Melanoma Incidence and Frequency Modulation Broadcasting

32Archives of Environmental Health

DOLK ET AL.

1

found that there was significantdecline in skin and bladder cancer incidence amongadults in England as distance from a frequency modula-tion (FM) broadcasting tower increased. In the secondpart of the study, these investigators reported a similartrend for individuals who lived various distances fromFM and TV towers, but the effect was less pronouncedthan in the first study.In the current study, we sought to determine if thefindings of Dolk et al.

1

on a local scale generally heldtrue for a country as a whole. If the findings held true,our second objective was to address possible improve-ments that could be initiated to reduce risk.We conducted the study in several steps. We initiallyaddressed the survival probability of the general popu-lation to correctly account for natural deaths in the dataanalysis. We then analyzed independent sets of expo-sure and incidents data from 4 different countries toextract each individual exposure–time-specific inci-dence function (Etsi). Published sets of age-specific inci-dence data (ASI) were reviewed for different years.Finally, the numbers of radio stations that were locallyreceivable were correlated with the melanoma inci-dence for the 288 communities in Sweden.

Method

The hypothesis we tested was that the increase inmelanoma incidence over the years could be explainedentirely by the expansion of the FM broadcasting net-work. If true, we should be able to calculate the num-ber of cases each year on the basis of some publiclyknown facts: (1) the total exposure history of the popu-lation (i.e., number of individuals covered by theexpanding FM broadcasting network each year); (2) thesurvival probability of the population (for the correctaccounting of the number of exposed individualsremaining alive over time); and (3) age-specific inci-dence data, which we used to determine the probabili-ty of contracting melanoma vs. exposure time.

Melanoma Incidence andFrequency Modulation (FM) Broadcasting

ÖRJAN HALLBERGOLLE JOHANSSONExperimental Dermatology UnitDepartment of NeuroscienceKarolinska InstituteStockholm, Sweden

ABSTRACT. The incidence of melanoma has been increasing steadily in many countries since1960, but the underlying mechanism causing this increase remains elusive. The incidence ofmelanoma has been linked to the distance to frequency modulation (FM) broadcasting tow-ers. In the current study, the authors sought to determine if there was also a related link ona larger scale for entire countries. Exposure–time-specific incidence was extracted fromexposure and incidence data from 4 different countries, and this was compared with report-ed age-specific incidence of melanoma. Geographic differences in melanoma incidencewere compared with the magnitude of this environmental stress. The exposure–time-specificincidence from all 4 countries became almost identical, and they were approximately equalto the reported age-specific incidence of melanoma. A correlation between melanoma inci-dence and the number of locally receivable FM transmitters was found. The authors con-cluded that melanoma is associated with exposure to FM broadcasting.

January/February 2002 [Vol. 57 (No. 1)]33

1.Survival statistics.

As time passes, exposed individ-uals will die from many causes, and they will be replacedby newborns. The out-dying function was calculated basedon survival data and the age distribution within theSwedish population. Thefunction is different from that ofnewborn babies inasmuch as it is based on a mixed pop-ulation: from 1 to 90 yr of age (Fig. 1). One must consid-er this function if the number of incidents by year are tobe calculated correctly, given that people who die will nolonger contribute to the incidence statistics.

2.Exposure–time-specific incidence.

It is necessaryto know the exposure–time-specific incidence when apossible effect from a continuous environmental stressis to be estimated. The ASI for melanoma in Sweden isshown in Figure 2. ASI increases from the age of puber-ty (as it also does in Norway). It is assumed that an adultpopulation follows the ASI curve (i.e., from 12 yr of age)immediately upon start of the exposure.If the measured incidence is a function of exposuretime only and the natural high-age incidence withoutthe environmental stress is low, then the ASI shouldapproach a constant level at high ages. This is sobecause old people have not been exposed for a longertime than has elapsed since the start of the exposure,irrespective of their age. Consequently, we expectedthis constant level to increase by calendar time.

3.Calculation of annual cases.

The annual numbersof the disease in a country can be calculated by multi-plying each year’s number of exposed people by the Etsiand by the relevant survival function. The response fromall exposed people is then added to give the total num-ber of incidents per year. The formal mathematics isprovided in the Appendix.

Results

1.A literature study revealed a link to the distancefrom an FM tower

1

(as was mentioned earlier). We alsofound a report that claimed that cancer could developfaster in a radiofrequency (RF) field.

2

Cancer incidence

0102030405060708090100020406080100

Years from start

S u r v i v i n g ( % )

AverageNewborn

Fig. 1. Survival function for the average population (solid line) and for newborn babies (dashedline).

01020304050600102030405060708090

Age (years)

A g e - s p e c i f i c i n c i d e n c e

7681869196

Fig. 2. Age-specific incidence of melanoma in Sweden, 1979

–

1996. As was expected, the high-age flat level is increasing by each year. The top curve (1996) was used for the calculation of theresponse in Sweden and Denmark.

Years from start of broadcastingAge (yr)

34Archives of Environmental Health

has also been associated with proximity to televisiontowers.

3

Other related reports were presented byAndersson et al

4

and by Westlund et al.

5

2.Data on the expansion of the FM broadcastingnetworks were collected for the United States (i.e., viaInternet), Norway

6

(i.e., direct contact with NRK [Nor-wegian Broadcasting]), Sweden,

7

and Denmark.

8

Wealso collected information on the frequencies used forFM in different countries. Western Europe and the Unit-ed States have been using 87

–

108 MHz, whereas theformer communist countries have used lower frequen-cies (i.e., approximately 70 MHz [Fig. 3]). Japan alsouses a lower frequency than Western Europe and theUnited States. Subsequent to 1992, many EasternEurope countries have transformed to the

“

Western

”

band of 87

–

108 MHz.3.The procedure we used to calculate the mela-noma incidence in Norway, Sweden, Denmark, and theUnited States is shown in Figure 4 for Norway.4.The calculations were based on the age-specificincidence reported in Norway (for Norway and theUnited States) and Sweden (for Sweden and Denmark),and they were not the result of curve-fitting. We corre-lated all calculated and measured incidence ofmelanoma, expressed in ppm per year of the respectivepopulations. The correlation was expressed by

R

2

=0.9788,

β

= 0.989, and

p

< .0001.5.For all 4 countries, the cumulative Etsi could bemodeled

9

by a log-normal distribution, sigma (base 10)of 0.52

–

0.59, and a time to 0.1% around 20 yr (Fig. 5).6.The melanoma incidences and the average FMpower densities in the 27 different counties of Swedenare shown in Figure 6.7.It was also of interest to us to determine if the total

Fig. 3. Melanoma incidences in different countries vs. frequency modulation, as was reported in1996. Some of the countries marked as

“

70 MHz

”

have unknown frequencies.

PtEsIrCeItFrUyBeAgGbDeAtFiDkUsChSeNoNzAu

P k K o O m I nI d M y I a C i A i T h K w P h S g J pE t P a K e M x H K U z P u C R G r L v M t B z A l S A R u P o E e

05101520253035

P t C e U y G bF i C h N z K o I d C i KJ p K e U z G r B z R u

Country

I n c i d e n c e 1 / 1 0 0 0 0 0

100MHz70 MHz

Fig. 4. Reported and calculated numbers of melanoma cases per year in Norway. The numbersare on top of a

“

natural

”

level of 100 cases per year that is not shown in the graph.

01002003004005006007008009001000

1 9 5 5 1 9 5 8 1 9 6 1 1 9 6 4 1 9 6 7 1 9 7 0 1 9 7 3 1 9 7 6 1 9 7 9 1 9 8 2 1 9 8 5 1 9 8 8 1 9 9 1 1 9 9 4 1 9 9 7 2 0 0 0 2 0 0 3 2 0 0 6 2 0 0 9