BAROMETRIC PRESSURE, EMERGENCY PSYCHIATRIC VISITS,

AND VIOLENT ACTS

Thomas J Schory, MD1, Natasha Piecznski, MA, MDiv, LMFT2,
Sunil Nair, MD3, Rif S El-Mallakh, MD
Background: Associations between human behaviour and
psychiatric decompensation and weather variables have been
inconsistent.
Objectives: We studied the association of certain weather variables
(specifically, humidity, wind speed, and barometric pressure) with
emergent psychiatric presentations, psychiatric admissions,
incidence of violent crimes, and suicides in a metropolitan area.
Method: We performed a retrospective study
mid-sized city. We included all documented
visits to the citys psychiatric emergency
violence data from the city police department
the county medical examiner.

for the year 1999 in a

emergent psychiatric
room. We obtained
and suicide data from

Results: The data suggest that total numbers of acts of violence and
emergency psychiatry visits are significantly associated with low
barometric pressure. Psychiatric inpatient admissions and suicides
are not associated with any of the weather variables investigated.
Conclusions: While alternate conclusions can be drawn, we propose
that the data support the interpretation that low barometric
pressure is associated with an increase in impulsive behaviours.
Additional investigation is warranted.
Clinical Implications
_ Behaviour is a product of multiple factors, some of which are
subtle.
_ Subjective patient reports of being affected by weather variables
may be valid.
_ Impulsive or violent behaviour may increase when barometric
pressure is low.
Limitations
_ This associational study provides no reliable information regarding
causality.
_ The effect size observed may be too small to be clinically useful,
and replication is essential to ensure that findings are not
accidental.
_ Elements of the study design may have reduced or magnified the
effect size.

Key Words: barometric pressure, emergency psychiatry, impulse

dyscontrol, violence, weather
At the dawn of modern medicine, Hippocrates observed that
cold and warm winds affected the physical and psychological
wellness of his patients (1). A small number of medical
conditions have been linked to weather variables,
particularly barometric pressure (2,3). In psychiatry,
seasonal variations in disease presentations are clearly part
of seasonal affective disorder (4) and bipolar illness (57).
More tenuous associations have also been suggested. It has
been argued that maximum temperature, rainfall, and cloud
cover are significantly correlated with parasuicide in women
(1). Geller and Shannon found an association between high
humidity and psychiatric presentations (8). Conversely,
Modai and others found no correlation between humidity or
other weather variables and psychiatric presentations (9).
The idea that cycles of the moon are associated with acts of
violence such as homicide and suicide (1) has been refuted in
careful studies (10).
Our experience in working in an emergency psychiatric service (EPS)
suggested that barometric pressure is associated with an increase in
the number of emergent visits. Consequently, we examined the
association of variables related to barometric pressure with the
number of EPS visits. Additionally, since abnormal behaviour may
also encompass nonpsychiatric presentations, we also investigated
the relation of weather variables to the incidence of violent crimes
and suicides.
Method
We obtained data for emergent psychiatric visits, 23-hour
psychiatric holding bed unit (HBU) admissions, and inpatient
psychiatric admissions from the University of Louisville Hospital EPS
records for 1999.
With the exception of February, we obtained weather data to reflect
the entire year from the weather division of a local television station
a total of 337 days. (February data had been collected in a
different location and were not available.) We collected data for
violent crimes from the Jefferson County Police Department. Suicide
data were obtained from the medical examiners office.
We applied a regression analysis for associational analyses. Chisquare was used for categorical analyses. Statistical significance
was set at P < 0.05. The primary outcome variable was set a priori
as the relation of behavioural variables and barometric pressure.
Since there were 15 primary comparisons, Bonferroni correction was
not needed.

Results
In 1999, there were 4717 visits to EPS, with 1165 patients placed in
the HBU and 1585 patients admitted to the hospital. There were
3804 violent criminal acts and 71 suicides reported in the 337 days
of the study period.
Barometric pressure was significantly correlated with the number of
EPS visits (r2 = 0.017, P = 0.0178) and total acts of violence (r2 =
0.015, P = 0.0257). There were no significant relations between
barometric pressure and HBU admissions (r2 = 0.005, P = 0.19),
inpatient admissions (r2 = 0.002, P = 0.36), or suicides (r2 =
0.003, P = 0.3).
Wind speed, which is related to periods of changing pressure, was
associated with total violence (r2 = 0.019, P = 0.012), but no other
associations were found (EPS visits r2 = 0.002, P = 0.39; HBU r2 =
0.001, P = 0.62; inpatient r2 = 0.01, P = 0.07; suicide r2 = 0.005,
P = 0.018). Relative humidity was not associated with any of the
measured variables (EPS visits r2 = 0.0003, P = 0.72; HBU r2 =
0.001, P = 0.66; inpatient r2 = 0.00007, P = 0.88; total violence r2
= 0.007, P = 0.14;
suicide r2 = 0.003, P = 0.34).
In 1999, barometric pressure in Louisville was significantly higher in
winter (92.59 [SD 0.191] inH2 O) than in spring (29.51 [SD 0.165]
inH2 O) (F = 3.85, P < 0.05) and summer (29.48 [SD 0.113] inH2
O) (F = 6.12, P < 0.05), but not autumn (29.957 [SD 0.156] inH2
O) (F = 0.26). EPS visits were greater in summer (mean visits
daily15.0, SD 4.3) than in winter (mean visits daily 12.8, SD 3.7) (F
= 4.223, P < 0.05), but not significantly different in spring (mean
visits daily 14.0, SD 4.0) or in autumn (mean visits daily 13.8, SD
4.4). The use of the HBU was least in spring (mean patients daily
3.0, SD 1.5), compared with summer (mean patients daily 3.7, SD
1.5) (F = 2.94, P < 0.05) and autumn (mean patients daily 3.8, SD
1.8) (F = 3.99, P < 0.05), but not when compared with winter
(mean patients daily 3.2, SD 1.5) (F = 0.213). There were no
seasonal differences in the rates of inpatient admissions daily
(winter mean 4.3, SD 2.2; spring mean 5.0, SD 2.4; summer mean
5.0, SD 2.4; autumn mean 4.5, SD 2.3; analysis of variance [ANOVA]
P = 0.12).
Discussion
This study found weak, but significant, correlations between low
barometric pressure and the total number of emergent psychiatric
visits and total reported acts of violent crime. There were no
significant relations between barometric pressure and HBU
admissions, inpatient admissions, or suicides. There were no
associations between any of the behavioural variables and wind

speed or humidity, other than an association between violence and

wind speed.
The larger sample size of the current study (n = 4717) may account
for our finding a relation while Modai and others (n = 393) did not
(9). Further, Modai and others only considered individuals with a
diagnosis of schizophrenia, unipolar depression, or bipolar disorder;
our study included all emergent visits, independent of diagnosis.
One interpretation of our observations is a possible relation between
barometric pressure and impulsivity. Subjects with impulsive
behaviour that does not represent a symptom of a more serious,
ongoing psychiatric illness are usually discharged. Alternatively,
when patients are admitted, it is usually for an illness that has been
worsening over days or weeks. Similarly, violent crimes are
frequently either impulsive or crimes of opportunity; violence is
rarely planned.
If this speculation is accurate, the mechanisms are not readily
apparent. High altitude is associated with changes in mental status
that have been attributed to the lower partial pressure of oxygen
(11). However, more subtle changes have been noted at lower
altitudes (for example, at 7500 ft, the pressure of commercial
aircraft cabins) (11). Changes in cerebral blood flow and physiologic
homeostasis have also been associated with barometric pressure.
For example, intracranial aneurysms are more likely to rupture
during fronts of low barometric pressure (2). Additionally, premature
labour (3) and premature rupture of membranes during pregnancy
(12), but not full-term delivery (13), may occur at a greater rate
during low-pressure fronts. Cerebrospinal fluid (CSF) concentrations
of the serotonin metabolite, 5-hydroxyindoleacetic acid (5-HIAA)
increase significantly with decreasing barometric pressure in men
and women with depression but not in healthy women (14,15).
Similarly, CSF concentration of the norepinephrine metabolite 4hydroxy-3-methoxyphenylglycol (HMPG) increases with increasing
atmospheric pressure in healthy women (15). Clearly, additional
investigation of the effect of barometric pressure on brain
physiology is warranted.
Low barometric pressures are usually associated with adverse
weather conditions (16), and inclement weather may change human
activities. Homeless men with psychiatric diagnoses are more
inclined to visit shelters during inclement weather (17). Thus, the
increase in EPS visits may be related to people seeking refuge from
the elements. Similarly, increased violent crimes may be related to
an increase in domestic violence attributable to people staying at
home in bad weather. Inclement weather is difficult to define. For
example, temperature measures are meaningful only as a deviation
from norms and can only be analyzed seasonally (for example, a

higher-than-typical temperature is good weather in winter but bad

weather in summer). Additionally, precipitation may occur at normal
temperatures or even favourable temperatures (for example, it
may be cooler during a summer rainstorm).
Consequently,
analyzing temperature can be misleading, and we chose not to do
so in this study. We attempted to investigate the issue of inclement
weather by looking at seasonal variation. We found that the greatest
number of EPS visits occurred in the summer, when the weather in
the Louisville area varies from pleasant to hot and is associated with
very little precipitation. The question can only be addressed in a
prospective study that creates season-specific definitions of
inclement weather.
In conclusion, our study showed a significant relation between low
daily barometric pressure and both the number of EPS visits and the
incidence of violent crimes. We propose that this relation is based
on a relation between barometric pressure and impulsivity, but we
acknowledge that other interpretations, including a spurious finding,
may explain the data. We hypothesize that barometric pressure may
alter the propensity toward impulsive behaviour through changes in
brain monoamines (14,15) or cerebral blood flow (2). Additional
study is required to further delineate this relation.