- Email: [email protected]
Thermal quantitative sensory testing: A study of 101 control subjects
Journal of Clinical Neuroscience 22 (2015) 588–591
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Laboratory Studies
Thermal quantitative sensory testing: A study of 101 control subjects Jessica Hafner a, Geoffrey Lee a, Jenna Joester a, Mary Lynch a, Elizabeth H. Barnes c, Paul J. Wrigley d,e, Karl Ng a,b,e,⇑ a
Department of Neurology and Clinical Neurophysiology, Clinical Administration 3E, Royal North Shore Hospital, Reserve Road, St Leonards, NSW 2065, Australia Office of Research and Research Training, Sydney Medical School, University of Sydney, NSW, Australia c NHMRC Clinical Trials Centre, University of Sydney, Camperdown, NSW, Australia d Pain Management Research Institute, Kolling Institute, Northern Sydney Local Health District, St Leonards, NSW 2065, Australia e Northern Clinical School, Sydney Medical School, University of Sydney, NSW 2006, Australia b
a r t i c l e
i n f o
Article history: Received 21 October 2014 Accepted 27 October 2014
Keywords: Marstock method of limits Quantitative sensory testing Small fibre neuropathy Thermal threshold Variability
a b s t r a c t Quantitative sensory testing is useful for the diagnosis, confirmation and monitoring of small fibre neuropathies. Normative data have been reported but differences in methodology, lack of age-specific values and graphical presentation of data make much of these data difficult to apply in a clinical setting. We have collected normative age-specific thermal threshold data for use in a clinical setting and clarified other factors influencing reference values, including the individual machine or operator. Thermal threshold studies were performed on 101 healthy volunteers (21 70 years old) using one of two Medoc Thermal Sensory Analyser II machines (Medoc, Ramat Yishai, Israel) with a number of operators. A further study was performed on 10 healthy volunteers using both machines and one operator at least 3 weeks apart. Thermal threshold detection increases with age and is different for different body regions. There is no significant difference seen in results between machines of the same make and model; however, different operators may influence results. Normative data for thermal thresholds should be applied using only age- and region-specific values and all operators should be trained and strictly adhere to standard protocols. To our knowledge, this is the largest published collection of normal controls for thermal threshold testing presented with regression data which can easily be used in the clinical setting. Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.
1. Introduction Dysfunction of small unmyelinated nerve fibres is thought to be responsible for many painful peripheral neuropathies. These nerve fibres are unable to be evaluated by conventional nerve conduction studies making confirmation of diagnosis, monitoring of disease progression and objective evaluation of therapies for their disorders difficult. Quantitative sensory testing (QST) is an automated psychophysical method used to indirectly test the function of these fibres. QST can be performed to assess various sensory modalities, including vibration, temperature, pinprick and pressure. Both stimulus detection and pain thresholds can be measured. QST is dependent upon the function of not only the peripheral nerve fibres but also the rest of the sensory pathway – including the dorsal root ganglia, spinal cord, thalamus and somatosensory cortex – as well as cognitive factors such as attention and reaction time.
⇑ Corresponding author. Tel.: +61 2 9463 1833; fax: +61 2 9463 1058. E-mail address: [email protected] (K. Ng). http://dx.doi.org/10.1016/j.jocn.2014.09.017 0967-5868/Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.
Our study focused on thermal threshold QST, which assesses both unmyelinated C-fibres (warm detection and heat pain) and myelinated Ad fibres (cold detection) [1]. Historically, thermal studies were designed to assess the warm-cold difference limen by the subject depressing a switch when a change in temperature was perceived. This is known as the Marstock method, named for the collaboration between the Marburg and Stockholm groups [2]. A modified Marstock method allows assessment of absolute thermal threshold by asking the subject to abort a gradually increasing thermal stimulus when it is perceived. The stimulus returns to a baseline of 32°C and the test is repeated a total of five times for warm and five times for cool sensation. The mean thermal threshold is recorded as the final value. The variability of the responses is also noted to ensure the test has been performed reliably. This is referred to as a method of limits. An alternate method, which presents the subject with a ‘‘yes/no’’ paradigm, known as the method of levels, may be more reliable but is more time consuming [3]. Our study aimed to collect normative age-specific data for thermal threshold detection using the modified Marstock method and to clarify whether other factors need to be taken into consideration
589
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
when using normative data, such as the individual machine or operator. 2. Methods One hundred and one normal volunteers (46 men, 55 women; aged 21–70 years [mean 43.4 years]) were studied. Data were collected by three operators using two different Medoc Thermal Sensory Analyser II machines (Medoc, Ramat Yishai, Israel) (operator G.L. used machine 1 and operators J.J. and M.L. used machine 2). The modified Marstock method of limits was used to record thermal thresholds for warm and cool detection at the thenar eminence of the right hand and over the dorsolateral aspect of the right foot. As this initial study demonstrated an apparent difference between the two machines, an additional study was performed with a single operator (J.H.) studying a further 10 subjects (six males, four females; aged 17–62 years [mean 39.7 years]) using both machine 1 and machine 2 on each subject with at least 3 weeks between tests. Half of the subjects were examined first on machine 1 followed by machine 2, and the other half underwent testing in the opposite order. In contrast to previously described methodology, we did not employ a training stimulus prior to commencing the formal testing. This reduced total test time. For all studies, a 3.0 3.0 cm thermode utilising the Peltier effect was used to provide thermal stimulation. A baseline adaptation temperature of 32°C was applied for 5 minutes to each site prior to commencing testing. The thermode changed temperature at 1°C per second, returned to baseline at 1°C per second following subject response and remained at baseline (32°C) for 4–6 seconds prior to delivering the next stimuli. The temperature range of the thermode was 0–50°C. Scripted verbal instructions were used to administer the test. Warm detection was measured five times at each site followed by cool detection measured five times at each site. The difference between the mean detection threshold and 32°C was recorded for warm and cool stimuli for each subject. The Medoc Thermal Sensory Analyser software produces a variance value (varp) for the five recordings from each modality. Varp levels greater than 5 in the hand and 10 in the foot were deemed to indicate unreliable testing from the spread of varp data, and any corresponding data were excluded from the final analysis. Data were also excluded if the subject reported paradoxical sensations (i.e. reporting heat sensation during a cool stimulus), if the response was noted as an outlier within the age-group set, or if the subject reported significant pain at the level of thermal detection. Statistical analysis was performed using Excel (Microsoft Corp., Redmond, WA, USA), SAS (SAS Institute Inc., Cary, NC, USA) and the Statistical Package for the Social Sciences software (SPSS, Chicago, IL, USA). To test for the effects of age, sex and operator/machine on detection thresholds, data were log-transformed and separate general linear models fitted in each combination of hand/foot and warm/cool threshold. The estimates from these models and the 95% confidence limits were back-transformed to calculate the effect of each variable as a factor. To predict upper limits of normal thresholds as a function of age, the raw detection thresholds were fitted to age only in linear regression models and the upper limit of the 99% prediction interval (corresponding to 2.5 standard deviations [SD]) calculated. In the 10 subjects who were tested by the same operator on the two different machines, paired differences were compared using t-tests to test for period and carry-over effects, and to estimate the effect of the different machines on the thresholds. 3. Results One hundred and one subjects were recruited; their demographics are shown in Table 1. Two subjects were excluded from the entire analysis as they reported paradoxical heat sensations.
Table 1 Demographics of the healthy subjects enrolled in this study Age, years
21–30
31–40
41–50
51–60
61–70
Total
Males Females Total
10 12 22
13 9 22
9 10 19
8 15 23
6 9 15
46 55 101
Two subjects with outlying results were excluded from the hand data. From the foot data, eight subjects were excluded from warm and three were excluded from cool analysis. This was due to pain at the thermal detection threshold (one warm, one cool); excessive variability of responses between five trials (seven warm; one cool); and outlying responses (one cool). Mean thermal detection thresholds were calculated for age in years with the upper limit of normal taken as the 99% prediction limit (2.5 SD) (Fig. 1). This confirmed previously reported increased thermal detection thresholds with age [4]. Regression analyses produced equations (Table 2) demonstrating that the effect of age is most notable in the foot, where warm thermal thresholds increase by 1°C per decade of age. The effect is still significant but of smaller magnitude for other thermal threshold parameters. Upper limits of normal were tabulated by age for easy use in the clinical setting (Supp. Table 1). The linear models fitted to log-transformed data showed there was strong evidence of a difference in thresholds in the hand depending on the machine used. This machine effect was not seen in the feet, which may be due to the larger thresholds being measured in this region. There was no evidence for an effect of sex on thresholds (Table 3). On machine 2, there was no evidence of a difference between the two operators (J.J. and M.L.) (p P 0.4 for all). The additional study to clarify this finding of an apparent machine difference (10 subjects studied on both machines by a sole operator [J.H.]) showed no evidence of an effect on intra-individual results of inter-test interval or learning (p P 0.36 for all), though small effects may not have been detected due to the size of this sample. There was also no evidence of any difference between the machines (Table 4). This part of the study was powered to detect a difference in thermal threshold between machines of 1 SD. This suggests that the initially observed machine difference may have been related differences between an individual operator (G.L.) and the two operators (J.J., M.L.). 4. Discussion Thermal threshold testing is a psychophysical method of assessing the function of the sensory pathways, including small nerve fibres, which are not assessed by traditional nerve conduction studies. The modified Marstock method is increasingly finding a role in the assessment and diagnosis of small fibre neuropathies including those seen in diabetes mellitus, renal failure and human immunodeficiency virus infection, as well as central sensory dysfunction (for example, post-stroke pain) and various pain syndromes (for example, angry backfiring C fibres syndrome, and cold hyperalgesia, cold hypoaesthesia, and cold skin syndrome) [5]. 4.1. Normative data for QST Normative data for thermal threshold testing have been reported previously, but differences in methodology, lack of agespecific values, and graphical rather than numerical presentation make much of these data difficult to apply to a modern clinical setting [1,2,4,6–10]. To our knowledge our study provides the largest single-centre cohort of normative thermal threshold detection data reported by age in years in a format that is easily applied to the clinical setting (Supp. Table 1). These data have been collected on
590
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
Fig. 1. Thresholds for the hand and the foot, showing individual results (diamonds) with the calculated mean ±2.5 standard deviations (solid lines; this is the upper limit of normal for a given age group). A tabulated version of these data is available for clinical application in Supplementary Table 1.
Table 2 Regression equations to calculate the mean temperature difference from 32°C (T) detected as a function of age in years Sensory modality
Site
Regression equation
95% CI (°C per year), p value
Warm
Foot Hand Foot Hand
T = 0.974 + (0.1 age) T = 1.053 + (0.012 age) T = 0.199 + (0.045 age) T = 0.751 + (0.011 age)
0.062–0.14, p < 0.001 0.004–0.02, p = 0.005 0.03–0.06, p < 0.001 0.002–0.02, p = 0.015
Cool
Table 4 Machine comparison study Sensory modality
Body region
Mean machine difference, °C (95% CI)
p value
Warm
Foot Hand Foot Hand
1.29 ( 0.57 ( 0.60 ( 0.19
p = 0.07 p = 0.40 p = 0.10 p = 0.14
Cool
0.11 to 2.70) 1.10 to 2.20) 0.14 to 1.36) ( 0.45 to 0.09)
CI = confidence interval.
CI = confidence interval.
a commercially available machine using standard software. Yarnitsky and colleagues [3] presented similar data in 20 year brackets. However, we have demonstrated that differences due to age are seen over shorter time spans. Yarnitsky’s normal values are similar to ours, except for the cool detection thresholds in the foot, where they report higher thresholds and a sex difference in all age groups. We have demonstrated the clear effect of age on normal thermal threshold detection, which has been shown previously in several studies [4,6,9–11], although this is not a universal finding [12]. The nature of this effect is a little variable between studies, with some modalities more susceptible than others. In general, there is an increase in thermal threshold detection with advancing age. In our study, this was most notable with warm detection in the foot. We have not found a significant effect of sex on thermal threshold detection. This is in keeping with other larger studies [4], but some studies do report a difference in thermal threshold detection between men and women. The effect of sex is variable when
present, with warm detection in the foot [10], cool detection in the foot [3] and cool detection in the hand [9] individually being described. There is a clear difference between body regions. This has been well established by previous studies [9] and has been confirmed by our data with clearly different threshold detection in the hand and the foot. The reasons for this are likely differences in skin innervation in different body regions and may also be influenced by longer distances, which inherently exaggerate reaction time artifact inherent to the method of limits testing. Our study initially suggested that there was a difference in the results derived from different machines. However, this apparent difference in results performed by mutually different operators was refuted when the machine was compared using the same subject and operator. Therefore, any machine difference that existed could be attributed to the difference in operators rather than machine within the power of our latter study. Machine 1 readings appeared to be generally larger than machine 2 except for that of
Table 3 Estimated effect of sex (male versus female) and machine (machine 1 versus machine 2) Sensory modality
Body region
Predictor
Estimated effect (95% CI)
p value
Warm
Foot
Sex Machine Sex Machine Sex Machine Sex Machine
1.06 1.21 0.91 0.69 1.11 1.23 0.86 0.55
p = 0.56 p = 0.08 p = 0.19 p < 0.0001 p = 0.34 p = 0.05 p = 0.05 p < 0.0001
Hand Cool
Foot Hand
CI = confidence interval.
(0.86–1.32) (0.98–1.50) (0.79–1.05) (0.60–0.79) (0.89–1.37) (1.00–1.52) (0.74–1.00) (0.47–0.64)
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
cool sensation in the hand, but the magnitude of these differences, inconsistent as they may be, can be partly appreciated by a comparison of the degree of change with variability studies that would be required in one subject by one operator on the same machine, a test not undertaken in this study. It is nevertheless advisable that a single machine be used wherever possible and laboratory normative values collected for that device [13]. One of the major challenges in the clinical utility of QST has been the significant variability seen between sessions in the same patient, leading to questions regarding reliability and reproducibility [14]. Fagius and colleagues reported up to 150% difference between sessions using the method of limits [6]. This problem appears less marked with the method of levels [7]. In their comprehensive comparative study, Yarnitsky and colleagues [3] found that thermal threshold detection in the hand using the method of limits had significant intersession differences. Results from the foot and using the more time consuming method of levels were not affected by significant intersession differences. Moravcova and colleagues demonstrated that use of coefficients of repeatability may be helpful in overcoming intra-individual variability; however, consideration must be given to the fact that patients with neuropathy have increased intra-individual variability compared to controls [11,14]. Markedly increased variability is helpful in detecting functional sensory disorders and results that are unreliable due to patient inattention. This variability is greater than that seen in neuropathy [15]. Using cut-off values for variability, as we did in our study, may be helpful in excluding erroneous abnormal results, but this requires further validation. Our study was not designed to detect intra-individual variability and this issue remains one for consideration, particularly when assessing disease progression or therapy responses in a given individual. QST of thermal thresholds is finding an increasing role in the diagnosis and monitoring of small fibre neuropathies given the paucity of other methods. Normative data for thermal threshold detection must be age- and body region-specific, and operators must be trained appropriately and adhere to protocols to ensure reliability. We present, to our knowledge, the largest single-centre study of normative data as a function of age in years in a format that can be easily applied in the clinical setting. Conflicts of Interest/Disclosures J.H. is supported by an unrestricted fellowship grant by Allergan and Ipsen Pharmaceuticals. J.H. and K.N. are supported by the
591
National Health and Medical Research Council (ID GNT1074648 and 512316), the Myositis Association Australia Inc. and the Brain Foundation Research Grants. Acknowledgements The assistance of Dr Christina Liang is appreciated. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jocn.2014.09.017. References [1] Rolke R, Magerl W, Campbell KA, et al. Quantitative sensory testing: a comprehensive protocol for clinical trials. Eur J Pain 2006;10:77–88. [2] Fruhstorfer H, Lindblom U, Schmidt WC. Method for quantitative estimation of thermal thresholds in patients. J Neurol Neurosurg Psychiatry 1976;39:1071–5. [3] Yarnitsky D, Sprecher E. Thermal testing: normative data and repeatability for various test algorithms. J Neurol Sci 1994;125:39–45. [4] Rolke R, Baron R, Maier C, et al. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006;123:231–43. [5] Yarnitsky D. Quantitative sensory testing. Muscle Nerve 1997;20:198–204. [6] Fagius J, Wahren LK. Variability of sensory threshold determination in clinical use. J Neurol Sci 1981;51:11–27. [7] Jamal GA, Hansen S, Weir AI, et al. An improved automated method for the measurement of thermal thresholds. 1. Normal subjects. J Neurol Neurosurg Psychiatry 1985;48:354–60. [8] Gruener G, Dyck PJ. Quantitative sensory testing: methodology, applications, and future directions. J Clin Neurophysiol 1994;11:568–83. [9] Harju EL. Cold and warmth perception mapped for age, gender, and body area. Somatosens Mot Res 2002;19:61–75. [10] Kemler MA, Schouten HJ, Gracely RH. Diagnosing sensory abnormalities with either normal values or values from contralateral skin: comparison of two approaches in complex regional pain syndrome I. Anesthesiology 2000;93:718–27. [11] Bravenboer B, van Dam PS, Hop J, et al. Thermal threshold testing for the assessment of small fibre dysfunction: normal values and reproducibility. Diabet Med 1992;9:546–9. [12] 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. [13] Shy ME, Frohman EM, So YT, et al. Quantitative sensory testing: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2003;60:898–904. [14] Moravcová E, Bednarik J, Svobodnik A, et al. Reproducibility of thermal threshold assessment in small-fibre neuropathy patients. Scr Med (BRNO) 2005;78:177–84. [15] Yarnitsky D, Sprecher E, Tamir A, et al. Variance of sensory threshold measurements: discrimination of feigners from trustworthy performers. J Neurol Sci 1994;125:186–9.
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Laboratory Studies
Thermal quantitative sensory testing: A study of 101 control subjects Jessica Hafner a, Geoffrey Lee a, Jenna Joester a, Mary Lynch a, Elizabeth H. Barnes c, Paul J. Wrigley d,e, Karl Ng a,b,e,⇑ a
Department of Neurology and Clinical Neurophysiology, Clinical Administration 3E, Royal North Shore Hospital, Reserve Road, St Leonards, NSW 2065, Australia Office of Research and Research Training, Sydney Medical School, University of Sydney, NSW, Australia c NHMRC Clinical Trials Centre, University of Sydney, Camperdown, NSW, Australia d Pain Management Research Institute, Kolling Institute, Northern Sydney Local Health District, St Leonards, NSW 2065, Australia e Northern Clinical School, Sydney Medical School, University of Sydney, NSW 2006, Australia b
a r t i c l e
i n f o
Article history: Received 21 October 2014 Accepted 27 October 2014
Keywords: Marstock method of limits Quantitative sensory testing Small fibre neuropathy Thermal threshold Variability
a b s t r a c t Quantitative sensory testing is useful for the diagnosis, confirmation and monitoring of small fibre neuropathies. Normative data have been reported but differences in methodology, lack of age-specific values and graphical presentation of data make much of these data difficult to apply in a clinical setting. We have collected normative age-specific thermal threshold data for use in a clinical setting and clarified other factors influencing reference values, including the individual machine or operator. Thermal threshold studies were performed on 101 healthy volunteers (21 70 years old) using one of two Medoc Thermal Sensory Analyser II machines (Medoc, Ramat Yishai, Israel) with a number of operators. A further study was performed on 10 healthy volunteers using both machines and one operator at least 3 weeks apart. Thermal threshold detection increases with age and is different for different body regions. There is no significant difference seen in results between machines of the same make and model; however, different operators may influence results. Normative data for thermal thresholds should be applied using only age- and region-specific values and all operators should be trained and strictly adhere to standard protocols. To our knowledge, this is the largest published collection of normal controls for thermal threshold testing presented with regression data which can easily be used in the clinical setting. Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.
1. Introduction Dysfunction of small unmyelinated nerve fibres is thought to be responsible for many painful peripheral neuropathies. These nerve fibres are unable to be evaluated by conventional nerve conduction studies making confirmation of diagnosis, monitoring of disease progression and objective evaluation of therapies for their disorders difficult. Quantitative sensory testing (QST) is an automated psychophysical method used to indirectly test the function of these fibres. QST can be performed to assess various sensory modalities, including vibration, temperature, pinprick and pressure. Both stimulus detection and pain thresholds can be measured. QST is dependent upon the function of not only the peripheral nerve fibres but also the rest of the sensory pathway – including the dorsal root ganglia, spinal cord, thalamus and somatosensory cortex – as well as cognitive factors such as attention and reaction time.
⇑ Corresponding author. Tel.: +61 2 9463 1833; fax: +61 2 9463 1058. E-mail address: [email protected] (K. Ng). http://dx.doi.org/10.1016/j.jocn.2014.09.017 0967-5868/Crown Copyright Ó 2014 Published by Elsevier Ltd. All rights reserved.
Our study focused on thermal threshold QST, which assesses both unmyelinated C-fibres (warm detection and heat pain) and myelinated Ad fibres (cold detection) [1]. Historically, thermal studies were designed to assess the warm-cold difference limen by the subject depressing a switch when a change in temperature was perceived. This is known as the Marstock method, named for the collaboration between the Marburg and Stockholm groups [2]. A modified Marstock method allows assessment of absolute thermal threshold by asking the subject to abort a gradually increasing thermal stimulus when it is perceived. The stimulus returns to a baseline of 32°C and the test is repeated a total of five times for warm and five times for cool sensation. The mean thermal threshold is recorded as the final value. The variability of the responses is also noted to ensure the test has been performed reliably. This is referred to as a method of limits. An alternate method, which presents the subject with a ‘‘yes/no’’ paradigm, known as the method of levels, may be more reliable but is more time consuming [3]. Our study aimed to collect normative age-specific data for thermal threshold detection using the modified Marstock method and to clarify whether other factors need to be taken into consideration
589
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
when using normative data, such as the individual machine or operator. 2. Methods One hundred and one normal volunteers (46 men, 55 women; aged 21–70 years [mean 43.4 years]) were studied. Data were collected by three operators using two different Medoc Thermal Sensory Analyser II machines (Medoc, Ramat Yishai, Israel) (operator G.L. used machine 1 and operators J.J. and M.L. used machine 2). The modified Marstock method of limits was used to record thermal thresholds for warm and cool detection at the thenar eminence of the right hand and over the dorsolateral aspect of the right foot. As this initial study demonstrated an apparent difference between the two machines, an additional study was performed with a single operator (J.H.) studying a further 10 subjects (six males, four females; aged 17–62 years [mean 39.7 years]) using both machine 1 and machine 2 on each subject with at least 3 weeks between tests. Half of the subjects were examined first on machine 1 followed by machine 2, and the other half underwent testing in the opposite order. In contrast to previously described methodology, we did not employ a training stimulus prior to commencing the formal testing. This reduced total test time. For all studies, a 3.0 3.0 cm thermode utilising the Peltier effect was used to provide thermal stimulation. A baseline adaptation temperature of 32°C was applied for 5 minutes to each site prior to commencing testing. The thermode changed temperature at 1°C per second, returned to baseline at 1°C per second following subject response and remained at baseline (32°C) for 4–6 seconds prior to delivering the next stimuli. The temperature range of the thermode was 0–50°C. Scripted verbal instructions were used to administer the test. Warm detection was measured five times at each site followed by cool detection measured five times at each site. The difference between the mean detection threshold and 32°C was recorded for warm and cool stimuli for each subject. The Medoc Thermal Sensory Analyser software produces a variance value (varp) for the five recordings from each modality. Varp levels greater than 5 in the hand and 10 in the foot were deemed to indicate unreliable testing from the spread of varp data, and any corresponding data were excluded from the final analysis. Data were also excluded if the subject reported paradoxical sensations (i.e. reporting heat sensation during a cool stimulus), if the response was noted as an outlier within the age-group set, or if the subject reported significant pain at the level of thermal detection. Statistical analysis was performed using Excel (Microsoft Corp., Redmond, WA, USA), SAS (SAS Institute Inc., Cary, NC, USA) and the Statistical Package for the Social Sciences software (SPSS, Chicago, IL, USA). To test for the effects of age, sex and operator/machine on detection thresholds, data were log-transformed and separate general linear models fitted in each combination of hand/foot and warm/cool threshold. The estimates from these models and the 95% confidence limits were back-transformed to calculate the effect of each variable as a factor. To predict upper limits of normal thresholds as a function of age, the raw detection thresholds were fitted to age only in linear regression models and the upper limit of the 99% prediction interval (corresponding to 2.5 standard deviations [SD]) calculated. In the 10 subjects who were tested by the same operator on the two different machines, paired differences were compared using t-tests to test for period and carry-over effects, and to estimate the effect of the different machines on the thresholds. 3. Results One hundred and one subjects were recruited; their demographics are shown in Table 1. Two subjects were excluded from the entire analysis as they reported paradoxical heat sensations.
Table 1 Demographics of the healthy subjects enrolled in this study Age, years
21–30
31–40
41–50
51–60
61–70
Total
Males Females Total
10 12 22
13 9 22
9 10 19
8 15 23
6 9 15
46 55 101
Two subjects with outlying results were excluded from the hand data. From the foot data, eight subjects were excluded from warm and three were excluded from cool analysis. This was due to pain at the thermal detection threshold (one warm, one cool); excessive variability of responses between five trials (seven warm; one cool); and outlying responses (one cool). Mean thermal detection thresholds were calculated for age in years with the upper limit of normal taken as the 99% prediction limit (2.5 SD) (Fig. 1). This confirmed previously reported increased thermal detection thresholds with age [4]. Regression analyses produced equations (Table 2) demonstrating that the effect of age is most notable in the foot, where warm thermal thresholds increase by 1°C per decade of age. The effect is still significant but of smaller magnitude for other thermal threshold parameters. Upper limits of normal were tabulated by age for easy use in the clinical setting (Supp. Table 1). The linear models fitted to log-transformed data showed there was strong evidence of a difference in thresholds in the hand depending on the machine used. This machine effect was not seen in the feet, which may be due to the larger thresholds being measured in this region. There was no evidence for an effect of sex on thresholds (Table 3). On machine 2, there was no evidence of a difference between the two operators (J.J. and M.L.) (p P 0.4 for all). The additional study to clarify this finding of an apparent machine difference (10 subjects studied on both machines by a sole operator [J.H.]) showed no evidence of an effect on intra-individual results of inter-test interval or learning (p P 0.36 for all), though small effects may not have been detected due to the size of this sample. There was also no evidence of any difference between the machines (Table 4). This part of the study was powered to detect a difference in thermal threshold between machines of 1 SD. This suggests that the initially observed machine difference may have been related differences between an individual operator (G.L.) and the two operators (J.J., M.L.). 4. Discussion Thermal threshold testing is a psychophysical method of assessing the function of the sensory pathways, including small nerve fibres, which are not assessed by traditional nerve conduction studies. The modified Marstock method is increasingly finding a role in the assessment and diagnosis of small fibre neuropathies including those seen in diabetes mellitus, renal failure and human immunodeficiency virus infection, as well as central sensory dysfunction (for example, post-stroke pain) and various pain syndromes (for example, angry backfiring C fibres syndrome, and cold hyperalgesia, cold hypoaesthesia, and cold skin syndrome) [5]. 4.1. Normative data for QST Normative data for thermal threshold testing have been reported previously, but differences in methodology, lack of agespecific values, and graphical rather than numerical presentation make much of these data difficult to apply to a modern clinical setting [1,2,4,6–10]. To our knowledge our study provides the largest single-centre cohort of normative thermal threshold detection data reported by age in years in a format that is easily applied to the clinical setting (Supp. Table 1). These data have been collected on
590
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
Fig. 1. Thresholds for the hand and the foot, showing individual results (diamonds) with the calculated mean ±2.5 standard deviations (solid lines; this is the upper limit of normal for a given age group). A tabulated version of these data is available for clinical application in Supplementary Table 1.
Table 2 Regression equations to calculate the mean temperature difference from 32°C (T) detected as a function of age in years Sensory modality
Site
Regression equation
95% CI (°C per year), p value
Warm
Foot Hand Foot Hand
T = 0.974 + (0.1 age) T = 1.053 + (0.012 age) T = 0.199 + (0.045 age) T = 0.751 + (0.011 age)
0.062–0.14, p < 0.001 0.004–0.02, p = 0.005 0.03–0.06, p < 0.001 0.002–0.02, p = 0.015
Cool
Table 4 Machine comparison study Sensory modality
Body region
Mean machine difference, °C (95% CI)
p value
Warm
Foot Hand Foot Hand
1.29 ( 0.57 ( 0.60 ( 0.19
p = 0.07 p = 0.40 p = 0.10 p = 0.14
Cool
0.11 to 2.70) 1.10 to 2.20) 0.14 to 1.36) ( 0.45 to 0.09)
CI = confidence interval.
CI = confidence interval.
a commercially available machine using standard software. Yarnitsky and colleagues [3] presented similar data in 20 year brackets. However, we have demonstrated that differences due to age are seen over shorter time spans. Yarnitsky’s normal values are similar to ours, except for the cool detection thresholds in the foot, where they report higher thresholds and a sex difference in all age groups. We have demonstrated the clear effect of age on normal thermal threshold detection, which has been shown previously in several studies [4,6,9–11], although this is not a universal finding [12]. The nature of this effect is a little variable between studies, with some modalities more susceptible than others. In general, there is an increase in thermal threshold detection with advancing age. In our study, this was most notable with warm detection in the foot. We have not found a significant effect of sex on thermal threshold detection. This is in keeping with other larger studies [4], but some studies do report a difference in thermal threshold detection between men and women. The effect of sex is variable when
present, with warm detection in the foot [10], cool detection in the foot [3] and cool detection in the hand [9] individually being described. There is a clear difference between body regions. This has been well established by previous studies [9] and has been confirmed by our data with clearly different threshold detection in the hand and the foot. The reasons for this are likely differences in skin innervation in different body regions and may also be influenced by longer distances, which inherently exaggerate reaction time artifact inherent to the method of limits testing. Our study initially suggested that there was a difference in the results derived from different machines. However, this apparent difference in results performed by mutually different operators was refuted when the machine was compared using the same subject and operator. Therefore, any machine difference that existed could be attributed to the difference in operators rather than machine within the power of our latter study. Machine 1 readings appeared to be generally larger than machine 2 except for that of
Table 3 Estimated effect of sex (male versus female) and machine (machine 1 versus machine 2) Sensory modality
Body region
Predictor
Estimated effect (95% CI)
p value
Warm
Foot
Sex Machine Sex Machine Sex Machine Sex Machine
1.06 1.21 0.91 0.69 1.11 1.23 0.86 0.55
p = 0.56 p = 0.08 p = 0.19 p < 0.0001 p = 0.34 p = 0.05 p = 0.05 p < 0.0001
Hand Cool
Foot Hand
CI = confidence interval.
(0.86–1.32) (0.98–1.50) (0.79–1.05) (0.60–0.79) (0.89–1.37) (1.00–1.52) (0.74–1.00) (0.47–0.64)
J. Hafner et al. / Journal of Clinical Neuroscience 22 (2015) 588–591
cool sensation in the hand, but the magnitude of these differences, inconsistent as they may be, can be partly appreciated by a comparison of the degree of change with variability studies that would be required in one subject by one operator on the same machine, a test not undertaken in this study. It is nevertheless advisable that a single machine be used wherever possible and laboratory normative values collected for that device [13]. One of the major challenges in the clinical utility of QST has been the significant variability seen between sessions in the same patient, leading to questions regarding reliability and reproducibility [14]. Fagius and colleagues reported up to 150% difference between sessions using the method of limits [6]. This problem appears less marked with the method of levels [7]. In their comprehensive comparative study, Yarnitsky and colleagues [3] found that thermal threshold detection in the hand using the method of limits had significant intersession differences. Results from the foot and using the more time consuming method of levels were not affected by significant intersession differences. Moravcova and colleagues demonstrated that use of coefficients of repeatability may be helpful in overcoming intra-individual variability; however, consideration must be given to the fact that patients with neuropathy have increased intra-individual variability compared to controls [11,14]. Markedly increased variability is helpful in detecting functional sensory disorders and results that are unreliable due to patient inattention. This variability is greater than that seen in neuropathy [15]. Using cut-off values for variability, as we did in our study, may be helpful in excluding erroneous abnormal results, but this requires further validation. Our study was not designed to detect intra-individual variability and this issue remains one for consideration, particularly when assessing disease progression or therapy responses in a given individual. QST of thermal thresholds is finding an increasing role in the diagnosis and monitoring of small fibre neuropathies given the paucity of other methods. Normative data for thermal threshold detection must be age- and body region-specific, and operators must be trained appropriately and adhere to protocols to ensure reliability. We present, to our knowledge, the largest single-centre study of normative data as a function of age in years in a format that can be easily applied in the clinical setting. Conflicts of Interest/Disclosures J.H. is supported by an unrestricted fellowship grant by Allergan and Ipsen Pharmaceuticals. J.H. and K.N. are supported by the
591
National Health and Medical Research Council (ID GNT1074648 and 512316), the Myositis Association Australia Inc. and the Brain Foundation Research Grants. Acknowledgements The assistance of Dr Christina Liang is appreciated. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jocn.2014.09.017. References [1] Rolke R, Magerl W, Campbell KA, et al. Quantitative sensory testing: a comprehensive protocol for clinical trials. Eur J Pain 2006;10:77–88. [2] Fruhstorfer H, Lindblom U, Schmidt WC. Method for quantitative estimation of thermal thresholds in patients. J Neurol Neurosurg Psychiatry 1976;39:1071–5. [3] Yarnitsky D, Sprecher E. Thermal testing: normative data and repeatability for various test algorithms. J Neurol Sci 1994;125:39–45. [4] Rolke R, Baron R, Maier C, et al. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006;123:231–43. [5] Yarnitsky D. Quantitative sensory testing. Muscle Nerve 1997;20:198–204. [6] Fagius J, Wahren LK. Variability of sensory threshold determination in clinical use. J Neurol Sci 1981;51:11–27. [7] Jamal GA, Hansen S, Weir AI, et al. An improved automated method for the measurement of thermal thresholds. 1. Normal subjects. J Neurol Neurosurg Psychiatry 1985;48:354–60. [8] Gruener G, Dyck PJ. Quantitative sensory testing: methodology, applications, and future directions. J Clin Neurophysiol 1994;11:568–83. [9] Harju EL. Cold and warmth perception mapped for age, gender, and body area. Somatosens Mot Res 2002;19:61–75. [10] Kemler MA, Schouten HJ, Gracely RH. Diagnosing sensory abnormalities with either normal values or values from contralateral skin: comparison of two approaches in complex regional pain syndrome I. Anesthesiology 2000;93:718–27. [11] Bravenboer B, van Dam PS, Hop J, et al. Thermal threshold testing for the assessment of small fibre dysfunction: normal values and reproducibility. Diabet Med 1992;9:546–9. [12] 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. [13] Shy ME, Frohman EM, So YT, et al. Quantitative sensory testing: report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology 2003;60:898–904. [14] Moravcová E, Bednarik J, Svobodnik A, et al. Reproducibility of thermal threshold assessment in small-fibre neuropathy patients. Scr Med (BRNO) 2005;78:177–84. [15] Yarnitsky D, Sprecher E, Tamir A, et al. Variance of sensory threshold measurements: discrimination of feigners from trustworthy performers. J Neurol Sci 1994;125:186–9.