Effect of mental stress on cold pain in chronic tension-type headache sufferers

Subjects were recruited via advertisements in local media and University of South Australia media requesting volunteers for a study on headaches. Written consent from each subject for study participation was obtained and the study was approved by the University’s Human Research Ethics Committee. Potential volunteers underwent a diagnostic interview based on the International Classification of Headache Disorders (ICHD-II) criteria [15]. Inclusion criteria for the CTH group were: satisfying ICHD-II criteria for CTH, aged 18–65 years, not currently receiving (or having received in the last 12 months) intervention for headache, no psychiatric or major medical condition currently or in the last 12 months, no concurrent headache or pain symptoms or diagnoses (other than CTH). Additionally, CTH subjects were required not to be taking, or not have taken in the last 3 months, regular analgesic medication other than ≤1,000 mg daily of aspirin or paracetamol. All subjects were required to have not taken any analgesic on the day of examination, and were required to be headache free at presentation for the experimental session. Inclusion in the Control group required additional criteria of no past or current chronic pain or headache diagnoses, fewer than five headaches in the last year, and none within the last 6 months. Additionally, Control subjects were recruited following headache subjects to allow matching for age and gender. All recruited subjects completed the study procedures.

The protocol involved exposing subjects to either an hour-long stressful mental task or an hour-long neutral condition, and measuring cold pressor responses before, during (30 min into the task), after task/neutral condition exposure. Headache subjects were randomly assigned to either the stress (CTH-S) or neutral (CTH-N) conditions, while all non-headache subjects (CNT) were exposed to the stress condition only, as we have previously shown that the same neutral condition does not effect changes in pain sensitivity in healthy subjects [16]. The sample contained N = 25 CTH-S, N = 23 CTH-N, and N = 23 CNT subjects. All sessions were conducted in an interview room at the School of Psychology, University of South Australia, between 9.00 a.m. and 5.00 p.m. on Monday to Friday. The room was a constant 23°C.

Prior to the experimental condition, subjects completed an in-house socio-demographic questionnaire and clinical interview detailed elsewhere [17]. Subjects also completed the State-Trait Anxiety Inventory [18], and the Centre for Epidemiological Studies-Depression Scale [19], due to possible effects of anxiety and depression on pain sensitivity.

The stress task was adapted from one previously demonstrated to induce mental stress and headache [20], and involved subjects solving anagrams and arithmetic problems, presented via computer monitor. Anagrams had three levels of difficulty: ‘Easy’ anagrams were words of 8–10 letters with two sets of adjacent letters presented in reverse order. ‘Difficult’ anagrams were long words with letters presented in random order, and ‘Insoluble’ anagrams were 9- to 11-letter words with letters presented in random order and one letter omitted so no solution existed. Previous research has shown subjects can solve most of the easy anagrams but few of the difficult anagrams [20]. The arithmetic problems have been used in previous research by our group [3], and involve subtraction or addition of two-digit (e.g. 73–58) and three-digit (e.g. 576–283) problems. There was an equal amount of anagrams and arithmetic problems in each block, which were presented in random order.

The task involved presentation of each problem for 10 s on the monitor, during which subjects were required to solve the problem in their head (without the use of paper or pencil). Following the problem presentation screen, a screen with the words ‘inter-trial interval’ was presented for 5 s, during which subjects verbalized their answer to the previous problem, or said ‘pass’ if they did not know the answer. The experimenter remained in the room with the subject for the entire time and pretended to enter their responses into a computer.

The task was divided into 12 ‘blocks’ of 10 problems. Subjects were told they would receive performance feedback via the screen after each block, advising of their accuracy relative to previous participants, in the form of ‘below average’, ‘average’, or ‘above average’. In fact, feedback was not based on performance, with subjects receiving ‘below average’ for eight blocks, and ‘average’ for four blocks. Subjects were also advised that the duration of the task was based on their performance, with the shortest possible duration being 30 min (if they got all problems correct), and the longest duration 1.5 h (if they got all of the problems incorrect). In fact, the task lasted for 1 h, regardless of performance.

The neutral condition involved subjects remaining seated in the interview room and browsing supplied magazines and newspapers for 60 min. The experimenter remained in the room working on a computer, and did not engage with the subjects during the task.

The cold pressor task involved subjects immersing their right hand up to the wrist in a circulating water bath containing water maintained at 0°C. Following immersion, subjects reported when the stimulation first became painful (pain detection threshold), then provided a rating of pain intensity every 10 s according to a visual analogue scale presented on a wall in front of them ranging from ‘0—no pain’ to ‘10—extremely painful’. Subjects were requested to keep their hand in the water for a long as possible, and to withdraw their hand when they could no longer tolerate the pain. The experimenter recorded the pain detection and tolerance times using a stopwatch, which was also used by the experimenter to prompt subjects for tonic pain ratings every 10 s during hand immersion.

Statistical analyses were conducted using the Statistical Package for the Social Sciences [21]. A two within [task (pre, during, post); cold pressor ratings (threshold, tolerance, rating)] and one between (CTH-S, CTH-N, CNT) factor repeated measures analysis of variance (RMANOVA) was used to explore group differences and effects of stress on cold pressor thresholds and pain ratings. Due to the majority of CTH sufferers reaching tolerance within 30 s of immersion, only pain ratings at 10 and 20 s were included in the tonic pain rating analyses, conducted thus, on N = 15 CTH-S, N = 13 CTH-N, and N = 13 CNT subjects. Significant RMANOVA results were followed up with examination of simple main effects between each pair of groups (CTH-S vs CTH-N, CTH-S vs CNT, CTH-N vs CNT). ANOVA and Chi-square test were used to compare groups on socio-demographic and clinical data. Finally, split-file Pearson’s r correlations were conducted to examine relationships between experimental (cold pressor) pain, clinical characteristics (headache history, average headache severity, headache frequency over the last month), anxiety, and depression.

The present results indicate reduced tolerance but not detection thresholds to cold induced pain at the hand in CTH sufferers. Such results are consistent with suggestions of central sensitization and an increased general pain sensitivity in CTH sufferers [1, 22, 23]. The results for detection thresholds are consistent with the only previous study to examine cold pressor exclusively in CTH sufferers [9], and are consistent with Marlowe [24], who reported decreased tolerance but not detection thresholds to ice placed at the temple in frequent tension-type headache sufferers. Similarly, differences between CTH and Control subjects on heat pain thresholds have been less often observed than have differences in pressure pain thresholds, particularly in CTH subjects from the general population, as assessed in the current study. It may be that such subjects represent less severe cases of CTH, with accordingly less sensitization [1]. In contrast to the present results, Bishop et al. [9] failed to find reduced tolerance to cold pressor in CTH subjects. The conflicting results may be due to method: Bishop et al’s [9] cold pressor used un-circulating water at 0–2°C, while we used circulating water at 0°C. Circulating water avoids heat build up around the hand [25], presenting a more painful (and possibly less tolerable) stimulus.

The present study found increased pain intensity ratings that approached significance (P = 0.05) in CTH sufferers compared to CNT. The result extends previous findings of increased sensitivity to cold pressor in migraine and mixed headache diagnostic groups [7, 8]. Increased pericranial muscle tenderness is a common finding in CTH sufferers [17, 26, 27], indicating sensitivity to supra-threshold mechanical stimulation is increased in pericranial region of CTH sufferers. The present results suggest CTH sufferers also have increased sensitivity to supra-threshold cold pain stimulation at extra-cephalic locations.

Ashina et al. [23] suggest supra-threshold testing may be more sensitive than pain detection thresholds for assessing central sensitization in CTH. The present findings of increased supra-threshold pain rating and decreased pain tolerance but not detection thresholds in CTH sufferers supports this suggestion. Bendsten et al. [28] found qualitatively altered stimulus–response function to supra-threshold pressure pain stimulation in CTH sufferers. The present results found no group differences in pain increase from 10 to 20 s during cold pressor immersion. This indicates that although overall rating of supra-threshold cold stimulation is increased in CTH sufferers, the stimulus–response function is not altered. However, a conclusion cannot be made due to the limited number of stimulus–response data points in the present study. Further research using more data points is needed to test this hypothesis.

Olesen [29] and others [1, 30] have suggested mental stress may contribute to CTH through aggravating already increased pain sensitivity in CTH sufferers. The present results support such a hypothesis, indicating mental stress reduces pain detection threshold and increases supra-threshold pain ratings more in CTH sufferers than healthy Controls.

A previous study found a brief (15 min) mental stress task decreased pericranial pressure pain thresholds but not cold pain thresholds more in CTH sufferers than in healthy controls [3]. In contrast, the present study found mental stress decreased cold pain detection thresholds more in CTH sufferers than controls. The difference may be due to the longer stress task used in the present study, or the use of cold pressor in the present study compared to an ice cube held against the wrist as the cold pain stimulus in the previous study [3].

The present study also extended previous findings by examining, for the first time, effects of induced mental stress on supra-threshold cold pain ratings in CTH sufferers. The results indicated that stress increased overall pain intensity from cold pressor more in the headache sufferers than in the control group. Further, the results indicated no difference in supra-threshold pain ratings between the healthy Controls exposed to the stress task and the headache group exposed to the neutral condition. This suggests that group differences in supra-threshold pain ratings may be due to effects of the mental task in the headache group. In contrast, mental stress did not affect the rate of pain increase (from 10 to 20 s) during cold pressor immersion in headache or control subjects, suggesting stress does not alter the stimulus–response function to cold induced pain. Mental stress-induced analgesia typically involves a rightward shift in the stimulus–response function of central nociceptive neurons [31], resulting in an increased threshold and decreased intensity with no alteration in the shape of the response function. It may be that hyperalgesic effects of mental stress, as observed in the present study, conversely involve a leftward shift in the response function. The present finding that stress reduced threshold and increased intensity, without altering rate of increase, is consistent with this suggestion. Again, a conclusion cannot be made due to the limited number of stimulus–response data points.

It is interesting to note that the present analyses found a group difference on pain tolerance but not detection threshold, and a group × task effect on pain detection but not tolerance threshold. This result suggests enhanced effects of mental stress on pain processing in CTH sufferers may partially operate through different pain mechanisms than those underlying increased pain sensitivity at rest in CTH sufferers. For example, increased baseline pain sensitivity in CTH sufferers may be due to sensitization (indexed preferentially in reduced tolerance and increased supra-threshold response), while stress may aggravate the increased pain sensitivity through reducing threshold to noxious input from the periphery, and increasing supra-threshold response. Further research to examine this hypothesis may aid in elucidating central mechanisms of stress-induced headache.

The relevance of experimental pain sensitivity to clinical pain has been widely debated [6, 32]. As chronic pain involves tonic sensation above threshold, tonic supra-threshold pain may be more relevant than pain thresholds to clinical pain [6]. Consistent with this, we found no correlation between pain thresholds and headache activity or psychological measures, but did find a correlation between pain intensity ratings and headache activity over the last month in the headache group. Previous studies report increased muscle tenderness is related to headache activity in CTH sufferers [17, 33, 34], indicating increased pain response in pericranial myofascia is of clinical relevance in CTH. The present results suggest a generalized increase in pain response may also be of clinical relevance to CTH. Conclusions are tentative however due to the lack of a consistent pattern of correlations across cold pressor and clinical pain measures in the present study.

A number of limitations to the present study warrant mention. We only used subjective measures of stress. However, the majority of research indicates that where differences between headache and Control subjects have been found, they have generally been found on subjective but not physiological indices of stress [35]. Similarly, we used self-report pain measures. However, self-report represents maximal integration of the stress and pain systems, and most previous research has used self-report to demonstrate increased pain sensitivity in CTH sufferers (e.g. [22, 25, 27]), and hyperalgesic effects of stress on pain sensitivity in healthy humans (e.g. [10, 11]). Indeed, previous research found effects of stress on pain report but not nociceptive reflex [36, 37]. Finally, we did not examine for possible gender effects in our study due to the sample size being too small for reliable analysis. However, as groups did not differ on measures of gender, such effects could not account for the observed group differences in pain sensitivity. Additionally, the issue of gender effects on pain sensitivity is at present unclear: Some authors have reported gender effects on pain sensitivity [38, 39], while others have not [40, 41]. Further research examining possible gender differences in the present protocol are required to address this issue.