Light pollution as a Risk Factor
for Breast Cancer: A GIS-Assisted
Case Study
I. Kloog, 1 B. Portnov 1, and A. Haim 2
1Department of Natural Resource and Environmental
Management,University of Haifa 2Department of Biology, University of
Haifa-Oranim
Presented 21 June 2005
Definition of Light Pollution :
Light pollution is “environmental pollution consisting of harmful or
annoying light from cities and outdoor lighting, which prevents
the observation of faint objects” (http://www.darksky.org).
Previous studies
Very few studies have been done to date to determine a
possible negative impact of artificial light on human
health. However, some indirect evidence in this
direction is nevertheless available:
• Brainard G. C. et al. (2001) show that there is a strong
correlation between the exposure to the prolonged photoperiod
and melatonin levels in blood.
• Davis et al. (2001) point up at a direct link between the lack of
melatonin caused by an exposure to a prolonged period of
artificial illumination and an increase in the breast cancer rate.
In particular, studies of night shift workers indicate higher rates
of breast cancer by 36-60% compared to the general population;
• Verkasalo et al. (1999) pointed out that the rate of breast cancer
in visually impaired women decreases as the degree of
impairment increases.
Goal of the Study:
• Although previous studies indicates that the exposure to
light pollution may be a risk factor in breast cancer
development, the empirical evidence in this direction is
rather fragmented and largely inconsistent.
• The goal of the present analysis is to Investigate the
relationship between exposure to artificial illumination
(light pollution) and breast cancer rates, using Israel as a
case study.
Novelty of the study
• To the best of our knowledge this is the first
study that uses macro-level remote sensing
data to link light pollution with the incidence
of breast cancer.
Research Phases
The study was carried out in three separate phases:
• A GIS assisted analysis of light intensity using satellite maps
- a low-resolution scale that covers the whole country;
• Light intensity and breast cancer rates in residential
neighborhoods - a high resolution scale covering four
residential neighborhoods in Tel-Aviv, where light intensity
was measured in-situ.
• Questionnaires: which were distributed among breast
cancer patients and among a control group of healthy
women from the same area.
Phase 1: Analysis of
Nightlight Map
• Data source: 1996-1997 satellite image
of radiance-calibrated night light
intensity supplied by the U.S. Defense
Meteorological Satellite Program
(DMSP).
• Data range: The light intensity is
measured in digital numbers (DN)
ranging from 8 to 113, which are
converted into nano-watts/cm2/sr as
follows:
Radiance=0.1*DN(2/3) (nanowatts/cm2/sr)
(The radiance in Israel ranges from 2.52 to
120 nano-watts/cm2/sr).
Breast Cancer Data
• Data source: Israel Ministry of Health
• Resolution: Small Statistical Areas
(SSA)
• Number of observations: ca. 214
• Time span: 1998 – 2001.
Data Matching
•The rates of breast cancer and nightlight intensity were merged
using the Spatial Join tool in ArcGIS.
•The result is the mean values, standard deviation of light
intensity against cancer rates for each locality (or SSA).
General Trends
The scatter plot reveals three distinct groups of towns – localities with
abnormally high, average and low cancer rates.
For each group of localities, the trend lines show a positive association
between night light intensity and breast cancer rates (R2=0.019-0.675).
• The localities with abnormally high cancer rates consist
mainly of towns that are located on the seam line
between the Palestinian autonomous areas and the
State of Israel (Tzoran, Tzor yigal, Meitar).
• In all these towns and their surrounding areas we found
extensive and large scale illumination systems (with very
high light intensity ranging between 0.29 and 0.47 micro-
lux), which were built by the state for security reasons.
• The localities with abnormally low cancer rates consist
mainly of towns with predominantly minority population :
Arab, Druze or Circassian (Kusife, Shfram, Umm Al-Fahm,
Baqa Al-gharbiyye etc.).
• These towns are characterized by relatively low average
incomes - low-income households and municipalities try to
minimize their outlays, by using (inter alia) less illumination
at both private homes and public domains.
• These localities are also characterized by the relatively low
rates of labor force participation by the minority population
(ca. 37%, as opposed to 52% among Jews), and specifically
by minority women, reduces their exposure to artificial light
at the work place.
• we then plotted the cancer rates in selected localities of each of these
groups as a function of the in-situ measured light intensity
200
180
160
Woman Breast Cancer rates
140
120
(per 100,00)
2
100 R = 0.9199
80
60
40
20
0
0 0.1 0.2 0.3 0.4 0.5
light intensity (micro-lux)
As shown, there is a positive correlation (R2=0.919) where higher light
intensities corresponded to higher rates of breast cancer irrespective of the two
group trends
Regression Analysis:
1. Dependent variable: per 100,000 breast cancer rates
(CR)
2. Main explanatory variable: the logarithm of the
average night light intensity in a locality (LI, nano-
watts/cm2/sr).
3. Controls:
Average per capita income (INC). The variable is
added as a proxy for illumination inside people’s
houses: As the average income increases, so does
the electricity consumption within the home, since
wealthier households can afford more illumination.
Population The variable is added since it adds missing
data, not captured by the satellite imaginary (artificial
illumination in public transport, shopping centers etc.),
that contributes to light pollution in a locality. The
bigger the population of a town, the more vehicles and
public lighted facilities it has.
G1: localities with abnormally high cancer rates.
G2: localities with abnormally low cancer rates.
Regression results
Explanatory B B VIF
OLS Model SL model
(Constant) -113.25 -143.23
(-4.17)* (-5.21)*
Mean light pollution 6.46 7.65 1.746
(4.24)* (4.38)*
Per Capita Income 21.02 24.39 1.454
(6.62)* (7.71)*
Localities with abnormally high cancer rates 41.05 40.73 1.089
(10.96)* (11.21)*
Localities with abnormally low cancer rates -35.35 -32.33 1.256
(-13.43)** (-12.48)*
Population size 2.96 2.93 1.586
(2.87)* (2.99)*
Number of observations 209 209
Residual standard error 12.08 11.64
Log-likelihood - -1075
R2 0.778 -
R2 adjusted 0.772 -
F 142.64 -
Explanation of the model
• More affluent and more illuminated localities tend to exhibit
(ceteris paribus) higher rates of breast cancer.
• While the average night light intensity (LI) is an indication of
artificial illumination of spaces outside people’s homes, the
income variable (INC) is a proxy for illumination intensity
inside dwellings.
Phase 2: Neighborhood Survey
• Study area: four neighborhoods in the
City of Tel Aviv.
• Selection criteria:
1. Breast cancer rates - highest and lowest rates across
the entire city).
2. Average incomes - above 5000 NIS and bellow 1500
NIS.
Haargaz – a low-income neighborhood Afeka – a high-income neighborhood
with low breast cancer rates (40 cases with low breast cancer rates (less than
per 100,000 women) 30 cases per 100,000 women)
Tikva - a low-income neighborhood Bavlei – a high-income neighborhood
with extremely high breast cancer with extremely high cancer rates (160
rates (150 cases per 100,000 women) cases per 100,000 per women)
in-situ Light Measurements
• Using a light-meter (LI-COR, LI-
189), night light intensity was
measured at 40 randomly
selected points in each
neighborhood.
• The light intensity was measured
in micro-Lux at the average
height of woman eyesight (1.7 m
above the ground).
Results:
Within each socio-economic group of neighborhoods, a similar significant
difference is reveled (p<0.01, p<0.05): Neighborhoods with high light
intensity show significantly higher breast cancer rates (p<0.01).
Phase 3: Questioners
• A specially designed questionnaire was distributed among
breast cancer patients in the Sheba Medical Center in the
Tel Aviv metropolis. The control group consisted of healthy
women.
• The data were collected using questioners (Helsinki
committee approved) that were filled anonymously by the
subjects with the care of the Department of Oncology of the
Sheba Medical Center.
• Data were collected between the dates of 1.1.2003 to
1.3.2005 from 100 breast cancer patients and 100 healthy
women.
Questioners Results
• The answers to each question in the questioner were
compared between the two groups, averaged and
subjected to a t-test.
• The analysis of answers of the two questioner groups
showed clear and highly significant differences
between the group of breast cancer patients and
healthy women in several important categories
Category Group mean std t sig.
Breast cancer occurrence in
the family with cancer 0.376 0.487 4.114 0.000
without cancer 0.115 0.321
Colorectal cancer occurrence
in the family with cancer 0.178 0.385 1.846 0.067
without cancer 0.082 0.277
Exposure to light through the
bedroom window at night with cancer 0.297 0.459 2.013 0.046
without cancer 0.164 0.373
Walking distance from
shopping centers with cancer 0.653 0.478 3.315 0.001
without cancer 0.393 0.493
Walking distance from cultural
centers with cancer 0.6 0.492 1.749 0.082
without cancer 0.459 0.502
Directions for future research
• This study has been a preliminary analysis which
included a limited set of case studies. Future
research needs to be carried out in both higher
resolution (a worldwide scale and other countries)
and low resolution scale (e.g., in-depth analysis of
individual urban localities such as Tel Aviv and
Haifa), to strengthen our results.
• In addition the relationship between light pollution
and other hormonal cancers (such as prostate
and colon cancers) needs to be also investigated.
Conclusions and Directions of
Future Research
• The survey thus reveals a strong association
between the exposure to high nightlight intensity
and the incidence of breast cancer.
• We thus suggest that municipalities should adopt a
smart policy of illumination. Such a policy should
reduce illumination when and where not absolutely
necessary, to both save energy (and money) and
prevent excessive light pollution which appears to
be a general environmental hazard to public health.
