Temporal stability and responsiveness of a conditioned pain modulation test

Participants

Participants were primarily recruited through local newspapers or the clinics’ newsletters, distributed via email or Facebook. Some subjects were recruited through other therapists at the clinics or were invited to the study when seeking care. Collaborating General Practitioners’ clinics were also asked to recommend this study to their patients.

The study population consisted of subjects with persistent or recurrent neck pain for at least six months, undergoing treatment for their neck pain.

Inclusion criteria were:

1. At least eighteen (18) years of age

2. Able to read and understand Swedish

3. No chiropractic treatment in the preceding three months

4. Not diagnosed with cardiovascular disease, hypertension, or diabetes

5. Not medicated with steroids, β-blockers, or antidepressants

6. Self-reported BMI below 30

7. Not pregnant

8. Not suffering from any other serious competing diagnoses or recent development of headache, dizziness, drop-attacks, or acute cervical radiculopathy

Responder status

Participants were categorized as either responders or non-responders on the basis of changes in clinical neck pain intensity as reported using a numeric rating scale (NRS-11) at baseline and at two-week follow-up (end of the intervention period). The cut-point for responder status was set on the basis of a minimal clinically important difference (MCID) [23] defined as a reduction of two or more NRS points on a 0–10 scale, after four treatments over a two-week period. In other words, an increase in pain or a reduction of less than two was considered a clinical non-responder.

Conditioned pain modulation

The specifics of the CPM test utilized in this study are similar to those described by O’Neill and O’Neill in 2015 [24]:

Test stimulus: The pressure pain intensity (PPI) was measured using a spring-operated clamp (two 2.8 cm × 1.9 cm pads). The clamp used is commercially available, procured by Clas Ohlson (article number 40-7211) and exerts a force of 7.3 kg at a 2.6 cm opening. The clamp was placed on the subject’s thumbnail for 10 s. Caution was taken to prevent the clamp from putting pressure on the eponychium but still placed as close to the eponychium as possible. The pain was rated on the NRS-11 from 0 (no pain) to 10 (worst imaginable pain) as an overall pain score of the PPI procedure. This was performed before (step 1) and immediately after (step 2) the cold pressor test.

Conditioning stimulus: The cold pressor test utilized a 40 L water container (Mobicool B4 230/12 V compressor) to keep the water at a stable temperature between 0 and 2 °C, monitored using a mercury thermometer. The container also contained a small submersible pump (Barwig Typ 03 0333, 720 L/h, 6 m), to keep the water circulating during tests. The subjects were asked to lower the opposite hand of the initial PPI test in the cold water and keep it there for at least 1 min or until the pain became unbearable. The pain intensity was recorded continuously as a VAS score (0 (no pain) to 100 (worst imaginable pain)) on a computer with a sampling rate of 1 Hz. Pain with the cold pressor test was summarized as the duration (tolerance) and summated pain scores over time, described as area under the curve.

The difference in NRS-11 between step 1 and step 2 of the test stimulus was calculated as the CPM response by subtracting step 1 from step 2.

All participants received the same instructions which followed a prepared manuscript. The measurement procedure and instructions were rehearsed by two researchers performing the measurements. The two testers also observed each other in a pilot study to ensure high and similar measurement quality.

Sample size calculation

The study sample size was determined a-priori as that of the RCT of which this study is a secondary analysis [25].

Statistical methods

Continuous variables were reported with means and standard deviations, and categorical variables were reported as counts and percentages.

The temporal stability of the first PPI, cold pressor test time, max cold pressor test pain, cold pressor test area under the curve, and CPM response, were presented as the mean and standard deviation for each repeat-test and combined.

The temporal stability of CPM for individuals and the study population is further illustrated using spaghetti plots (all participants) and box plots (population).

The individual CPM change between repeat-tests is tabulated is frequency tabulation.

Temporal stability was analyzed using ICC3 and CPM data were analyzed as a multivariate linear regression (repeated measures MANOVA type III), with five CPM variables (first PPI, cold pressor test time, max cold pressor test pain, cold pressor test area under the curve, and CPM response) as dependent variables and with clinical responder status, RCT group allocation, and test day as independent variables.

Ethics

This study was approved by The Regional Ethical Review Board (Stockholm) (2018/2137-31). A written informed consent form was signed by all subjects. The study was conducted in accordance with the Helsinki declaration [26].

Participants

One hundred thirty-one subjects were recruited of which four subjects dropped out during the intervention period. A further two participants had incomplete data, leaving one-hundred-and-twenty-five (n=125) for analyses.

As evident from Table 1, 47 participants (38%) were categorized as clinical responders compared to 78 non-responders (62%). For a summary of descriptive variables sub-grouped by responder status, please see Table 1.

Quantitative sensory testing and temporal stability of CPM

Raw data for cold pressor test, mechanical pressure pain intensity and conditioned pain modulation are summarized in Table 2. Changes over time in the primary outcome for this study (conditioned pain modulation) are presented in Table 3. The raw CPM data and change over time are illustrated in Figure 1 as a spaghetti plot. CPM data population distribution is illustrated in Figure 2 as box plots.

Figure 1:

Conditioned pain modulation over time. Spaghetti plot of conditioned pain modulation for each individual participant over time. Lines are slightly jittered and semi-transparent to allow for illustration of multiple superimposed lines as darker lines. The y-axis CPM represents the difference in pressure pain intensity (ΔNRS) before/after cold pressor test. Linear regression is illustrated as a blue line with grey confidence interval.

Figure 2:

Distribution of conditioned pain modulation at baseline at follow-ups. Boxplot (median, inter-quartile range and 1.5xIQR) illustration of the distributions of conditioned pain modulation, i.e. the difference in pressure pain intensity (ΔNRS) before/after cold pressor test. . Raw data are superimposed and presented as slightly jittered circles to allow for visualization of overlying observations.

The large majority (81%) were observed to change by no more than a single NRS point, i.e. ±1 NRS in CPM response over time, and less than 6% changed by 3 NRS points or more.

The mean change in individual CPM responses was 0.22 from baseline to Week 1 (SD=1.35), and −0.15 from Week 1 to Week 2 (SD=1.24). An Intra-class Correlation Coefficient (ICC3 – single, fixed rater) for CPM across the three time points yielded a moderate or fair reliability coefficient of 0.54 (p<0.001) – see Table 4.

Factors affecting CPM response

Study population

Arguably, the study population was internally heterogeneous in so far as participants had received one of two different treatments, but we suggest this was justified as there is no reason to expect that the treatments would have any direct inherent effects on the CPM responses. Rather, any effect of the treatments on CPM response would likely be the result of changes in clinical pain, but no significant difference in change in clinical pain was observed between treatment groups [27]. As reported above, the results of a linear regression model confirmed that treatment group allocation did not influence CPM responses.

It should be noted, that as no pain-free control group was included in the current study we can make no inferences about the magnitude of the observed CPM responses in our population, only that the CPM responses were not related to clinical response to treatment.

Temporal stability of CPM

The linear regression modeling indicated that the occasion (test days 1-to-3) was also not a significant effect/predictor for CPM response. In other words, the distribution of CPM responses on a population level remained stable over the study period.

Arguable this does not preclude greater changes over time within individuals if those changes are roughly equal in either direction. The analysis of variance, however, demonstrated that this was not the case, and the mean change in CPM response for individuals between tests was found to be relatively small.

A systematic literature review from 2016 [13] reported reliability between test sessions as ranging from fair to excellent in six different studies, but also noted that these employed different CPM test paradigms. Conversely, a systematic review and meta-analysis from 2022 [28] reported that whilst intra-session reliability was good-to-excellent with a mechanical pressure and cold-pressor test paradigm, the inter-session reliability was fair or less for all modalities and for both healthy participants and pain patients. There is also evidence, that the CPM effect is more pronounced or robust when employing cold-pressor test as the conditioning stimulus, as the case was in this study, compared to other methods [29]. The current data are thus in line with these previous findings and demonstrate a fair degree of stability over time, despite participants being engaged in a treatment program.

Responsiveness of CPM to changes in clinical pain

We hypothesized that patients who experienced a clinically important improvement in neck pain would also exhibit increases in CMP responses compared to those without clinical improvement. This hypothesis is not supported by our findings.

At face value, this suggests that CPM response is in fact not affected by clinical improvement in patients with persistent or recurrent neck pain.

However, there are a few caveats that should temper our interpretation: We defined a minimum clinically important change as a reduction in NRS of at least two on the 11-NRS scale and this was observed in 38% of participants with a mean change of −3.3 (0.2). It is possible that clinical pain intensity and/or duration in this cohort of neck pain patients was insufficient to lead to reduced CPM responses in the first place. However, with a mean baseline clinical pain intensity of 5.5 (1.6) and a neck pain duration of longer than six months, this is less likely. We can only speculate, but as generalized hyperalgesia seems common in patients with persistent pain, we would expect a degree of CPM attenuation in this population as well.

It is also possible that the changes in clinical pain following treatment were insufficient to effect changes in CPM responses robust enough to be statistically detectable.

QST technical considerations

As described in the Introduction, CPM testing is more complicated than simpler QST test paradigms and is sensitive to differences in methodology. This might go some way to explain conflicting findings in CPM research [13, 14]. Whilst CPT is commonly used as the conditioning stimulus in CPM paradigms, the use of a mechanical clamp as a test stimulus is not. Whilst CPM responses have previously been demonstrated using such a setup [24, 30], it is possible that the test stimulus was insufficiently intense to make a CPM response detectable (i.e. a flooring effect). However, with a clamp force of 7+ kg at 2.4 cm opening and overall mean pain of 4.41 at the first measurement, this seems unlikely.