Bright Ideas? Ultraviolet Radiation, Weather, and the Seasonality of Invasive Bacterial Disease in North America

1
Bright Ideas? Ultraviolet Radiation, Weather,
and the Seasonality of Invasive Bacterial Disease
in North America

• David N. Fisman, MD MPH FRCP(C)
• Medical Epidemiologist, Ontario Public Health
  Laboratory
• Scientist, Research Institute of the Hospital for
  Sick Children
• TIBDN Research Day, Mt. Sinai Hospital, November
  27, 2008

2
Ultraviolet (U.V.) Radiation
Image source NASA via Wikimedia commons
3
Ultraviolet (U.V.) Radiation (2)
Image source NASA via Wikimedia commons
4
Ultraviolet (U.V.) Radiation (3)
5
U.V. Radiation Protection Against Invasive
Bacterial Disease?

• Historical perspectives (sunlighthealth).
• Apparent linkages between U.V. radiation and
  (prevention of) infection
• U.V. and tuberculosis (Finsen, Riley).
• U.V. and flu season.
• Biological mechanisms
• Effects on pathogens, effects on hosts.
• Approaches to epidemiological links.
• Confounding and ecological fallacy.
• IGAS, IPD, IMD.

6
Historical Perspectives

• Notion of season and weather in disease causation
  central to Hippocratic medicine (ca. 400 B.C.).
• Interaction between environment and individual
  constitutions results in disease.
• Tracts include Airs, Waters, Places Epidemics

Image National Library of Medicine
7
DarknessDisease

• An English word falls upon the ear almost with a
  sense of shock...In the midst of it all there is
  a sudden wild scattering, a hustling of things
  from the street into dark cellars
• Jacob Riis in How the Other Half Lives (1890) on
  visits by NYC Health Inspectors to Lower East
  Side, 1890. Quoted in Markel, Quarantine!
  (1997), page 35.

Jacob Riis, ca. 1890. Museum of the City of New
York. Reproduced from Markel, Quarantine! (1997).
8
Light (and Enlightenment)Health

• Sanitorium movement Dr. E.L. Trudeau founds
  first N. American TB san in Saranac Lake, NY
  ca. 1875.
• Treatment focuses on outdoor exposure, fresh air,
  sunlight.

• Parallels between sanitorium movement and
  Victorian improvement movement generally
• You have no zymotic infectious disease, no
  poverty, no drunkennesswhat human society might
  be, were it all light, with no suffering and dark
  corners.
• William James, on the Chautauqua Institution (for
  improvement of Sunday School teachers), 1896.
  Quoted in Markel, Quarantine! (1997).

9
Effects of U.V. on Active Tuberculosis

• E.L. Trudeau (U.S.) and Nils Finsen (Denmark)
• Empirical observations of behavior of own
  illnesses lead to experimentation with outdoor
  exposure (ELT) and direct application of U.V.
  radiation (NF).
• Finsens phototherapy effective in cure of
  lupus vulgaris (cutaneous T.B.).
• Recent investigation indicates lenses resulted in
  UVA dosing.
• Nobel Prize for Medicine or Physiology, 1903.

10
Effects of U.V. on Survival of M. tuberculosis in
Droplet Nuclei
Riley R.L. et al., American Review of
Tuberculosis 1957 American Journal of Hygeine
1959. Image from Nardell and Dharmadhikari,
IUATLD Vancouver 2008.
11
Seasonality, Influenza and U.V.
Hope-Simpson, 1981 reproduced in Cannell JJ et
al., Epidemiology and Infection, 2006.
12
U.V. and Reduced Infection Risk

• Direct damage to pathogens.
• Mutagenesis.
• Porphyrins (TB).
• Improvement of host immune function.
• Vitamin D and immune function.

13

• Aberrant covalent bonds formed between adjacent
  cytosines.
• Dimers read as AA, not CC.
• Repaired with TT (classical C-T mutation).

Image source Wikimedia commons
14
Mechanism of Finsens Lenses
Photic stimulation leads to O radicals.
15
Effects of Light on Immune Response

• Dowell 2001 reviews association between
  seasonality and photoperiod.
• Numerous physiological changes (psychological,
  sexual, immunological, pathological) associated
  with shortened (wintertime) photoperiods in
  humans and animals.
• For example, Siberian hamsters exposed to
  short-day photoperiod demonstratephagocytosis
  and oxidative burst.
• Dowell, Emerging Infectious Diseases 2001.

16
Vitamin D

• Steroid hormone, synthesis depends on cleavage of
  DHC from skin in response to U.V.
• Vitamin D deficiency associated with numerous
  defects of macrophage maturation and function.
• Relative deficiency associated with numerous
  adverse health outcomes in observational studies
  (cancer, heart disease).
• Recently described relationship between Toll-like
  receptor, vitamin D, and cathelicidin Liu P et
  al., Science 2006.

Source Cannell JJ et al., Epidemiol Infect 2006
17
Vitamin D (2)
Zasloff M, Nature Medicine 2006
18
Is it really the season?

• Establishing causal links between environmental
  factors and disease occurrence may be difficult
  when the disease is seasonal.
• Relationships may be confounded with underlying
  factors
• e.g. increased incidence during certain types of
  weather might just reflect population risk
  behaviour
• Strong correlation is necessary but NOT
  sufficient
• Aggregation of exposures may lead to ecological
  fallacy.

Slide courtesy of Laura Kinlin and Alexander
White
Cases per week
19
Seasonally Oscillating Environmental Exposures,
Philadelphia
40
30
20
10
0
01/1994
01/1996
01/1998
01/2000
01/2002
01/2004
01/2006
01/2008
Date
TMAX (C)
MAXCIE/10
Delaware River dissolved O2 (2)
20
Seasonality, Environment, and Infectious Disease

• 2 year project funded by U.S. NIAID
  (R21-AI065826).
• Cooperative work performed by SickKids,
  Philadelphia Department of Public Health, and
  Ontario Public Health Laboratory.
• Evaluate weekly or monthly disease incidence
  using Poisson models with smoothers
  (de-trended).
• Use of novel case-crossover method to evaluate
  acute effects of exposures on disease risk.
• Both methods should largely control for
  confounding by nonspecific seasonal oscillation
  in exposures and disease incidence.
• Evaluate effects over different time scales.

21
Houston, Texas
22
Residual (Excess) Deaths, Relative to Model
23
Supplementary Figure Schematic diagram of
control selection strategy for case-crossover
study. Each row represents a 3-week time block.
Hazard and control periods (matched by
day-of-week) are selected from the 3-week time
block, resulting in random directionality of
control selection.
24
PneumococcusPhiladelphia
Source White ANJ et al., BMC Infectious
Diseases, submitted
25
PneumococcusPhiladelphia
Univariable Models Univariable Models Univariable Models
Meteorological Element IRR (95 CI) P
Cooling-degree Days (oC) 0.92 (0.90 0.94) lt0.001
Mean Temperature (oC) 0.96 (0.95 0.97) lt0.001
Relative Humidity () 0.98 (0.97 0.99) 0.002
UV Index 0.89 (0.87 0.92) lt0.001
Sulphur Oxides (ppm x 100) 1.73 (1.27 2.37) 0.002
Average Wind Speed (km/h) 1.01 (1.006 1.015) lt0.001
Multivariable Modelsa Multivariable Modelsa Multivariable Modelsa
IRR (95 CI) P
0.97 (0.94 1) 0.06
... ... ...
... .. ...
0.74 (0.59 0.83) 0.01
... ... ...
... ... ...
Source White ANJ et al., BMC Infectious
Diseases, submitted
26
PneumococcusPhiladelphia
Source White ANJ et al., BMC Infectious
Diseases, submitted
27
Invasive Meningococcal Disease
Slide courtesy of Ms. Laura Kinlin, Hospital for
Sick Children/University of Toronto SPH
28
Invasive Meningococcal Disease Philadelphia
Multivariable Models Including Cubic Splines Multivariable Models Including Cubic Splines
IRR (95 CI) P
- -
0.98 (0.96-0.996) 0.02
1.04 (1.003-1.07) 0.03
- -
- -
- -
- -
- -
- -
Environmental Exposure Univariable models Univariable models
Environmental Exposure IRR (95 CI) P
Wind speed, km/h 1.14 (1.06-1.23) 0.001
Mean temperature, C 0.97 (0.95-0.99) 0.001
Maximum relative humidity 1.05 (1.01 - 1.08) 0.01
Snowfall, mm 1.05 (1.02-1.07) 0.001
UV Index Unit 0.92 (0.86-0.99) 0.02
Total ozone, ppm 0.85 (0.72-1.01) 0.06
Carbon monoxide, ppm 2.25 (1.18-4.27) 0.01
Oxides of nitrogen, ppm 1.72 (1.23-2.39) 0.002
Oxides of sulphur, ppm 2.52 (1.34-4.74) 0.004
Multivariable Models Including Oscillatory Seasonal Smoothers and Annual Trend Multivariable Models Including Oscillatory Seasonal Smoothers and Annual Trend
IRR (95 CI) P
- -

1.04 (1.004-1.08) 0.03
- -
- -
- -
- -
- -
- -
Kinlin L. et al., American Journal of
Epidemiology 2009, in press
29
Results Case-Crossover Analysis

• An acute protective association was identified
  for UV index during the period 1-4
• days prior to onset of cases (OR, 0.55 95 CI,
  0.36-0.84)
• - A significant dose-response relationship
  between UV index and disease risk was
• detected at this lag (Wald chi-squared for
  trend, 4.22 1 df P 0.04)

Kinlin L. et al., American Journal of
Epidemiology 2009, in press
30
Dose Response Relationship

• Future directions currently evaluating effects
  in 4 temperate cities in northern and southern
  hemispheres (Toronto, Philadelphia, London, and
  Sydney).

Quintile of UV Radiation Odds ratio (95 CI)
1st (referent) 1
2nd 0.82
3rd 0.91
4th 0.68
5th 0.62
Wald chi-squared for trend, 4.14 1 df P0.04
Kinlin L. et al., American Journal of
Epidemiology 2009, in press
31
Summary

• Ambient U.V. radiation has harmful effects
  (mutagenesis) but may also have salubrious
  effects related to infectious diseases.
• Biologically plausible relationships effects on
  pathogens and hosts.
• U.V. associated with reduced risk of invasive
  bacterial disease (IPD, IMD) in Philadelphia,
  after controlling for seasonality.
• Effects over both short and long time-scales
  direct effects on pathogen and vitamin D related
  effects?

32
Acknowledgements

• Funders U.S. NIAID, SickKids SSuRE and SickKids
  Foundation, Ontario Early Researcher Program.
• The (Dream) Team Amy Greer, Victoria Ng, Laura
  Kinlin, Alexander N.J. White.
• Co-investigator Dr. Caroline Johnson (PDPH).
• Collaborators Frances Jamieson, Natasha
  Crowcroft, Elizabeth Brown (OAHPP/OPHL), C.
  Victor Spain (PDPH?Merck Frosst), Graham Fraser,
  Julia Granerod (UK HPA), Murray Mittleman, Greg
  Wellenius (Harvard SPH).