatology (ACR) criteria 1 (the FM group); the other eight were healthy sedentary controls matched for age, gender and physical activity 19 (MCG). Pain was assessed using the McGill Pain Questionnaire Short Form (MPQ-SF) 20 and the severity of FM symptoms by means of the FibroFatigue Scale (FFS) 21. (Table 1).

The possible non-homogeneity of the FM population has been reported since the early nineties 2223, but the subgroup distribution of the whole FM population has not yet been fully studied. Possible differentiations have been related to the onset of the disease: reactive traumatic (both physical and emotional) or primitive, but there are no substantial clinical differences between the two groups 24. This has also been also confirmed by more recent data from Riberto et al. 25. For these reasons the sample was considered clinically homogeneous and subjects were only excluded if they were simultaneously affected by any other rheumatological disease or high blood pressure (or receiving anti-hypertensive treatment such as clonidine); any infectious or viral diseases during the previous three months; metabolic or endocrine diseases; neuromuscular diseases or any other neurological or psychiatric disorder that may have hindered them from doing the required task; or if they were regularly receiving antidepressants, opiates or anti-epileptic drugs. All of the enrolled subjects underwent 48 hours' washout of analgesic drugs. Although paracetamol was allowed as a rescue drug, none of the subjects were under the influence of any drug during the s-EMG recording session.

The thickness of the subcutaneous tissue under the electrodes was estimated using a skin fold caliper (Gima, Gessate, Milan, Italy) in order to eliminate the volume conductor between the signal source (the muscle) and the recording device (the electrodes) as a confounding factor between the groups 26.

The study was approved by the "Foundation Salvatore Maugeri", Pavia Italy, Ethics Committee, and was conducted in accordance with the principles of the Declaration of Helsinki; all of the subjects gave their informed consent.

The myoelectrical signals were detected from the biceps brachii muscle of the dominant side, which was selected because: 1) it is a muscle, and not a site of any of the 18 myalgic points used to classify FM, and was not a site of spontaneous pain in our patients, and so the presence of incident pain could not be considered the local cause of an altered muscle contraction; 2) it is easily accessible for electrode placement; 3) it does not cause patient discomfort; 4) it allows the recording of high quality signals; and 5) published normative data are available 27.

After being picked up by a linear array of 16 electrodes (1 mm diameter silver contacts positioned 10 mm apart) in single differential configuration 28, the myoelectrical signals were passed through a 10–450 Hz bandwidth filter, amplified (EMG16-16 channel amplifier, LISiN Bioengineering Centre, Turin Polytechnic, Italy), sampled at 2048 Hz, digitised by a 12-bit A/D converter (DAQCARD-6024E National Instruments, Austin, Texas, USA), and stored on a personal computer.

Each subject lay supine on a bed with the dominant arm horizontal and abducted to 90°. The forearm was 120° flexed (180° being full extension), with the hand supinated. The arm was placed in an isometric brace (MISO1, LISiN Bioengineering Centre, Turin Polytechnic, Italy), which was equipped with two torque transducers (one on each side of the arm) and connected to a display that provided the subject with visual feedback of the torque level produced (Figure 1).

The subjects were asked to perform a brief (3–5 second) isometric flexion contraction that allowed the quality of the myoelectrical signals to match the criteria described in detail by Rainoldi et al. 29, and then performed three 3-second maximum voluntary contractions (MVCs) under isometric conditions, separated by 5-minute intervals. The contraction showing the highest force value was selected as the reference MVC, thus allowing sub-maximal targets to be set on the visual feedback display.

After the MVC had been assessed, the motor points of the biceps brachii were identified and marked on the skin, and the one with the greatest mechanical response and lowest current intensity was chosen for the electrically elicited contraction session.

After placing an adhesive stimulation electrode (area = 9 cm², Spes Medica, Vignate, Milan, Italy) on the selected motor point, a rectangular current pulse was applied using a time width of 0.3 ms and a frequency of 25 Hz; the stimulation was supramaximal (about 10% above the level generating the peak M-wave or the maximum level tolerated by the subjects). Two electrically elicited 30-second contractions separated by a 10-minute rest were performed at each experimental session. The 25 Hz stimulation was judged not to be unbearable or uncomfortable by all of the subjects in both groups.

After a further five minutes' rest, the subjects were asked to perform two voluntary contractions at 30% MVC and two at 60% MVC in a randomised sequence; the contractions lasted 30 seconds and were separated by 5-minute rest periods. This procedure is usually adopted in our lab to allow subjects to familiarise themselves with the protocol without inducing fatigue or a learning effect; only the data relating to the second contractionswere recorded and analysed.

The initial values and rates of change in mean spectral frequency (MNF), average rectified value (ARV) and conduction velocity (CV) were computed off-line by means of numerical algorithms 30 using non-overlapping signal epochs of 0.5 seconds, thus generating 60 estimates of each variable during the 30-second contractions.

The correlation coefficient (CC) between the two adjacent double differential (DD) signals was used to ensure correct electrode positioning and the reliability of the CV estimates.

Linear regression analysis was used to calculate the initial values and rates of change in MNF, ARV and CV during the voluntary and electrically elicited contractions. The normalised rate of change (the ratio between the rate of change and the initial value) was also calculated for all variables. As demonstrated elsewhere 27, this normalisation highlights differences in the myoelectrical manifestations of fatigue that may be related to different pools of activated motor units.

Motor unit action potentials (MUAPs) were extracted from the 13 DD channels only during the voluntary contraction 13. For each identified MUAP, the muscle fibre CV was estimated from the surface EMG signal using 3–7 adjacent DD channels in order to obtain a CC of > 0.75 and a CV > 8 m/s 31. The selected channels included those used to estimate the global EMG variables.

The MUAP CV distributions were calculated for each contraction using epochs of one second. In this way, the initial values and rates of change were obtained by pooling the behaviour of each MU, and not by using the whole signal providing the "global" variables.

In order to compare general behaviour in the MCG and FM group, the data were pooled at each contraction level.

The non-parametric Mann-Whitney U test for independent samples was used to identify any significant between-group differences in the s-EMG variables, and the non-parametric Kolmogorov-Smirnov test was used to compare CV distributions. A p value of < 0.05 was considered statistically significant.

Functional changes in FM muscles are often assessed by measuring strength and endurance, and the most frequent findings are reduced exercise endurance 32, reduced voluntary muscle force 67, and a perceived greater effort during fatigue 32. However, other studies have found no differences in maximal voluntary strength between FM and healthy subjects 1633, and their results are supported by our own MVC findings, which suggest that no strong and unequivocal muscle fibre modifications (in distribution, size or type) occur in FM (Table 2).

Using both voluntary and electrically elicited contractions makes it possible to identify the site of impairment along the neuromuscular chain: differences in the myoelectric manifestations of fatigue observed only during voluntary contractions can be related to an altered motor control strategy, whereas differences observed only during electrically elicited contractions can be considered as being mainly due to membrane properties 15.

As we found between-group differences in the EMG variables assessed on the basis of signals recorded during voluntary but not electrically elicited contractions, it is possible to localise functional impairment in FM at a higher level than that of the muscle membrane.

In line with other studies 1840414243, although adopting a different approach, we can conclude that the lack of myoelectrical fatigue and the associated muscle symptoms in FM syndrome are more likely to be generated by an alteration in central nervous system motor recruitment strategies than by peripheral mechanisms within the muscle itself 4344. The reduced EMG manifestations of fatigue and the type II fibre hypotrophy shown histologically in FM 104546474849. have also been found in chronic fatigue syndrome and interpreted as a sign of de-conditioning 50.

The CV rate of change and the normalised rate of change estimated from the extracted MUAP pool were in line with the MNF findings. As CV estimates provide a clear physiological description of the fatiguing mechanism, these results confirm the MNF findings.

Differences in the distribution and skewness of CV are usually explained as a consequence of different MU pool recruitment, with increases in average CV being due to an increase in faster MU recruitment 51. The mean CV distribution and skewness values in our FM patients were higher than those in the MCG during the 60% MVC, which leads us to conclude that FM patients recruit more of the faster MUs, and so an increase in the s-EMG signs of muscle fatigue can be expected. Similar observations were made by Gerdle et al. in their study of trapezius muscle 52: the CV values in the FM subjects were higher than those in the controls during low level contractions (no load, 1 kg, and 2 kg).

The dissociation between the curve of CV distribution and the myoelectrical manifestations of fatigue in our FM patients can be interpreted on the basis of an anomalous sensory-motor system pattern and the use of non-physiological strategies in managing functional motor tasks, as seen in chronic fatigue syndrome 53.

The CV distribution curves and estimates from each extracted MUAP offer two not necessarily matching descriptions because fast MUAPs can be recruited frequently (thus increasing the number of bins on the right side of the distribution: see Figure 2, upper row) but only for short periods without increasing the myoelectrical manifestations of fatigue (see Table 2).

The theoretical possibility of an altered order of recruitment/de-recruitment has recently been experimentally demonstrated by means of a technique based on EMG signals 54 and, given the findings described above, it can be speculated that patients affected by FM use compensatory motor strategies to recruit MUs in a different manner from that expected predicted by Henneman's principle. The wider CV distributions and fewer myoelectrical manifestations of fatigue observed in our FM group may therefore be due to the recruitment/de-recruitment of a partially or non-fatigued MU.

In an attempt to confirm this speculative explanation, the coefficient of variation (COV) of the force generated during isometric exercise at 60% MVC was calculated with the assumption that the hypothesised higher level of recruitment/de-recruitment in the FM group is related to a higher COV in mechanical output. The COVs in the two groups were not statistically different, but did show a trend in the hypothesised direction.

An altered recruitment/de-recruitment pattern can be considered a motor response to an alteration in the central integration of sensory inputs 5155, and the presence of central hyper-excitability has also been observed by others as a neurophysiological mechanism of muscle pain in FM 5405156 Our findings confirm the concomitant presence of central sensitisation and functional abnormalities in muscle contraction (i.e. the recruitment and de-recruitment of muscle fibres during a 60% MVC) in FM during maximal/sub-maximal load and static contraction observed by other authors 57.