Repeatability of cold pain and heat pain thresholds: The application of sensory testing in migraine research

Abstract

Normal heat pain threshold (HPT) and cold pain threshold (CPT) repeatability should be estimated in order to identify thermal allodynia in longitudinal studies, but such data are scarce in the literature. The aim of our study was to estimate normal HPT and CPT repeatability in the face, forehead, neck and hand. In addition, we reviewed briefly normative studies of thermal pain thresholds relevant for headache research. Thermal pain thresholds were measured on three different days in 31 healthy headache-free subjects. Coefficients of repeatability and normal limits were calculated. HPT and CPT were lowest in the face. Pooled across regions, the lower repeatability limit for the test/retest ratio was 63% for HPT and 55% for CPT. The upper normal CPT limit varied between 24.5°C and 29.7°C. Lower HPT limits ranged between 35.5°C and 40.8°C. Quantitative sensory methods provide useful information about headache and pain pathophysiology, and it is important to estimate the normal test/retest repeatability range in follow-up studies.

Keywords

Quantitative sensory tests cold pain threshold heat pain threshold repeatability reliability headache

Introduction

Thermal quantitative sensory thresholds (QSTs) are useful in detecting peripheral neuropathy in diabetes (1) and for quantification of thermal allodynia in painful conditions like migraine (2,3) or other hyperalgesic conditions (4). In order to measure significant clinical disease- or condition-related changes (e.g., the onset of thermal allodynia), it is important to know the normal day-to-day variability of HPT and CPT. Moreover, it is mandatory to report reference limits (‘the normal range’) for this variability. One of the most useful measures of the normal day-to-day variability range is the coefficient of repeatability (CR) (5). The heat pain threshold (HPT) CR has been quantified as 5.9°C in the hand (6) while data on cold pain threshold (CPT) repeatability is particularly scarce in the literature. Even the most recent papers on thermal test repeatability in healthy subjects do not report CR (7,8). In order to use thermal pain thresholds in headache research, it is also mandatory to quantify normal repeatability in the face and neck regions. The present study on healthy subjects was part of a larger study to investigate thermal QST in migraineurs (9). The main aim of our analysis was to estimate HPT and CPT repeatability in the face, forehead, neck and hand in healthy subjects.

Subjects

Thirty-one healthy controls (28 women and 3 men), with a mean age of 40 years (standard deviation [SD] 11.4, range 19–64 years), mean height 168.5 (SD 7.4) cm, weight 70.8 (SD 10.2) kg and days from the last menstruation 14.3 (SD 5.9), were recruited from a local workplace and among blood donors. They were interviewed by an experienced research nurse to exclude headache, neurological disorders, psychiatric disorders, significant painful conditions, traumatic injury to the central or peripheral nervous system, cancer, arthritis, cardiovascular disease and chronic medication use. The age and sex distribution reflect that these subjects were recruited as controls for the migraine study (9). The test was repeated on three different visit days (days A, B and C) at the same time of the day. The interval between consecutive tests was 3 to 10 days (77% of intervals were between 6 and 8 days). Caffeine-containing beverages and tobacco were not allowed after midnight on the day of the investigation. Subjects signed a written informed consent form and received the equivalent of 150 US dollars after completing the three sessions to cover expenses. The study was approved by the Regional Committee for Research Ethics and Norwegian Social Science Data Services.

Methods

HPT and CPT were measured on SOMEDIC SenseLab equipment (Somedic Sales AB, Stockholm). This device uses the method of limits (ML) to determine pain thresholds (10,11). Although using ML to measure pain thresholds depends on reaction time, as opposed to the method of levels, ML is faster, widely used in research, and fully adequate for clinical use. Four sites were stimulated with a hand-held, rectangular 25 × 50 mm Peltier element thermode (Somedic Sales AB, Stockholm): the forehead (frontal region above the eyebrows aligned with the inner canthus), face (maxilla 1 cm below the lower orbital rim close to the nasal border), neck (at the insertion of the sternocleidomastoid muscle just below the lower margin of the processus mastoideus) and hand (thenar eminence overlying the abductor pollicis brevis muscle). The maximal temperature range was 5°C–50°C and the rate of change was 1°C per second. The baseline thermode temperature was set to 32°C.

The subject was lying supine on a couch. Three heat stimuli were followed by three cold stimuli with a random interstimulus interval of between 4 and 6 seconds. Stimuli were applied consecutively to the left and right hand, left and right forehead, left and right face, and left and right neck. This fixed order was used in all patients to reduce inter-individual variability. The subject was instructed to press the ‘stop’ button immediately when pain appeared. The button was held in the right hand, except for during the right thenar examination, when the left hand was used.

Data analysis

Mean (N = 3 stimuli per site within the session) HPT and CPT were calculated as the average of right- and left-sided thresholds. In subjects who did not report pain within the 5°C–50°C range, we defined HPT as 50°C (in 13% of measurements) and CPT as 5°C (in 27% of measurements).

Day A/day B and day B/day C test/retest differences for HPT and CPT, and the standard deviation (SD) of these differences (SD_dif) were calculated. The day-to day difference was assessed by Wilcoxon’s test. A clinically useful measure of the normal day-to-day variability range is the CR, defined as 2 × SD_dif (5). In addition, we calculated the difference thresholds (HPT_d =HPT − 32 and CPT_d = 32 − CPT) in order to detemine the ln(PT_ddayA/PT_ddayB) and ln(PT_ddayB/PT_ddayC) test/retest ratios (12,13). Ratios were calculated with random denominator, either ln(A/B) or ln(B/A) and either ln(B/C) or ln(C/B) to control for possible order-effects. ln-transformed variables are preferred when distributions are skewed; for example, when test/retest differences increase as the pain thresholds increase. Since ln(A/B) = ln(A)−ln(B), it is easy to realize that SD of an ln-transformed ratio is equivalent to SD_dif. Mean and SD were calculated for each ln-transformed ratio, and repeatability limits (mean ± 2SD) were retransformed back to the ratio scale and tabulated as a percentage. These retransformed limits are analogous with ±2SD_dif = ±CR. SYSTAT version 11 (Systat Software Inc, Chicago) and SPSS version 13 (SPSS Inc, Chicago) were used for statistical analysis.

Results

Thermal pain thresholds and repeatability in healthy subjects are reported in Table 1. Pooled across body regions, the lower repeatability limit was 63% for HPT and 55% for CPT. Pooled CR was 8.4°C for CPT and 4.9°C for HPT. In other words, HPT would have to decrease from a normal baseline by more than 4.9°C before one could conclude that thermal heat allodynia had appeared in the subject.

Table 1.

Heat and cold pain thresholds and repeatability in 31 headache-free controls

	Pooled mean and SD across days (°C)¹			Asymmetry variation (°C)	Variation beween days
	³PT mean	⁴PT_d mean	SD between subjects	⁶2×SD_(R-L)dif	CR² (°C)	⁵Upper day1/day2 ratio in %	⁵Lower day1/day2 ratio in %
Heat pain threshold
Forehead	46.1	14.1	3.6	2.4	5.6	169	61
Face	43.9	11.9	4.2	2.9	4.5	159	67
Neck	47.1	15.1	3.2	3.2	5.5	164	61
Hand	46.1	14.1	3.3	1.9	3.8	161	63
Pooled across regions	45.8	13.8	3.6	2.7	4.9	163	63
Cold pain threshold
Forehead	12.0	20.0	7.6	4.5	10.6	219	48
Face	12.9	19.1	8.4	5.2	6.2	156	65
Neck⁷	11.0	21.0	7.9	4.7	9.3	217	50
Hand	11.0	21.0	6.7	2.8	6.8	153	64
Pooled across regions	11.7	20.3	7.7	5.1	8.4	187	55

SD = standard deviation. PT = pain threshold. R = right. L = left. ¹R + L average (°C) thresholds were determined on days A, B and C, 3–10 days apart. ²Coefficient of repeatability (CR) = 2 SD_dif = 2 × SD of day A/B and day B/C differences. ³PT = thermal pain threshold on Celsius scale. ⁴PT_d: PT difference (HPT_d = HPT−32°C. CPT_d = 32°C−CPT. ⁵LN(PT_ddayA/PT_ddayB) and LN(PT_ddayB/PT_ddayC) ratios: upper limit = mean + 2SD, lower limit = mean−2SD (retransformed to ratio scale; ratio = 1 is reported as 100%). ⁶SD of individual R-L differences. ⁷One outlying asymmetry (14°C) is excluded.

Pooled day A thresholds were slightly lower than day B thresholds (HPT mean 45.0°C [SD 3.8] and 46.1°C [SD 3.4]; Wilcoxon p < .0005 and CPT mean 12.6°C [7.6] and 14.5°C [7.6]; Wilcoxon p = .02). Pain thresholds did not differ on day B and day C.

Repeatability for CPT was lower in the forehead and the neck compared to the face and the hand. In the forehead and neck, HPT repeatability was slightly better than CPT repeatability. The lowest HPT and the highest CPT values were found in the face.

Essential thermal pain threshold studies using ML have been reviewed (Table 2). The normal upper limit for CPT varied between 17°C and 32°C in the literature (see the ‘CPT_UL’ column in Table 2). Some control subjects report heat pain thresholds below 40°C, and the normal lower limit for HPT varied between 36°C and 43°C according to the literature (see ‘HPT_LL’ column in Table 2).

Table 2.

Review on normal arm and head thermal pain thresholds with methods of limits

Author	^aSubjects	Mean age (range)	^bBaseline (°C)	Area (cm²)	Range (°C)	Rate in °C	Site	HPT (°C) mean (SD)	CPT °C mean (SD)	Limit method	HPT_LL °C	CPT_UL °C	Comments
Verdugo [27]	61 (37F)	38.8 (17–72)	32	12.5	5–50	4	Thenar	44.6 (1.9)	12.3 (4.4)	2 SD	40.8	21.1	NR = 5(50)°C
Yarnitsky [6]	106 (65 F)	(20–79)	32	13.8	0–50	2	Thenar	45.7 (3.2)		1.96 ± root (BSS + ESS)	39.5		CR = 5.85°C
Claus [11]	77	19–40	35	12.5	10–50	7.5	Wrist	43.7 (2.0)		2.5 SD	38.8		CPT < 10°C in 15%
		40–78					Wrist	43.8 (3.0)			36.3
Yarnitsky [30]	16 (10 F)	27.8 (18–42)	35	12.5		1.2	Dorsal forearm	42.8 (2.0)
Hilz [31]	20 (10F)	25.1 (20–30)	35			2	Thenar	42.0		Figure	38.0		Increase of 1.5°C after caffeine
							Volar wrist	42.0			38.0
Hagander [32]	46 (30F)	33.3 SD 9.6	32	13.3		1	Thenar	43.4	9.8	5–95 p	36.3	24.5
							Volar wrist	42.9	14.2		37	27.6
							Dorsal hand	43.9	16.2		37.2	27
Navarro [33]	75 (45 F)	32.8 (17–62)	33	5.0	1–50	1.5	Dorsal digit II	44.8 (3.2)	10.6 (8.9)	2 SD calc	38.4	28.4	95 p UL
							Dorsal hand	43.9 (2.9)	11.8 (8.5)		38.1	28.8
Meier [34]	101 (53 G)	11.5 (6–17)	32	9.0	0–50	1.5	Thenar	^c41.7	^c14.9	2.5–97.5 p	35.8	28
Kelly [35]	50 (31 F)	34 (19–59)	32	6.3	0–50	1.5	Thenar	45.6	6.7	0–90p 10–100 p	43.3	16.9	Very low thenar CPT (small thermode?)
							Medial forearm	44.5	11.8		40.9	24.5
^dRolke [25]	180 (110 F)	38.4 (17–75)	32	9/12.5	0–50/5–50	1	^eFace, hand	43,44	14,12.5	1.96 SD	35,37	32,28	Threshold estimated from figures
Wasner [7]	20 (10 F)	35.4 (12.5)	32	9	0–50	1	Dorsal hand	43.5 (F), 43.9 (M)	20.3 (F) 17.7 (M)	2.5–97.5 p	37.9¹ 35.5²	28¹ 30.5²	¹Standard protocol ²After test trials Day 0,1, and 21: good ICC, CR not reported.
Agosthino [8]	39 (20 F)	42 (20–77)	32			1.5	Thenar	45.8 (3.2)³ 46.5 (2.7)⁴	9.2 (6.1)³ 10.8 (6.6)⁴	2 SD calc (max/min)	40.2 (38.5)	22.7 (18.7)	³Day 1, ⁴Day 2, CR not reported, ICC not reported
Sand	31 (28 F)	40 (19–64)	32	12.5	5–50	1	Forehead	46.1 (3.6)	12.0 (7.6)	2 SD	38.9	27.2	NR = 5(50)°C
(present study)							Thenar	46.1 (3.3)	11.0 (6.7)		39.4	24.5	CR: See Table 1

BSS = between subjects sum-of-square. CPT = cold pain threshold in °C. CR = coefficient of repeatability. ESS = Error sum-of-square. HPT = Heat pain threshold in °C. ICC = intraclass correlation coefficient. LL = lower limit (reported, calculated or estimated from figures). M = mean. ML = method of limits. NR = no response. p = percentile. RT = reaction time. SD = standard deviation. UL = upper limit (reported, calculated or estimated from figures). ^aSubjects: number of healthy control subjects (F = female; G = girls; M = men). ^b Baseline: neutral starting temperature for the thermode. ^cMedian. ^d10 centers. ^eSites not specified further.

Discussion

In order to determine the normal range for day-to-day change in HPT and CPT, the CR variable is recommended (5). However, most authors use neither this measure nor its ratio-scale analogue. For example, Burstein et al. (2) did not report data on normal repeatability in their migraine study, and they defined heat allodynia as a change larger than or equal to the between-subjects SD (i.e., ≥3.8°C HPT and ≥6.8°C CPT) change. The definition of limits will obviously affect the prevalence of allodynia. Previously, the heat pain repeatability limit has been determined as 5.9°C in the hand (6). In the present study we extend these measurements to include both heat and cold pain thresholds at several body sites and we report CR for the hand, face, forehead and neck. An individual decrease in forehead HPT greater than 5.6°C or a relative decrease in HPT_d to 67% or less would accordingly suggest definite thermal hyperalgesia (or allodynia) with our method. We recommend the use of such limits to define the development of hyper- or hypoalgesia in future longitudinal clinical studies.

Thermal allodynia in migraine has also been defined in other ways—for example, as absolute thresholds below 40°C or above 20°C (14) or below 42°C and above 18°C (15), or just by reporting the statistical group differences (16). It is apparently justified to score heat allodynia for thresholds below 40°C, because it is consistent with the observation that C polymodal nociceptors normally do not fire until a temperature of about 42°C–47°C has been reached (17). However, it has repeatedly been demonstrated (Table 2) that the lower normal HPT limit can be below 40°C. Even the most recent attempt to narrow the normal range for HPT and CPT by training sessions and elimination of expectation failed in this respect (7).

In addition, it is important to assess the influence of methodologically important details such as baseline temperature, thermode area, location, rate of change, ML calculation and substitution of floor- and ceiling-effect values (Table 2). An increasing rate of change may, for example, reduce pain thresholds with a reaction time (RT)-independent method (18), but this effect is counteracted by the RT-induced threshold elevation by the ML. It is accordingly recommended that each laboratory construct its own repeatability coefficients, particularly if methodological details differ from those used in published studies.

Repeatability between and within raters for cephalic thermal detection thresholds has previously been quantified and reported to be adequate (13). Reliability data on brush allodynia, also reported to be abnormal in migraine (19), is apparently unavailable, and repeatability data for mechanical punctate allodynia are also suprisingly sparse when we consider the reputed value of the latter method in pain research. Indeed, only a site × side × monofilament parameter seems to have been formally evaluated (20).

Test/retest correlation was reported to be similar in healthy subjects and nociceptive pain patients (8), although this method is inappropriate as a measure of repeatability (5). In addition, one study suggested that the chronic pain intensity level does not affect patients’ HPT and CPT (8). Our data suggest that pain thresholds habituate slightly when repeated after about one week; pain threshold habituation has also been found by others (21–24).

Our results are representative for women, and the data we present were collected as a normal range for a migraine study. This could be a limitation, but inconsistent data about possible sex difference in thermal thresholds have been reported in the past. Rolke et al. (25) found significantly lower HPT and CPT in women (effect size not reported), whereas other researchers did not report sex differences for thresholds obtained with ML (6,26).

We conclude that it is important to use appropriate methods to quantify normal test/retest reliability both for thermal testing and for the other sensory thresholds. Such information is mandatory to define cold and heat hyperalgesia or allodynia in the individual patient. Sensory thresholds cannot distinguish between central and peripheral hyperalgesia (or hypoalgesia). These measurements are nevertheless very useful in pain and headache research as well as for various clinical differential-diagnostic work-ups (27,28). More research is needed (29) in order to increase the reliability and specificity of sensory system quantification methods.

Footnotes

Acknowledgements

Marit Stjern, Grethe Helde, Gøril Gravdahl and Nikita Zhitniy provided technical assistance.

References

American Diabetes Association and American Academy of Neurology. Report and recommendations of the San Antonio conference on diabetic neuropathy. Ann Neurol 1988; 24: 99–104.

Burstein

Yarnitsky

Goor-Aryeh

Ransil

Bajwa

. An association between migraine and cutaneous allodynia. Ann Neurol 2000; 47: 614–624.

Burstein

Cutrer

Yarnitsky

. The development of cutaneous allodynia during a migraine attack clinical evidence for the sequential recruitment of spinal and supraspinal nociceptive neurons in migraine. Brain 2000; 123: 1703–1709.

Cruccu

Anand

Attal

. EFNS guidelines on neuropathic pain assessment. Eur J Neurol 2004; 11: 153–162.

Bland

Altman

. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; i: 307–310.

Yarnitsky

Sprecher

Zaslansky

Hemli

. Heat pain thresholds: normative data and repeatability. Pain 1995; 60: 329–332.

Wasner

Brock

. Determinants of thermal pain thresholds in normal subjects. Clin Neurophysiol 2008; 119: 2389–2395.

Agostinho

Scherens

Richter

. Habituation and short-term repeatability of thermal testing in healthy human subjects and patients with chronic non-neuropathic pain. Eur J Pain 2009; 13(8)779–785.

Sand

Zhitniy

Nilsen

Helde

Hagen

Stovner

. Thermal thresholds are decreased in the migraine preattack phase. Eur J Neurol 2008; 15: 1199–1205.

10.

Fruhstorfer

Lindblom

Schmidt

. Method for quantitative estimation of thermal thresholds in patients. J Neurol Neurosurg Psychiatry 1976; 39: 1071–1075.

11.

Claus

Hilz

Hummer

Neundõrfer

. Methods of measurement of thermal thresholds. Acta Neurol Scand 1987; 76: 288–296.

12.

Sand

Zwart

Helde

Bovim

. The reproducibility of cephalic pain pressure thresholds in control subjects and headache patients. Cephalalgia 1997; 17: 748–755.

13.

Becser

Sand

Zwart

. Reliability of cephalic thermal thresholds in healthy subjects. Cephalalgia 1998; 18: 574–582.

14.

Burstein

Collins

Jakubowski

. Defeating migraine pain with triptans a race against the development of cutaneous allodynia. Ann Neurol 2004; 55: 19–26.

15.

Jakubowski

Silberstein

Ashkenazi

Burstein

. Can allodynic migraine patients be identified interictally using a questionnaire? Neurology 2005; 65: 1419–1422.

16.

Kitaj

Klink

. Pain thresholds in daily transformed migraine versus episodic migraine headache patients. Headache 2005; 45: 992–998.

17.

Price

. Psychophysical measurement. In: Bovie

Hansson

Lindblom

(eds) Touch, temperature and pain in health and disease: mechanisms and assessments: progress in pain research and management. Vol 3, Seattle: IASP Press, 1994, p.3–25.

18.

Tillman

Treede

Meyer

Campbell

. Response of C fibre nociceptors in the anaesthetized monkey to heat stimuli: correlation with pain threshold in humans. J Physiol 1995; 485: 767–774.

19.

LoPinto

Young

Ashkenazi

. Comparison of dynamic (brush) and static (pressure) mechanical allodynia in migraine. Cephalalgia 2006; 26: 852–856.

20.

Valk

de Sonnaville

van Houtum

. The assessment of diabetic polyneuropathy in daily clinical practice: reproducibility and validity of Semmes Weinstein monofilaments examination and clinical neurological examination. Muscle Nerve 1997; 20: 116–118.

21.

Neisser

. Temperature thresholds for cutaneous pain. J Appl Physiol 1958; 14: 368–372.

22.

Taylor

McGilis

SLB

Greenspan

. Body site variation of heat pain sensitivity. Somatosens Mot Res 1992; 10: 455–465.

23.

Smith

Tooley

Montague

Robinson

Cosper

Mullins

. Habituation and sensitization to heat and cold pain in women with fibromyalgia and healthy controls. Pain 2008; 140: 420–428.

24.

Palmer

Martin

. Thermal perception thresholds recorded using method of limits change over brief time intervals. Somatosens Mot Res 2005; 22: 327–334.

25.

Rolke

Baron

Maier

. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006; 123: 231–243.

26.

Fillingim

Maddux

Shackelford

. Sex differences in heat pain thresholds as a function of assessment method and rate of rise. Somatosens Mot Res 1999; 16: 57–62.

27.

Verdugo

Ochoa

. Quantitative somatosensory thermotest. A key method for functional evaluation of small calibre afferent channels. Brain 1992; 115: 893–913.

28.

Chong

Cros

. Technology literature review: quantitative sensory testing. Muscle Nerve 2004; 29: 734–747.

29.

Shy

Frohman

. Quantitative sensory testing: report of the therapeutics and technology assessment subcommittee of the American Academy of Neurology. Neurology 2003; 60: 898–904.

30.

Yarnitsky

Ochoa

. Studies of heat pain sensation in man: perception thresholds, rate of stimulus rise and reaction time. Pain 1990; 40: 85–91.

31.

Hilz

Claus

Balk

Neundörfer

. Influence of caffeine, sweating and local hyperemisation on “Marstock” thermotesting. Acta Neurol Scand 1992; 86: 19–23.

32.

Hagander

Midani

Kuskowski

Parry

. Quantitative sensory testing: effect of site and skin temperature on thermal thresholds. Clin Neurophysiol 2000; 111: 17–22.

33.

Navarro

Kennedy

. Evaluation of thermal and pain sensitivity in type I diabetic patients. J Neurol Neurosurg Psychiatry 1991; 54: 60–64.

34.

Meier

Berde

Dicanzio

Zurakowski

Sethna

. Quantitative assessment of cutaneous thermal and vibration sensation and thermal pain detection thresholds in healthy children and adolescents. Muscle Nerve 2001; 24: 1339–1345.

35.

Kelly

Cook T Backonja

. Pain ratings at the thresholds are necessary for the interpretation of quantitative sensory testing. Muscle Nerve 2005; 32: 179–178.