In Vitro Study of the Interaction of
Wireless Phones and
Implantable Cardioverter Defibrillators

Executive Summary



This report summarizes the results of an in vitro investigation of the interaction between wireless phones and implantable cardioverter defibrillators (ICDs) by the Center for the Study of Wireless Electromagnetic Compatibility at the University of Oklahoma. The results of this first phase of the Wireless Phone - ICD Study were initially presented at the Fourth Annual EMC Forum in Dallas, Texas (Grant and Schlegel, 1997b) and at the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society in Chicago, Illinois (Grant and Schlegel, 1997a) in October of 1997. Specifically, the in vitro laboratory tests were designed to determine the extent of interaction between 10 ICDs from 4 major companies and 9 phone technologies used in the United States, Europe, and other parts of the world.

This study comprised Phase I of the Center’s ICD research, and focused on (1) determining the amount of interaction between wireless phones and ICDs, and (2) identifying those factors which had a significant influence on the level of interaction. The phone technologies used in this study transmit in either the "low" frequency cellular bands of 800 MHz (AMPS, CDMA, TDMA-50 Hz, TDMA-11 Hz) or 900 MHz (TDMA-217 MHz), or the "high" frequency PCS bands of 1800 MHz (TDMA-217 MHz) or 1900 MHz (CDMA, TDMA-50 Hz, TDMA-217 Hz). Tests were conducted in the presence and absence of a simulated ECG signal. Relative orientation and coplanar separation distance of the phone and ICD were also considered.

This project was undertaken to determine if the electromagnetic interaction sometimes observed between wireless phones and cardiac pacemakers also exists between wireless phones and ICDs. In addition, the project was designed to more precisely classify the modes of ICD interaction and to identify which wireless phone variables affected the amount of interaction. Finally, the study documented the conditions which might promote interaction in order to increase the current understanding of the mechanism of the interaction.

A peer review team, named the ICD Study Design Group, helped define the scope of the project and provided input with respect to the test protocol, design of the torso simulator, and other parameters of the study. The group consisted of representatives from ICD manufacturers, wireless phone manufacturers, wireless carriers, government agencies, and other researchers (see Appendix A).


Testing was conducted in a closed, electromagnetically shielded room at the Lucent Technologies Inc. Test Facility in Oklahoma City, Oklahoma. The functionality of the heart/ICD system was simulated through the use of a torso simulator and various electronic equipment items which generated and monitored electrical signals. The test equipment comprised:

  1. a torso simulator consisting of a saline bath and grid supports for the ICD, ICD leads, and wireless phones,
  2. signal monitoring equipment for acquiring the waveforms from the ICD and the ECG signal when injected,
  3. ECG signal injection equipment,
  4. various ICDs,
  5. various wireless phones, and
  6. wireless phone base station simulators for the various phone technologies.

All testing was conducted under worst-case conditions with the phones at their highest output power and ICDs set to the greatest sensitivity and the minimum refractory period permitted for each unit. In all tests, the ICD case and the phone keypad were positioned in the horizontal plane, with the top surface of the ICD 0.5 cm below the surface of the saline solution and 1 cm from the surface plane of the phone.

The test factors were divided into two categories: ICD variables and wireless phone variables. The ICD variables consisted of the ICD model and the presence or absence of an injected ECG signal. Ten ICD models were provided by the four major manufacturers of ICDs. All tests were conducted twice, once with no injected ECG signal (typical asynchronous pacing) and once with an injected ECG signal (typical inhibition). All units were tested in the VVI mode during bradycardia pacing (the Ventricle chamber is paced, the Ventricle chamber is sensed, and the response to sensing is Inhibited).

The phone variables consisted of the phone technology, the phone model, and phone orientation with respect to ICD header alignment. Nine phone technologies provided by five major manufacturers of digital wireless phones were used in the testing as shown in Table 1.

Phone Technology

Frequency Band


Analog (AMPS)

800 MHz


CDMA (Cellular)

800 MHz



1900 MHz


TDMA-11 Hz

800 MHz


TDMA-50 Hz (Cellular)

800 MHz

TIA/EIA 627 (IS-54)

TDMA-50 Hz (PCS)

1900 MHz


TDMA-217 Hz (GSM)

900 MHz

ETSI GSM 11.22

TDMA-217 Hz (PCS 1900)

1900 MHz


TDMA-217 Hz (DCS 1800)

1800 MHz

ETSI GSM 11.22

Table 1: Information on Wireless Phones Used in the Study.

The TDMA-11 Hz phone, the CDMA (cellular) phone, and the TDMA-50 Hz (PCS) phone were tested in an open-loop transmit mode (i.e., enabling a test mode to produce the typical pulsing format of that technology without being in communication with an active cell site or base station simulator). For the TDMA-11 Hz model, this was accomplished through hardware modification of the phone by the manufacturer. For the CDMA (cellular) and the TDMA-50 Hz (PCS) models, keypad programming was used to configure each phone in the full-power transmit mode (with variable vocoder rate for CDMA). All other phones were tested while in communication with a base station simulator in the "connect" mode (i.e., the phone is answered after it rings). The TDMA-50 Hz dual-mode phones were also tested in the analog "connect" mode to examine any interaction with the analog format.

Four different relative orientations of the ICD and phone under test were examined. In all test conditions, the ICD case and the phone keypad were in parallel planes with the phone keypad facing up. The ICD was oriented such that the leads exited the header to the bottom (south) when viewed from above with the long axis of the torso simulator going from left to right (west to east). In other words, the leads exited the case parallel to the short axis of the torso simulator. The four orientations of the phone antenna from base to tip were (1) south to north (0°) with the antenna unbent, (2) east to west (90°) with the antenna unbent, (3) south to north (0°) with the antenna pressed against the grid, or (4) east to west (90°) with the antenna pressed against the grid.

Various levels of the test factors were combined to define a single test run (e.g., ICD model 01 with the injected ECG signal in conjunction with phone Y in a South-North/Unbent Orientation). Each run consisted of individual tests at those grid points determined by the decision rules of the test. Not all combinations of phone technology and ICD model were tested. The results of 496 test runs are reported here.

At any test point on the grid, a simple determination was made as to whether an interaction event occurred or did not occur. If an event occurred, the event was classified as to interaction mode based on the identification of six different modes of interaction (e.g., defibrillation therapy, inhibition). Various interaction measures were then computed. For example, the total number of interaction events was calculated as a function of the ICD unit, phone unit, and other variables. This measure has the distinct disadvantage of not accounting for varying opportunity for an event to occur, that is, the number of test runs involving a specific level of a factor.

To properly adjust for the varying number of test runs involving a specific level of a particular factor, the total number of interaction events within a given category must be adjusted by the number of test runs conducted in that category. This adjusted measure has been termed the Incidence of interaction. This Incidence measure can be further decomposed into two elements, Rate and Density. Rate refers to the proportion of runs in a given category in which at least one interaction event was observed. In other words, Rate is the ratio of the number of runs with at least one observed interaction to the total number of runs in that category. Density refers to the number of interaction events divided by the number of runs with any observed interaction. Thus, Density provides a measure of the average number of interactions for a test run that had interaction. Note that Incidence = Rate ´ Density.

Finally, the maximum distance of interaction was obtained by identifying the farthest X-Y grid point from the ICD header at which an interaction event was observed. This identified the location of the base of the phone antenna without regard to the relative phone orientation or the point of maximum field strength along the antenna axis. This X-Y grid point was converted to a Euclidean distance and used as a measure of interaction susceptibility.

Scope of this Study

It is important to clarify the scope of this study and the degree to which the test results can be extrapolated to evaluating interaction possibilities in the general population of ICDs. By design, this study considered a limited number of ICDs (10 ICDs compared with almost 30 pacemakers in the Center’s pacemaker interaction study). This allowed testing to be completed in a timely fashion (a few months vs. more than a year in the pacemaker study).

Therefore, this study presents general guidelines regarding the nature of interaction and provides a basis for additional research. It was not designed to be an exhaustive study considering all types of ICDs. It does represent a sample from the top companies which provide the vast majority of devices. It also includes all available phone technologies in the U.S. and Europe. It should also be noted that the study did not include a complete test of all combinations of wireless phones and ICDs, but rather a sample which covered approximately 69% of the possible combinations.


This study was conducted under worst-case conditions with phones operating at their highest power levels and each ICD programmed to its maximum sensitivity setting and minimum refractory period. All tests were conducted with the phone in close proximity to the ICD, representing a phone being carried in clothing pockets or held adjacent to the body in the vicinity of an implanted ICD. Caution must be exercised in using these results to directly contrast one phone technology with another due to differences in the frequency bands used and possible differences in the implementation of these technologies by particular manufacturers. Hence, the results should be interpreted carefully.

The following general conclusions can be drawn from the study:

1. Throughout all testing, there was no instance of ICD re-programming.

2. Whenever interaction was observed during a test, the ICD returned to its normal operation as soon as the phone was turned off.

3. More than half of the ICDs (6 out of 10) were completely immune to interaction from all of the nine wireless technologies tested. Only ICD's #58 and #12 (from the same manufacturer) had interactions with more than one phone technology.

4. Under worst-case conditions, there was substantial interaction of ICD's with the TDMA-11 Hz phone, with 25.0% of the test runs resulting in some interaction (and a Density of 60.3 points of interaction per run with interaction). This phone technology is typically used for specialized business applications and currently operates using a lower mobile transmitter power (0.6 W) than when the ICD testing was performed (1 W). This lower power should reduce the incidence of interaction with this system.

5. If ICD's #58 and #12 are excluded from the results, along wit tests involving the TDMA-11 Hz phone technology, the remaining tests produced no interaction.

6. The overall Rate of interaction for the ICD's was 9.5%, and the overall average Density was 46.3 interaction points per run with interaction. If the worst offending ICD (#58) is excluded, the Rate is reduced to 3.8%, and the average Density drops to 19.6 interaction points per run with interaction.

7. Only ICD #58 interacted with the analog phone technology. This ICD is currently marketed under additional FDA labeling restrictions regarding separation between the ICD and wireless phone. No other ICD's interacted with analog phones.

8. No interaction events were observed for the PCS band (1800-1900 MHz) phones using any of the phone signal formats.

9. The majority of interactions consisted of unwarranted delivery of defibrillation therapy or inhibition.

10. The number of defibrillation therapies was unaffected by the presence or absence of an ECG signal. Since inhibition therapy can only be observed in the absence of an ECG signal, there were more inhibition interaction points observed per run for ECG Off.

11. The relative orientation of the phone and the ICD case was a significant factor in terms of the amount of interaction that occurred.

12. Many but not all ICD-phone combinations were tested. It is thus important to use proper summary measures for comparison (i.e., Rate, Density, and Maximum X-Y Distance). The worst offending ICD (#58) and phone (TDMA-11 Hz) were tested with every phone technology and ICD, respectively. In addition, TDMA-217 Hz (1900) and Analog were both tested with every ICD.

13. The maximum coplanar (X-Y) distance between the ICD header and the base of the phone antenna at which interaction occurred for any ICD was 7.4 inches for the highest power TDMA-11 Hz phone. For the remaining phones, the maximum interaction distance was 6.2 inches. If measured from the ICD header to the tip of the antenna or to the point of maximum RF field strength, the maximum distance of interaction is substantially less. These results support HIMA-recommended implantable device labeling to "maintain a minimum separation of 6 inches."

14. The test results were repeatable to the level of identifying the specific location of interaction points and the mode of the interaction. Further comparisons with clinical data are needed to determine the correlation of in vitro testing results with in vivo interaction events.

Future Research

Future testing will examine the maximum height (Z-axis separation) at which interaction events occur. Additional testing will also be conducted as new wireless technologies and ICD prototypes become available.

Copyright © 2010 The Center for Study of Wireless Electromagnetic Compatibility, University of Oklahoma. All Rights Reserved.