Algometry with a clothes peg compared to an electronic pressure algometer: a randomized cross-sectional study in pain patients

The study protocol was approved by the local ethics committee.

At the Bern University Hospital 170 consecutive in-patients suffering from pain were asked to participate, 157 of them consented to participate in the algometric study.

Our aim was to explore the two test methods in a wide range of pain types. Therefore, we recruited patients in different clinical departments, i.e. the orthopaedic department (orthopaedic pain group) and the medical-psychosomatic department (medical-psychosomatic pain group). Eligibility criterion was an acute or chronic pain condition as the primary clinical problem. Patients with pain disorders of neuropathic origin (e.g. peripheral neuropathy or radiation syndromes) were not included because of expected inhomogeneous peripheral pain sensitivity condition. We also excluded patients with craniocerebral trauma and those with an inflamed or traumatised algometric measuring site (e.g. infected ear lobe, broken hand).

In order to determine the correlation between the algometric methods, the tests were performed using a randomized cross-sectional design. Moreover, both tests were carried out during the same session. The order of the tests was individually determined by flipping a coin.

To compare test-retest reliability (repeatability) of the two methods, clothes peg test and pressure algometry were repeated in a consecutive subgroup of chronic patients of the medical-psychosomatic department with pain duration of more than 6 months. Patients needed to stay in the hospital for at least 7 days to qualify for repeatability analysis. The retest assessments were performed exactly the same way as the previous test session, i.e. at the same time of the day, under identical conditions, and by the same tester.

To assess the psychological status of the patients, they were asked to fill in a self-rating test each time they underwent pain testing. We applied the validated German Version of the depression subscale of the Hospital Anxiety and Depression Scale (HADS-D), which is designed to rate symptoms and levels of depression in non-psychiatric populations [8]. Total HADS-D scores (range 0-21) were interpreted either as clinically relevant or clinically not relevant according to a cut-off-value of ≥ 9 for clinically relevant depression [9].

For our pain provocation test we used polypropylene clothes pegs (type MaxiMedium 2083, size 78 × 10 mm, imported 2004 by C.C. Hansen, Denmark, produced in Hangzhou, China). For algometric procedures we selected clothes pegs with a clamping force of exactly 10 Newton at an extension of 5 mm (Figure 1). Pilot testing showed stable measurement readings between 0 and +35 degrees centigrade. Repeated use (more than 2,000 consecutive tests) did not result in any decrease in clamping force. As the manufacturer of MaxiMedium in the meantime (2010) uses a thinner wire for this clothes peg type, we had to regalvanize the springs of the new series with nickel to obtain the original clamping force of 10 Newton (Algopeg, galvanized by Rolf Helbling, Biel, Switzerland).

A separate sample of clothes pegs with lower or higher clamping force (4, 6, 14 Newton, respectively) was selected for training purposes (see below).

For testing PPdt and PPtt, we used an established electronic pressure algometer with a probe of 1.0 cm² (Somedic Type II, size 161 × 170 × 30 mm, Somedic Production AB, Hörby, Sweden). The instrument was calibrated following the standard protocol as recommended by the manufacturer: A 1-kg weight provided by the manufacturer was applied to the 1.0 cm² diameter applicator head attached to the end of the nozzle. The acceptable calibration values ranged from 99 to 101 kPa.

Before algometry was started, the tester explained the procedure to the patients. The experiment was performed in a quiet setting without any interruptions and shielded from other patients. Each of the two tests was once performed for training purposes as a trial run. To avoid any local pain sensitization, the trial run tests were performed on the index finger, the subsequent actual testing was performed on the middle fingers. Additionally, the trial run was deliberately carried out with several "demo clothes pegs" of different clamping forces (calibrated for 4, 6 or 14 Newton). This preliminary challenge with clothes pegs of obviously varying forces was done with the intention to let the patients assume that test pegs can objectively cause different degrees of pain ("blinding"). The use of clothes pegs of varying forces was important especially with regard to the retest-setting. It follows that the patients could not automatically presume that every test peg causes the same pain intensity.

All actual test series were carried out with a 10 Newton-calibrated clothes peg. For each measurement the test pin was applied on the middle fingers and ear lobes for 10 seconds. The patient afterwards indicated the pain intensity on a numerical pain rating scale (NRS) on which 0 stands for „no pain" and 10 for „the most intense pain imaginable". Since pain increases during the 10 seconds stimulation, patients were explicitly asked about the intensity of pain they perceived at the end of the test (i.e. at 10 seconds).

The tests were performed on both sides of the body to average out possible side-specific differences in perception [10]. According to the protocol, both instruments were applied successively to the right middle finger, the left middle finger, the right ear lobe and the left ear lobe. The electronic algometer and the clothes peg were applied to the nails of the middle fingers without touching the nail fold (Figures 2a and 2b). Pain tests on the ear lobes were performed on the central soft tissue part without touching the ear cartilage (Figure 2c).

For the analyses we used SPSS 17 for Windows. Variables were expressed as percentages, mean values with standard deviation (SD) and because some variables showed a skewed distribution, also as median plus interquartile range (IQR). The level of significance was set at p < 0.05 (two-tailed). As the data were not normally distributed, correlation with Spearman's rho was calculated to measure the correlation between the electronic and clothes peg algometric method. According to Cohen's conventions, r = 0.3-0.5 corresponds to a correlation size of a medium effect and r > 0.5 corresponds to a large effect [7].

Repeatability coefficients and plotted test-retest differences against test-retest means using the Bland Altman technique were computed for those patients who qualified for repeatability analysis (cf. chapter "Patients and design") [11].

In addition, we plotted test-retest differences against test-retest time in order to investigate whether repeatability was time-dependent.

The PPdt and clothes peg tests could be performed in all of the 157 patients. Six patients refused to test PPtt because of fear of severe pain.

Main characteristics of the patients are shown in Table 1. Seventy (45%) patients were recruited from the surgical orthopaedic department and 87 patients (55%) from the medical-psychosomatic department. In both departments the treating physicians made diagnoses according to the usual standards and classification systems (ICD-10). Twenty-six patients showed more than one diagnosis (20% of the orthopaedic patients, 13% of the medical-psychosomatic patients).

We found a wide range of different pain types: Approximately half of the orthopaedic patients (n = 34) suffered from acute traumatic pain: 12 patients had a trauma of upper extremities or shoulders, 17 had bone fractures or joint lesions of the lower extremities, 9 had a trauma in the thoracic, spinal or pelvic region. Seven patients had a polytrauma with ≥ 3 lesions. The other half of the orthopaedic patients (n = 36) suffered from pain of osteoarthritic degenerative origin: 11 had hip arthritis, 11 had knee arthritis, 7 had degenerative shoulder problems, and 10 had degenerative low back pain. The vast majority of the patients of the medical-psychosomatic department suffered from chronic pain syndromes which were incompletely explained by persistent peripheral tissue damage: 5 patients suffered from chronic tension headache, 11 had chronic cervical pain syndromes, 3 suffered from chronic temporomandibular or atypical facial pain syndromes, 27 had chronic low or upper back pain, 8 had chronic abdominal or pelvic pain, 6 suffered from functional hemisided pain syndromes, and 16 had chronic atypical postsurgical pain syndromes. Twenty-three of the medical-psychosomatic patients suffered from pain without any structural findings or any history of peripherally induced pain: 15 of them were diagnosed with fibromyalgia, 8 with trunk accentuated somatoform pain syndrome.

Forty-one (47%) of the medical-psychosomatic pain patients showed clinically elevated depression scores (HADS-D ≥ 9 points), whereas only 6 patients (9%) had elevated depression-scores in the orthopaedic group.

Pain sensitivity data are described in Table 2. The distribution of baseline pain values and clothes peg data (ear lobe) are illustrated with box-and-whisker plots (Figure 3).

The comparison between clothes peg values and PPdt (respectively PPtt) revealed an inverse correlation across all patients, i.e. the higher the pain in the clothes peg test, the lower the pressure pain threshold, and vice versa. Table 3 shows partial correlation coefficients for the associations between the two pain measurement methods. It can be seen that the absolute size of the correlation coefficients is similar in both patient groups and both anatomical test sites. According to Cohen's conventions only one value corresponds to a correlation size of a medium effect (r = 0.3-0.5), all other correspond to a large effect (> 0.5).

Forty-seven patients from the medical-psychosomatic department qualified for repeatability analysis. Test-retest time varied between 7 and 41 days (mean: 21; SD 8). For both algometric methods, repeatability coefficients were smaller than one SD of test-retest means, indicating a more than four-fold lower within-individual variance over time than between individuals (Table 4). Test-retest difference was neither related to pain intensity (Figure 4) nor to test-retest time (Figure 5).