The nociceptive withdrawal reflex: Normative values of thresholds and reflex receptive fields

2.1. Participants 

The study protocol was approved by the ethics committee of the University of Bern (KEK 147/04) and was in accordance with the Declaration of Helsinki. Between April 2006 and December 2007, a total of 300 pain-free subjects were investigated. They were recruited by advertising in local newspapers, announcements at the Universities of Bern, Basel, Zürich and Fribourg, in various Swiss public institutions and at senior residences. Exclusion criteria were: no knowledge of German language, any acute or chronic pain condition, intake of any pain medication for less that 24h before the investigation, pregnancy (as ruled out by pregnancy test in women in reproductive age) and breast feeding.

Consecutive subjects who replied to the announcements were selected so that 75 subjects were analyzed within each of the following age categories: 20–34, 35–49, 50–64 and 65–80years. Furthermore, the male: female ratio within each age category had to be 1:1 (±1).

All volunteers provided written informed consent before participating.

2.2. Demographic and psychological data 

Gender, age, height, weight and body-mass index (BMI) were recorded. Then the volunteers were asked to fill four questionnaires: Beck Depression Inventory (BDI), State-Trait-Anxiety-Inventory (STAI), Catastrophizing Scale of the Coping Strategies Questionnaire and Short-Form 36 (SF-36).

The BDI is a 21-item self-report measure assessing affective, cognitive and somatic symptoms of depression. Higher scores indicate higher levels of depressive symptoms (Beck et al., 1996).

The STAI is a 40-item self-report questionnaire designed to assess symptoms of anxiety. It consists of two independent scales: a state anxiety scale and a trait anxiety scale, each with 20 items, leading to a score between 20 and 80. Higher scores indicate greater levels of anxiety. The state and trait scales explore anxiety as a current emotional state and as a personality trait, respectively (Spielberger et al., 1979, Laux et al., 1981).

The 6-item catastrophizing scale of the CSQ was used to assess pain catastrophizing cognitions (Rosenstiel and Keefe, 1983). The subscale score is the mean of all six items, and higher scores indicate higher degrees of pain catastrophizing.

The SF-36 questionnaire is a self-administered, 36-item questionnaire that measures health-related functions in eight domains: physical functioning (PF), role limitations due to physical problems (RP), bodily pain (BP), vitality (VT), general health perceptions (GH), social functioning (SF), role limitations due to emotional problems (RE) and mental health (MH). These eight domains were grouped into two health dimension scales: physical (PF, RP, BP, VT) and mental (SF, GH, RE, MH) (Ware and Sherbourne, 1992). The total score was also calculated. Each scale ranges from 0 (lowest level of functioning) to 100 (highest level) (Ware et al., 1993).

2.3. Electrophysiological pain tests 

2.4. Data analysis 

The 80%, 90% and 95% confidence intervals (CI) were calculated for all the tests.

Backward stepwise regression analyses were conducted on each test. Because a very high proportion of subjects had normal health status, BDI, STAI, catastrophizing and SF-36 were not analyzed as continuous variables but were dichotomized as described below. The cut off values for each of these variables were chosen to best distinguish normal from abnormal values for our specific sample.

In all the regressions, the following independent (explanatory) variables were analyzed: gender, age, interaction of gender with age (sex*age), BMI, body side of testing (right vs. left and dominant vs. non-dominant), BDI (cut off 11), STAI state scale (cut off 35), STAI trait scale (cut off 35), catastrophizing (cut off 3), SF-36 physical dimension (cut off 90), SF-36 mental dimension (cut off 90) and SF-36 total score (cut off 90). Concerning body side, the regression analyses were more significant when dominant vs. non-dominant was used, instead of right vs. left. Only the former analyses are presented. The regressions were performed on the following dependent variables: single stimulus reflex threshold, single stimulus pain threshold, temporal summation reflex threshold, temporal summation pain threshold, area of RRF, volume of RRF.

A P value <0.05 was considered as significant. In the final regression models all the variables with a P<0.1 were included in order to provide information on the variables that were only marginally statistically insignificant.

4.1. Descriptive characteristics of the tests 

The distributions of single stimulus and temporal summation thresholds were not too dissimilar from a normal distribution, although with slight asymmetries (Fig. 1, Fig. 2). Area and volume of RRF displayed a decreasing frequency of observations with increasing area and volume; the decrease was much more marked for volume, where most observations lay in the range 0.0–0.3 (Fig. 3).

The threshold for evoking reflexes was higher than the pain threshold after single stimulus (Table 2). Previous studies have found identical thresholds (Willer, 1977, Chan and Dallaire, 1989), while in other studies the reflex threshold was lower than the pain threshold (Bromm and Treede, 1980, Micalos et al., 2008). This is probably related to different test sites and/or different definitions of the reflex threshold (Rhudy and France, 2007). In this study, a demand of fairly long lasting EMG burst might explain the relatively higher reflex thresholds to single electrical stimulation. On the other hand, the pain and reflex thresholds to repeated electrical stimulation were almost identical, in agreement with previous observations (Banic et al., 2004).

4.2. Influence of demographic variables 

When the tests are used for clinical purposes, not only the statistical significance but also the quantitative impact of the explanatory variables are important. The quantitative impact is determined by the regression coefficients (Table 3, Table 4, Table 5) and provide indications on the magnitude of clinical relevance of the correlations.

Several investigations have shown that pain thresholds are lower in women than in men across various stimulus modalities (Chesterton et al., 2003, Ge et al., 2004). However, gender was not a predictor of any outcome measure in the present study. In previous investigations, the nociceptive reflex threshold to single stimulus was either not affected by gender (Willer, 1990) or lower in women than in men (France and Suchowiecki, 1999). The temporal summation reflex threshold was slightly lower in women than in men (Serrao et al., 2004). Unlike these investigations, our finding on the lack of gender effect resulted from the analysis of a large sample size and was consistent across the different tests, suggesting that electrical tests are probably insensitive to gender differences.

The influence of age on pain sensitivity is still controversial and the mechanisms underlying the correlation are poorly understood. The influence of age seems to be strongly dependent on the stimulus modality (Gibson and Farrell, 2004, Lautenbacher et al., 2005). In the present study, age was related with different assessment modalities (Table 6). The highest quantitative impact was observed with the single stimulus pain threshold, with a correlation coefficient of 0.0463. This means that for an increase in 10years of age the threshold increases by 0.463mA, i.e. by 4.2% in relation to the mean value of the threshold. For the temporal summation assessments, the correlation was negative, but the quantitative impact was negligible: for an increase in 10years of age the threshold increases by 0.183 and 0.184mA for pain and reflex thresholds, respectively. The same negligible quantitative impact was observed for area and volume of RRF. The generally low quantitative influence of age on the assessments probably explains the inconsistent findings of previous investigations on the nociceptive reflex, which were conducted on smaller sample sizes and did not cover the whole range of age (Sandrini et al., 2005). A less efficient endogenous inhibitory control has been detected in elderly compared with young subjects, which may partly explain the increased pain sensitivity that we found with single stimulus pain threshold (Edwards et al., 2003). For practical purposes, we suggest that the confidence intervals presented in Table 2 are used as reference values independent of age.

The body side was related significantly with reflex threshold to single stimulus. Measurements on the dominant side had a threshold lower than on the non-dominant side by 0.9696mA, i.e. 6.0% lower in relation to the mean value of the threshold. A study on non-nociceptive reflexes revealed no side differences, but because only 11 subjects were investigated the study probably did not have sufficient power to detect differences (Sakamoto et al., 2006). We are not aware of studies analyzing the effect of body side on nociceptive reflex parameters. In the absence of such investigations, explanations for our finding remain speculative. Differences in sensory and motor conduction velocities of peripheral nerves between dominant and non-dominant arm have been documented (Colak et al., 2004): it can be postulated that the preferential use of the dominant limb may lead to a subclinical sensitization that is reflected by lowered reflex thresholds. A further possible explaining factor is the greater strength and muscle mass of the dominant side, leading perhaps to a lower activation threshold of the muscles.

4.3. Influence of psychological and health-related parameters 

The analyses on psychological and health-related parameters should be evaluated under the consideration that we studied almost only healthy subjects. Only a small number of them displayed disturbances in the investigated dimensions, so that the variables had to be dichotomized in order to measure their potential role. This implies that the effects of the variable under consideration are only significant beyond a critical threshold of the psychological and health-related parameters.

The importance of depression in pain syndromes is well-known, but it is still unclear whether depression is a determinant or a cause of pain (Angst et al., 2008). There are few and inconsistent data on the influence of depressive symptoms on pain thresholds. Depression as assessed by the BDI affected pressure (Petzke et al., 2003) and heat pain (de Zwaan et al., 1996). Conversely, in an early study on chronic pain patients, BDI was not related to any experimental pain modality including the nociceptive reflex (Boureau et al., 1991). In the present study, BDI was a predictor only of the single stimulus pain threshold, with a correlation coefficient of −2.5943. This means that subjects with depression scores ⩾11 have estimated pain thresholds 2.5943 lower than those subjects with scores <11, reflecting a 23.8% decrease in relation to the mean value of the threshold. The fact that depression affected only a subjective pain threshold and not the reflex assessments suggests that pure spinal nociceptive processes may be independent of the influence of depression. The same can be said for the subjective pain threshold to repeated stimulation (temporal summation), which was not affected. This model may therefore reflect spinal integrative mechanisms, rather than supraspinal pain processing.

State and trait scales of STAI were not significantly correlated with any test (Table 6). In a previous study, inducing anxiety experimentally in healthy volunteers decreased heat pain thresholds (Rhudy and Meagher, 2000). In contrast, anxiety did not affect the nociceptive reflex threshold after single electrical stimulation (French et al., 2005). In an early study, anxiety influenced electrical pain tolerance, but not pain detection threshold (Robin et al., 1987). Our findings suggest that anxiety is not a relevant contributor of electrical pain thresholds and of pain tests involving spinal nociceptive reflexes.

Our findings confirm the lack of correlation between catastrophizing and nociceptive reflex threshold in both healthy volunteers (France et al., 2002, Rhudy et al., 2007) and patients with neck pain after whiplash injury (Sterling et al., 2008).

We are not aware of studies that correlated parameters of the SF-36 or similar scales with pain thresholds. Among the different SF-36 parameters, the only statistical significance was observed on the area of the RRF for mental health. The correlation was negative, reflecting a decrease in pain sensitivity for scores ⩾90: the RRF area decreases by 0.0577, which represents 17.5% of the mean value of RRF area. This finding suggests a possible modest influence of general health status on spinal nociceptive processes, but the fact that only one parameter was affected render an interpretation of this result difficult.

4.4. Meaning of the limited influence of the variables on the tests 

The limited influence of the predictors on the tests that we analyzed can be considered in two ways. The lack of effect of factors that are known to influence pain sensitivity, such as gender or certain psychological factors, indicates that such electrical tests explore only part of the complex sensory and affective experience of pain. On the other hand, the relative robustness of the tests may be used advantageously when the influence of confounding parameters is unwanted. This may be the case for pharmacological studies conducted on small samples, in which it may be difficult to control for confounding factors. In a clinical setting, the evaluation of nociceptive processes that are unaffected by demographic and psychological factors may be useful in different situations: for instance, to make inferences on central plasticity processes leading to generalized central hypersensitivity, independent of the influences of higher cognitive and affective components.