Mobile phone interference with medical equipment and its clinical relevance: a systematic review

In Australia, 72% of households own a mobile phone, an increase of more than 10% in 2 years.1 Mobile phone use is restricted within hospitals because of a perceived risk of interference with medical equipment which could endanger the safety of patients. Many of these restrictions relate to outdated mobile phone technology, leading to debate as to their validity.2-6 Digital mobile phones are now used in Australia, and we need to ascertain if interference caused by these phones is clinically relevant.

Digital mobile phones in Australia (GSM and CDMA) currently operate at frequencies around 900 MHz with single-band technology and between 800 MHz and 1800 MHz with dual-band technology; there are plans to extend the band spectrum to 2100 MHz for GSM mobile phones.7,8 The amount of electromagnetic radiation (EMR) emitted is low, and is governed by the inverse square law (exponential decay of radiation with distance from the emitting object). Typically, even in standby mode, mobile phones produce EMR of up to 42 V/m at 0.1 m, dropping to below 7 V/m at 1 m.9 Phones overseas may operate at different frequencies and have different EMR profiles.

The first issue concerning mobile phones is the EMR they produce, and the safe distance they should be used from medical equipment that may be susceptible to EMR. The second issue is the ability of medical equipment to resist EMR. Most world standards relate to the United States Food and Drug Administration, which issued a voluntary standard in 1979 specifying that medical equipment should be immune from interference in fields of up to 7 V/m within the frequency range of 450–1000 MHz.9 Clearly, mobile phones may breach these limits if operated within one metre, but this does not automatically translate to electromagnetic interference (EMI).

Finally, hospital construction needs to take into account EMR from different areas within the hospital, as well as external sources, to limit interference with medical equipment.10 For example, allowing mobile phone use in a hospital corridor adjacent to a ward with sensitive medical equipment susceptible to EMR could be problematic.

The focus of this review is to assess whether studies can establish if the interference created by EMR from mobile phones is clinically relevant in Australia.

After all potentially relevant studies had been found, those that did not fulfil our selection criteria were excluded (Box 1). No studies testing only analogue phones were found. One study was excluded because it did not differentiate between digital and analogue phones,11 and another two were excluded because ambiguous reporting of results made drawing conclusions difficult.12,13 Of the studies testing phones at frequencies used in Australia, one tested only at 900 MHz,9,14 four tested at both 900 MHz and 1800 MHz,15-18 and two tested only at 1800 MHz19 (Box 2 and Box 3). Thus, seven studies fulfilled our selection criteria.9,14-19

The 29 categories of equipment tested and the presence of clinically relevant EMI are summarised in Box 4. The maximum distance from a device at which EMR caused clinically relevant EMI ranged from 5 cm to 200 cm (Box 2) at 900 MHz and 5 cm to 80 cm at 1800 MHz (Box 3). Most clinically relevant EMI occurred at less than 1 m, with three ECG monitors, two external pacemakers and two infusion pumps recording EMI at up to 2 m.

The studies confirm that EMI occurs when mobile phones operate close to sensitive medical equipment. However, clinically relevant EMI is much less frequent. In the two largest studies, 23%–53% of individual devices were affected by EMI, but only up to 10% were clinically relevant.15,18 Conclusions were similar among the studies, with all recommending some form of restriction.

Only one study13 fulfilled all our ideal criteria for testing, but it was excluded because of ambiguous reporting of results. Further, low numbers of tests were done, weakening study validity in some cases. This may explain the heterogeneity of results and reflects the lack of standardised testing, recording and reporting procedures.

There is no standard policy on mobile phone use in hospitals in Australia. In Victoria, hospitals set their own restrictions, and, in New South Wales, hospitals follow a 2003 Health Department directive restricting use in clinical areas.20 There has been only one report of a death, which was caused by a respirator being switched off by EMI from a mobile phone.21 Two articles involved monitoring the use of mobile phones by hospital clinical staff for up to 6 months, and found no clinically adverse effects.16,22 Finally, two-way radios produce similar EMR and have been used for decades in hospitals, with no reports of medical equipment interference or malfunction.18

Not having a standard definition of “clinically relevant” EMI makes study comparison difficult. We chose a broad definition in an attempt to overcome the different definitions used. Irnich and Tobisch described clinically relevant EMI as that potentially causing “realistic danger”, with individual event explanations.15 The UK Medical Devices Agency classified EMI as events having a direct effect on the patient and those not directly affecting the patient.18 The remaining five articles considered EMI as individual events, categorising them as those resulting in “interrupted function” “faulty action”, such as altering of an infusion pump rate; or “switching-off” of a device.9,14,16,17,19 Simple screen interference or sounds not altering recordings were not considered clinically relevant. Finally, malfunctions caused by EMI may be “silent” (ie, the equipment may operate slightly outside its normal range, but not enough to trigger an alarm).23 These events are difficult to record and quantify.

Only one study serviced equipment before and after testing (but was excluded from review because of ambiguous reporting of results)13 and one study used a simulator instead of a handset.14 Despite variable equipment and testing conditions, the presence of any recorded EMI for individual devices was greater than 10% in all studies. A much wider range occurred with clinically relevant EMI and, although pooling of data was not the object of this study, it is worth noting that, overall, clinically relevant EMI was observed in 45 of 479 devices tested (9%) at 900 MHz, and in 14 of 457 devices tested (3%) at 1800 MHz.

One should be cautious about drawing conclusions from overall study results. Of 29 categories tested, none were consistent across all studies (Box 4), and it is impossible to ascertain (and unlikely) that any two devices tested were identical. Devices were not matched and numbers of devices in each category were different. For example, 66 infusion pumps were tested by Irnich and Tobisch, with none having clinically relevant EMI,15 whereas Clifford et al found half of their six infusion pumps showed clinically relevant EMI.9 Attempting to minimise bias between individual devices, we also report results by category (Box 4). However, some studies chose only one representative device in a category, and again these were not matched, potentially skewing overall category results in either direction.

Based on the available evidence and accepting its limitations, some form of restriction on mobile phone use in hospitals in Australia seems sensible. It appears at least 4% of medical devices could experience clinically relevant EMI when a mobile phone is within 2 m. However, the number of events when mobile phones are used more than 1 m from medical equipment is almost negligible, with only 0.01% (7/479) of all devices tested at 900 MHz showing clinically relevant EMI at this range.

Mobile phones operating at 1800 MHz appear to cause less clinically relevant EMI and need to be closer to medical equipment to affect it. However, in Australia, dual-band mobile phones operate at both frequencies, and it is impossible to be sure which frequency is in use at any one time. In contrast, in Europe and Asia, phones operate predominantly in the 1800 MHz band. This information will be more relevant when phones operating in the 2000 MHz band become available in Australia, although further testing will be necessary.

We do not believe that mobile phone detection devices that alert or verbally warn owners to switch off phones (used in some hospitals in the United Kingdom) are warranted.4 The safest option is the “1 m rule” proposed by Irnich and Tobisch15 — restrict mobile phone use to greater than 1 m from equipment. Staff, patients and visitors need to be made aware of the risk to patient safety if mobile phones are used, particularly within 1 m of sensitive medical equipment. To avoid confusion, it is probably best to restrict use by patients and visitors unless they are in mobile phone “friendly zones” within a hospital.16 Although policies have been recommended in Australia to take into account changing mobile phone technology and relax some restrictions, their local implementation is difficult and may be costly, involving education of clinicians, patients and the public to be successful.20

Future research into clinically relevant EMI in medical equipment from mobile phones needs to be standardised to enable adequate comparison. Firstly, with testing, the 18-step method proposed by Tri and colleagues could be adopted.13 Furthermore, the definition of clinically relevant EMI needs to be clearly stated. Definitions should be standardised across studies and a large sample of devices tested in each study — only then can further conclusions be drawn. Ultimately, medical equipment needs to be manufactured to resist EMI from mobile phones, with standards that take into account new phone technology.