Modern Wireless Telecommunication Technologies and Their ...

TECHNOLOGY, COMPUTING,

AND

SIMULATION

SOCIETY

FOR

TECHNOLOGY

IN

ANESTHESIA

SECTION EDITOR STEVEN J. BARKER

Modern Wireless Telecommunication Technologies and Their Electromagnetic Compatibility with Life-Supporting Equipment Mats K. E. B. Wallin,

MD, MSc*†,

Therese Marve,

MSc†,

and Peter K. Hakansson,

MSc†

*Department of Anesthesiology and Intensive Care, Karolinska Hospital, and †Med. Lab. Sci. and Tech, Karolinska Institute, Stockholm, Sweden

Hospitals rely on pagers and ordinary telephones to reach staff members in emergency situations. New telecommunication technologies such as General Packet Radio Service (GPRS), the third generation mobile phone system Universal Mobile Telecommunications System (UMTS), and Wireless Local Area Network (WLAN) might be able to replace hospital pagers if they are electromagnetically compatible with medical devices. In this study, we sought to determine if GPRS, UMTS (Wideband Code Division Multiple AccessFrequency Division Duplex [WCDMA FDD]), and WLAN (IEEE 802.11b) transmitted signals interfere with life-supporting equipment in the intensive care and operating room environment. According to United States standard, ANSI C63.18 –1997, laboratory tests

A

well-functioning communication system is essential in an emergency hospital. The first commercial pagers were used in St. Thomas Hospital in London, England, in the mid 1950s (1). Today, most hospitals still rely on alphanumeric pagers and ordinary telephones to reach staff members during an emergency situation. In general, daily use of mobile phones and other wireless devices had increased rapidly. In hospitals, however, restrictions on the use of wireless telecommunication products are common after reports in the late 1980s and early 1990s of malfunctions in medical devices (MD) because of electromagnetic interference from various electronic equipment and cellular phones (2). The risk of interference between mobile phones and MDs mainly

Supported, in part, from Ericsson AB and TeliaSonera Sverige. Accepted for publication April 29, 2005. Address correspondence and reprint requests to Mats Wallin, MD, MSc, Department of Anesthesiology and Intensive Care, Karolinska Hospital, S-171 76 Stockholm, Sweden. Address e-mail to [email protected]. DOI: 10.1213/01.ANE.0000180216.83554.00 ©2005 by the International Anesthesia Research Society 0003-2999/05

were performed on 76 medical devices. In addition, clinical tests during 11 operations and 100 h of intensive care were performed. UMTS and WLAN signals caused little interference. Devices using these technologies can be used safely in critical care areas and during operations, but direct contact between medical devices and wireless communication devices ought to be avoided. In the case of GPRS, at a distance of 50 cm, it caused an older infusion pump to alarm and stop infusing; the pump had to be reset. Also, 10 cases of interference with device displays occurred. GPRS can be used safely at a distance of 1 m. Terminals/cellular phones using these technologies should be allowed without restriction in public areas because the risk of interference is minimal. (Anesth Analg 2005;101:1393–1400)

depends on transmission power, distance to the transmitter, and immunity (construction) of the MD, but signal characteristics, such as frequency and modulation (pulsing), are also important (3). Today, there are many different telecommunication systems for mobile phones—Wide Area Networks (WANs)— around the world. The predominant second-generation digital telecommunications system used for mobile phones is Global System for Mobile communication (GSM). It is used in the United States, Europe, parts of Asia, and other countries. GSM uses a variant of the Time Division Multiple Access (TDMA) standard. These WAN technologies are also suitable for telecommunication in larger local area networks such as factories, offices, and hospitals. Several studies have been performed testing medical equipment with signals from analog and GSM signals (4 – 6). Interference tests have also been performed with mobile phones based on other telecommunication technologies (6,7) such as Code Division Multiple Access (CDMA) and other TDMA standards, mostly used only in the United States. For a more accurate and detailed description of all these different telecommunication technologies for mobile phones, we suggest the appendix in the article in Health Physics Anesth Analg 2005;101:1393–1400

1393

1394

TECHNOLOGY, COMPUTING, AND SIMULATION WALLIN ET AL. TELECOMMUNICATION AND LIFE-SUPPORTING EQUIPMENT

by Morrissey (8). The safety distances recommended in some articles vary; however, electromagnetic interference generally occurs only when the cellular phones are in proximity to the MDs (3). New technologies keep entering the market. Wireless Local Area Networks (WLANs), primarily designed for data communication and often used in laptops and some handheld pocket personal computers, are now partly implemented in a few hospitals, although only a few studies concerning electromagnetic compatibility have been published (9,10). In addition, terminals with phone function using WLAN (802.11.b) technology have recently been released on the market. No compatibility studies on MDs have been published in English-language journals for General Packet Radio Service (GPRS)—the data communication protocol for GSM—which can be implemented in the existing GSM network, and the third-generation mobile phones system in Europe Universal Mobile Telecommunications System (UMTS) using the Wideband Code Division Multiple Access (WCDMA) technology with capability to send videophone calls and data. If these new wireless communication systems can operate in an environment with several MDs without any electromagnetic interference, it will be possible to replace the pager systems and to introduce new mobile services in hospitals. The aim of this study was to investigate whether electromagnetic interference with the MDs in an intensive care unit (ICU) and operating room (OR) could be noticed when transmitting GPRS, WCDMA, and WLAN signals.

Methods This study was performed in the OR and ICU at the Karolinska Hospital in Stockholm, Sweden. The study protocol was approved by our local institutional ethics committee. Provocative laboratory tests were made on 76 medical apparatuses (Table 1). In addition, clinical tests were performed during 11 operations and almost 100 h in the ICU. About one-third of the equipment was tested in the clinical setting; the ones tested depended on the type of operation and intensive care given in the hospital during the weeks of the tests. The test procedure was based on the American standard ANSI C63.18 –1997 (11), which provides health care organizations with a reproducible test method for evaluating the electromagnetic immunity of existing MDs to different radio-frequency (RF) transmitters. The recommended practice applies to most MDs and all portable RF transmitters with output power levels up to 8 W. The standard describes how to evaluate the performance of a MD when being exposed to RF fields. The deviations from normal performance, i.e., alarms, error messages, or distortions of displayed waveforms, are divided into 20 different categories.

ANESTH ANALG 2005;101:1393–1400

The signals used in our tests were WCDMA-Frequency Division Duplex, GSM/GPRS (1800 MHz) transmitting at 4 time slots, and WLAN (IEEE 802.11b). The signal characteristics and output power levels can be found in Table 2. The WCDMA and GPRS signals were simulated using base band and signal generators (AMIQ and SMHU 58; Rohde & Schwarz, GmbH, Munich, Germany) connected to a dipole antenna (738 454, 1710 – 2170 MHz, gain 2 dBi; Kathrein, Rosenheim, Germany) via a RF amplifier (LS Elektronik AB, Spanga, Sweden) and a 5-m-long cable (Fig. 2). Using this setup, it was easier to control the signal modulation, frequency, and output power compared with using commercial products. Besides, at the time for the study, no commercial phones were available using the specified characteristics. Before the tests, the output power levels for the GPRS and WCDMA signals were verified using a Rohde & Schwarz power meter. The WLAN signals were generated by commercial Personal Computer Memory Card International Association (PCMCIA) 802.11b network adapters (Symbol Spectrum 24™, Holtsville, NY). During the WLAN tests, a large file was sent, via a peer-to-peer connection, between two laptop computers. All signals were verified with a spectrum analyzer (Tektronix 7601, Bracknell, UK) during all laboratory measurements. Moreover, before and after every laboratory measurement session, the background electric field strength was measured using a broadband electromagnetic field analyzer (EMR-20; Wandel & Goltermann, Research Triangle Park, NC) to ensure that no strong fields from other sources were present.

Laboratory Tests The laboratory tests gave an opportunity to test the interference of the MDs in their normal environment without exposing patients to potential danger. Only after ensuring that none of the signals interfered with the tested MDs in a critical way, clinical tests during surgery and intensive care were then performed. The 76 different MDs (Table 1) included in the tests were all equipment used in the central ICU, the dialyzing unit, and ORs at the Karolinska Hospital. The MD under test was in operating mode and centered in an empty room, either on its own stand or, when possible, on a nonconducting table. The transmitting antenna or laptop computer was moved in a circle around the MD. The initial radius of the circle was 2 m. If no interference appeared, the radius was decreased and the test repeated. The same procedure was performed with smaller radii until the transmitting antenna was as close as possible to the MD, which exposed the MD to the highest electric field strength levels because the

ANESTH ANALG 2005;101:1393–1400


1395

Table 1. Medical Devices Tested for Interference with UMTS (WCDMA, FDD), WLAN (802.11b), and GPRS in the Study Type of equipment Anesthesia machine

Blood warmer Defibrillator Dialyzer Electrosurgery Perfusion system Infusion pump

Laparoscopic equipment Light source Operation stand Monitor

Producer/name Siemens SC 900 XL (KION) Datex ADU Dra¨ger Julian plus Dameca Dream Dameca MCN 590 and 10590 Anmedic JB1 Biegler BW 585 HP CodeMaster WL Gambro AK100 Hospal Prisma CFM Fresenius 4008 E Valleylab Force FX Valleylab Force 10 Sto¨ckert S3 (ECMO) Sto¨ckert CAPS 28-60-00 (ECMO) Sto¨ckert CAPS Alaris P6002 Baxter Flo-Gard 6201 Baxter Colleague Sandoz Compat nr 199235 IVAC 591 Braun IMED 960 IVAC 7101 EX Braun Infusomat fmS IVAC 531 IVAC 571 Sabratek 3030 Plus Storz Thermoflator Storz Xenon 300 Luxtec 9100 xenon Transferis ALM Datex GM 13 Ultima Agilent M3046A Datex AS/3 (Nokia CRT screen) Datex AS/3 (Hitachi CRT screen) Datex Cardiocap 5 Datex S/5 HP 78361A (EKG)

Purchasing year Manufactured Manufactured 2001 2001 1993 1998 2000 1995 Manufactured Manufactured Manufactured 2000 1993 1994 1998 1993 1994 1994 Manufactured

Comments

1999 1998

X X X X

1995 1994 1994

X X X

X 2001

1991

Alarmed and stopped, GPRS 50 cm

2001 1982 1998 2001

X

1994 2000 2001 2001 2000

X X X

1990 Manufactured 1997 1998 Interference on screen, GPRS 30 cm 1993 2001 Manufactured 2001

HP M1094A/Model 68S

1991

Datex PM8 Cardiocap II Hellige Servomed SMS 151

1991

Siemens Sirecust 732 (EKG)

1993

Colin Press Mate 8800 (NIBP) Marquette Solar 8000 (m. CRT)

1996 1999

Medair Life Sense Protocol Propaq 104

2001 1998

Protocol Propaq 106 EL Datex Light (F-LMP141-00) Datex Satlite Trans (SpO2)

1993 1998 1993

Clinically tested

X X X

Distortion of presented ECG curve, GPRS 10 cm Interference on screen, GPRS 5 cm X Distortion of presented ECG curve, GPRS 15 cm Distortion of presented ECG curve, GPRS 10 cm Interference on screen, GPRS 5 cm Distortion of presented ECG curve, GPRS 10 cm X

1396


ANESTH ANALG 2005;101:1393–1400

Table 1. Continued Type of equipment

Patient heater X-ray Telemetry Tourniquet TV-monitor

Ultrasonic doppler

Ultrasound Ultrasound equipment Vacuum suction apparatus Ventilator

Video equipment Incubator

Producer/name

Purchasing year

Datex TuffSat (SpO2) Ohmeda 3800 (SpO2) Mallinckrodt Warm touch 5800 ZIEHM Exposcop CB7-D Swemac Biplanar 400E GE Light speed GE Apex pro Zimmer ATS 2000 Panasonic M2080 Sony PVM-1443MD Sony PVM-2043MD Sony PVM-1453MD Parks Flow detector 811-b Huntleigh MiniDOPPLEX D900

2001 Manufactured 2001 1998

Aloka SSD-1400 Aloka Echo Camera SSD 680 USSC Autosonics Generator Egnell Surgical Suction Pump Dra¨ger Evita 4 Siemens Servo Ventilator 300 Respironics BIPAP Vision ST Siemens Servo 900 C Olympus CLV-U40, light source Storz Tricam 20 222 020 Storz telecam SL PAL 202120 20 Dra¨ger Babyterm 8010

Comments

Clinically tested

X Manufactured 1998 Manufactured 1999 1998 1996 1991 1992 1997 1990 1997

X X Interference noise, GPRS/UMTS/WLAN 100/1/5 cm, Interference noise, GPRS/UMTS/WLAN 400/25/50 cm, not CE-marked Interference noise, GPRS 10 cm

1998 1992 2001 1994 2001 1988 1998 2001 1997 1998

X X X X X X

UMTS ⫽ Universal Mobile Telecommunications System; WCDMA FDD ⫽ Wideband Code Division Multiple Access, Frequency Division Duplex; GPRS ⫽ General Packet Radio Service; WLAN ⫽ Wireless Local Area Network.

electromagnetic field is strongest close to the antenna and then rapidly decreases with distance. We also ensured that the transmitting antenna or WLAN adapter was as close as possible to sockets and cable ports because the MDs were expected to be especially sensitive to electromagnetic fields at these points. The flowchart in Figure 1 describes the test procedure. If interference occurred, the transmitting source was moved outwards until the interference stopped. To be considered as an established interference effect from the test signal, the distance to the antenna and the type of interference were to be the same in three consecutive measurements.

Clinical Tests in the Operation Unit For tests during surgery, we chose operations where as many MDs as possible were used. The tested signals were alternated and transmitted in periods of 15–20 min during the entire operation, sometimes up to 4 h. During surgery, it was not possible to move the

transmitting antenna in circles around the MDs as in the laboratory tests. Instead, the transmitting antenna, or laptop computer, was moved around in the OR to cover as many positions as possible in distances as close as possible to the MDs.

Clinical Tests in the ICU The tests in the ICU were conducted similarly to the tests during surgery. The transmitting antenna and the laptop computer were placed at various locations in the ICU during the course of a day. The tested signals were alternated. In every room, there were between one and three patients, and all patients had several different MDs connected to them. In a normal situation, 3 surveillance monitors, 10 infusion pumps, 2 ventilators, and 1 dialysis machine could be operating simultaneously in each ICU room. Apart from this equipment, diagnostic devices such as ultrasound and radiograph were used during the course of a day.

ANESTH ANALG 2005;101:1393–1400


1397

Figure 1. Flowchart of the test procedure.

Results Laboratory Tests The laboratory tests showed that 65 of the 76 MDs (85%) were immune to all signals tested, and only 2 devices showed interference effects caused by the WCDMA and WLAN sources. For WCDMA and WLAN, interference noise was only detected from the loudspeakers of the two handheld diagnostic ultrasonic Dopplers tested (Table 1). For WCDMA, the distances between the transmitting source and the Dopplers were 1 and 25 cm, respectively, and for WLAN, the distances were 5 and 50 cm,

respectively. These interferences were not considered critical because the Dopplers are diagnostic equipment, and the noise cannot be misinterpreted. For GPRS, one of the affected MDs malfunctioned in a critical way. At a distance of 50 cm to the antenna, an older infusion pump alarmed, stopped, and had to be reset. The other 10 cases of interference (Table 1) were display distortions and interference noise. The noted interference deviations were interference pattern on three cathode ray tube screens, distortion of electrocardiogram curves on four older patient monitors, and characteristic interference noise from the three ultrasonic devices. The

1398


ANESTH ANALG 2005;101:1393–1400

Table 2. Technical Characteristics of the Signals in the Study Technology

UMTS (WCDMA FDD)

Frequency Output power

1952.6 MHz 250 mW

Transmission

5-MHz bandwidth

Other information

No power control simulated

WLAN (IEEE 801.11b)

GPRS 1802.1 MHz 1 W maximum 500-mW average Pulse repetition frequency 217 Hz 200-kHz bandwidth Transmission on four time slot in a row No power control simulated

2.4–2.5 GHz 100 mW Direct sequence spread spectrum modulation Maximum transmission rate 11 Mbit/s Ad-hoc network

UMTS ⫽ Universal Mobile Telecommunications System; WCDMA FDD ⫽ Wideband Code Division Multiple Access, Frequency Division Duplex; GPRS ⫽ General Packet Radio Service; WLAN ⫽ Wireless Local Area Network.

Figure 2. The test set up.

distances between the GPRS transmitting antenna and the affected MDs varied from 5 to 50 cm except for the ultrasonic Dopplers, where the interference distances were as long as 1 and 4 m.

Clinical Tests In the clinical tests, no new types of interference were noticed. During surgery, the only interference noticed was distortion of the curves on a monitor when the GPRS signal transmitting antenna was very close. During intensive care, there was no interference registered for any signal.

Discussion The most important finding in our study is the minimal interference seen from WCDMA and WLAN (standard 802.11b) signals in these sensitive hospital environments (OR and ICU). Only two diagnostic ultrasonic Dopplers were affected, although we used signals with the highest possible average power levels to create worst-case scenarios. The WCDMA signal transmitted 250 mW, which mostly is intended for data traffic and is twice the power level of most WCDMA terminals. The GPRS signal in our tests used the maximal number of time slots (4), which give an average power level of 0.5 W. Most GPRS phones transmit at 1 or 2 time slots, which gives lower average

output power levels. Complementary tests were for made on 4 MDs, using GPRS with only 1 active time slot (pulse frequency, 217 Hz; average power, 0.125 W). The distances when interference occurred decreased between 30% and 80% compared with the 4-time slot signal. Both UMTS and GSM/GPRS have a feature so the mobile terminals can reduce their output power to the minimum the level required to reach the closest radio base station. In an office building with a well designed in-building network and with a distributed multiantenna system, the GSM mobile terminals average output power levels were reduced to approximately 4% (15 mW) of the maximum level (12). This power reduction depends on the network coverage and signal conditions that are affected by motion, but because the mobility inside a hospital is limited, this power fluctuation was not considered in this study. Similar reductions can be expected for GPRS and WCDMA signals. Consequently, the risk for electromagnetic interference in a real network will be less than reported in this study. In 1993, the International Electrotechnical Commission (IEC) adopted an electromagnetic compatibility (EMC) standard for MDs, IEC 60601–1–2, (13). It was specified for RFs up to 1 GHz, and not until September 2001 was a new revision published that extended to cover frequencies up to 2.5 GHz (14). Hence, many of the electronic MDs used in hospitals today were manufactured before 2001 and have therefore not been tested for the higher frequencies used by, for instance, GSM/GPRS 1800 MHz, WLAN, and WCDMA. All MDs tested in this study were purchased or manufactured before 2001, and the old infusion pump, which was the only MD that malfunctioned critically, was manufactured even before the first publication of the IEC standard. The current EMC standard for MDs specifies a 10 V/m immunity level for life-supporting equipment and a 3 vol/m immunity level for non–lifesupporting equipment. The electric field strength decreases exponentially with distance. Figure 3 presents

ANESTH ANALG 2005;101:1393–1400


Figure 3. The calculated (unbroken line) and measured electric field strength levels (points) at different distances from the dipole antenna used (1952.3 MHz; 250 mW). The immunity levels for medical devices (MDs; 10 and 3 vol/m) are the 2 horizontal lines. The dotted line is an example from a well designed in-building network where the maximal output power of the mobile terminals is reduced to 15 mW. The field strength levels were measured in the electromagnetic field research laboratory at Ericsson in Stockholm, Sweden. WCDMA ⫽ Wideband Code Division Multiple Access.

the calculated (unbroken line) and measured electric field strength levels (points) at different distances from the dipole antenna used (1952.3 MHz; 250 mW). The immunity levels for MDs are the two horizontal lines. As can be seen, the field strength is less than both immunity levels at distances more than 1 m, which means that as long as the MDs are in compliance with the IEC standard, there should be no risk of interference. The dotted line in Figure 3 is shown as an example from an in-building network where the terminals maximal output power is reduced to 15 mW. At this power level, the critical distance for lifesupporting equipment is ⬍20 cm. Regarding the curves in Figure 3, it was surprising that an ultrasound Doppler from 1997 gave interference noises at a distance as far as 4 m from the antenna. However, this device was not CE-marked, which means that it need not have been in compliance with the IEC standard. Otherwise, all life-supporting equipment tested was electromagnetic compatible with WCDMA and WLAN, even with signal sources in proximity. Consequently, both theoretical data and the present study indicate that WCDMA and WLAN have the capability to replace old-fashioned pager systems. However, this study cannot conclude that interference between the tested signals and MDs can never occur. Thus, we recommend that all hospitals conduct tests according to the American ANSI C63.18 –1997 standard (11) before implementing new telecommunication technologies. The fact that well-functioning communication systems are of highest importance for a hospital was revealed in an official Australian survey of hospital

1399

admissions in the mid 1990s (15). Communication problems, in a context more inclusive than just technical communication devices, were the most common single reason responsible for (11%) preventable disabilities and deaths and were nearly twice as common as those caused by inadequate medical skill (6%). The numbers of deaths caused by communication problems were approximately 1500 per year. Even if modern telecommunication technologies could contribute to decreasing this figure by only a small percent, many lives can be saved. We have not found any published reports about patients who have died because of an incidence caused by electromagnetic interference in an ICU or OR. It is thus time to put the risk for electromagnetic interference into a proper perspective and instead use the many possibilities offered by new telecommunication technologies. New services could be introduced in hospitals, e.g., constant access to telephones, smart alarm systems, radiograph pictures directly to the doctor via cellular phones, and video consultations with paramedics at an accident. According to one report (16), there could also be substantial economic advantages. In this report it is suggested that the American health system could save $30 billion per year with improved telecommunications. These figures may be optimistic, but money could be saved if a modern platform for mobile communication were implemented in emergency hospitals. However, the telecommunication companies have to develop terminals designed for hospital applications containing just those functionalities important for hospital use. Compared with WLAN, UMTS (WCDMA) has the advantage of being designed primarily for both speech and text communication. To receive alarms as text-message and then be able to call back immediately with the same device will be an important improvement (17). It has a larger data transmission capacity, causes less interference than GSM/GPRS, and is, thereby, suitable to be the technology that replaces the hospital pager systems. We conclude that these new wireless technologies can be used in most sensitive parts of hospital environments. WCDMA and WLAN (802.11b, 2.4 GHz) cause minimal interference and can be used in ICUs and ORs. Direct contact between MDs and terminals ought to be avoided. For GPRS (1800 MHz), it is recommended that the hospital staff keep a distance of 1 m between the terminals and the MDs. However, medical staff must always be aware of the fact that interference can occur and should know how to recognize it. Hospitals are recommended to have a management protocol determining how to test their MDs regarding EMC and, if necessary, replace old equipment with low immunity. In public areas, such as cafeterias, corridors, and waiting areas, UMTS (WCDMA), WLAN, and GPRS (1800 MHz) mobile

1400


phones/terminals should be allowed without restrictions because the risk for interference is minimal. This study was conducted with technical and financial support from Ericsson AB and TeliaSonera Sverige.

References 1. Available at: http://www.wirelessmessaging.org/media/ info2.0.html. Accessed September 16, 2005. 2. Silberberg JL. Performance degradation of electronic medical devices due to electromagnetic interference. Compliance Engineering 1993;10:1– 8. 3. Klein AA, Djaiani GN. Mobile phones in the hospitals: past, present and future. Anaesthesia 2003;58:353–7. 4. Hietanen M, Sibakov V, Ha¨llfors S, von Nadelstahl P. Safe use of mobile phones in hospitals. Health Phys 2000;79:S77– 84. 5. Barbaro V, Bartolini P, Baenassi M, et al. Electromagnetic interference by GSM cellular phones and UHF radios with intensive care and operating room ventilators. Biomed Instrum Technol 2000;34:361–9. 6. Morrisey JJ, Swicord M, Balanzo Q. Characterization of electromagnetic interference of medical devices due to cell phones. Health Phys 2002;82:45–51. 7. Baba I, Furuhata H, Kano T, et al. Experimental study of electromagnetic interference from cellular phones with electronic medical equipment. J Clin Engl 1998;23:122–34. 8. Morrissey JJ. Mobile phones in the hospital: improved mobile communication and mitigation of EMI concerns can lead to an overall benefit to healthcare. Health Phys 2004;87:82– 8.

ANESTH ANALG 2005;101:1393–1400

9. Rice WP. 2.4 GHz WLAN EMI in medical devices. J Clin Engl 2000;25:260 – 4. 10. Tan KS, Hinberg I. Effects of wireless local area network (LAN) system, a telemetry system, and electrosurgical devices on medical devices in a hospital environment. Biomed Instrum Technol 2000;34:115– 8. 11. Institute of Electrical and Electronics Engineers, Inc. American national standard recommended practice for on-site ad hoc test method for estimating radiated electromagnetic immunity of medical devices to specific radio-frequency transmitters. (standard C63.18). Piscataway, NJ: IEEE, 1997. 12. Persson T, To¨rnevik C, Larsson LE, Love´n J. GSM mobile phone output power distribution by network analysis of all calls in some urban, rural and in-office networks, complemented by test phone measurements. Bioelectromagnetics (BEMS) Abstract book (p 182) from the Twenty-Fourth Annual Meeting, Quebec City, Quebec, Canada, June 23–27, 2002. 13. IEC 606–1–2, collateral standard: electromagnetic compatibility— requirements and tests. 1st ed. Geneva: International Electrotechnical Commission, 1993. 14. IEC 606–1–2, collateral standard: electromagnetic compatibility— requirements and tests. 2nd ed. Geneva: International Electrotechnical Commission, 2001. 15. Zinn C. 14,000 preventable deaths in Australian hospitals. BMJ 1995;310:1487. 16. Little AD. Telecommunications: Can it help solve America’s health care problem? Cambridge, MA: Arthur D. Little, 1992. 17. Halkes M. A response to “Mobile phones in the hospital: past, present and future”, Klein AA, Djaiani GN, Anaesthesia 2003; 58:353–7. Anaesthesia 2003;58:1147.