Before this study, a survey was conducted among experts within KAEND and the field of rehabilitation medicine to gauge the interest in utilizing standardized reference standard data. The findings indicated a strong preference for age and sex-specific average and cut-off data. Consequently, this study is dedicated to establishing age and sex-specific reference standards for nerve conduction, with a specific focus on individuals in their 50s—the most commonly tested age group in Korea based on the 2016 Health and Medical Bigdata Open portal—alongside healthy representative adults in their 20s for comparative analysis [26]. The number of subjects was determined based on previous reports, ensuring accurate estimation and representation of 3rd (97th) percentiles and 2 standard deviation values [6,27].

During 2018–2021, 349 healthy volunteers participated in this study after responding to recruitment ads posted at four hospitals or on the website of the clinical trial center. The following were the criteria for inclusion: (1) reside in Korea, (2) are in good health, and (3) are at the ages of 20–39 or 50–69 years old. The following were the criteria for exclusion: (1) patients on pacemakers, (2) diagnosed with diabetes, (3) having hypothyroidism or hyperthyroidism, (4) having cancer or on chemotherapy or external beam radiation, (5) pregnant or planning to become, (6) significantly reduced cognitive function making communication difficult, (7) numbness or pain in all or part of the upper or lower limbs for more than a month in the last three months, (8) diagnosed with psychiatric or neurological disease or on medications, (9) having other medical conditions or taking medications that affect the peripheral nerve function as judged by the researchers.

NCS was conducted from the left and right upper extremities of each of the 349 subjects according to the predetermined protocol. Out of a total of 349 subjects, 324 subjects conducted median motor NCS, and 301 subjects conducted median sensory NCS. Among the 648 median motor nerves that were analyzed, 13 were excluded as outliers. As a result, a total of 635 median motor nerves were included in the statistical analysis and reference standard production for distal motor NCS parameters. However, only 629 nerves were analyzed for proximal motor NCS and nerve conduction velocity (NCV) parameters due to the presence of Martin-Gruber anastomosis (MGA) in 6 nerves. Similarly, among the 602 median sensory nerves analyzed, 18 were excluded as outliers, and a total of 584 were included in the analysis. Outliers are determined by more than 1.5 interquartile range from the 3rd quartile or less than 1.5 interquartile range from the 1st quartile according to the distribution characteristics of the parameters from each of the four subgroups by age and sex. All subjects have read and signed an informed consent form and were rewarded after the examination. The present study protocol was reviewed and approved by the Institutional Review Board of Seoul National University Hospital (No. 1804-125-940, 2104-107-1212), Seoul National University Bundang Hospital (No. B-1809-493-301, B-2106-689-402), SMG-SNU Boramae Medical Center (No. 10-2018-82, 10-2021-76), and National Traffic Injury Rehabilitation Hospital (No. NTRH-20017, NTRH-21008).

The parameters of compound motor action potential (CMAP) were as follows: onset latency (L_onset); baseline to negative peak amplitude (Amp_b-p); negative peak to positive peak amplitude (Amp_p-p); negative spike area (A_neg); negative spike duration (D_neg). The L_onset from CMAP elicited by proximal stimulation was only used in calculating the NCV parameter and is not presented as reference standard. The motor NCV was determined by dividing the distance between the proximal and distal stimulation sites by the time difference between the proximal and distal latencies (Fig. 1A). For the sensory nerve action potential (SNAP), L_onset, peak latency (L_peak), Amp_b-p and Amp_p-p were presented as parameters. The NCV, A_neg, and D_neg were not presented as they were not included in the current reference standard (Fig. 1B).

The NCS data were produced using six NCS instruments (Nicolet EDX^®; Natus) from four hospitals. Each instrument consisted of a base unit, a hand-held stimulator probe, an amplifier, a control panel, PC and an LCD monitor. The NCS instruments, soft tape measures, thermometers, and indoor thermo-hygrometers had been calibrated annually. The frequency filtration range for motor NCS was 10–10,000 Hz, sweep speed was 2 ms/division, and sensitivity was 2–5 mV/division. The frequency filtration range for sensory nerves was 20–2,000 Hz, sweep speed was 1–2 ms/division, and sensitivity was 5–20 μV/division.

The stimulation intensity was determined by progressively raising the intensity until no further amplitude rise was possible. In order to ensure that the stimulation is supramaximal, 10% to 20% of stimulation intensity was added. If maximal stimulation intensity could not be attained, with stimulation duration of 0.1 ms, even at 100 mA, the process was repeated with the stimulation duration of 0.2 ms.

First, the surface temperature of both hands was measured to ensure that it is between 33 and 37 degrees Celsius. If necessary, the use of a water bath, a heating lamp, a hot pack, or a hair dryer was permitted prior to beginning NCS. The skin was cleaned with a swab containing 70% isopropyl alcohol to increase electrode contact and decrease impedance. Although the influence of ambient temperature and humidity on NCSs has not yet been precisely established, these variables were likewise kept within tolerable limits of 15 to 25 degrees Celsius and 40 to 60 percent, respectively.

The posture, stimulation locations, and recording sites for median motor NCS were determined according to the predetermined in-house protocol, which is in accordance with widely accepted methodology in Korea and other countries [1,27,28]. In brief, for median motor NCS, the active electrode was attached to the midpoint of the line between the first metacarpophalangeal joint and the median point of the distal wrist crease, on the most prominent belly of the abductor pollicis brevis muscle. It can be moved to obtain maximum negative deflections to get as close to the motor point as possible. The reference electrode was attached to the interphalangeal joint of the thumb. The ground electrode was attached adjacent to the palm or the dorsum of hand. The cathode of the stimulator was placed 8 cm proximal to the active electrode along the following path: (1) from the active electrode to the midpoint of the distal volar wrist crease and (2) upward along the midline between the flexor carpi radialis and palmaris longus tendons. Proximal stimulation was just medial to the brachial artery pulsation on the antecubital fossa (Fig. 2A). The MGA was defined as a nerve where Amp_p-p is 0.3 greater than Amp_b-p. Considering that MGA is an anatomical variant occurring in the forearm region, the upper limbs in which MGA was confirmed were only excluded from the analysis of proximal motor NCS parameters and NCV.

For median sensory NCS, the active electrode was attached to the skin over the second proximal phalanx. The reference electrode was attached at least 4cm distal to the active electrode. The ground electrode was attached adjacent to the active electrode between the active electrode and the cathode of the stimulator or on the hand dorsum. The cathode of the stimulator was placed 14 cm proximal to the active electrode between the flexor carpi radialis and palmaris longus tendons (Fig. 2B).

The evaluation of uncertainty begins with the definition of measurand. In this study, the measurands included all NCS parameters which mentioned in the previous section. The process of evaluating uncertainty is carried out according to the Guide to the expression of uncertainty in measurement (GUM), which is the international standard for uncertainty evaluation [29]. The evaluation of uncertainty followed the steps outlined in the GUM, including the expression of model equation; investigation of uncertainty components and standard uncertainty for each component; calculation of combined standard uncertainty and expanded uncertainty. The entire procedure for evaluating uncertainty is publicly available in Korean at the Data Center for Korean Reference Nerve Conductions website (https://emgreference.org/43) and we attached additional supplement materials with an English translation for international readers (Supplementary Material S1). Additionally, this approach is akin to that employed in the paper on reference standard data for the other nerves, conducted parallelly during 2018–2019 [24,25].

Anthropometric data were summarized by age and sex, then compared using the student-t test. To assess the impact of age and sex on NCS parameters, both univariable and multivariable linear mixed models (LMMs) were employed to effectively address the data gathered from two separate instances within both extremities. The univariable LMM was adjusted only for the side (left or right), while the multivariable LMM accounted for height, weight, soft lean mass, body fat mass, and skeletal muscle mass. Additionally, the effect of stimulation intensity on NCS parameters was also analyzed using a multivariable LMM that accounted for the aforementioned anthropometric factors, as well as sex and age. Normality was confirmed in each model through residual plots, with no models failing to meet this criterion. Statistical significance was established at p-value<0.05. All statistical analyses were conducted using R statistical software (version 3.5.3; R Foundation for Statistical Computing).

The anthropometric data, detailed in Table 1, reveal significant differences in measured parameters based on sex within the same age group or age within the same sex group. Specifically, obesity rates, defined as a body mass index exceeding 25, demonstrate notable disparities. In the 20s–30s age group, 51.0% of male and 6.8% of female are classified as obese. In the 50s–60s age group, these rates shift to 46.2% for male and 27.2% for female. Notably, only one subject required an increased stimulation duration of 0.2 ms for supramaximal stimulation.

Tables 2 and 3 comprehensively summarize the NCS parameters, with Table 2 summarizes the upper limits of latency parameters and Table 3 summarizes the lower limits of the rest parameters. All parameters are presented as mean with expanded uncertainty and mean with two standard deviations.

Due to the vast amount of data, this section will present the cut-off values with expanded uncertainty on L_onset and NCV which hold clinical significance in the diagnosis of carpal tunnel syndrome. In case of L_onset of SNAP, the cut-off value was 3.13 ms for male in their 20s–30s and 3.45 ms for those in their 50s–60s. For female, these values were 3.20 and 3.33 ms for the respective age groups. Regarding CMAP, the cut-off values were 4.20 ms for male in their 20s–30s and 4.57 ms for those in their 50s–60s. For female, these values were 4.06 and 4.41 ms for the respective age groups. As for the cut-off values for NCV, the values were 46.34 and 43.71 m/s for male in their 20s–30s and 50s–60s, respectively. For female, these values were 43.97 and 41.85 m/s for the respective age groups.

In the CMAP parameters, such as L_onset, Amp_b-p, Amp_p-p, A_neg, and NCV showed significant age effects in both univariable and multivariable analyses, with p-values less than 0.001. However, D_neg did not exhibit a significant age effect in either analysis. In the SNAP parameters, L_onset, L_peak, Amp_b-p, and Amp_p-p showed significant age effects in both analyses. In particular, SNAP amplitude parameters exhibited significant effect sizes, presenting -18.49 (p<0.001) for Amp_b-p and -27.86 (p<0.001) for Amp_p-p.

Regarding the effect of sex on all NCS parameters, there were no parameters that showed the sex effect, except for A_neg and D_neg, which were observed as significant in both types of analyses. In case of L_onset of CMAP and Amp_b-p and Amp_p-p of SNAP, they displayed statistical significance in the univariable analysis. However, these effects did not persist in the multivariable analysis once anthropometric factors were controlled for. These findings are summarized in Table 4.

When exploring the effects of stimulation intensity on CMAP parameters, L_onset showed a negative effect size of -0.002. Amp_b-p demonstrated a effect in both distal and proximal CMAPs with effect sizes of -0.011 and -0.018, respectively. D_neg also showed significant differences with effect sizes of 0.004 in distal and 0.003 in proximal CMAP. These parameters show p-values indicating statistical significance. However, Amp_p-p in the distal CMAP did not reach statistical significance (-0.015, p=0.077), contrasting with its significant effect in the proximal CMAP (-0.028, p<0.001). In the SNAP parameters, L_onset showed a significant effect size of -0.003 (p=0.004), and L_peak was also significant with an effect size of -0.003 (p=0.032).

The remaining parameters did not show significant differences, and the results are summarized in Table 5.

In this study, we produced the NCS dataset comprising 648 median motor nerves and 602 sensory nerves, along with 14 NCS parameters and 6 anthropometric factors, making it the largest normative dataset currently available. A few previous studies in Korea have presented normative data for the median nerve, however they were subject to certain limitations.

For example, in 1982, Hahn and Chang [30] reported normative data on distal latency, Amp_b-p, and conduction velocity for both motor and sensory median nerves. However, their study was limited by a small sample size of only 120 subjects, with some subgroups containing as few as 10 subjects. Later, in 1998, Lee et al. [31] reported the terminal latency and amplitudes of median motor and sensory nerves in 242 normal Korean subjects. Yet, their study did not provide normative data segmented by age and sex, nor did it examine the effects of these factors. In 2002, a meta-analysis conducted by Lee and Ra [32] reviewed data from 36 articles, with subject populations ranging from 296 to 500. However, only one of these articles focused on the Korean population, which was the study conducted by Hahn and Chang [30]. Our study addresses these gaps by providing a more extensive and detailed dataset, featuring over 100 nerves in each groups, categorized by sex and age. This extensive dataset could enriches the understanding of median nerve characteristics in the Korean population.

Furthermore, this study not only establishes the largest reference standard for median NCS but also introduces the concept of “expanded uncertainty,” a value commonly used in other medical fields [17-23], yet novel in the context of NCS. Previous attempts by physicians, researchers, and electrodiagnostic societies to create normative data involved skilled practitioners and standardized procedures [6-15]. However, achieving “metrological traceability” remains. Metrological traceability can be achieved by developing measurement standards that include measurement uncertainty and by establishing an unbroken calibration chain to the international unit system through collaboration with the National Metrology Institute in Korea. For instance, we have developed a standard signal generator that mimics neural waveforms, used for calibrating electrodiagnosis machines. This signal generator undergoes yearly calibration by a nationally certified institution and is accompanied by calibration certificates. In our pursuit of robust NCS reference data, we calculated the expanded uncertainty by considering a wide range of comprehensive uncertainty components. These components were meticulously integrated and presented in a formula that encompasses each standard uncertainty. Considering these various potential variabilities could enhance reliability across different laboratories. Electrodiagnostic laboratories not only refer to these standards for clinical tests but also have the flexibility to adapt this reference data to various laboratory environments by adjusting the standard uncertainty with their specific uncertainty components from our openly disclosed protocol. Consequently, reference data presented with expanded uncertainty becomes versatile, making it applicable to diverse situations. To achieve this goal, more hospitals must actively participate, labs must become proficient in evaluating nerves with uncertainties, and international collaboration is necessary to establish international reference standards.

Our study was conducted on adult populations with two age groups: those in their 20–30’s and those in their 50–60s. The latter group is commonly tested for NCS in Korea, and we sought to examine the impact of age by comparing them with the younger group. Thus, the age effects reported in our study are based on the comparison between these distinct adult groups. Although direct comparisons with studies from other age populations are limited, discovering similar trends could still hold significant implications. Our data could be compared with two previous studies, due to the appropriateness of the study populations and the methodological completeness. The first study, conducted by Hahn and Chang [30], was the only one that attempted to produce the reference of median NCS for a healthy Korean population. The second study, conducted by Buschbacher [27], was the only study to meet the NDTF criteria set by AANEM for the median motor nerve. Hahn and Chang [30] compared NCS parameters of individuals in their 10–40s with those in their 50–60s, while Buschbacher [27] compared individuals in their 20–40s with those in their 50–70s.

In our study, the effect of aging on all NCS parameters, except D_neg, was significant in the univariable analysis. Upon further multivariable analysis, all parameters remained statistically significant, but the effect sizes of CMAP parameters were slightly reduced, and the effect sizes of SNAP parameters were slightly increased.

At first, all latency parameters and NCV had an aging effects that delayed and slowing down. Previously, NORRIS et al. [33] have explained the aging changes of conduction velocity in terms of changes in the endoneurial permeability because of vascular changes and local ischemia, WAGMAN and LESSE [34] have explained it in terms of changes in metabolism and circulation. The previous Korean study showed a similar trend but reported sensory NCV, which was not included in our study and was not statistically significant. It is possibly due to the smaller sample size [30].

All amplitudes parameters and A_neg are larger in younger group. This age effect on Amp_b-p are previously reported and appear to be relevant [15]. However, reports regarding Amp_p-p and A_neg are scarce. The previous Korean data, which only reported statistically significant aging effects on the Amp_b-p of SNAP, contrasts with our study [30]. Nevertheless, our findings, with its sufficient sample size and inclusion of various covariates, demonstrated aging effects on all amplitude parameters, corroborating with Buschbacher's [27] data.

The absence of an age effect in D_neg suggests a relatively consistent aging process in nerve fibers, because inhomogeneous aging in either the fastest or slowest fibers could conceivably result in a decrease or increase in D_neg. Additionally, the observed decrease in amplitudes with aging suggests that it is primarily attributable to a reduction in the source of action potentials, rather than dispersion of conduction duration. Previous findings revealed that aging, including in adults ranging from their 40s to 70s, had no impact on D_neg, further reinforcing the idea of a homogeneous aging process in nerves post-adulthood [27,30]. Conversely, some studies have shown that children have shorter durations compared to adults, which suggests that immature nerves may conduct electrical signals more synchronously [30,35].

In the previous studies, sex effects have been inconsistent. Notably, when considering anthropometric covariates comprehensively, the apparent effect of sex tends to vanished [3,36]. There have been reports of minor morphological and certain electrophysiological variations based on sex [5,37,38]. A particular study on cadavers focusing on sexual dimorphism in morphometric characteristics found no significant differences in the total count, average cross-sectional area, and circularity ratio of myelinated axons [37]. Gøransson et al. [38] noted a higher density of epidermal nerve fibers in females, while Meh and Denišlič [39] observed a heightened sensitivity to minor temperature shifts in females. Some researchers have hypothesized that these differences might be attributable to environmental factors, such as higher rates of alcohol consumption, occupational neurotoxic exposure, and physical trauma, which are more common in males [5,38]. In our study, all six examined anthropometric parameters showed substantial sex-based differences, necessitating multivariate analysis in assessing the impact of sex on NCS data, to adequately account for physical disparities between the sexes.

Initially, multivariable analysis, accounting for anthropometric factors, revealed no sex differences in all latency parameters of motor and sensory nerves. This suggests that median nerve fibers transmit electrical signals at equivalent speeds in both sexes, given similar physical traits. There was no sex effect in NCV, as it is calculated based on latency. These findings align with previous Korean data, however, contrast with the study by Buschbacher [27], which reported sex differences in L_onset and NCV [30]. This inconsistency could stem from different racial backgrounds of subjects. However, Buschbaher's [27] study only considered height as a covariate, overlooking other sex-specific anthropometric variables like lean muscle mass, body fat mass, and appendicular muscle mass.

Among all amplitude parameters, only SNAP amplitudes were significantly larger in female in univariable analysis. However, this significance vanished after adjusting for anthropometric factors in multivariable analysis. A_neg was significantly small in females. Although A_neg decreases in both females and the aging group, the result in females is due to shorter durations without a difference in amplitude, while in the aging group, it is due to smaller amplitudes without a difference in duration, presenting a contrasting appearance. In case of D_neg, it was the only parameter displaying no aging effect but a significant sex effect, even after adjusting for anthropometric factors. This implies that besides the physique differences, other neuromuscular physiologic factors such as dimorphic neurotransmitter release at the neuromuscular junction [40] might contribute to the observed disparities between sexes.

In summary, latencies and NCV were faster, and amplitudes and A_neg were higher at the subjects in their 20–30’s. The smaller A_neg and shorter D_neg of female subjects show the statistical significance but the difference of other parameters were vanished after adjusting for anthropometric factors.

In our previous study, we firstly describe supramaximal stimulus intensity on Asian populations, and stimulation intensity show correlation with body fat mass and faster L_onset [25]. The present study also found statistical significance between stimulation intensity and NCS parameters such as all latency parameters, all duration parameters, and CMAP amplitude parameters, although the effect sizes were small. A_neg were same on both distal and proximal stimulations, so that whole median nerve fibers are depolarized and supramaximal stimulus are appropriately achieved. Interestingly, at higher stimulation intensities, durations were longer and amplitudes were smaller. This could be due to the requirement for higher intensity to achieve supramaximal stimulation, indicating the presence of more conducting soft tissue between the stimulator and the median nerve, leading to slight but definite unsynchronized excitation. Furthermore, even after achieving supramaximal stimulation, higher intensity shortened all latency parameters. Nevertheless, the effect size is notably small, amounting to less than one percent compare to age effects, suggesting that the clinical significance may be limited. Therefore, while these results are intriguing, further comprehensive research is necessary to fully understand their clinical implications.

There are several dominant uncertainty components. The most dominant source of uncertainty came from the “difference between subjects including repeated observation on a subject.” For example, the expanded uncertainties of distal motor L_onset ranged from 0.89 to 1.03 according to groups, and the uncertainty components from “difference between subjects” accounted for 45.04% to 50.03%.

Meanwhile, the second most dominant source of uncertainty, the “selection of points on the waveform,” accounted for 23.93% to 26.32%. Notably, in NCV parameter where these components are calculated twice, the expanded uncertainty becomes enlarged compared to other parameters. This outcome could be an inherent aspect of the current practice, where practitioners manually identify the inflection points of each L_onset. Factors such as using a consistent scale of sensitivity and sweep speed, as well as ensuring similar skill levels among practitioners could help mitigate these uncertainty components.

The uncertainty components of “measurements from skin temperature” accounted for 12.39%–13.63%. To address this aspect, the conversion coefficients were adopted to convert surface temperature and distance into milliseconds, as reference in existing literatures [1,4]. For example, Kouyoumdjian et al.’s [4] study stated that a 1°C difference in skin temperature corresponds to a 0.2-millisecond change in motor latency. To aid in comprehension, the uncertainty components of input quantities for each measurand of the median motor nerve are presented as an example (Supplementary Table S1).

This study faces several limitations. Firstly, it examines non-continuous adult age groups, omitting data for individuals in their 40s, those over 70, and those under 20. However, through preliminary expert surveys, the study primarily presented data comparing standards for individuals aged 50 to 60, where median nerve conduction studies are most prevalent, to those in the 20 to 30 age group. Secondly, the actual distance and composition of the conducting soft tissue between the stimulator and the median nerve are not directly measured. However, this component was at least indirectly reflected in the differences between subjects, and while direct measurement using imaging technique such as ultrasonography is feasible, it is time-intensive and thus hinders clinical application. In the future, such detailed measurement standards might be considered for specialized research situations requiring precise data. Lastly, employing a single device type suggests the potential for variability in uncertainty components when using devices from other manufacturers. Nevertheless, the overall resemblance in the setup of NCS machines implies that these disparities may be insignificant. If necessary, the expanded uncertainties using other NCS devices could be determined by substituting their specific uncertainty components and standard uncertainties, given that reference standard have traceability. Future research could expand to include NCS devices from multiple manufacturers. Additionally, it is feasible to develop a model formula that accounts for the uncertainties of each separate element of NCS machines.

In conclusion, this study produced the measurement and reference standards for median NCS in the Korean population, considering appropriate uncertainty components. It also revealed effects of age, sex and stimulation intensity among Korean population beside physique difference from the produced reference standard data. All measurement tools are connected to the calibration chain, the protocol is reproducible, and the data is traceable to the international measurement standard. The resulting data has been approved as the national standard reference data of Korea, and it is expected to improve reliability among laboratories, decrease the need for redundant re-tests, reduce medical costs, and be utilized in various clinical situations.