Abstract: The slow spread of residential electrification in the US in the first half of the 20th century from urban to rural areas resulted by 1940 in two large populations; urban populations, with nearly complete electrification and rural populations exposed to varying levels of electrification depending on the progress of electrification in their state. It took until 1956 for US farms to reach urban and rural non-farm electrification levels. Both populations were covered by the US vital registration system. US vital statistics tabulations and census records for 1920–1960, and historical US vital statistics documents were examined. Residential electrification data was available in the US census of population for 1930, 1940 and 1950. Crude urban and rural death rates were calculated, and death rates by state were correlated with electrification rates by state for urban and rural areas for 1940 white resident deaths. Urban death rates were much higher than rural rates for cardiovascular diseases, malignant diseases, diabetes and suicide in 1940. Rural death rates were significantly correlated with level of residential electric service by state for most causes examined. I hypothesize that the 20th century epidemic of the so called diseases of civilization including cardiovascular disease, cancer and diabetes and suicide was caused by electrification not by lifestyle.
Tags: Residential Electrification, Cardiovascular Disease, Malignant Disease, Diabetes, Suicide, Compact Fluorescent Light Bulbs, Halogen Lamps, Wireless Routers, Dimmer Switches, Switching Power Supplies, Radiofrequency Radiation, High Frequency Voltage Transient, Cell Phones, Cell Towers, Terrestrial Antennas, Wi-Fi, Wi-Max.
Historical evidence that electrification caused the 20th century epidemic of "diseases of civilization" †
Samuel Milham *
Washington State Department of Health, Olympia, WA, USA
† Supported by a small grant from Children with Leukemia.
* Address: 2318 Gravelly Beach Loop NW, Olympia, WA 98502-8837, USA.
Article history: Received 14 August 2009 / Accepted 18 August 2009 / Available online xxxx
Please cite this article in press as: Milham S. Historical evidence that electrification caused the 20th century epidemic of "diseases of civilization". Med Hypotheses (2009), doi:10.1016/j.mehy.2009.08.032
The slow spread of residential electrification in the US in the first half of the 20th century from urban to rural areas resulted by 1940 in two large populations; urban populations, with nearly complete electrification and rural populations exposed to varying levels of electrification depending on the progress of electrification in their state. It took until 1956 for US farms to reach urban and rural non-farm electrification levels. Both populations were covered by the US vital registration system. US vital statistics tabulations and census records for 1920–1960, and historical US vital statistics documents were examined. Residential electrification data was available in the US census of population for 1930, 1940 and 1950. Crude urban and rural death rates were calculated, and death rates by state were correlated with electrification rates by state for urban and rural areas for 1940 white resident deaths. Urban death rates were much higher than rural rates for cardiovascular diseases, malignant diseases, diabetes and suicide in 1940. Rural death rates were significantly correlated with level of residential electric service by state for most causes examined. I hypothesize that the 20th century epidemic of the so called diseases of civilization including cardiovascular disease, cancer and diabetes and suicide was caused by electrification not by lifestyle. A large proportion of these diseases may therefore be preventable. (c) 2009 Elsevier Ltd. All rights reserved.
In 2001, Ossiander and I  presented evidence that the childhood leukemia mortality peak at ages 2–4 which emerged in the US in the 1930s was correlated with the spread of residential electrification in the first half of the 20th century in the US. While doing the childhood leukemia study, I noticed a strong positive correlation between level of residential electrification and the death rate by state due to some adult cancers in 1930 and 1940 vital statistics. At the time, a plausible electrical exposure agent and a method for its delivery within residences was lacking. However, in 2008 I coauthored a study of a cancer cluster in school teachers at a California middle school  which indicated that high frequency voltage transients (also known as dirty electricity), were a potent universal carcinogen with cancer risks over 10.0 and significant dose–response for a number of cancers. They have frequencies between 2 and 100 kHz. These findings are supported by a large cancer incidence study in 200,000 California school employees which showed that the same cancers and others were in excess in California teachers statewide . Power frequency magnetic fields (60 Hz) measured at the school were low and not related to cancer incidence, while classroom levels of high frequency voltage transients measured at the electrical outlets in the classrooms accurately predicted a teacher’s cancer risk. These fields are potentially present in all wires carrying electricity and are an important component of ground currents returning to substations especially in rural areas. This helped explain the fact that professional and office workers, like the school teachers, have high cancer incidence rates. It also explained why indoor workers had higher malignant melanoma rates, why melanoma occurred on part of the body which never are exposed to sunlight, and why melanoma rates are increasing while the amount of sunshine reaching earth is stable or decreasing due to air pollution. A number of very different types of cancer had elevated risk in the La Quinta school study, in the California school employees study, and in other teacher studies. The only other carcinogenic agent which acts like this is ionizing radiation.
Among the many devices which generate the dirty electricity are compact fluorescent light bulbs, halogen lamps, wireless routers, dimmer switches, and other devices using switching power supplies. Any device which interrupts current flow generates dirty electricity. Arcing, sparking and bad electrical connections can also generate the high frequency voltage transients. Except for the dimmer switches, most of these devices did not exist in the first half of the 20th century. However, early electric generating equipment and electric motors used commutators, carbon brushes, and split rings, which would inject high frequency voltage transients into the 60 Hz electricity being generated and distributed.
With a newly recognized electrical exposure agent and a means for its delivery, I decided to examine whether residential electrification in the US in the first half of the last century was related to any other causes of death. Most cancers showed increasing mortality in this period, and many are still increasing in incidence in the developed world.
Thomas Edison began electrifying New York City in 1880, but by 1920, only 34.7% of all US dwelling units and 1.6% of farms had electric service (Table 1). By 1940, 78% of all dwelling units and 32% of farms had electric service . This means that in 1940 about three quarters of the US population lived in electrified residences and one quarter did not. By 1940, the US vital registration system was essentially complete, in that all the 48 contiguous United States were included. Most large US cities were electrified by the turn of the century, and by 1940, over 90% of all the residences in the northeastern states and California were electrified. In 1940 almost all urban residents in the US were exposed to electromagnetic fields (EMFs) in their residences and at work, while rural residents were exposed to varying levels of EMFs, depending on the progress of rural electrification in their states. In 1940, only 28% of residences in Mississippi were electrified, and five other southern states had less than 50% of residences electrified (Table 2). Eleven states, mostly in the northeast had residential electrification rates above 90%. In the highly electrified northeastern states and in California, urban and rural residents could have similar levels of EMF exposure, while in states with low levels of residential electrification, there were potentially great differences in EMF exposure between urban and rural residents. It took the first half of the 20th century for these differences to disappear. I examined US mortality records by urban and rural residence by percent of residences with electric service by state.
Table 1 - Growth of residential electric service US 1920–1956 percent of dwelling units with electric service.
Table 2 - Percent of residences with electric lighting 1930 and 1940 by state.
The diseases of civilization or lifestyle diseases include cardiovascular disease, cancer and diabetes and are thought to be caused by changes in diet, exercise habits, and lifestyle which occur as countries industrialize. I think the critical variable which causes the radical changes in mortality accompanying industrialization is electrification. Beginning in 1979, with the work of Wertheimer and Leeper , there has been increasing evidence that some facet of electromagnetic field exposure is associated epidemiologically with an increased incidence of leukemia, certain other cancers and non-cancers like Alzheimer’s disease, amyotrophic lateral sclerosis, and suicide. With the exception of a small part of the electromagnetic spectrum from infra red through visible light, ultraviolet light and cosmic rays, the rest of the spectrum is man-made and foreign to human evolutionary experience. I suggest that from the time that Thomas Edison started his direct current electrical distribution system in the 1880s in New York City until now, when most of the world is electrified, the electricity carried high frequency voltage transients which caused and continue to cause what are considered to be the normal diseases of civilization. Even today, many of these diseases are absent or have very low incidence in places without electricity.
Evaluation of the hypothesis
To evaluate the hypothesis, I examined mortality in US populations with and without residential electrification. Vital statistics tabulations of deaths , US census records for 1920–1970 , and historical US documents [8,9] were examined in hard copy or downloaded from the internet. The same state residential electrification data used in the childhood leukemia study  was used in this study. Crude death rates were calculated by dividing number of deaths by population at risk, and death rates by state were then correlated with electrification rates by state using downloaded software . Time trends of death rates for selected causes of death by state were examined. Most rates were calculated by state for urban and rural residence for whites only in 1940 deaths, since complete racial data was available by urban/rural residence by state for only 13 of 48 states. Data was available for 48 states in the 1940 mortality tabulations. District of Columbia was excluded because it was primarily an urban population. Excel graphing software  and "Create a Graph"  software was used.
I had hoped to further test this hypothesis by studying mortality in individual US farms with and without electrification, when the 1930 US census 70 year quarantine expired in 2000. Unfortunately, the 1930 US farm census schedules had been destroyed.
Rural residential electrification did not reach urban levels until 1956 (Table 1). Table 2 shows the level of residential electrification for each state for 1930 and 1940. In 1930 and 1940 only 9.5% and 13%, respectively, of all generated electricity was used in residences. Most electricity was used in commercial and industrial applications.
Figs. 1–4 were copied and scanned from "Vital statistics rates in the United States 1940–1960", by Robert Grove Ph.D. and Alice M. Henzel. This volume was published in 1968. Fig. 1 shows a gradual decline in the all causes death rate from 1900 to 1960 except for a spike caused by the 1918 influenza pandemic. Death rates due to tuberculosis, typhoid fever, diphtheria, dysentery, influenza and pneumonia and measles all fell sharply in this period, and account for most of the decline in the all causes death rate. Figs. 2–4 show that in the same time period when the all causes death rate was declining, all malignant neoplasms (Fig. 2), cardiovascular diseases (Fig. 3), and diabetes (Fig. 4) all had gradually increasing death rates. In 1900, heart disease and cancer were 4th and 8th in a list of 10 leading causes of death. By 1940 heart disease had risen to first and cancer to second place, and have maintained that position ever since. Table 3 shows that for all major causes of death examined, except motor vehicle accidents, there was a sizable urban excess in 1940 deaths. The authors of the extensive 69 page introduction to the 1930 mortality statistics volume noted that the cancer rates for cities were 58.2% higher than those for rural areas. They speculated that some of this excess might have been due to rural residents dying in urban hospitals. In 1940, deaths by place of residence and occurrence are presented in separate volumes. In 1940 only 2.1% of all deaths occurred to residents of one state dying in another state. Most non-resident deaths were residents of other areas of the same state. Table 4 presents correlation coefficients for the relationship between death rates by urban rural areas of each state and the percent of residences in each state with electric service. In 1940 urban and rural residence information was not available for individual cancers as it was in 1930, but death rates for each cancer were available by state. They were used to calculate correlations between electric service by state and respiratory cancer, breast cancer and leukemia mortality.
All causes of death
There was no correlation between residential electrification and total death rate for urban areas, but there was a significant correlation for rural areas (r = 0.659, p = <0.0001). Fig. 5 shows the 1940 resident white death rates for urban and rural areas of states having greater than 96% of residences electrified and states having less than 50% of residences electrified. In the highly electrified states, urban and rural death rates were similar, but in low electrification states, the urban death rates were systematically higher than the rural death rates. The urban death rates were similar in both high and low electrification states.
All malignant neoplasms
In 1940, the urban total cancer rate was 49.2% higher than the rural rate. Both urban and rural cancer deaths rates were significantly correlated with residential electrification. Fig. 6 shows the 1940 resident white total cancer rates for urban and rural areas of states having greater than 96% of residences electrified and states having less than 50% of residences electrified. Four of the five high electrification states had similar urban and rural total cancer rates, while all the low electrification states had urban rates about twice as high as rural rates. Both urban and rural total cancer rates were lower in low electrification states than in high electrification states. Fig. 7 shows the time trend of the total cancer rate between 1920 and 1960 for Massachusetts (1940 electrification rate = 97.6%) and Louisiana (1940 electrification rate = 48.9%). The Massachusetts cancer rate was about twice that of Louisiana between 1920 and 1945. The Massachusetts rate leveled off in 1945, but the Louisiana rate increased steadily between 1920 and 1960. A declining urban–rural gradient for cancer is still evident in 1980– 1990 US cancer incidence data . Swedish investigators  have reported increasing cancer mortality and incidence time trend breaks in the latter half of the 20th century.
Fig. 1. Death rates: death registration states, 1900–32, and United States, 1933–60.
Fig. 2. Death rates for malignant neoplasms: death registration states, 1900–32, and United States, 1933–60.
Fig. 3. Death rates for major cardiovascular renal diseases: death registration states, 1900–32, and United States, 1933–60.
Fig. 4. Death rates for diabetes mellitus: death registration states, 1900–32, and United States, 1933–60.
Table 3 - 1940 US white resident crude death rates per 100,000 by urban/rural residence.
Table 4 - Correlation coefficients (r) 1940 crude US death rates by state by electrification for white resident deaths.
Fig. 5. All causes death rates by urban rural status and electrification in the US for white residents in 1940.
Fig. 6. Total cancer death rates by urban rural status and electrification in the US for white residents in 1940.
Fig. 7. US white resident total cancer death rates for Massachusetts (97.6% elect.) and Louisiana (48.9% elect.) by year.
Fig. 8. US white resident breast cancer death rates for California (96% elect.) and Tennessee (50% elect.) by year.
Fig. 9. Total diabetes rates by urban rural status and electrification in the US for white residents in 1940.
Fig. 10. Total heart disease rates by urban rural status and electrification in the US for white residents in 1940.
Fig. 11. Total suicide death rates by urban rural status and electrification in the US for white residents in 1940.
Fig. 12. 1940 US white resident urban rural suicide death rates by state and electrification.
No urban rural information was available for respiratory cancer, but the correlation between residential electrification and state death rates was r = 0.611; p = <0.0001. This cancer is etiologically strongly related to cigarette smoking, so the correlation with electrification is surprising. A large electrical utility worker cohort study found a high respiratory cancer incidence related to high frequency EMF transient exposure independent of cigarette smoking with a significant dose–response relationship .
Although urban/rural information was not available for breast cancer, the 1940 state breast cancer death rates have a correlation of r = 0.794; p = <0.0001 with residential electrification. Fig. 8 shows the typical time trend of breast cancer death rates for a state with a high level of electrification (96%) and one with a low level of electrification (<50) in 1940. The California breast cancer death rate increased from 1920 to 1940, and then gradually decreased until 1960. The Tennessee breast cancer death rate is less than half of the California rate in 1920 and continues a steady increase until 1960.
This cause has a 66% urban excess. In spite of this, the correlation coefficients for urban and rural areas are similar at r = 0.66; p = <0.0001. There is some animal and human evidence that EMFs can effect insulin production and blood glucose levels . Fig. 9 shows that in states with low levels of electrification in 1940, the urban diabetes death rates are consistently higher than the rural rates, but are always lower than the urban and rural rates in the high electrification states.
Since the childhood leukemia age peak is strongly associated with residential electrification, it was interesting that the all leukemia death rate correlation was r = 0.375; p = 0.0042. Most of these deaths are in adults and are of different types of leukemia. A study of amateur radio operators showed a selective excess only of acute myelogenous leukemia .
Coronary artery disease and other heart disease
These two cause groups had the same percentage urban excess (33%), and very similar patterns of urban and rural correlation coefficients with residential electrification. The urban correlations were about r = 0.4 and rural deaths had correlations of 0.78 and 0.79, respectively. Fig. 10 shows the 1940 resident white coronary artery disease death rates for urban and rural areas of states having greater than 96% of residences electrified and states having less than 50% of residences electrified. Four of the five high electrification states had similar urban and rural total cancer rates, while all the low electrification states had urban rates about twice as high as rural rates. Urban and rural coronary artery death rates were lower in low electrification states than in high electrification states.
The urban suicide death rate is about 30% higher than the rural rate. The urban suicide rate is not correlated with residential electrification (r = 0.077; p = 0.299), but the rural death rate is correlated with 1940 state residential electrification levels (r = 0.729; p = <0.0001). Fig. 11 shows the 1940 resident white suicide for urban and rural areas of states having greater than 96% of residences electrified and states having less than 50% of residences electrified. In four of five high electrification states, rural suicide rates are higher than the urban rates. In all of the low electrification states, the urban rate is higher. The rural rates in the high electrification states are higher than the rural rates in the low electrification states. Fig. 12 shows X Y scatter plots for urban and rural suicide by electrification for 48 states. Suicide has been associated with both residential  and occupational  EMF exposure. Suicide is probably the visible peak of the clinical depression iceberg.
Motor vehicle accidents
Although the mortality rates are similar in urban and rural areas, the correlations with residential electrification levels are different. There is a slight negative correlation (r = -0.254) in urban areas and a positive correlation (r = 0.451) in rural areas. Since motor vehicle fatality is related to access to a vehicle and to speed. It may be that in the larger cities it was difficult to go fast enough for a fatal accident, and in rural areas especially on farms, a farmer who could afford electrification could also afford a car.
When Edison and Tesla opened the Pandora’s box of electrification in the 1880s, the US vital registration system was primitive at best, and infectious disease death rates were falling rapidly. City residents had higher mortality rates and shorter life expectancy than rural residents . Rural white males in 1900 had an expectation of life at birth of over 10 years longer than urban residents. Although the authors of the 1930 US vital statistics report noted a 58.2% cancer mortality excess in urban areas, it raised no red flags. The census bureau residential electrification data was obviously not linked to the mortality data. Epidemiologists in that era were still concerned with the communicable diseases.
Court Brown and Doll reported  the appearance of the childhood leukemia age peak in 1961, forty years after the US vital statistics mortality data on which it was based was available. I reported a cluster of childhood leukemia  a decade after it occurred, only because I looked for it. Real time or periodic analysis of national or regional vital statistics data is still only rarely done in the US.
The real surprise in this data set is that cardiovascular disease, diabetes and suicide, as well as cancer seem to be strongly related to level of residential electrification. A community-based epidemiologic study of urban rural differences in coronary heart disease and its risk factors was carried out in the mid 1980s in New Delhi, India and in a rural area 50 km away . The prevalence of coronary heart disease was three times higher in the urban residents, despite the fact that the rural residents smoked more and had higher total caloric and saturated fat intakes. Most cardiovascular disease risk factors were two to three times more common in the urban residents. Rural electrification projects are still being carried out in parts of the rural area which was studied.
It seems unbelievable that mortality differences of this magnitude could go unexplained for over 70 years after they were first reported and 40 years after they were noticed. I think that in the early part of the 20th century nobody was looking for answers. By the time EMF epidemiology got started in 1979 the entire population was exposed to EMFs. Cohort studies were therefore using EMF-exposed population statistics to compute expected values, and case-control studies were comparing more exposed cases to less exposed controls. The mortality from lung cancer in two pack a day smokers is over 20 times that of non-smokers but only three times that of one pack a day smokers. After 1956, the EMF equivalent of a non-smoker ceased to exist in the US. An exception to this is the Amish who live without electricity. Like rural US residents in the 1940s, Amish males in the 1970s had very low cancer and cardiovascular disease mortality rates .
If this hypothesis and findings outlined here are even partially true, the explosive recent increase in radiofrequency radiation, and high frequency voltage transients sources, especially in urban areas from cell phones and towers, terrestrial antennas, wi-fi and wi-max systems, broadband internet over power lines, and personal electronic equipment, suggests that like the 20th century EMF epidemic, we may already have a 21st century epidemic of morbidity and mortality underway caused by electromagnetic fields. The good news is that many of these diseases may be preventable by environmental manipulation, if society chooses to.
Conflicts of interest statement
 Milham S, Ossiander EM. Historical evidence that residential electrification caused the emergence of the childhood leukemia peak. Med Hypotheses 2001;56(3):290–5.
 Milham S, Morgan LL. A new electromagnetic field exposure metric: high frequency voltage transients associated with increased cancer incidence in teachers in a California school. Am J Ind Med 2008;51(8):579–86.
 Reynolds P, Elkin EP, Layefsky ME, Lee JM. Cancer in California school employees. Am J Ind Med 1999;36:271–8.
 US bureau of the Census. The statistical history of the United States from colonial times to the present. New York: Basic Books; 1976.
 Wertheimer N, Leeper E. Electrical wiring configurations and cancer. Am J Epidemiol 1979;109(3):273–84.
 Vital statistics of the United States (annual volumes 1930, 1940). Washington, DC: US Government Printing Office.
 US Census Bureau. Washington, DC: US Department of Commerce; 2009. <www.census.gov>.
 Vital statistics rates in the US 1940–1960, National Center for Health Statistics. Washington, DC: US Government Printing Office.
 Historical statistics of the United States colonial times to 1970. US Bureau of the Census. Washington, DC: US Commerce Department.
 Vassar statistical computation web site. Authored by Richard Lowry, Emeritus professor of psychology. Vassar College, Poughkeepsie NY, USA; 2009. <<http:// faculty.vassar.edu/lowry/VassarStats.html>>.
 Microsoft Excel 2007. Microsoft Corporation. One Microsoft Way, Redmond, WA, 98052-6399.
 National Center for Education Statistics (NCES) web site. US Department of Education. Washington, DC: Institute of Educational Sciences; 2009. <<http:// nces.ed.gov/nceskids/createAgraph/>>.
 Howe HL, Keller JE, Lehnherr M. The relation of population density and cancer incidence. Illinois 1986–1990. Am J Epidemiol 1993;138:29–36.
 Hallberg O, Johannson O. Cancer trends during the 20th century. J Aust College Nutr Environ Med 2002;21(1):3–8.
 Armstrong B, Theriault G, Guenel P, Deadman J, Goldgerg M, Heroux P. Association between exposure to pulsed electromagnetic fields and cancer in electric utility workers in Quebec, Canada, and France. Am J Epidemiol 1994;140(9):805–20.
 Navakatikian MA, Tomashevskaya LA. Phasic behavior and endocrine effects of microwaves of nonthermal intensity. In: Carpenter DO, editor. Biological effects of magnetic fields, vol. 1. San Diego, CA: Academic Press; 1994. p. 333–42.
 Milham S. Increased mortality in amateur radio operators due to lymphatic and hematopoietic malignancies. Am J Epidemiol 1988;127(1):50–4.
 Perry S, Reichmanis M, Marino AA, Becker RO. Environmental powerfrequency fields and suicide. Health Phys 1981;41(2):267–77.
 van Wijngaarden E, Savitz DA, Kleckner RC, Cai J, Loomis D. Exposure to electromagnetic fields and suicide among electric utility workers: a nested case-control study. West J Med 2000;173(2):94–100.
 Court Brown WM, Doll R. Leukemia in childhood and young adult life: Trends in mortality in relation to aetiology. BMJ 1961;26:981–8.
 Milham S. Leukemia clusters. Lancet 1963;23(7317):1122–3.
 Chadna SL, Gopinath N, Shekhawat S. Urban-rural difference in the prevalence of coronary heart disease and its risk factors. Bull World Health Org 1997;75(1):31–8.
 Hamman RF, Barancik JJ, Lilienfeld AM. Patterns of mortality in the Old Order Amish. Am J Epidemiol 1981;114(6):345–61.