epilepsy - hospitalization - air pollution - fertilizers - nitrous oxide
epilepsia - hospitalização - poluição do ar - fertilizantes - óxido nitroso
The link between air pollution and epilepsy has come under recent scrutiny. Scorza
et al.[1 ] have proposed that urban air pollution may be a significant factor influencing human
health and, particularly, sudden unexpected death in epilepsy. As part of their analysis,
the authors suggested nutritional interventions that may ameliorate epilepsy risk,
including intake of omega-3 fatty acids[1 ]. Empirical evidence in support of a link between air pollution and epilepsy has
emerged in recent years. Cakmak et al.[2 ] performed a daily time-series analysis of the association between gaseous and particulate
matter air pollution and hospitalization for epilepsy in Chile. The authors reported
pooled city estimates of relative risk for hospitalization for epilepsy associated
with changes in concentrations of various pollutants including nitrogen dioxides (NO2 ), carbon monoxide (CO), sulfur dioxide (SO2 ), ozone (O3 )), and particulate matter (both PM10 ) and PM2.5 ). The confidence intervals for all the identified pollutants indicated that air pollution
may be a risk factor for hospitalizations for epilepsy. A hospital-record based investigation
on the role of urban air pollution in epilepsy attack in a Chinese population has
shown similar result, with a possible protective effect of ozone exposure[3 ]. Animal studies indicate that ozone exposure may exert a protective effect against
pentylenetetrazole-induced seizures through the restoration of cellular redox balance
and regulation of the A1 adenosine receptor[4 ]. These studies cumulatively provide evidence in support of the claims made by Scorza
et al.[1 ] for a role of air pollution in modulating the risk of hospitalization for epilepsy.
Moreover, the work by Cakmak et al.[2 ] was cited in a recent review by Chilean health authorities and government officials
conducting a workshop to better understand how the economic reclassification of the
country has impacted environmental health with a goal of identifying important areas
for future improvement[5 ]. One specific recommendation made by the expert panel was to incorporate aspects
of climate change in the future study of the human health effects of air pollution.
Consistent with these stated goals, we have recently proposed, through empirical investigation
and review, that exposure to air pollution and specifically the pervasive air pollutant
and greenhouse gas, nitrous oxide (N2 O), may provoke susceptibility to neurodevelopmental disorders, like attention-deficit
hyperactivity disorder (ADHD) and autism spectrum disorders (ASD)[6 ],[7 ],[8 ],[9 ]. Epilepsy is a recognized comorbidity in both ADHD and ASD[10 ],[11 ],[12 ]. We have previously highlighted clinical case reports[13 ], as well as several animal studies[14 ],[15 ],[16 ], which indicated the role of N2 O exposure in reducing epileptiform seizure activity[17 ], while withdrawal from the drug has been shown to induce seizure-like activity[14 ],[15 ],[16 ]. Our novel environmental N2 O-mediated hypothesis of neurodevelopmental impairment is supported by recent evidence
indicating N2 O hotspots in central Chile, given the following statement from the work:
“…in our study area (off central Chile), they have ΔN2 O three times higher than the average monthly ΔN2 O…the increased N2 O production seemed to be caused by the addition of anthropogenic NO3
- associated with strong river runoff and a subsequent reduction to N2 O (i.e., partial denitrification)[18 ].”
However, the aforementioned analyses by Cakmak et al.[2 ], Xu et al.[3 ], and Mallok et al.[4 ] do not address the air pollutant, N2 O, in the discovered association between air pollution and risk for epilepsy hospitalization.
Consistent with the future goals identified by Chilean health and government officials[5 ], we attempt to qualify the prior work by these research groups by performing regression
analyses using a state-based total of farm use of anthropogenic nitrogen fertilizer,
which is thought to be the major environmental contributor to N2 O emissions[18 ],[19 ]. While we attempt to document prior associations, by Cakmak et al.[2 ] and others, between certain air pollution exposures and hospitalization for epilepsy
for years 2001 to 2005, we also wish to test our hypothesis that increasing use of
anthropogenic nitrogen-based fertilizers in agriculture (the most recognized causal
environmental contributor to N2 O emissions) may be significantly and negatively associated with hospitalization for
epilepsy. We rationalize this hypothesis by emphasizing prior clinical reports and
animal studies indicating a N2 O-induced suppression of epileptiform activity. Coupled with our prior conclusions,
this hypothesis, if supported, may support the co-morbid link between epilepsy and
other neurodevelopmental disorders by suggesting both are related to exposure to (i.e.,
ADHD and ASD) or withdrawal from (i.e., epilepsy) environmental N2 O[6 ],[7 ],[8 ],[9 ],[17 ].
METHODS
For years 1997 to 2006, inclusive, we have gathered data on the state sum totals of
farm use of nitrogen fertilizers (in kilograms)[20 ], described in the authors’ prior epidemiological studies[7 ],[9 ]. Although the United States Geological Survey of the U.S. Department of the Interior
states that “no warranty expressed or implied is made by the U.S. Geological Survey
as to the accuracy of the data,” we believe the use of county data on a state aggregated
level is consistent with the recommendations of the United States Geological Survey[20 ],[21 ]. Moreover, other government agencies, like the Environmental Protection Agency,
make similar disclaimers regarding data contained on their organization’s website[22 ], yet researchers continue to utilize the Environmental Protection Agency data on
air quality emissions[23 ].
To replicate the work by Cakmak et al.[2 ], the authors averaged all locale state estimations of the annual air concentration
of the pollutants between 2001 to 2005 (NOx and SO2 ) [one hour] in ppb, O3 and CO [eight hour run average] in ppm; PM10 total 0-10mm standard temperature and pressure [24-hour] and PM2.5 mm [local conditions, 24 hour] in micrograms/cubic meter), using air quality data
from the Environmental Protection Agency[24 ]. Our dependent condition of interest is hospitalization for both all-listed and
principal diagnoses of epilepsy, derived from data using the Healthcare Cost and Utilization
Project (HCUPnet)[25 ], as has been performed previously[7 ],[9 ].
We conducted a Poisson regression methodology including two-way fixed effects. Briefly,
a random variable Y is said to have a Poisson distribution with parameter m if it
takes integer values y = 0, 1, 2, … with probability
for m > 0. The mean and variance of this distribution can be shown to be E(Y) = var(Y)
= m. We have a sample of n observations of discharges related to epilepsy, y
1 , y2 , …, y
n, which are treated as realizations of independent Poisson random variables, with Y
ij , ~ P (m
ij ), where i represents a state and j an observation year. In order to test the robustness of hypothesized associations,
we selected, a priori, several categories of epilepsy discharge diagnoses, including the clinical classification
software category designations provided by HCUPnet (CCS 83: epilepsy, convulsions),
using both the all-listed total and principal discharge diagnoses. The ICD9 codes
that comprise this CCS category include 345.0, 345.00, 345.01, 345.1, 345.10, 345.11,
345.2, 345.3, 345.4, 345.40, 345.41, 345.5, 345.50, 345.51, 345.6, 345.60, 345.61,
345.7, 345.70, 345.71, 345.8, 345.80, 345.81, 345.9, 345.90, 345.91, 780.3, 780.31,
780.32, 780.33, 780.39. Additionally, we tested the independent ICD9 diagnostic code
345.xx for epilepsy (all-listed discharge diagnoses). We let the logarithm of the
mean depend on a vector of time-varying explanatory variables, x
ij, such that the log-linear model is the following: log (m
ij ) = x
ij
’b
1 . Exponentiating, we have a multiplicative model for the mean discharges: m
ij = exp{x
ij
’b
1 }. In each case, the exponentiated regression coefficient exp{b
1ijk } yields an incidence rate ratio, which represents a multiplicative effect of the
k th predictor on the mean. Increasing x
k by one log-unit multiplies the mean by a factor exp{b
1k }. All regression analyses and related diagnostics were performed using R (packages ggplot2 and sandwich with robust standard errors), as we have described
previously[7 ],[9 ].
Although we are unable to directly replicate the methodology of Cakmak et al.[2 ], we believe the strength of our two-way fixed effects Poisson statistical approach
lies in the fact that we have both accounted for individual state-level heterogeneity
among a greater number of locations (30 selected HCUP states versus seven Chilean
urban centers) and controlled for correlations across a longer time series (10 years
versus five years in the case of farm use of nitrogen fertilizers), thus reducing
the likelihood for omitted variable bias in the modeling. While not considered by
Cakmak et al.[2 ], this approach may be important, given the role that edaphology may play in the
flux of environmental air contaminants[26 ], as well as the evolving legislation, especially in the United Sates, surrounding
the regulation of air pollution over many decades[27 ].
RESULTS
The percent changes of the various air pollutants studied in this investigation are
presented in [Table 1 ]. All pollutants decreased in concentration level from 2001 to 2005, with carbon
monoxide, nitrogen dioxide, and sulfur dioxide decreasing the most. [Figure 1 ] shows the increase in the annual sum of farm use of nitrogen fertilizers for the
selected HCUP available states between 1997 and 2006. [Figure 2 ] presents the annual sum of hospitalization discharges associated with all three
categories of epilepsy included in this brief report in the selected HCUP states.
Table 1
Average (standard deviation) annual pollutant concentration across all reporting HCUP
states, 2001-2005, with percent change.
Year
Carbon monoxide (ppm)
Nitrogen dioxide (ppb)
Ozone (ppm)
PM10 (mg/m[3 ])
PM2 (mg/m[3 ])
Sulfur dioxide (ppb)
2001
0.62 (0.22)
29.02
0.04630
23.74
11.89
11.80
(7.84)
(0.00561)
(5.28)
(2.36)
(6.68)
2002
0.63 (0.21)
27.19
0.04695
23.47
11.36
10.64
(7.66)
(0.00497)
(5.68)
(2.17)
(6.02)
2003
0.60 (0.18)
25.79
0.04582
23.35
10.87
10.31
(7.60)
(0.00411)
(4.81)
(2.25)
(5.86)
2004
0.55(0.17)
24.04
0.04293
22.00
10.75
9.72
(7.31)
(0.00382)
(4.50)
(1.98)
(5.95)
2005
0.51(0.16)
24.74
0.04625
23.31
11.55
10.09
(7.07)
(0.00475)
(4.93)
(2.38)
(5.50)
% change
-17.74
-14.75
-0.11
-1.81
-2.86
-14.49
HCUP: Healthcare Cost and Utilization Project.
Figure 1 Annual sum of farm use of nitrogen fertilizers across all HCUP reporting states included
in this analysis for all years available.HCUP: Healthcare Cost and Utilization Project
Figure 2 Annual sum of hospital discharge diagnoses across all HCUP reporting states included
in this analysis for all years available. Categories include all-listed epilepsy discharges
(Clinical Classification Software: 83), principal epilepsy discharges (Clinical Classification
Software: 83), and all-listed epilepsy discharges using only ICD 9 classification
code: 345.xx.HCUP: Healthcare Cost and Utilization Project.
[Table 2 ] shows the results of our two-way fixed effects Poisson regression models. The data
indicate that the air pollutants identified by Cakmak et al.[2 ] were not statistically significantly associated with hospitalization for all-listed
diagnoses of epilepsy during the same time period after multiple comparison correction
for the seven pollutants studied for each hospitalization category (0.05/7 = p < 0.007).
Nitrogen dioxide concentration showed a slight protective effect against hospitalization
for epilepsy (345.xx), particularly when the condition was the principal diagnosis,
but the results did not retain statistical significance when adjusting for multiple
pollutant comparisons. However, a one-unit log increase in total annual farm use of
nitrogen-based fertilizers in each state is significantly negatively associated with
hospitalization for all three epilepsy categories studied. Further quartile analysis
revealed that states above the 50th percentile of farm use of nitrogen fertilizer (i.e., heavy agricultural states) saw
the greatest protective effect against hospitalization for epilepsy. States below
this percentile had no protective effect (data not shown ).
Table 2
Incident Rate Ratio of Hospitalization for Epilepsy Associated with a One Log-unit
Change in Average Pollutant Concentrations using state and time fixed effects Poisson
regression in R (packages ggplot2 and sandwich with robust standard errors) for all
available HCUPnet states, United States, 2001–2005.
Pollutant (state-time obs)
Carbon monoxide (n = 131)
Ozone (n = 131)
Sulfur dioxide (n = 129)
Nitrogen Dioxide (n = 124)
PM10 (n = 131)
PM2.5 (n = 128)
Farm nitrogena,b (n = 216)
Epilepsy
IRR
95%CI
IRR
95%CI
IRR
95%CI
IRR
95%CI
IRR
95%CI
IRR
95%CI
IRR
95%CI
Diagnosis
All-listed
1.00
0.98–1.04
0.91
0.83–1.00
0.99
0.94–1.03
1.02
0.97–1.08
0.97
0.93–1.02
0.96
0.87–1.06
0.96
.93–.98*
Principal
1.02
0.96–1.08
0.91
0.78–1.07
0.98
0.93–1.03
0.93
0.88,
0.99
0.92–1.06
0.98
0.83–1.15
0.92
.88–.96*
0.98
ICD9 (345.xx)
1.09
0.98–1.21
0.76
0.59–0.97
0.99
0.91–1.08
0.85
0.74–0.97
0.95
0.85–1.07
0.95
0.77–1.17
0.82
0.74–0.90*
*p <.007, adjusted for multiple pollutant comparisons (.05/7 =.007); aperiod 1997–2006
to match the time period utilized in prior work9 and controlling for non-farm use
of nitrogen fertilizers; bas this variable acts as a proxy for air emissions of the
agricultural pollutant, nitrous oxide (N2O), which depend upon the total nitrogen
applied, sum state totals were used and log transformed.
HCUP: Healthcare Cost and Utilization Project.
DISCUSSION
The results of this brief report suggest a possible link between agricultural air
pollution, particularly the farm use of anthropogenic nitrogen fertilizers and associated
emissions of N2 O, and hospitalization for epilepsy in 30 states included in the HCUP database. However,
this finding is reliant on the assumption that farm use of anthropogenic nitrogen
fertilizers represents an accurate estimation of environmental N2 O emissions. Although others have acknowledged this relationship[19 ], we must exert some hesitation with the underlying interpretation of our results.
For example, the potential for bias in our results cannot be discounted, as the finding
that the protective effect of farm use of nitrogen fertilizers against hospitalization
for epilepsy may be associated with a more rural lifestyle that is less afflicted
by stress, a recognized trigger for epilepsy[28 ]. However, this explanation is not consistent with our prior work showing a disproportionate
burden in hospitalization for other mental health comorbidities, like ADHD (1997-2006),
that occurs in these less populated areas, as we have discussed[7 ],[9 ]. A more consistent explanation that our collective studies now afford is that exposure
to environmental anthropogenic nitrogen fertilizers used in agriculture and associated
emissions of N2 O may facilitate a continuum of psychiatric impairments; that is, increasing use of
anthropogenic nitrogen fertilizers in agriculture and exposure to N2 O leads to ADHD symptoms (i.e., working memory impairments and inattention), and withdrawal
and/or tolerance facilitates epileptiform activity. Molecular evidence of N2 O toxicity indicates an antagonism of N-methyl-D-aspartate receptors, which may undergo
a rapid upregulation in response to N2 O withdrawal and/or tolerance, resulting in receptor hyperactivation and possible
seizure generation[29 ], as has previously been cited and discussed[6 ],[17 ].
We have attempted to replicate prior associations among other commonly studied air
pollutants, like CO, O3 , NOx, SOx, and particulate matter but were not able to find consistent, statistically
significant associations across the pre-specified hospitalization categories during
the same time period studied by Cakmak et al.[2 ]. Additional analyses including a longer time series (1997-2006) were not able to
alter these conclusions. These results suggest that future investigations seeking
to elucidate the role of air pollution in risk of hospitalization for epilepsy should
be expanded to include environmental emissions of the agricultural and combustion
pollutant, N2 O, especially given the clinical evidence and preclinical animal studies directly
associating N2 O exposure with epileptiform activity[13 ],[14 ],[15 ],[16 ]. While a protective effect of ozone did not meet our statistical significance after
multiple pollutant comparison correction, epidemiological evidence indicating a potential
protective effect of ozone in epilepsy should also consider that the effect may have
to do with the possible role of N2 O as an agent of ozone depletion[30 ], in addition to any independent cellular mechanisms possibly at play[4 ]. Similarly, while nitrogen amended soils may leach N2 O emissions, nitrogen dioxide fluxes may be minimized under conditions of agricultural
fertilization[31 ].
The current finding merges critical concepts of climate change into the growing burden
of chronic human neurological illness and, in particular, suggests an important direction
to take in the study of epilepsy and related comorbidity. Recent reports are indicating
that current Intergovernmental Panel on Climate Change guidelines are underreporting
indirect emissions of N2 O by a magnitude from three- to nine-fold[32 ],[33 ], while Intergovernmental Panel on Climate Change methodologies presuming linear
increases in direct soil N2 O emissions as a function of nitrogen inputs may be missing true exponential rises
in global emissions, especially under conditions of heavy fertilizer use (i.e., agriculture)[19 ]. Our data may be particularly revealing in this regard, considering that the greatest
protective effect against hospitalization for epilepsy occurs in states that report
the greatest amount of nitrogen fertilizers used in agriculture.
It is also interesting to further speculate on the association between environmental
sources of N2 O and risk of hospitalization for epilepsy by recognizing that diurnal variation in
N2 O flux[34 ] (i.e., greater in the afternoon hours) may be negatively related to the early morning
risk of seizure in generalized epilepsy[35 ]. Though, more research is needed to understand what role, if any, environmental
exposure to N2 O may have in human health assessments. While our presumption that the anthropogenic
use of nitrogen fertilizers is directly reflective of environmental N2 O burden is a limitation of the analysis, it remains our suggestion that scientists
may need to reconfigure their thinking about particular environmental pollutants,
like N2 O, recognizing that agents long thought to carry little-to-no harm in the controlled
and acute medical setting may not exist in the daily environment in a similar manner.
