Quantitative and Qualitative Responses to Topical Cold in Healthy Caucasians Show Variance between Individuals but High Test-Retest Reliability

Abstract

Increased sensitivity to cold may be a predictor of persistent pain, but cold pain threshold is often viewed as unreliable. This study aimed to determine the within-subject reliability and between-subject variance of cold response, measured comprehensively as cold pain threshold plus pain intensity and sensation quality at threshold. A test-retest design was used over three sessions, one day apart. Response to cold was assessed at four sites (thenar eminence, volar forearm, tibialis anterior, plantar foot). Cold pain threshold was measured using a Medoc thermode and standard method of limits. Intensity of pain at threshold was rated using a 10cm visual analogue scale. Quality of sensation at threshold was quantified with indices calculated from subjects' selection of descriptors from a standard McGill Pain Questionnaire. Within-subject reliability for each measure was calculated with intra-class correlation coefficients and between-subject variance was evaluated as group coefficient of variation percentage (CV%). Gender and site comparisons were also made. Forty-five healthy adults participated: 20 male, 25 female; mean age 29 (range 18–56) years. All measures at all four test sites showed high within-subject reliability: cold pain thresholds r = 0.92–0.95; pain rating r = 0.93–0.97; McGill pain quality indices r = 0.87–0.85. In contrast, all measures showed wide between-subject variance (CV% between 51.4% and 92.5%). Upper limb sites were consistently more sensitive than lower limb sites, but equally reliable. Females showed elevated cold pain thresholds, although similar pain intensity and quality to males. Females were also more reliable and showed lower variance for all measures. Thus, although there was clear population variation, response to cold for healthy individuals was found to be highly reliable, whether measured as pain threshold, pain intensity or sensation quality. A comprehensive approach to cold response testing therefore may add validity and improve acceptance of this potentially important pain measure.

About the Authors:

Penny Moss

* E-mail: [email protected]

Affiliation: School of Physiotherapy and Exercise Science, Curtin University of Technology, Perth, Western Australia

Jasmine Whitnell

Affiliation: School of Physiotherapy and Exercise Science, Curtin University of Technology, Perth, Western Australia

Anthony Wright

Affiliation: School of Physiotherapy and Exercise Science, Curtin University of Technology, Perth, Western Australia

Introduction

Increased sensitivity to cold may be a robust predictor of persistent pain [1,2,3] or of increased post-operative pain [4]. Cold hyperalgesia has been reported as an important characteristic of neuropathic pain [5,6,7] and also in some individuals with less clearly neuropathic disorders such as fibromyalgia [8,9] whiplash associated disorder, [10,3], spinal pain [11,12] and osteoarthritis [13,14].

Despite this increasingly use of cold response data, literature reporting the reliability of cold pain measures is limited. A recent systematic review of quantitative sensory testing (QST) concluded that, in contrast to other QST measures, cold pain threshold (CPT) reliability is not yet well established, largely due to limited published data [15]. This lack of data may be partly explained by the floor effect from the 5°C cut-off temperature of thermodes used by many studies, which means that significant numbers of healthy participants do not reach cold pain threshold, resulting in exclusion of CPT from analysis [16].

The limited available CPT reliability data is ambiguous, complicated by considerable methodological variability between studies, particularly as regards thermode type, test sites and statistical analysis. For example, Heldestad et al. [17] investigated variation in volar forearm CPT in healthy subjects, reporting very low intra-subject coefficients of variation (CV) of 0.73%, 1.6% and 0.63%. However, investigators were required to use a 10°C cut off temperature, thus artificially limiting variability. Geber et al. [18] used the TSA II thermode (Medoc, Israel) which has a cut-off temperature of 0°C and reported within-subject test-retest Intra-Class Correlation Coefficients (ICC) for CPT of between r = 0.86 and r = 0.79 across non-standardised test sites in 60 participants. Moloney at al. [19] also used the Medoc thermode and reported intra-subject ICCs of between r = 0.87 and r = 0.94 for CPT at the second digit. However, this study also reported very high variability between individuals with between-subject CVs of between 84.9% and 90.2%. There is clearly a need for additional CPT reliability data, which clearly delineates within-subject and between-subject results.

Additional sensory evaluation of cold response, such as intensity or quality of sensation experienced during cold stimulus would offer a more comprehensive assessment of cold sensitivity. However, only a very few studied have included any such additional sensory measures, with only two considering test-retest reliability. Wasner and Brock [20] asked subjects to quantify intensity of pain and cold or heat during CPT on numerical rating scales and found high test-retest reliability in both (ICCs between r = 0.80 and .r = 0.94). Geber et al. [18] recorded number of paradoxical heat sensations experienced during cold stimuli, reporting a less strong correlation between two test sessions (r = 0.351). However, no previous study has analysed the test-retest reliability of both pain intensity and a comprehensive assessment of sensation quality at cold pain threshold.

The current study therefore aimed to determine more comprehensively the within-subject test-retest reliability and between-subject variability of sensory response to cold in healthy adults over three separate test sessions at two upper limb and two lower limb sites. Response was measured using conventional CPT temperature and pain intensity sensation quality ratings at CPT as well as cold detection thresholds (CDT).

Methods

Study Design

A repeated measures design was used, with response to cold (CPT, pain intensity at CPT and pain quality at CPT) tested at four body sites on three separate occasions. Cold detection thresholds were also tested on each occasion and at each site. Using data from a previous study [13] it was calculated that a sample size of between 40 and 50 subjects would provide 80% power (α = 0.05).

Participants

Subjects were recruited voluntarily, via word-of-mouth and advertisements, from the Perth W.A. metropolitan area. All subjects were adults over 18 years of age, with normal sensation at the test sites and able to read and understand written and spoken English. Only volunteers of Caucasian ethnicity were recruited in order to limit ethnic variability in cutaneous pain thresholds [21,22]. Exclusion criteria included the presence of any acute pain, history of any chronic pain condition or neurological disorder and intake of any pain medication within 24 hours of testing. All volunteers provided written informed consent prior to participating. Approval was obtained from the Curtin University Human Research Ethics Committee (Approval Number PT0188).

Procedure

All subjects attended three test sessions, each separated by at least 24 hours. All testing was performed by the same investigator and standardized, scripted instructions given for each task. Four sites on the non-dominant side were tested in randomized order (Fig 1).

[Figure omitted. See PDF.]

Fig 1. Test sites.

https://doi.org/10.1371/journal.pone.0151972.g001

Outcome Measures

Cold detection and cold pain thresholds.

Cold detection threshold (CDT) was tested first at each site. Any subject who failed to detect cold within 10°C of the starting temperature was excluded from further testing. Cold pain threshold (CPT) tests were then completed at each site. All assessment was performed with a Thermal NeuroSensory Analyzer (TSA) 2001-II (MEDOC, Israel) and stimulus delivery controlled by TSA software. A thermode probe with a contact area of 15cm² was securely strapped to the test site. All CDT and CPT testing followed the standard Method of Limits protocol [23]. The temperature decreased at a rate of 1°C/s from a baseline of 32°C to a cut-off of 0°C. For CDT, subjects were instructed: “Please press your hand-held switch as soon as you sense any cold sensation”. For CPT subjects were instructed: “Please press your switch as soon as the sensation of cold changes to one of uncomfortable or painful cold”. Once the subject pressed the control switch, or once 0°C was reached, the temperature returned to baseline. For each site and each threshold type, a practice was followed by three trials, each separated by a variable pause of between 4 and 6 seconds. The mean value was used for analysis. For subjects who did not press the control switch before the cut-off temperature, CPT value was defined as 0°C.

Pain intensity of sensation at CPT.

On completion of CPT testing at each site subjects were asked to rate the intensity of pain experienced at CPT, using a 100mm visual analogue scale (VAS). Visual analogue scales have demonstrated concurrent validity with other pain measurement methods [24] and good reliability [25].

Quality of sensation at CPT.

Participants were also asked to select words from the McGill Short-Form Questionnaire (SF-MPQ) (Part 2) [26] to describe sensation quality at CPT. This was completed once per test session. The SF-MPQ (part 2) includes a list of 70 sensory, affective and general descriptors. Two indices can be calculated: the Pain Rating Index (PRI) is the sum of the values allocated to each word [27] the number of words counted (NWC) is the total number of words chosen [26]. The SF-MPQ has been deemed valid and reliable (r = .96) [28] and has previously been used in experimental pain studies with healthy subjects [29].

Sample Size and Data Analysis

Data were analyzed using SPSS for Mackintosh v22 (IBM Corp., NY), with alpha set at p≤0.05. Shapiro-Wilk analysis showed that only CPT values at the thenar eminence and foot were normally distributed. All non-normal data was log transformed for further analysis. Within-subject reliability across the three test sessions was calculated for each test site using Intra-Class Correlation Coefficients (ICC) (2-way absolute model) for mean CPT, VAS pain and McGill indices at CPT and for mean CDT. Within-days variance for each measure at each site was expressed as Coefficient of Variation percentage (CV%), using the standard deviation (SD) of between-days means divided by the between-days mean x100. Between-subject variance was explored through calculation of group means, standard deviations and inter-quartile ranges for each site at each test session. Group CV% was also calculated for each test session at each site ((group SD/group mean)x100). Skewness and kurtosis values were also calculated for each measure at each site on each test occasion to further explore data distribution.

Influence of site and gender were also analysed. Differences between sites were analysed using either Repeated Measures ANOVA or Friedman’s 2-way Analysis of Variance (session and site, or session and gender as factors). Independent t-tests / Mann-Whitney U test evaluated whether there was a gender difference in CPT, VAS and McGill scores. Gender-separated ICCs, within-days CV% and group CV% were calculated to evaluate differences in within-subject test-retest reliability and between-subjects variance between genders.

Results

Forty-five participants completed all three testing sessions: 20 male and 25 female; mean age 29 (range 18–56) years. There were no significant differences in age between genders (p = 0.466). No subjects were withdrawn from the study.

Cold Pain Threshold

Within-subject reliability for CPT across the three test sessions was high at each test site. ICCs ranged from r = .92 to r = .95 (p < .001) (Table 1). Mean CPT was consistently higher at the two upper limb sites (mean 11.21°C) compared with the lower limb sites (mean 9.07°C). Between-days variance for CPT across the three test sessions ranged from CV% of 1.16% for the forearm to 9.08% for the foot.

[Figure omitted. See PDF.]

Table 1. Within-subject test-retest reliability data.

https://doi.org/10.1371/journal.pone.0151972.t001

Between-subjects variance for CPT was considerably larger, with group standard deviations and interquartile ranges consistently high relative to the population mean (Table 2).

[Figure omitted. See PDF.]

Table 2. Between-subject means, standard deviations (SD), interquartile ranges and group coefficient of variation (CV%) for cold response measures.

https://doi.org/10.1371/journal.pone.0151972.t002

Pain intensity at CPT

Within-subjects reliability for VAS rating of pain intensity at CPT was also consistently high (ICCs r = 0.93 thenar to r = 0.97 tibialis anterior (Table 1). Between-days variance ranged from CV% of 1.15% at tibialis anterior to 6.20% at the forearm. VAS pain ratings at CPT were relatively low, averaging 3/10. The thenar eminence exhibited greatest pain at CPT (mean VAS 3.6 ± 0.1) with the least pain experienced at the tibialis anterior site (mean VAS 2.6 ± 0.03).

Between-subjects variance for pain intensity at CPT was similarly greater than within-subject variance (Table 2). At each site, VAS intensity ranged from 0/10 to 8.5/10. Group data were also positively skewed at all sites and kurtosis values indicated a generally flat data distribution (Table 3).

[Figure omitted. See PDF.]

Table 3. Within-subject test-retest reliability, comparing males and females.

https://doi.org/10.1371/journal.pone.0151972.t003

Sensation quality at CPT

Within-subject reliability for sensation quality was also high (ICC for NWC r = 0.87 (p < .001); ICC for PRI r = 0.85 (p < .001)) and between-days variance was low (CV% 2.51% for NWC and 1.31% for PRI) (Table 1). Fig 2 illustrates the consistency of word choice between test sessions. The basic sensation of cold was almost universally experienced at CPT. Paradoxical heat-type words, usually associated with an abnormal response, were chosen by 12 of these pain-free participants (27%), almost always in combination with a strong cold word such as "freezing".

[Figure omitted. See PDF.]

Fig 2. Sensation quality descriptors at each test session.

There was high consistency of the most frequently chosen descriptors of sensation quality at cold pain threshold across the three test sessions.

https://doi.org/10.1371/journal.pone.0151972.g002

As for CPT and VAS, between-subject variance for sensation quality was wide, with large group SDs and interquartile ranges for both descriptor index values (Table 2). Data for both indices were positively skewed with low kurtosis values (Table 3). Fig 1 illustrates the range of sensory qualities reported, each word selected by at least 3 participants. Although 93% of participants selected at least one cold-type word, most selected ‘cold’ or ‘freezing’. Paradoxical heat in combination with cold was reported by 15% of participants. Three subjects (7%) reported paradoxical burning with no cold.

Cold detection threshold

Within-subject reliability for cold detection threshold was considerably lower than for cold pain threshold, with ICCs ranging from r = 0.75 for the foot to r = 0.56 for the forearm. In contrast to CPT, between-subjects variance was less, with smaller SDs and interquartile ranges relative to the mean. CDT distribution was negatively skewed and generally exhibited higher Kurtosis values than for CPT.

Influence of test site location

Repeated Measures ANOVA was applied to investigate differences between test sites in CPT values. There were differences between sites on all test occasions, with upper limb sites consistently exhibiting higher CPT values than lower limb sites (Table 1). This difference reached statistical significance on Days 2 (p = 0.002) and 3 (p = 0.001) although not on Day 1 (p = 0.102). Pain ratings at CPT also varied between test sites, with upper limb sites consistently evoking significantly higher pain ratings than lower limb sites (combined upper limb means: 3.63, 3.44 and 3.46 /10; combined lower limb means: 2.83, 2.70, 2.70 / 10). Investigation of CDT values showed that the foot was consistently less sensitive than the other test sites on all test occasions (all p<0.001).

Influence of gender

CPT values were significantly higher for females than for males, at all but the tibialis anterior site, although there was no significant gender difference in pain rating at CPT at any site (p = 0.270 to p = 0.435), or in sensation quality (NWC p = 0.747; PRI p = 0.814).

Female subjects showed greater reliability for CPT testing at all sites (ICCs r = 0.94–0.97) compared with males (ICCs r = 0.90–0.95), for sensation quality (NWC: ICC males r = 0.66, female r = 0.95; PRI ICC males r = 0.70, females r = 0.93) and slightly higher reliability for pain rating (ICCs males r = 0.90–0.95; females r = 0.92–0.96) (Table 3). Between-subjects variance also tended to show a gender differences, with females exhibiting lower group variance than males for both CPT values and pain ratings at CPT (Tables 4 and 5): mean group CV% for CPT across all test sessions, females 64.5%, males 79.2%; mean group CV% for pain rating across all sessions, females 73.3%, males 81.0%.

[Figure omitted. See PDF.]

Table 4. Between-subject variance for cold response measures, females only (n = 25).

https://doi.org/10.1371/journal.pone.0151972.t004

[Figure omitted. See PDF.]

Table 5. Between-subject variance for cold response measures males only (n = 20).

https://doi.org/10.1371/journal.pone.0151972.t005

Discussion

This study evaluated the within-subject reliability and between-subject variance of response to cold over three days in healthy individuals.

Cold pain threshold

Within-subject reliability for CPT across three sessions was excellent, with ICCs between r = 0.93 and r = 0.94 and between-day CV% less than 10% for all sites (Table 1). Comparable test-retest studies using ICCs report similarly high CPT coefficients: from r = 0.87 to r = 0.94 for finger, hand and foot values [19, 20, 30]. Cumulatively, these results suggest that, for an individual, CPT is a reliable measure over time. In contrast, between-subjects variation in CPT was wide. Group CV% on Day 1 ranged from 62% at the hand to nearly 92% at tibialis anterior (Table 2). This distinction between high individual reliability yet wide variance between individuals is supported by previous studies [19, 20, 30]. It is this considerable population variability that has resulted in the view that CPT is an unusable QST measure [16,19]. Yet CPT consistently shows high individual consistency for repeated measurement at a specific site.

Pain intensity and sensation quality at CPT

A similar pattern of individual consistency but population variance was shown for both VAS pain intensity and sensation quality (NWC, PRI) at CPT (Tables 1 and 2). There are very few studies with which to compare these data. For VAS rating, Wasner and Brock [20] used numeric rating scales for pain at CPT at the hand, reporting similar means and a similarly high correlation between ratings of r = 0.90 (p < .001) over 24 hours. For between-subjects variability, Kelly at al. [31] reported within-subjects standard deviation of 1.4/10 for VAS pain at CPT, yet a large between-subjects standard deviation of 5.7/10, also supporting the current study. Very few studies have assessed quality of sensation at CPT and none have reported it in sufficient depth for meaningful comparison. The current study is therefore the first to investigate comprehensively sensation quality at CPT for healthy individuals. Clearly more data is needed, but this study suggests that both pain rating and sensation quality at CPT are individually determined and reliable over time. However, as for CPT, there is considerable variation in these ratings between individuals. Indeed a ‘normal’ response for some may even include paradoxical or noxious qualities normally assumed to reflect pathology. The nature of this variability in pain-free individuals warrants further clarification and investigation.

Influence of test site

Whilst reliability at each site was high, there was a clear difference between sites for CPT, with values consistently higher at upper limb compared with lower limb sites (mean 11.2°C versus 9.1°C). This site difference is consistent with many other studies [19, 32, 33]. Cold detection in the current study was also significantly less sensitive at the foot and this is supported by the only other comparable study [20]. Strong within-subject reliability for each site appears to indicate the robustness of measures of cold pain and detection.

Influence of gender

This study found that healthy females reported CPT at higher temperatures, although the actual sensory experience does not appear to differ, with no gender difference in pain rating or sensation quality at CPT. Many factors have been suggested to explain the influence of gender, such as differences in central pain modulation [34] or different societal mores [35]. Previous data regarding the effect of gender on CPT is extensive but inconsistent [36], with some supporting a significant gender difference [37] and some disagreeing [31,20]. The current study also found that females were more reliable in all cold response measures. Wasner and Brock [20] also reported that females were more reliable than males for CPT, even over 21 days.

Study limitations

A number of limitations must be acknowledged. Although only Caucasian participants were included in order to limit ethnicity as a confounder, the study needs to be validated with different ethnic samples. Due to testing time restrictions, McGill descriptors were only assessed once on each test occasion at the end of CPT testing. It is therefore still unclear whether cold pain sensation quality is a reflection of central sensory interpretation and so consistent across body regions. This should be further investigated as exploration of widespread CPT sensation quality may be important in understanding the mechanisms involved in central pain processing.

Conclusion

Cold pain testing has been discounted by some as of limited value due to poor reliability [16,19,31,32]. However this study found that, although population variance was high, individual subject reliability for all measures of cold response was excellent. The addition of pain intensity and sensation quality measures at threshold provides a more comprehensive characterization of response and may assist future interpretation of the mechanisms influencing normal and abnormal pain processing.

Author Contributions

Conceived and designed the experiments: PM AW JW. Performed the experiments: PM JW AW. Analyzed the data: PM JW AW. Wrote the paper: PM JW AW.

Citation: Moss P, Whitnell J, Wright A (2016) Quantitative and Qualitative Responses to Topical Cold in Healthy Caucasians Show Variance between Individuals but High Test-Retest Reliability. PLoS ONE 11(3): e0151972. https://doi.org/10.1371/journal.pone.0151972

References

1. Ritchie C, Hendrikz J, Kenardy J, Sterling M. Derivation of a clinical prediction rule to identify both chronic moderate/severe disability and full recovery following whiplash injury. Pain 2013;154:2198–2206. pmid:23831865

2. Sterling M, Hendrikz J, Kenardy J, Kristjansson E, Dumas JP, Niere K, et al. Assessment and validation of prognostic models for poor functional recovery 12 months after whiplash injury: a multicentre inception cohort study. Pain 2012;153:1727–34. pmid:22658881

3. Goldsmith R, Wright C, Bell SF, Rushton A. Cold hyperalgesia as a prognostic factor in whiplash associated disorders: A systematic review. Man Ther 2012;17:402–10. pmid:22464187

4. Ahmad S, De Oliveira GS, Bialek JM, McCarthy RJ. Thermal Quantitative Sensory Testing to Predict Postoperative Pain Outcomes Following Gynecologic Surgery. Pain Medicine; 15: 857–864. pmid:24517836

5. Pfau DB, Rolke R, Nickel R, Treede RD, Daublaender M. Somatosensory profiles in subgroups of patients with myogenic temporomandibular disorders and fibromyalgia syndrome. Pain 2009;147:72–83. pmid:19767146

6. Backonja MM, Attal N, Baron R, Bouhassira D, Drangholt M, Dyck PJ, et al. Value of quantitative sensory testing in neurological and pain disorders: NeuPSIG consensus. Pain 2013; 154:1807–19. pmid:23742795

7. Freeman R, Baron R, Bouhassira D, Cabrera J, Emir B. Sensory profiles of patients with neuropathic pain based on the neuropathic pain symptoms and signs. Pain 2014; 155:367–76. pmid:24472518

8. Smith BW, Tooley EM, Montague EQ, Robinson AE, Cosper CJ, and Mullins PG. Habituation and sensitization to heat and cold pain in women with fibromyalgia and healthy controls. Pain 2008; 140: 420–428. pmid:18947923

9. Lee YC, Nassikas NJ, Clauw DJ. The role of the central nervous system in the generation and maintenance of chronic pain in rheumatoid arthritis, osteoarthritis and fibromyalgia. Arthritis Res Ther 2011;13:211–221. pmid:21542893

10. Sterling M, Jull G, Vicenzino B, Kenardy . Sensory hypersensitivity occurs soon after whiplash injury and is associated with poor recovery. Pain 2003;104:509–517. pmid:12927623

11. O'Sullivan P, Waller R, Wright A, Gardner J, et al. Sensory characteristics of chronic non-specific low back pain: A subgroup investigation. Man Ther 2014; 19:311–318. pmid:24731602

12. Steinmetz A, Jull GA. Sensory and sensorimotor features in violinists/violists with neck pain. Arch Phys Med Rehabil 2013; 94:2523–2528. pmid:23664957

13. Moss P, Benson H, Will R, Wright A. Cold hyperalgesia is associated with altered pain quality and reduced function in people with knee osteoarthritis. Europ J Pain Supp. 2011;5:128.

14. Hochman J, Davis A, Elkayam J, Gagliese L, Hawker GA. Neuropathic pain symptoms on the modified painDETECT correlate with signs of central sensitization in knee osteoarthritis. Osteoarthritis Cartilage 2013; 21:1236–42. pmid:23973136

15. Moloney NA, Hall TM, Doody CM. Reliability of thermal quantitative sensory testing: A systematic review. J Rehabil Res Dev 2012; 42:191–208.

16. Wylde V, Palmer S, Learmonth ID, Dieppe P.Test-retest reliability of Quantitative Sensory Testing in knee osteoarthritis and healthy participants. Osteoarthritis Cartilage 2011;19:655–8. pmid:21329759

17. Heldestad V, Linder J, Sellersjo L, Nordh E. Reproducibility and influence of test modality order on thermal perception and thermal pain thresholds in quantitative sensory testing. Clin Neurophysiol 2010;121:1878–1885. pmid:20478739

18. Geber C, Klein T, Azad S, Birklein F, Gierthmuhlen J, Huge V, et al. Test–retest and interobserver reliability of quantitative sensory testing according to the protocol of the German Research Network on Neuropathic Pain (DFNS): A multi-centre study. Pain 2011;152:548–556. pmid:21237569

19. Moloney NA, Hall TM, O'Sullivan TC, Doody CM. Reliability of thermal quantitative sensory testing of the hand in a cohort of young, healthy adults. Muscle Nerve 2011;44:547–552. pmid:21826684

20. Wasner GL, Brock JA. Determinants of thermal pain thresholds in normal subjects. Clin Neurophysiol 2008;119:2389–2395. pmid:18778969

21. Rowell LN, Mechlin B, Ji E, Addamo M, Girdler SS. Asians differ from non-Hispanic Whites in experimental pain sensitivity. Eur J Pain 2005;15:764–771.

22. Rahim-Williams B, Riley J, Williams AKK, Fillingham RB. A Quantitative Review of Ethnic Group Differences in Experimental Pain Response: Do Biology, Psychology, and Culture Matter? Pain Medicine 2012; 13: 522–540. pmid:22390201

23. Fruhstorfer H, Lindblom U, Schmidt WC. Method for quantitative estimation of thermal thresholds in patients. J Neurol Neurosurg Psychiatry 1976; 39:1071–1075. pmid:188989

24. Downie WW, Leatham PA, Rhind VM, Wright V, Branco JA, Anderson JA. Studies with pain rating scales. Ann Rheum Dis 1978; 37:378–381. pmid:686873

25. Boeckstyns ME, Backer M. Reliability and validity of the evaluation of pain in patients with total knee replacement. Pain 1989; 38:29–33. pmid:2780060

26. Melzack R. The short-form McGill pain questionnaire. Pain 1987; 30:191–197. pmid:3670870

27. Burckhardt CS, Jones KD. Adult Measures of Pain. Arthritis Rheum 2003; 49:S96–S104.

28. Grafton KV, Foster NE, Wright C.C. Test-retest reliability of the Short-Form McGill Pain Questionnaire: assessment of intraclass correlation coefficients and limits of agreement in patients with osteoarthritis. Clin J Pain 2005; 21:73–82. pmid:15599134

29. Law LAF, Sluka KA, McMullen T, Lee J, Arendt-Nielsen L, Graven-Nielsen T. Acidic buffer induced muscle pain evokes referred pain and mechanical hyperalgesia in humans. Pain 2008; 140:254–264. pmid:18835099

30. Knutti IA, Suter MR, Opsommer E. Test-retest reliability of thermal quantitative sensory testing on two sites within the L5 dermatome of the lumbar spine and lower extremity. Neurosci Lett 2014; 579;157–62. pmid:25064700

31. Kelly KG, Cook T, Backonja MM. Pain ratings at the thresholds are necessary for interpretation of quantitative sensory testing. Muscle Nerve 2005; 32:179–84. pmid:15937874

32. Rolke R, Baron R, Maier C, Tolle TR, Treede RD, Beyer A, et al. Quantitative sensory testing in the German research network on neuropathic pain (DFNS): standardized protocol and reference values. Pain 2006; 123:231–243. pmid:16697110

33. Sand T, Nilsen KB, Hagen K, Evers S, Marziniak M. Repeatability of cold pain and heat pain thresholds: the application of sensory testing in migraine research. Cephalalgia 2010; 30:904–909. pmid:20656701

34. Martin VT. Ovarian hormones and pain response: a review of clinical and basic science studies. Gend Med 2009;6:168–192. pmid:19406368

35. Thorn BE, Clements KL, Ward LC, Dixon KE, Kersh BC, Boothby JL, et al. Personality factors in the explanation of sex differences in pain catastrophizing and response to experimental pain. Clin J Pain 2004; 20:275–282. pmid:15322433

36. Fillingim RB, King CD, Ribeiro-Dasilva MC, Rahim-Williams B, Riley JL. Sex, gender, and pain: a review of recent clinical and experimental findings. J Pain 2009; 10;447–485. pmid:19411059

37. Neziri AY, Scaramozzino P, Andersen OK, Dickensen AH, Arendt-Nielsen L, Curatolo M. Reference values of mechanical and thermal pain tests in a pain-free population. Eur J Pain 2011; 15: 376–83. pmid:20932788