Radiation Protection Dosimetry
Vol. 83, Nos 1–2, pp. 165–169 (1999)
Nuclear Technology Publishing
EXPOSURE METRICS FOR RF EPIDEMIOLOGY: CELLULAR
8000 W. Sunrise Blvd
Plantation, FL, USA
Abstract — The parameters are described that characterise the exposure of the users of cellular phones. The parameters are
distinguished in two classes: the human and the cell phone parameters. Among the human parameters the following are discussed:
size and shape of head and neck, manner of holding the phone (left vs. right, ﬁnger tips vs. palm contact) and phone position
on the face of the user. The cell phone parameters causing the largest exposure variations are: antenna geometry (size, shape,
extended or retracted) and matching conditions; operating RF power level; proximity of tissue to RF currents on metal parts,
channel access method (analogue, pulsed, CDMA). The large variability of the RF exposure is further expanded by the variety
(ever increasing) of phone models available to users who may change service frequently or sporadically. After a brief discussion
of possible dose deﬁnitions and the uncertainty of the ‘user’ of a cell phone for a speciﬁc call, the paper analyses the critical
exposure parameters that should be investigated to characterise statistically the RF exposure of the subjects of an epidemiological
study. The improved exposure assessment of the users of cellular phones requires the cooperation of network operators and
equipment manufacturers. The statistics of the most critical parameters, those with variability greater that 10:1, can be collected
by modifying the software and hardware of the cell phone equipment. The paper suggests base station software modiﬁcations
and the introduction of cell phone ‘dosemeter’ devices that record some of the critical exposure parameters. A certain number
of these ‘dosemeters’ should be distributed among subscribers to determine the statistical variations of the RF exposure from
cell phones. The paper concludes by recommending a pilot dosimetric study independent from any epidemiological study.
INTRODUCTION On the whole, exposure assessment has been rudi-
Epidemiological studies of users of wireless tech- mentary and unsatisfactory. In particular, epidemiolog-
nology hold the promise to yield deﬁnitive information ical studies of cellular telephone users have relied upon
about the long-term safety of this widespread and rela- data on call duration and user identiﬁcation from billing
tively new form of exposure to electromagnetic ﬁelds. records (22). To date there has been no capability to
The paramount value of epidemiologic research is ensure that the billed customer was the person exposed
recognised by various authorities, including prac- during the phone call nor to measure the level of
titioners of public health (1,2), the guidelines of govern- exposure by either direct or indirect methods. Ideally,
ment agencies (3,4), the risk assessment guidelines of the an epidemiological study would measure dose to the
International Agency for Cancer Research (5). user by objective methods, but even the deﬁnition of
Despite the importance of epidemiological data, for dose for users of wireless devices still need to be estab-
assessment of risk for users of wireless equipment, there lished.
is at present only one directly relevant published study. This paper addresses many of the gaps in understand-
In that paper (6) Rothman et al. present preliminary data ing the electromagnetic ﬁeld exposure to users of cellu-
showing no effects on overall mortality. lar telephones. The following sections identify key fac-
As can be seen from recent reviews (e.g. Ref. 7) most tors that determine user exposure, place them in relative
epidemiologic research on radiofrequency electro- order of signiﬁcance, propose a deﬁnition of dose, and
magnetic ﬁelds concerned other technical uses of micro- recommend several technical means (constituting dose-
waves such as radar (8–10) and, to a limited degree, meter-equipped phones and radios) to advance capabili-
exposures near broadcasting transmitters (11–13), and ties for exposure assessment in the context of epidemio-
microwave towers (14) Exposures in these environments logical research.
are very different from exposures in close proximity to
the RF sources of a cellular telephone. Several research RELEVANT HUMAN AND CELL PHONE
papers have presented data on persons whose exposure EXPOSURE PARAMETERS
was evaluated by job titles (10,15,16) and hobbies (17–19).
The results provide a mixture of possible associations The RF exposure of cellular phone users occurs in the
with disease and evidence against such associations. In immediate vicinity of the electromagnetic ﬁeld sources,
some instances there have been efforts to estimate the which are the currents and the charges on the metal parts
level of exposure based on measurements or estimates of the device. In electrodynamical terms, the exposure
of radiofrequency exposure for particular jobs (8,20,21). happens in the reactive near-ﬁeld of the RF sources. In
this condition, there is an energy interaction between the power levels. The output power of the phone is con-
RF sources (the phone) and the parts of the body in trolled by the base station and most cell phone networks
contact or in close proximity to the phone. ‘Close prox- have a 100:1 operating range for the RF power of the
imity’ means a distance no greater than /2 , or about mobile units. Cellular networks are designed to operate
6 cm in the band 800–900 MHz. In physical terms, the so that the voice quality of one channel (one frequency)
energy interaction, in matched conditions, minimises the is limited by the interference of other signals using the
electromagnetic energy stored and maximises that emit- same frequency in other parts of the cellular system. For
ted by the user’s body and cellular phone. this reason the network reduces the RF power of each
The energy interaction is substantially determined by roaming unit at the minimum level compatible with the
the geometrical shape of the phone components and the voice quality required for a phone conversation.
proximity of body organs, as well as their dielectric and
dissipative characteristics. From these general consider-
ations it is possible to identify the relevant human para- DEFINITION OF DOSE
meters: Since the early 1980s, the deﬁnition of speciﬁc
(a) Size and shape of the head and neck. absorption rate (SAR) has gained substantial support as
(b) Internal anatomy (fat, muscle, bone content) of a valid metric of human exposure to RF electromagnetic
head and neck. energy. The SAR is deﬁned as the ratio of the rate of
(c) Manner of holding the phone (ﬁnger tip or palm RF energy absorption in a very small volume of tissue
contact, left or right hand). and the mass of that volume. SAR is measured in watts
(d) Dielectric characteristics of human tissue. per kilogram (W.kg−1). The absorbed energy increases
(e) Phone position on the face of the user. the kinetic energy of the random motion of the atoms
(f) Hand size and sweatiness. and ions of the exposed biological tissue.
With the given deﬁnition of exposure metric, it is
Other human geometric or anatomical parameters could easy to deﬁne the dose as the sum of the products
cause minor variations in exposure, but the list above
has proven theoretically and experimentally (23,24) suf-
ﬁcient to characterise the user exposure. Dose = SAR (ti) ti
The list of cell phone parameters is long because of
the variety of phone shapes, sizes and antenna designs where ti is the time interval of exposure with dose
as well as the number of RF wireless technologies avail- rate SAR(ti).
able to the consumer: Since SAR is a point and an instant function, it
remains to specify which space- and time-averaging fea-
(a) Antenna geometry (extended, retracted, size and
tures of the SAR metric must be selected for use in the
dose formula. The time averaging should be performed
(b) Antenna isolation from phone case.
over a period of the RF signal or over a time interval
(c) Distance of metal parts (including the antenna)
during which the phone output power is constant. There
from the body of the user.
is also the possibility of considering only the SAR
(d) Operating RF power level and frequency.
values above a threshold: a dichotomous function which
(e) Channel access technology (Frequency Domain
requires a detailed knowledge of the statistics of human
Multiple Access, Code Domain Multiple Access,
exposure, which may not be available at the present
Time Domain Multiple Access).
The reactive coupling between the body and the The volumetric averaging of SAR has several poss-
phone determines the antenna impedance. The power ible options. SAR can be averaged over a small or a
level and the radiation impedance establish the intensity large volume of tissue. With the most reﬁned theoretical
of the RF currents on the metal parts of the phone and and experimental tools of RF dosimetry, it is possible
on the body of the user; these currents induced in the to average SAR over a volume of about 0.1 cm3 or
body constitute the human exposure. approximately 100 mg of tissue shaped like a cube.
Assessment of user exposure from cellular phones is With this building block, it is possible to average SAR
difﬁcult to model and to measure. Theoretical dosimetry over an entire organ or major parts of an organ. The
using the most recent numerical methods requires all the type and size of volume averaging should be selected
storage and numerical/algorithmical resources of high after the speciﬁc pathologies of an epidemiological
power computer workstations. Experimental dosimetry study have been identiﬁed so their aetiologies can be the
(direct measurement) requires specialised RF sensors basis of SAR averaging criteria. This selection requires
and expensive robotic positioning systems because of cooperation among epidemiologists, pathologists and
the critical distance factors between RF sources and experts in RF dosimetry. In fact, the importance of the
observation points in simulated tissue. close cooperation between epidemiologists, pathologists
An important dosimetric variable that also must be and RF dosimetry experts to evaluate the feasibility of
considered is the fact that the phone operates at different an epidemiologic study of the users of cellular phones
EXPOSURE METRICS FOR RF EPIDEMIOLOGY
cannot be overstated. Without close consultation, the ‘critical’ metric in the sense deﬁned above, but can be
selection of the exposure metrics might be inappropriate controlled as will be explained in the next section.
and defeat or dilute the purpose of a large study. The variability of exposure with head shape, size and
tissue types can be bracketed by using numerical or
experimental dosimetry. Results to date show that the
WHO IS CALLING? WHO IS USING THE PHONE? maximum variation of exposure parameters from strictly
The RF exposure of the individuals in a prospective anatomical features is less than 2:1 (25) and thus is not a
cohort or a case-control study must be measured over a critical parameter.
period of time, in order to evaluate the dose. Is there
available a methodology or a means to ensure that a IMPROVED DOSIMETRIC DATA COLLECTION
speciﬁc phone is used only by one person or that a call
is performed only by one individual? At this time, the At the present time there is readily available tech-
answer must be: No. nology to enhance user device identiﬁcation; there is
People sometimes share a phone as part of an ofﬁce also the possibility of a novel ‘dosemeter phone’ that
or family plan or just during a call if the need arises. uses state of the art instrumentation and software to
While ofﬁce sharing is decreasing with the cost of the characterise cell phone exposure.
subscriber units, family sharing is becoming more com-
mon with the penetration of the cell phones in the larg- User and device identiﬁcation
est strata of the population. For example, teenagers on
a weekend or evening outing often use a parent’s phone. Uncertainty about user identiﬁcation can be decreased
From billing records, where only monthly air time by use of phone ‘smart cards’, which are plastic cards
and the number of calls are collected, it is not possible with read-only memory (ROM) electronics. When the
to reconstruct with certainty who made the phone calls. ‘smart card’ is inserted in the cellular phone, the device
The following sections discuss the development of an reads the identiﬁcation of the card owner. This infor-
available personal identiﬁer for call time charging pur- mation is transmitted to the base station for billing pur-
pose, which will be helpful to reduce the uncertainties poses and/or to decrease the prepaid value of the card
about user identity. at the end of the conversation. Multiple users of a single
In the next few years, using voice recognition algor- phone can have individual cards.
ithms, it will be possible to disable the phone if a user Call duration is recorded by current technology using
with a voice different from that which ‘trained’ the beginning and ending time of the RF link during a con-
voice recogniser, tries to use the device. As cell phones versation. This information can be correlated to phone
become the storage device of private data, a voice rec- or ‘smart card’ number. The phone electronic serial
ogniser will be an effective way to prevent the misuse number serves to identify the phone model. These ident-
of sensitive information stored in the phone memory. ifying data can be recorded and made available for
The voice recognition feature also can be used to ident- exposure assessment purposes. The various carriers
ify subscriber air time, if the change in voice character- need to establish uniform procedures and formats for
istics and the call duration are recorded by the com- data collection and reporting.
A ‘dosemeter’ phone
CRITICAL EXPOSURE METRICS At this time there is no reliable statistical information
From the discussion in the previous sections it is clear on the operational power level of phones, how people
that cellular phone user exposure involves a large num- switch the phone from the left to the right ear and how
ber of parameters, each with its range of variability. We they position or reposition the phone at the same side
deﬁne as ‘critical’ the parameters that change the of the head. This information can be collected by trained
exposure of a user over a range of 10:1 or greater. With observers or by instrumenting cellular phones to become
this criterion one can identify the following as para- ‘dosemeters’. However, the ‘dosemeter’ phones cannot
meters to be controlled: be substantially different in weight, size and shape from
real phone models.
(1) User identiﬁcation and call duration. In brief, these are the technical features of a poss-
(2) RF power level from cellular phone (100:1). ible ‘dosemeter’:
(3) Left or right hand use (side exposure 10:1).
(4) User positioning of phone at the face (10:1). (1) Pressure points on the narrow sides of the phone.
(5) Phone model (20:1). These are simple elastomeric pressure switches,
which identify the position of the thumb of the user.
The possibility of an antenna extended or retracted The side with the lower number of pressure points
causes a variation of about 5:1 in the SAR of certain identiﬁes the thumb location. One can imagine situ-
regions of the head. Therefore antenna position is not a ations where this device is defeated, e.g. right hand
against left ear, but these should be a statistical min- microcomputer memory and input/output ports can
ority of the cases. If the phone is operated from a be used to store and transmit periodically all the
holster on the belt, all pressure points are equally information collected by the other sensors in the
switched on. ‘dosemeter’.
(2) Operating power level data. The phone records the
It is clear that the instrumentation of a number of
power level and the duration of the RF transmission
‘dosemeter’ phones is a substantial task, although well
during a conversation. The information can be
within the present technology of cellular telephony. In
stored and transmitted to the base station during or
fact the prototype is currently under development for the
at the end of a call.
activities of IEEE Standard Coordinating Committee 34.
(3) Antenna up or down position information. This is
easily achieved by reading the state of a switch con-
nected to the base of the antenna. The information CONCLUSION
can be transmitted at the beginning of the RF link. This paper has outlined methodologies which, if feas-
Changes in the conﬁguration of the antenna during ible,hold considerable potential to increase the statistical
connect time will be transmitted and recorded at the certainty of validating the RF exposure of users of cellu-
base station. lar phones. The very deﬁnition of ‘dose’ requires close
(4) Phone position at the face of the user. This infor- cooperation between pathologists, epidemiologists and
mation is the most difﬁcult to instrument for collec- experts in RF dosimetry after the speciﬁc pathologies
tion. It requires a set of pressure sensors around the of an epidemiologic study have been identiﬁed. This
ear piece of the phone and a device to measure the point is stressed because of its fundamental importance,
distance between the axis of the antenna and the given the fact that there is no proven pathway to any
head of the user. The distance is measured at the speciﬁc pathology by RF energy at the low levels of
base of the antenna, so the distance metering instru- SAR typical of cellular phone exposure.
ment is mounted on top of the phone case. There To increase the statistical validity of the measure of
are at least two available prototype devices that can the RF currents in the users of cellular devices, ‘dose-
be used for this purpose. They are currently being meter’ phones should be developed and tested for data
developed at the MIT Media Laboratory in Cam- collection purposes. This entails a dosimetric pilot study
bridge, MA, and at the Lawrence Livermore that should be conducted independently of any epidemi-
National Laboratories in Livermore, CA. ologic study. Such a pilot study is a substantial effort
(5) Antenna matching conditions. This information that requires close cooperation between manufacturers
complements the data provided by the positioning and cell phone service providers, as well as pathologists,
instrumentation. The antenna matching conditions epidemiologists and experts in RF dosimetry.
can be measured by resorting to a relatively simple If statistically valid RF exposure data can be gener-
RF circuit board currently being developed as part ated by ‘dosemeter phones’, then it may be possible to
of the activities of IEEE Standard Coordinating design prospective epidemiologic studies to determine
Committee 34. The RF board output DC voltages whether cell phones cause adverse health effects in
(giving the amplitude and phase of the antenna humans. Without the many preparatory efforts outlined
reﬂection coefﬁcient) are read once a minute and in this paper, epidemiologic studies of users of cellular
stored in the memory of a microcomputer. The phones run the risk of being incorrect or futile.
1. Pike, M. C. Epidemiology and Risk Assessment: Estimation of GI Cancer Risk from Asbestos in Drinking Water and Lung
Cancer Risk from PAHs in Air. In: Risk Quantitation and Regulatory Policy. Eds D. G. Hoel, R. A. Merrill and F. P. Perera
(NA: Cold Spring Harbor Laboratory) pp. 55–64 (1985).
2. Kaldor, J. and Day, N. The Use of Epidemiologic Data for the Assessment of Human Cancer Risk. In: Risk Quantitation
and Regulatory Policy. Eds D. G. Hoel, R. A. Merrill and F. P. Perera (NA: Cold Spring Harbor Laboratory) pp. 79–87 (1985).
3. Albert, R.E. U.S. Environmental Protection Agency Revised Interim Guideline for the Health Assessment of Suspect Carcino-
gens. In: Risk Quantitation and Regulatory Policy. Eds D. G. Hoel, R. A. I. Merrill and F. P. Perera (NA: Cold Spring
Harbor Laboratory) pp. 307–329 (1985).
4. USEPA. Proposed Guidelines for Carcinogen Risk Assessment (Washington, DC: Ofﬁce of Research and Development,
5. IARC. On the Evaluation of Carcinogenic Risks to Humans. Monographs Vol. 52, pp 15–36 (1991).
6. Rothman, K. J., Loughlin, J. E., Funch, D. P. and Dreyer, N. A. Overall Mortality of Cellular Telephone Customers. Epidemi-
ology 7(3), 303–305 (1996).
7. Bergvist, U. Review of Epidemiological Studies In: Mobile Communications Safety. Eds N. Kuster, Q. Balzano, J. C. Lin
(London: Chapman & Hall) pp. 147–170 (1997).
8. Robinette, C. D., Silverman, C. and Jablon, S. Effects upon Health of Occupational Exposure to Microwave Radiation
(Radar). Am. J. Epidemiol. 112, 39–53 (1980).
EXPOSURE METRICS FOR RF EPIDEMIOLOGY
9. Lester, J. R. and Moore, D. F. Cancer Incidence and Electromagnetic Radiation (meeting abstract). Cancer Res. 21 319
10. Hill, D. G. A Longitudinal Study of a Cohort with Past Exposure to Radar: the MIT Radiation Laboratory Follow-up Study
PhD Dissertation, The Johns Hopkins University (University of Michigan Dissertation Service, Ann Arbor) (1988).
11. Maskarinec, G., Cooper, J. and Swyger, L. Investigation of Increased Incidence in Childhood Leukemia near Radio Towers
in Hawaii: Preliminary Observations. J. Environ. Pathol. Toxicol. Oncol. 13(1), 33–37 (1994).
12. Dolk, H., Shaddick, G., Walls, P., Grundy, C., Thakrar, B., Kleinschmidt, I. and Elliot, P. Cancer Incidence near Radio and
Television Transmitters in Great Britain, Part 1. Sutton Coldﬁeld Transmitter. Am. J. Epidemiol. 145, 1–9 (1997).
13. Dolk, H., Elliot, P., Shaddick, G., Walls, P. and Thakrar, B. Cancer Incidence near Radio and Television Transmitters in
Great Britain, Part II. All High-power Transmitters. Am. J. Epidemiol. 145, 10–17 (1997).
14. Selvin, S., Schulman, J. and Merrill, D. Distance and Risk Measures for the Analysis of Spatial Data: a Study of Childhood
Cancers. Soc. Sci. Med. 34(7), 769–777 (1992).
15. Milham, S. Jr Mortality from Leukemia in Workers Exposed to Electrical and Magnetic Fields. N. Engl. J. Med. 307,
16. Thomas, T. L., Stolley, P. D., Sternhagen, A., Fontham, E. T. H., Bleecker, M. L., Stewart, P. A. and Hoover, R. N. Brain
Tumor Mortality Risk among Men with Electrical and Electronics Jobs: a Case-control Study. J. Natl Cancer Inst. 79(2),
17. Milham, S. Silent Keys: Leukaemia Mortality in Amateur Radio Operators. Lancet 1, 812 (6 April 1985).
18. Milham, S. Increased Mortality in Amateur Radio Operators due to Lymphatic and Hematopoietic Malignancies. Am. J.
Epidemiol. 127(1), 50–54 (1988).
19. Milham, S. Mortality by License Class in Amateur Radio Operators. Am. J. Epidemiol. 128(5), 1175–1176 (1988).
20. Grayson, J. K. Radiation Exposure, Socioeconomic Status, and Brain Tumor Risk in the US AirForce: a Nested Case-control
Study. Am. J. Epidemiol. 143(5), 480–486 (1996).
21. Szmigielski, S. Cancer Morbidity in Subjects Occupationally Exposed to High Frequency (Radio frequency and Microwave)
Electromagnetic Radiation. Sci. Total Environ. 180(1), 9–17 (1996).
22. Funch, D. P., Rothman, K. J., Loughlin, J. E. and Dreyer, N. A. Utility of Telephone Company Records for Epidemiologic
Studies of Cellular Telephones. Epidemiology 7(3), 299–302 (1996).
23. Kuster, N. and Balzano, Q. Energy Absorption Mechanism by Biological Bodies in the Near Field of Dipole Antennas above
300 MHz. IEEE Trans Veh. Technol. VOVT-41, 17–23 (1992).
24. Hombach, V. et al. The Dependence of EM Energy Absorption upon Human Head Modelling at 400 MHz. IEEE Trans.
Microwave Theory Technol. MTT-44, 1865–1873 (1996).
25. Schonborn, F. et al. Differences in Energy Absorption between Heads of Adults and Children in the Near Field of Sources
Health Phys. 74(2), 160–168 (1998).