Simplified electrophysiological approach combining a point‐of‐care nerve conduction device and an electrocardiogram produces an accurate diagnosis of diabetic polyneuropathy

Abstract

Aims/Introduction

This study aimed to investigate the diagnostic potential of two simplified tests, a point‐of‐care nerve conduction device (DPNCheck™) and a coefficient of variation of R‐R intervals (CV_R‐R), as an alternative to traditional nerve conduction studies for the diagnosis of diabetic polyneuropathy (DPN) in patients with diabetes.

Materials and Methods

Inpatients with type 1 or type 2 diabetes (n = 167) were enrolled. The study population consisted of 101 men, with a mean age of 60.8 ± 14.8 years. DPN severity was assessed using traditional nerve conduction studies, and differentiated based on Baba's classification (BC). To examine the explanatory potential of variables in DPNCheck™ and CV_R‐R regarding the severity of DPN according to BC, a multiple regression analysis was carried out, followed by a receiver operating characteristic analysis.

Results

Based on BC, 61 participants (36.5% of the total) were categorized as having DPN severity of stage 2 or more. The multiple regression analysis yielded a predictive formula with high predictive power for DPN diagnosis (estimated severity of DPN in BC = 2.258 – 0.026 × nerve conduction velocity [m/s] – 0.594 × ln[sensory nerve action potential amplitude (μV)] + 0.528In[age(years)] – 0.178 × ln[CV_R‐R], r = 0.657). The area under the curve in receiver operating characteristic analysis was 0.880. Using the optimal cutoff value for DPN with severer than stage 2, the predictive formula showed good diagnostic efficacy: sensitivity of 83.6%, specificity of 79.2%, positive predictive value of 51.7% and negative predictive value of 76.1%.

Conclusions

These findings suggest that DPN diagnosis using DPNCheck™ and CV_R‐R could improve diagnostic efficiency and accessibility for DPN assessment in patients with diabetes.

Graphical abstract Text

The gold standard electromyography system diagnosis of diabetic polyneuropathy can be replicated using two simple quantitative tests: DPNCheck™ and coefficient of variation of R‐R intervals. By combining these tests, we have developed an estimation formula with excellent diagnostic performance. The use of DPNCheck and electrocardiogram would simplify the diagnosis of diabetic polyneuropathy, making it more accessible, reproducible and reliable.

INTRODUCTION

Diabetic polyneuropathy (DPN) is the most common neuropathy in developed countries, affecting approximately 50% of diabetes patients ¹ , ² , ³ , ⁴ . Patients with diabetes present with several clinical presentations, including distal symmetric polyneuropathy, autonomic neuropathy and mononeuropathy ⁴ , ⁵ . In this report, we conceptualize DPN as a combined pathophysiological condition involving both distal sensorimotor polyneuropathy and autonomic neuropathy. Patients with DPN often suffer from infections, ulcers and amputations of the lower extremities. The lifetime incidence of foot ulcers in DPN patients is 34%, accounting for a large proportion of non‐traumatic amputations ⁶ , ⁷ . The risk of lower extremity amputation is 10‐fold higher in patients with diabetes ⁸ . The 5‐year survival rate for above‐knee amputees with diabetes was 31% (51% in non‐diabetic patients) ⁹ . The Toronto Consensus, neuropathy disability score and nerve conduction study (NCS) are used to diagnose DPN. Although the Toronto Consensus and neuropathy disability score are widely used, they include subjective assessments, which make it difficult to evaluate objectively and reproducibly. Traditionally, as the gold standard diagnostic criteria, DPN is diagnosed with two or more impaired nerves proven by NCS. However, it is prohibitive to examine NCS at many facilities. because it requires skilled technicians to carry out the test. DPNCheck™, a device for NCS of the sural nerve, requires no specific skill to examine. The validity of the device has been reported by Kamiya et al ¹⁰ .

Cardiovascular autonomic neuropathy (CAN), which is part of DPN, causes higher mortality in patients with diabetes ¹¹ . CAN is assessed for cardiovascular risk stratification of patients. Various CAN evaluation methods have been proposed: the standard deviation of the R‐R interval, coefficient of variation of R‐R intervals (CV_R‐R), low‐frequency spectra/high‐frequency spectra, percent of difference between adjacent normal R‐R intervals >50 ms, root mean square successive difference and Cardiovascular Autonomic Reflex Test ¹² , ¹³ , ¹⁴ . Among them, CV_R‐R is commonly utilized to evaluate autonomic function in Japan.

Previously, we verified that DPNCheck™ can effectively diagnose moderate to severe DPN ¹⁰ . In the present study, we hypothesized that the combination of CV_R‐R with DPNCheck™ might further improve the diagnostic performance of DPN. Here, we investigated whether the combination could reproduce the diagnosis of DPN using the conventional NCS.

MATERIALS AND METHODS

Eligibility criteria

A total of 167 participants with diabetes, who were admitted at Aichi Medical University Hospital in Nagakute, Japan, to improve their hyperglycemia from August 2014 to October 2019, were included. This was a retrospective cohort study using electronic health records. Presumed consent called opt‐out consent using the website of the hospital was applied to the participants. They were provided the right of withdrawal from the study at any time. Patients were excluded if they had a history of other causes of peripheral neuropathy; with body mass index >35; no sural nerve action potentials elicited by DPNCheck™, including patients who have undergone lower limb amputation; or had diabetic ketoacidosis, severe infection or severe injuries. Study procedures were approved by the ethics committee of Aichi Medical University Hospital (approval number: 2019‐133).

NCS

To assess the sensory nerve conduction velocity and amplitude of sensory nerve action potential (SNAP) in the sural nerves, bilateral sural nerves were studied using both the DPNCheck™ device (NeuroMetrix Inc., Waltham, MA, USA) and a conventional electromyography system (EMGS; Neuropack X1, MEB‐2312; Nihon Kohden, Tokyo, Japan), as previously documented ¹⁰ . The conventional EMGS was also utilized to carry out NCS on bilateral tibial nerves. The peak‐to‐peak amplitudes were used to measure the compound muscle action potential (CMAP) of the tibial nerve and the SNAP of the sural nerve. In the process of data interpretation, the NCS data acquired through the standard EMGS were utilized to evaluate the severity of DPN ¹⁴ . This evaluation involved using a classification method known as Baba's classification (BC), which categorized the participants into five stages, as outlined below:

• Stage 0: Participants with normal NCS findings and no abnormalities.

• Stage 1: Participants with mild neuropathy, characterized by the presence of any delay in tibial motor nerve conduction velocity (<40 m/s), sural sensory nerve conduction velocity (<40 m/s), tibial minimal F‐wave latency (>{12.8 + 0.22 × Height (cm)} ms) or the presence of A wave.

• Stage 2: Participants with moderate neuropathy, marked by a decrease in sural SNAP amplitude to <5 μV.

• Stage 3: Participants with neuropathy ranging from moderate to severe, manifested by a decrease in sural SNAP amplitude to <5 μV and a decrease in tibial CMAP amplitude ranging from ≥2 to <5 mV.

• Stage 4: Participants with severe neuropathy, indicated by a decrease in sural SNAP amplitude to <5 μV, and a decrease in tibial CMAP amplitude to <2 mV.

In the present study, we opted to exclude the consideration of the “presence of A wave” item due to insufficient agreement among raters.

For the analysis of diagnostic accuracy, participants were classified into two groups; participants with stages 2–4 in BC were classified as DPN, and participants with stages 0 or 1 were classified as non‐DPN.

The CV_R‐R

The measurement of CV_R‐R was carried out using established techniques, as reported previously ¹⁵ . To evaluate the CV_R‐R, electrocardiogram recordings were obtained from participants in the supine position after a 5‐min period of bed rest. A 1‐min resting electrocardiogram recording was initially obtained, followed by an additional 1‐min recording during deep breathing at a rate of six breaths per minute. The CV_R‐R was calculated using the formula: CV_R‐R (%) = (standard deviation of R‐R intervals) / (mean R‐R intervals) × 100.

Statistical analysis

During data analysis, the data were analyzed using lower values of SNAP amplitudes, CMAP amplitudes and NCVs, or higher values of minimal F‐wave latency in the bilateral nerve responses. The statistical software used for the analysis was SPSS Statistics version 20 for Windows, provided by IBM SPSS (Chicago, IL, USA). The analyses were carried out by personnel who were unaware of the identities of the participants. To examine differences in continuous and categorical variables, Student's t‐tests and χ²‐tests with Yates' correction were used, respectively. Spearman's correlation coefficients were utilized to analyze correlations. The neurological parameters that showed significant correlations with the severity in BC were included in multiple regression models to develop an effective prediction model of DPN. The diagnostic validity was assessed by constructing a receiver operating characteristic (ROC) curve and calculating the area under the ROC curve. The optimal threshold was determined using the highest Youden Index, which is calculated as (sensitivity + specificity) − 1.

RESULTS

Clinical characteristics

The clinical characteristics of the participants are presented in Table 1. In total, 167 persons were included in the analyses (101 men, 66 women; aged 60.8 ± 14.8 years). Based on BC, 36.5% of the participants (n = 61) were classified as DPN (stage 2–4). Although the mean age and duration of diabetes significantly increased in the DPN participants, glycosylated hemoglobin and body mass index showed no significant change compared with non‐DPN participants. The parameters of DPN – that is, tibial CMAP, and sural NCV and SNAP amplitude using EMGS; sural NCV and SNAP amplitude using DPNCheck™ – showed development of dysfunction in the peripheral nervous system of the DPN participants. The correlation coefficient between SNAP amplitude using EMGS and DPNCheck™ was 0.690 (P < 0.001). The CV_R‐R recorded during deep breathing also decreased in the DPN participants. The parameters of diabetic nephropathy, estimated glomerular filtration rate and urinary albumin‐to‐creatinine ratio also showed significant deterioration in the DPN group.

Table 1.

Clinical characteristics

	Total	Non‐DPN	DPN	P‐value
N	167	106	61
Age (years)	60.9 ± 14.8	58.6 ± 15.2	64.9 ± 13.3	<0.01**
Male (%)	63.5	61.3	60.7
Duration of diabetes (years)	9.9 ± 11.1	7.5 ± 9.1	14.1 ± 12.8	0.001**
Body mass index (kg/m2)	23.8 ± 4.3	23.9 ± 4.2	23.5 ± 4.4	0.490
Systolic blood pressure (mmHg)	128.0 ± 18.5	125.1 ± 14.9	132.3 ± 22.2	0.06
Diastolic blood pressure (mmHg)	76.6 ± 10.7	75.7 ± 9.3	78.6 ± 13.2	0.272
Casual blood Glucose (mg/dL)	197.5 ± 148.2	192.2 ± 166.3	206.5 ± 112.6	0.577
Glycosylated hemoglobin (%)	10.1 ± 2.4	10.3 ± 2.6	9.9 ± 2.0	0.171
Glycoalbumin (%)	28.6 ± 10.2	28.3 ± 10.6	29.1 ± 9.5	0.652
Creatinine (mg/dL)	0.85 ± 1.11	0.84 ± 1.28	0.87 ± 0.78	0.871
eGFR (mL/min/1.73m2)	85.1 ± 34.4	88.5 ± 33.2	79.7 ± 35.8	0.126
uACR (mg/g)	115.4 ± 680.1	46.3 ± 166.8	232.7 ± 1092.3	<0.01* , **
Total cholesterol (mg/dL)	190.5 ± 50.8	195.8 ± 54.1	178.9 ± 41.2	0.146
High‐density lipoprotein (mg/dL)	49.0 ± 19.6	48.2 ± 19.7	50.7 ± 19.5	0.572
Low‐density lipoprotein (mg/dL)	108.2 ± 44.5	113.1 ± 38.2	99.1 ± 53.4	0.07
Triglyceride (mg/dL)	154.0 ± 136.9	161.9 ± 154.9	137.5 ± 87.8	0.431
Mean IMT (mm)	1.04 ± 0.43	0.99 ± 0.43	1.11 ± 0.44	0.106
Ankle‐Brachial Index	1.13 ± 0.09	1.13 ± 0.08	1.15 ± 0.11	0.188
Toe‐Brachial Index	0.76 ± 0.14	0.78 ± 0.15	0.73 ± 0.13	0.06
baPWV (cm/s)	1,616 ± 373	1,554 ± 352	1747 ± 377	0.001**
CVR‐R, resting (%)	2.77 ± 1.87	2.83 ± 1.68	2.66 ± 2.16	0.580
CVR‐R, deep breathing (%)	4.86 ± 4.28	5.48 ± 4.95	3.78 ± 2.40	0.01**
QTc, Bazett (ms)	0.42 ± 0.04	0.42 ± 0.02	0.43 ± 0.02	0.047**
QTc, Fridericia (ms)	0.41 ± 0.05	0.41 ± 0.02	0.43 ± 0.09	0.07
Tibial nerve
NCV (m/s)	42.3 ± 3.2	43.2 ± 2.9	40.9 ± 3.0	<0.001**
CMAP amplitude (mV)	16.3 ± 7.4	17.7 ± 7.1	14.0 ± 7.3	0.002**
F wave latency (ms)	48.5 ± 4.7	48.72 ± 5.0	48.2 ± 4.3	0.592
Sural nerve, EMGS
NCV (m/s)	46.3 ± 4.8	47.5 ± 4.3	44.2 ± 4.9	<0.001**
SNAP amplitude (μV)	9.0 ± 5.8	11.9 ± 5.3	4.0 ± 2.3	<0.001**
Sural nerve, DPNCheck™
NCV (m/s)	49.9 ± 6.6	51.8 ± 5.3	46.5 ± 7.4	<0.001**
SNAP amplitude (μV)	14.7 ± 8.7	18.1 ± 8.5	8.8 ± 5.4	<0.001**
Stages of DPN based on Baba's classification		Stage 0: 31 Stage 1: 75	Stage 2: 53 Stage 3: 7 Stage 4: 1

^*Significant in logarithm conversion. P‐value: between patients with and without diabetic polyneuropathy.

^**Significant P‐values. baPWV, brachial‐ankle pulse wave velocity; CMAP, compound muscle action potential; CV_R‐R, coefficient of variation of R‐R intervals; DPN, diabetic polyneuropathy; EMGS, electromyography system; eGFR, estimated glomerular filtration rate; IMT, intima‐media thickness; NCV, nerve conduction velocity; QTc, corrected QT interval; SNAP, sensory nerve action potential; uACR, urine albumin‐to‐creatinine ratio.

Correlation of DPN severity and clinical parameters

The correlations between each clinical parameter and the severity of DPN in BC were analyzed. DPN progression showed significant correlations with the following parameters: age, duration of diabetes, systolic blood pressure, creatinine, estimated glomerular filtration rate, urinary albumin‐to‐creatinine ratio, low‐density lipoprotein, intima‐media thickness, ankle‐brachial index and brachial‐ankle pulse wave velocity. In the neurological parameters, DPN progression had significant correlations with the following parameters: deep breathing CV_R‐R, and sural NCV and SNAP amplitude using DPNCheck™ (Table 2).

Table 2.

Correlation of diabetic polyneuropathy severity and clinical parameters

	Correlation coefficient	P‐value
Age (years)	0.300	<0.001*
Duration of diabetes (years)	0.378	<0.001*
Body mass index (kg/m2)	−0.071	0.364
Systolic blood pressure (mmHg)	0.238	0.012*
Diastolic blood pressure (mmHg)	0.123	0.297
Casual blood glucose (mg/dL)	0.056	0.505
Glycosylated hemoglobin (%)	−0.072	0.371
Glycoalbumin (%)	0.048	0.567
Creatinine (mg/dL)	0.169	0.042*
eGFR (mL/min/1.73m2)	−0.204	0.011*
uACR (mg/g)	0.210	0.012*
Total cholesterol (mg/dL)	−0.126	0.238
High‐density lipoprotein (mg/dL)	0.091	0.372
Low‐density lipoprotein (mg/dL)	−0.202	0.014*
Triglyceride (mg/dL)	−0.102	0.338
Mean IMT (mm)	0.281	0.001*
Ankle‐brachial index	0.196	0.015*
Toe‐brachial index	−0.159	0.051
baPWV (cm/s)	0.380	<0.001*
CVR‐R, resting (%)	−0.184	0.017*
CVR‐R, deep breathing (%)	−0.335	<0.001*
QTc, Bazett	0.154	0.051
QTc, Fridericia	0.122	0.207
Tibial nerve
NCV (m/s)	−0.402	<0.001*
CMAP amplitude (mV)	−0.381	<0.001*
F wave latency (ms)	0.040	0.604
Sural nerve, EMGS
NCV (m/s)	−0.297	<0.001*
SNAP amplitude (μV)	−0.708	<0.001*
Sural nerve, DPNCheck™
NCV (m/s)	−0.360	<0.001*
SNAP amplitude (μV)	−0.587	<0.001*

^*Significant P‐values. baPWV, brachial‐ankle pulse wave velocity; CMAP, compound muscle action potential; CV_R‐R, coefficient of variation of R‐R intervals; DPN, diabetic polyneuropathy; EMGS, electromyography system; eGFR, estimated glomerular filtration rate; IMT, intima‐media thickness; NCV, nerve conduction velocity; QTc, corrected QT interval; SNAP, sensory nerve action potential; uACR, urine albumin‐to‐creatinine ratio.

Estimation of the severity of DPN using DPNCheck™ and CV_R‐R

We examined whether DPNCheck™ and CV_R‐R assists a diagnosis of DPN. By using both univariate regression analysis and ROC analysis, critical cutoff values were determined: CV_R‐R at 4.05%, sensory nerve conduction velocity (evaluated using DPNCheck™) at 49.5 m/s, and SNAP (evaluated using DPNCheck™) at 12.5 μV. For the multiple regression analysis, the neurological parameters, which significantly correlated with the severity of DPN, were selected as independent variables. In the multiple regression analysis, using the stage numbers of BC as the dependent variable, and the values transformed by the natural logarithm of age, the values transformed by the natural logarithm of sural SNAP amplitude using DPNCheck™, the sural NCV using DPNCheck™ and the values transformed by the natural logarithm of deep breathing CV_R‐R as independent variables, the estimated severity of DPN in BC (eBC) was obtained as follows: eBC = 2.258 – 0.026 × NCV (m/s) – 0.594 × ln(SNAP amplitude [μV]) + 0.528ln(age[years]) – 0.178*ln(CV_R‐R), r = 0.657 (Table 3). ROC analysis to classify DPN showed that the eBC had a moderate discriminative power to predict DPN (area under the ROC curve = 0.880; Figure 1). The eBC cutoff value yielding the highest accuracy was determined to be 1.26. The diagnostic performance demonstrated favorable results, with a sensitivity of 83.6%, specificity 79.2%, positive predictive value 69.9.7%, negative predictive value 89.4%, positive likelihood ratio 4.03 and negative likelihood ratio 0.21. These diagnostic accuracies are comparable to those of the sural SNAP amplitude using DPNCheck™ or the combination of sural SNAP amplitude and NCV using DPNCheck™, and superior to the sural NCV using DPNCheck™ or deep breathing CV_R‐R (Table S1).

Table 3.

Univariate and multivariate regression analysis

	Univariate			Multivariate
	Coefficients	P‐value	Standardized coefficients	Coefficients	P‐value	Standardized coefficients
(Constant)				2.228	0.14
ln (age)	0.931	<0.001	0.322	0.528	0.006	0.186
NCV, DPNCheck™	−0.049	<0.001	−0.372	−0.026	0.005	−0.197
ln (SNAP amplitude, DPNCheck™)	−0.825	<0.001	−0.590	−0.594	<0.001	−0.426
ln (CVR‐R, deep breathing)	−0.429	<0.001	−0.344	−0.178	0.038	−0.142

CV_R‐R, coefficient of variation of R‐R intervals; ln, natural logarithm; NCV, nerve conduction velocity; SNAP, sensory nerve action potential.

Figure 1.

The receiver operating characteristic curve. The receiver operating characteristic curve shows the diagnostic ability for the severity of diabetic polyneuropathy using the estimated formula retrieved by the multiple regression analysis. The area under the receiver operating characteristic curve was 0.880.

DISCUSSION

We attempted to replicate the diagnosis for DPN based on the gold standard system EMGS using two simple quantitative tests, DPNCheck™ and CV_R‐R. These two tests can evaluate the function of two different nervous systems: the sensory nervous system and the autonomic nervous system. As a result, we successfully verified that the diagnosis using the estimation formula obtained from DPNCheck™ and CV_R‐R could reproduce the diagnosis based on EMGS.

The prevalence of DPN among all patients in the present study was 36.5%. The prevalence of DPN varies depending on the diagnostic method. In a study using vibration perception thresholds for the diagnosis, the prevalence of neuropathy was 28.5% ¹⁶ . In our previous report, the prevalence of DPN in Japan was 35.8%, in which DPN was diagnosed using the simple diagnostic criteria by the Diabetic Neuropathy Study Group in Japan, which is also comprised of physical symptoms and signs. Partanen et al. ¹⁷ , who utilized EMGS for the DPN diagnosis, reported a prevalence of 41.9% for DPN in type 2 diabetes patients after 10 years with diabetes. Given the differences in the prevalence based on the differences in diagnostic methods, here, we selected an electrophysiological diagnostic approach, which has better objectivity and reproducibility, for assessing DPN. The clinical significance of DPN determined by electrophysiological tests has not been fully supported worldwide. However, as the diagnoses based on physical symptoms and signs, which are inferior in reproducibility and objectivity, should be gradually phased out in research settings, the current method could become one of the most promising diagnostic approaches in the future.

Among the collected clinical data, there were significant differences between the groups with or without DPN in CV_R‐R recorded during deep breathing and parameters of nerve conduction function, which were consistent with previous reports ¹⁰ , ¹⁸ . Additionally, the current study showed the significant correlation between brachial‐ankle pulse wave velocity and DPN severity, which is consistent with a previous study reported by Kase et al ¹⁹ . As the brachial‐ankle pulse wave velocity can identify patients at high risk for cardiovascular events and death ²⁰ , the correlation of brachial‐ankle pulse wave velocity and DPN severity might reinforce the hypothesis that the progression of DPN relates to cardiovascular events and death.

The assessment of DPN in BC primarily relies on NCS, which focuses on evaluating the severity of sensory and motor nerves while excluding autonomic neuropathy. However, considering that DPN consists of sensorimotor and autonomic neuropathy, we incorporated CV_R‐R, a parameter of autonomic neuropathy, in our evaluation. Compared with the multiple regression equation in our previous report ¹⁰ , which attempted to diagnose DPN using NCV and SNAP amplitude of DPNCheck™, the contribution of NCV and SNAP amplitude of DPNCheck™ was similar in this study: in the results of the multiple regression analyses, the standardized coefficients for SNAP amplitudes and NCV were −0.485 (previous) vs −0.426 (current) and −0.215 (previous) vs −0.197 (current), respectively. Additionally, the area under the curve by ROC curve showed consistent findings with the previous report (0.871, previous) compared with the current value of 0.880 ¹⁰ . These findings in the current study, which incorporated the assessment of autonomic neuropathy, showed comparable results to the method of evaluating sensory neuropathy using DPNCheck™. At this stage, it remains uncertain whether including the evaluation of autonomic neuropathy is advantageous. One clinical study showed that the combination of tests to detect nerve conduction abnormality and autonomic dysfunction was the most sensitive and objective, and suitable to distinguish subclinical neuropathy ²¹ . As more data are gathered, the importance of including autonomic neuropathy in the diagnosis of diabetic neuropathy will become increasingly evident.

We found that the CV_R‐R, a parameter of CAN, is useful in the diagnosis of DPN. A task force of the European Society of Cardiology and the Electrophysiology Society of the North American Pacing Society recommended evaluating heart rate variability using a 24‐h Holter electrocardiography. Compared with these cumbersome tests, the examination of CV_R‐R can be carried out in a short time, does not require special equipment and is easy to repeat. Although CV_R‐R has moderate reproducibility, the test can be carried out even in facilities without skilled technicians. Especially in Japan, CV_R‐R has been widely recognized and used as an indicator of diabetic autonomic neuropathy. However, although Wheeler et al. ²² found loss of respiratory variability in R‐R interval in patients with diabetic neuropathy as far back as 1973, its usefulness has not been fully exploited to date. We hope that the current findings will make the CV_R‐R test more versatile in the future.

There are three limitations in the current study. First, NCS including DPNCheck™ are strongly influenced by obesity and edema. Considering this fact, patients with a body mass index >35 were excluded from the study. Therefore, the current findings can be applied to patients with a body mass index <35. The widespread adoption of DPNCheck™ and further research examining the impacts of obesity would facilitate the assessment of diabetic neuropathy in individuals with obesity. Second, as only inpatients who aimed to improve their hyperglycemia were included in this study, most of the cases suffered from hyperglycemia at the time of admission. Therefore, the hyperglycemia might contribute transient deterioration in the function of the peripheral nervous system. To verify the current findings, we will carry out research including patients with well‐controlled glucose profiles in the future. Third, we collectively examined a cohort comprising patients with either type 1 or type 2 diabetes. It is acknowledged that the etiology of DPN might vary between patients with type 1 and those with type 2 diabetes. As a result, we intend to carry out analyses on more extensive cohorts in future studies, aiming to facilitate a more comprehensive exploration of distinct diabetes types.

We have attempted to replicate the gold standard EMGS diagnosis of DPN using two simple quantitative tests. These two tests can evaluate the function of two different nervous systems, namely, the sensory and autonomic nervous systems, which are susceptible to early impairment in DPN. By combining these tests, we obtained an estimation formula with good diagnostic performance.

The use of DPNCheck™ and electrocardiogram would make the diagnosis of DPN simple, ubiquitous, and with high reproducibility and reliability.

DISCLOSURE

Hideki Kamiya and Jiro Nakamura are Editorial Board members of Journal of Diabetes Investigation and co‐authors of this article. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication. The other authors declare no conflict of interest.

Approval of the research protocol: This study was approved by the ethics committee of Aichi Medical University Hospital (approval number: 2019–133, approval date: 10 January 2020).

Informed consent: N/A.

Approval date of registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

Supporting information

Table S1. Sensitivity, specificity, positive predictive value, and negative.