Risk of occurrence of electromagnetic interference from the application of transcutaneous electrical nerve stimulation on the sensing function of implantable defibrillators

Abstract

Background

Transcutaneous electrical nerve stimulation (TENS) is an established method for pain relief. But electrical TENS currents are also a source of electromagnetic interference (EMI). Thus, TENS is considered to be contraindicated in implantable cardioverter-defibrillator (ICD) patients. However, data might be outdated due to considerable advances in ICD and cardiac resynchronization therapy (CRT) filtering and noise protection algorithm technologies. The aim of this pilot safety study was to re-evaluate the safety of TENS in patients with modern ICDs.

Methods and results

One hundred and seven patients equipped with 55 different models of ICD/CRT with defibrillators from 4 manufacturers underwent a standardized test protocol including TENS at the cervical spine and the thorax, at 2 stimulation modes—high-frequency TENS (80 Hz) and burst-mode TENS (2 Hz). Potential interference monitoring included continuous documentation of ECG Lead II, intracardiac electrograms and the marker channel. Electromagnetic interference was detected in 17 of 107 patients (15.9%). Most frequent were: interpretations as a premature ventricular beats (VS/S) in 15 patients (14%), noise reversion in 5 (4.6%) which resulted in temporary asynchronous pacing in 3 (2.8%), interpretation as ventricular tachycardia/ventricular fibrillation in 2 (1.9%), and premature atrial beat in 2 (1.9%) patients. Electromagnetic interference occurrence was influenced by position (chest, P < 0.01), higher current intensity (P < 0.01), and manufacturer (P = 0.012).

Conclusion

Overall, only intermittent and minor EMI were detected. Prior to the use of TENS in patients with ICDs, they should undergo testing under the supervision of a cardiac device specialist.

Graphical abstract

Open in new tabDownload slide

Introduction

Implantable cardioverter-defibrillators (ICDs) are recommended for a variety of indications in routine practice for primary and secondary prevention of malignant ventricular tachyarrhythmias.¹ The number of patients with an ICD is already large and is further growing due to the ageing population and a wider indication for implantation of ICDs.

Transcutaneous electrical nerve stimulation (TENS) is a widely used complementary treatment for the management of acute and chronic non-malignant pain. It is a low-cost, easily accessible, effective, and a self-applied at home treatment without serious adverse effects.² Both the requirements of an ICD and chronic pain increase with age. Therefore, it is more and more common to see elderly patients that have an ICD and require additional pain treatment. Electromagnetic interference (EMI) can occur while using TENS which can cause a dangerous dysfunction of the ICD leading to inadequate or inhibited therapy as well as inhibition of pacing.³ Inappropriate ICD shocks are not just psychologically and physically disturbing but cause a higher overall mortality.⁴

Electromagnetic interference between TENS and ICDs has been reported in a systematic study done in 2008 by Holmgren et al.⁵ and Crevenna et al.⁶ in 2003. Based on this data, current medical guidelines do not recommend the use of TENS in patients with ICDs or have extremely strict safety requirements which are difficult to fulfil in real-life practice.⁷ As a result and due to concerns from physicians and physiotherapists in routine clinical practice, patients with an ICD are generally excluded from the potential benefits of TENS treatment.⁸

However, studies which identify TENS as a potential source of EMI were conducted over 12 years ago.^5,6 Due to the considerable advances in ICD technology, including filtering characteristics and noise protection algorithms, the current recommendations might be outdated.⁹

We therefore aimed to re-evaluate the risk of potentially dangerous interference between modern ICD/CRT with defibrillators (CRT-Ds) and TENS in this pilot safety study by means of a standardized test protocol.

Methods

Study design and population

In this prospective, single-centre study, we aimed to evaluate the risk for EMI in patients with ICD systems and TENS application on the cervical spine and chest.

One hundred and seven patients were included after signing an informed consent form. Fifty-five different models of single-chamber, dual-chamber, and triple-chamber ICD systems were included. Implantable cardioverter-defibrillator systems from four different manufacturers (St. Jude Medical, Sylmar, CA, USA; Guidant/Boston Scientific Corp., St. Paul, MN, USA; Biotronik, Berlin, Germany; and Medtronic, Minneapolis, MN, USA) were included. Detailed patient and device characteristics are provided in Table 1.

The primary endpoint was the occurrence of EMI. Secondary endpoints were the distribution of the different entities of EMI and device integrity pre- and post-TENS testing. We included all patients who signed the informed consent, were aged >18 years, and were not pregnant. Exclusion criteria were contraindications for TENS application. We recruited patients from the cardiac device outpatient clinic of the Department of Cardiology, at the Charité Campus Benjamin Franklin from January 2018 until May 2018 that presented for routine device check-up.

This study conforms to the guiding principles of the Declaration of Helsinki of 2014 and was approved by the local ethics committee of the Charité (EA4/130/17). All authors had full access to the data and have read and agreed to the manuscript as written.

Study protocol

In the present study, the transcutaneous nerve stimulator model ‘TENS Eco 2’ (Schwa-Medico GmbH, Ehringhausen, Germany) was used. The stimulation was carried out using the self-adhesive electrode patches model ‘STIMEX’ with the dimensions 50 × 50 mm from the same company. Impulse form was biphasic asymmetrically compensated rectangular pulse with a pulse width of 180 µs. The current was increased to maximum comfortable level, and duration of stimulation was 1 min for each test.

1. High-frequency TENS: Continuous impulses with a pulse frequency of 80 Hz

2. Low-frequency TENS: Burst mode, pulse salvos with a repetition frequency of 2 Hz

TENS testing was performed in two positions:

• Position 1 (P1): The cervical spine, paravertebral, at the level of the spinous process of the 7th cervical vertebra, electrode distance ∼10 cm with the cathode to the left

• Position 2 (P2): The chest at the level of the 5th intercostal space, electrode spacing ∼10 cm, with the cathode to the right (Table 2)

A baseline recording with the patches placed at each position (P1 and P2) but without current flow was done. This resulted in three tests for each position and six total tests (2 at baseline and 4 during TENS stimulation). For assessment of EMI, the surface ECG Lead II, all intracardiac electrograms, and the marker channel were recorded continuously for 1 min for each test and later saved, printed, and analysed.

The test protocol was performed with the patient in the awake and unsedated state. A supervising cardiologist (C), a resident physician (R), and a device clinic nurse (DN) were present in the outpatient clinic. A routine device interrogation was performed before testing to assure device integrity and analyse the heart rhythm. The ICD programming was not changed. The device programming for arrhythmia detection was according to current guidelines and underlying disease. Detection zones (heart rate), detection time (either actual time in seconds or number of consecutive beats) varied depending on the manufacturer. Sensitivity was also manufacturer-dependent. Sensitivity and tachycardia detection programming are summarized in Table 3.

Only the ICD therapy function was manually deactivated through the device programmer for the time of testing. The recording and interpretation functions stayed intact. The patient was under electrocardiogram (EKG) and clinical monitoring throughout the testing.

The patient was placed in a recumbent position, and EKGs were recorded continuously for 1 min at baseline. Transcutaneous electrical nerve stimulation electrodes were placed immediately in P1, and TENS was started at 80 Hz, and the intensity was slowly increased to the maximum comfortable level and held there for the continuous recording time of 1 min. This same procedure was then followed with TENS at 2 Hz.

The TENS electrodes were then moved to P2, and the above described testing protocol with baseline recording and testing at 80 and 2 Hz was repeated. After the conclusion of testing, the ICD therapy function was reactivated, and a device interrogation was performed. A schematic workflow is shown in Figure 1.

Statistical methods

Descriptive statistics are presented as number (n) and per cent (%) for qualitative variables and as median and interquartile range for quantitative variables (not normally distributed).

 A two-sided P-value < 0.05 was considered statistically significant. McNemar test was used to test the difference of occurrence of EMI between electrode positions and types of stimulation. Chi-square and Fisher’s exact tests were used to test for differences in qualitative characteristics and Mann–Whitney U test was used for characterizing quantitative variables between the subgroups with and without EMI. Lead parameters pre- and post-testing were analysed with the Wilcoxon test. As this is an explorative study, no adjustment for multiple significance testing was applied. Sample size estimation: in order to achieve an estimate of the probability of occurrence for EMI of 3.6% with a 95% confidence interval, the TENS must be tested on 103 patients.

Results

Demographics: patient and device characteristics

One hundred and seven patients were included in this study. The group included 93 (86.9%) men with a median age of 73.5 years (range 55–82). A total of 75 (70.1%) patients had received their ICD for primary prevention, and the main underlying disease was an ischaemic cardiomyopathy (n = 55, 51.4%). The implantation site was mainly left chest (n = 103, 96.3%). The ICDs were implanted at least 2 months before the investigation. Detailed patient and device characteristics are provided in Table 1. Detailed device and lead characteristics of all 17 patients with EMI are shown in Supplementary material online, Tables S1–S3.

Interference

In our study, EMI occurred in 17 out of 107 (15.9%) patients undergoing TENS testing. More than 1 EMI could occur in 1 patient resulting in a total number of 53 EMIs in these 17 patients. Gender, age, BMI, underlying heart disease, NYHA class, indication for implantation, patients’ heart rhythm, and device localization did not differ in the groups with or without EMI. The sensing configuration (true bipolar vs. integrated bipolar) and number of leads [single- or dual-chamber or CRT (three leads)] did not influence the occurrence of EMI (Table 1). A statistical significant difference in the occurrence of EMI was seen by manufacturer (P = 0.012). Twenty-seven out 107 patients (25.2%) had an implanted device from Medtronic. None of these 27 patients had EMI (P = 0.012). Programming sensitivity and tachycardia detection programming did not differ in the groups with or without EMI (Table 3).

Interference was detected significantly more often with the chest rather than at the cervical position [15/107 (14%) vs. 3/107 (2.8%), P < 0.01]. The sub-entities of all EMI are summarized in Table 4. The most frequent entities of EMI were right ventricular sensing (RVS: 15/107, 14%), noise (5/107, 4.7%), and asynchronous stimulation (3/107, 2.8%). Right ventricular sensing was seen significantly higher at the chest level compared to cervical testing [14/107 (13.1%) vs. 1/107 (0.9%), P < 0.01]. The stimulation frequency and pattern with 80 Hz continuous stimulation and at 2 Hz with burst stimulation did not influence the occurrence of EMI [15/107 (14%) vs. 14/107 (13.1%), P = 1.0] (Table 5). The applied electrical current was individually determined for each of the four tests separately by slowly increasing the intensity to the maximum comfortable level.

Table 3 shows the median of the current intensities. The current intensities in the group with EMI were significantly higher compared to the group without EMI (Test 1.1: P = 0.001, Test 1.2: P = 0.025, Test 2.1: P = 0.026, Test 2.2: P = 0.019). Overall, the median current intensity of TENS testing at the chest was lower than on the cervical spine. This is most likely due to a lower tolerance threshold due to denser innervation in the chest area.

Lastly, for our secondary endpoint, we compared lead integrity parameters before and after TENS testing to rule out a negative effect of the electrical current on the leads. Except for RVS (P = 0.025) and for impedance (P = 0.005), all the other parameters showed no statistical significance before and after TENS testing. When calculating the mean values, the RVS before and after TENS is 13.04 and 13.09 mV, respectively. The difference of only 0.05 mV can be regarded as negligible. The same applies to the mean values ​​of the high-voltage impedance before and after TENS, which are 67.3 Ω and 66.3 Ω. On average, the high-voltage impedance after TENS testing is only 0.98 Ω lower. This difference can also be seen as negligible. Furthermore, probe measurements do slightly vary in repeated measurements, which is considered normal unless there is a continuing trend towards higher or lower values. An influence of TENS on the probe measurements of stimulus thresholds, impedances, and sensing is therefore not to be assumed.

Discussion

A potential risk of interference with TENS and cardiac implantable electronic devices (CIED) will continue to be a frequent question by patients and caregivers. Potential EMI can put the patient at risk.¹¹ This fear and uncertainty need to be addressed and answered by evidence-based research. A general exclusion of all patients with devices from treatments with TENS puts unjustified restrictions on these patients and likely reduces patient satisfaction.

To date, this is the largest study that investigated the occurrence of EMI due to TENS application in patients with ICDs and CRT-Ds. We saw EMI in 15.9% of all patients (17/107). This is a significant reduction compared to previous studies from Holmgren et al.⁵ from 2008 (n = 30, EMI in 16/30, 53.3%) and Crevenna et al.⁶ from 2003 (n = 8, EMI 6/8, 75%). The general rule is to position the TENS electrodes over any area of pain. We hypothesized that given the advancements in noise discrimination algorithms that no EMI will occur. To prove our hypothesis, we tested at a position that is used in clinical practice (neck) and at a high-risk test setting at the anterior left chest with minimal distance to the CIED and bilateral application of energy (i.e. device within the flow of current) and therefore highest-risk factors for occurrence of EMI. This is not a test setting that is used in clinical practice. Keeping this in mind, we saw a further reduction in EMI with only 2.8% (3/107) in a real-life test setting position compared to our constructed high-risk test setting at the chest level (15/107, 14%).

The most frequent entities of EMI were RVS (15/107, 14%), noise (5/107, 4.7%), and asynchronous stimulation (3/107, 2.8%). Right ventricular sensing was seen significantly more at the chest level compared to cervical testing (14/107 (13.1%) vs. 1/107 (0.9%), P ≤ 0.01). A possible explanation is the smaller distance between the TENS patches, and the ICD leads at the chest level.

A measurement of the absolute distance between the patches and CIED at axial level was not possible because that would require X-ray documentation. Radiation exposure was not feasible or justified for our study purpose. The main factors contributing to an increased distance between patch and CIED are weight, height, and subcutaneous fat. For this reason, we analysed BMI as a surrogate parameter and found no differences in groups with or without EMI (Table 1, P = 0.997).

The intermittent detection and misinterpretation of EMI as premature ventricular beats (VS/S) would not cause any relevant risk for the patient. A continuous misinterpretation of EMI as premature ventricular beats in pacemaker-dependent patients, however, can cause an inhibition of pacing as shown in Figure 2. This phenomenon has been reported in a case report from 1990.¹²

Interpretation as noise was the second most common EMI (5/107, 4.7%). Although this interpreted correctly as an artefact and therefore relatively benign, it can still harm the patient by switching into asynchronous pacing mode such as V00 or D00 mode if this programmed. Asynchronous stimulation appeared in three patients and is especially relevant for patients with heart failure and CRT due to interruption of biventricular pacing. The response to noise varies depending on the device programming. Figure 3 shows an example of noise detection. During the noise interval, an intrinsic QRS complex was not detected which means it can cause undersensing. In the worst-case scenario, the undersensing could lead to fixed ventricular stimulation which could lead to induction of ventricular tachyarrhythmias. Furthermore, device interpretation of TENS–EMI as noise only appeared intermittently. As seen in Figure 3, the same single TENS signal was initially interpreted as noise and, in another cycle interval, falsely interpreted as a premature ventricular beat (VS), ventricular tachycardia (VT), or ventricular fibrillation (VF). An explanation might be that noise intervals must be triggered repeatedly. Another possibility might be a slight fluctuation of the TENS amplitude below the programmed sensitivity due to minimal variations in the current application of the pulse TENS generator or changes in the TENS electrodes impedance due to minimal subconscious patient movement or even respiratory chest movements.¹³ In conclusion, it can be said that even correct detection of noise holds a certain risk for patients.

Interpretation of EMI as ‘VT/VF’ is the most feared and dangerous form because it can lead to inappropriate ICD shocks.^14–17 Annotation as VT/VF was seen in two patients (2/107, 1.8%): one during testing on the chest and another while testing on the cervical spine. We do have to note that as shown in Figure 3, we only saw intermittent annotation as VT/VF and never a consecutive annotation as VT/VF. This is of importance because though an intermittent annotation does lead to activation of tachycardia detection algorithms, it would not lead to an inadequate shock because cycle length is only one detection criterion. Most modern ICDs require other VT/VF criteria such as onset, stability, morphology, and, most of all, a continuous detection over a set period of time to be fulfilled before delivering ICD therapy. We conclude that even the most dangerous type of EMI only occurred in 2/107 patients and only occurred intermittently and did not cause any harm. Potential harm in a real-life scenario cannot be excluded a 100%, but the likelihood seems to be very low.

We further analysed TENS parameters, device characteristics, and programming that might have an impact on the detection of EMI. The occurrence of EMI was not influenced by the stimulation frequency (80 Hz vs. 2 Hz, P = 1.0) or pattern (continuous vs. burst stimulation, P = 1.0). No difference was seen based on lead characteristics (integrated vs. dedicated bipolar, P = 0.539), the ICD system (single-/dual-chamber or CRT–ICD, P = 0.08), nor the sensitivity programming (Table 3, P > 0.05) or arrhythmia detection programming (Table 3, P > 0.05) defers in the group with EMI. These results differ from Holmgren et al.⁵; findings with differences were described for EMI occurrence between integrated vs. bipolar leads and stimulation frequency 80 Hz vs. 2 Hz.

A smaller distance of the patches to the ICD system and a higher current intensity led to more EMI. Electromagnetic interference was detected significantly more often at the chest than at the cervical position [15/107 (14%) vs. 3/107 (2.8%), P = < 0.01]. This is in accordance with results from previous studies and a systematic review from 2017.^3,5,18 The current intensities in the group with EMI were significantly higher compared to the group without EMI (Test 1.1: P = 0.001, Test 1.2: P = 0.025, Test 2.1: P = 0.026, Test 2.2: P = 0.019). Despite the overall median of current intensity of TENS testing at the chest being lower than on the cervical spine, we saw significantly more EMI at the chest level. This implies that the main factor for appearance of EMI is distance to electrodes and secondly the intensity of applied current. However, we did see one case of false interpretation of TENS–EMI as ‘VT/VF’ at the cervical level despite the longer distance. This implies that distance alone is not a sufficient factor. These results are in accordance with the EHRA consensus paper that describes distance and bipolar application pattern close to the CIED as the main risk factors for EMI.¹⁸ Below the current intensity of 17 mA, no EMI was detected. We are hesitant though to define this as a cut-off for safe TENS use because depending on the body physique, axial distance to the device and impedance may vary significantly.

Curwin et al.¹⁴ described a patient in whom inappropriate shock delivery occurred after repositioning TENS electrodes, suggesting that changing the orientation of the electrodes led to altered detection and sensing in the same patient. It is possible that the vector used plays a role. We cannot conclude that one position without EMI might have EMI if the position is kept the same and the electrode vector is slightly altered.

Transcutaneous electrical nerve stimulation has been in routine clinical use for over 30 years now. Upon review of the literature, there have been no major developments in the application sites and the current (high vs. low frequency and burst vs. continuous stimulation) used. We believe that our results are generalizable for only the current settings we have chosen high frequency (80 Hz, continuous) and low frequency (2 Hz, burst). These are the two most frequently used settings.² Other settings are possible but were not tested. We used one TENS device from one manufacturer. The TENS machine is just a technical tool to generate electrical current with different frequency and pattern. Given the device is functional, the findings are generalizable.

Transcutaneous electrical nerve stimulation applications in physical therapy have a duration of on average 20 min per session.¹⁹ We had to do an abbreviated testing protocol of 1 min per test. The rationale behind it was, firstly, the estimated time the device algorithms need to make a definitive interpretation of the detected signals. The longest detection delay is up to 1 min. Secondly, we had to compromise for a shorter testing duration considering our resources in the device clinic (time, personnel, etc.), stress on our volunteering elderly patients, their willingness to participate in the study, and also potential CIED battery consumption (continuous interrogation/telemetry contact). In retrospect, we do acknowledge that this is a limitation and we cannot exclude occurrence of EMI with a longer testing period.

Furthermore, we did see a significant difference in the occurrence of EMI between the manufacturers. However, since this was an exploratory pilot study with low numbers of EMI overall and not blinded or randomized and not powered to conclude a correct statistical difference, these results need to be interpreted with caution, and no conclusions should be drawn. There was no cases of EMI in patients with a device from Medtronic (0/27, P = 0.012). Electromagnetic interferences in patients with devices from Biotronik (5/34, 14.7%), Boston Scientific (6/25, 24%), and Abbott (6/21, 28.6%) were equally distributed. A possible explanation might be the introduction of the lead integrity algorithm in leads from Medtronic after reporting of inappropriate shocks caused by oversensing from a lead fracture, which occurred in up to 50%.²⁰ We strongly have to emphasize that we cannot conclude that devices from Medtronic are safe for TENS use in general. A multivariate analysis is not reasonable due to the small number of patients and EMI and the large number of variables that need to be included in testing such as device characteristics (model and number of electrodes) and patient characteristics (i.e. BMI).

Limitations

This was a single-centre exploratory pilot study that aimed to document any clinically important EMI. Both physicians and patients were not blinded. The patients and devices represent the local patient population and device pool from our outpatient clinic. We used a standardized but nonrandomized order of testing. We do not know whether the results would have been different with a randomized order. Only two electrode positions in one standardized orientation were tested, and only two stimulation frequencies and patterns were tested with one TENS device model from one manufacturer. Transcutaneous electrical nerve stimulation application in the distal extremities possibly behaves differently with fewer or no instances of EMI since the main risk factor for EMI was distance. This site was not tested in our protocol. We used an abbreviated test protocol with a testing duration of only 1 min. This is a limitation because real-life applications take 20 min, and we cannot exclude that a longer application would lead to more EMI. Further statistical analysis with a multivariate analysis was not reasonable because of the small number of EMI and the large number of variables that potentially can have an influence on the occurrence of EMI. The study was conducted in 2018. Accordingly, technical advancements since then are not represented in this study.

Clinical implications

In our study we did see EMI due to TENS use. The incidence of EMI was significantly lower compared to previous studies, which does reflect advancement in ICD technology. Overall, even in the highest-risk setting with the smallest possible distance to the ICD and bipolar current application at the left anterior chest, only intermittent, minor, and not clinically relevant EMI were detected and led to no harm. The use of TENS in patients with ICDs should undergo prior testing under supervision of a cardiac device specialist. Maximum current intensity, different types of stimulation frequencies and pattern, and different orientation of the TENS patches for the site of application should all be tested.

Supplementary material

Supplementary material is available at Europace online.

Acknowledgements

We thank all the technicians and engineers from St. Jude Medical, Sylmar, CA, USA; Guidant/Boston Scientific Corp., St. Paul, MN, USA; Biotronik, Berlin, Germany; and Medtronic, Minneapolis, MN, USA for their assistance in interpreting device annotations and understanding the device algorithms for arrhythmia and noise discrimination.

Data availability

Raw data were generated at Charité—Universitätsmedizin Berlin, Deutsches Herzzentrum der Charité—Campus Benjamin Franklin, Department of Cardiology. Derived data supporting the findings of this study are available from the corresponding author (S.S.A.) on request.

References

Author notes