Brit. J. Psychiat. (I976), 529, 472—5
Seasonal
Patterns of Suicide, Depression Electroconvulsive Therapy By M. R. EASTWOOD
Summary. tions
This study
for suicide,
hospital
examined
and J. PEACOCKE
the hypothesis
admissions
and
that there
for depressive
illnesses
are seasonal
and
varia
electroconvulsive
therapy and that these are inter-related. It was found that the seasonal variations did coincide, with peaks occurring in the Ontario spring and autumn. The importance of depression as a cause of suicide and the effect of electroconvulsive therapy upon the suicide rate are discussed.
Davies(1973)discovered no seasonalvariation
INTRODUCTION
The seasonalnatureof suicidehas been re cognized since the time of Durkheim (1897). Thus, Dublin (1963)noted a springpeak for suicidefor severalcountriesin Europe, the USA and Japan and an additionalautumn peak in the USA. Lester (1971) found peaks in May and October in upstate New York, in contrast to deaths from all other causes which had the highest rates in December. Depressive illnesses are often thought to have a seasonal variation, popularly assumed to be in the winter,
but
this is less well documented
than
in the
excretion
of catecholamines
and
Carney
et al (i@@) obtained inconclusive results in theirstudies ofmagnesium levels forpsychiatric diagnoses
at different
times
of the
year.
The
prescription of electroconvulsive therapy (ECT),
which has been suggestedas the treatmentof choice for endogenous also seems
to vary
depression during
(d'Elia,
the year
1975),
(Eastwood
and Peacocke,1975). It was hypothesized that there are seasonal variations for suicide, hospital admissions for depressive illnesses and electroconvulsive
that of suicide. Zung (i@i@,)and Kraines (i@@7), in the USA, both showed that hospital admissions for depression had a non-significant tendency to peak in March/April, and Kraines furtherfound a statistically significant peak in
therapy,
shown
(295 . 7),
and that these are inter-related. METHOD
The admissiondate and dischargediagnosis for all psychiatric admissionsto provincial September. In Germany, Faust (i@7@) showed in-patient facilities were obtained from the that hospital patients with schizophrenia and OntarioMinistryofHealthfortheperiod1969 endogenous depression were admitted in ‘¿warm' to 1973 inclusive. The principaldiagnoses weather, while those with neurotic depression examined were the depressive psychoses (296 and did not have this seasonal variation. Hospital 2g8@o) and depressive neurosis (@
[email protected])@ Other diagnoses studied were schizo-affective disorders studies from the southern United States have during
an
the
increase
summer
in
psychiatric
months
admissions
(Cerbus
other
schizophrenias
(295)
and
the
and
neuroses other than depressive neurosis (300). Dallara, 1975; Cerbus, 1970). The studies cited The admissions for these diagnoses and the for Ontario (suppliedby Stat tendeithertobe unclearin defining depression suicidefigures
or to be on a small scale. Community data are meagre, although Kellner (1966) found no seasonal variation in the frequency of neurosis
istics Canada)
seen by general
fiveyearsof the studythe averagenumber of
practitioners.
More indirectly, 472
were examined
by month
for the
five-year period. The province
of Ontario
has
a population
and during
the
of over 8,ooo,ooo
M. R. EASTWOOD
admissions
to Ontario
psychiatric
facilities
AND
was
6000
approximately 43,000. The ECT data, which were not available for the province, were obtained from three Toronto hospitals; a post graduate hospital
473
J. PEACOCKE
5900
psychiatric institute, a city mental and a psychiatric unit of a general
5800
hospital. These facilities together had approxim ately 750 beds, with 4,000 admissions a year 5700
during the study period. Since there was con sistency for the five years for suicide, depression
and ECT, the following results are expressed as averages for each month. RESULTS
@ @
5500
Suicide The seasonal variation forsuicide isshown in Figure @.The peaks for males are in May and Octoberand forfemalesinMarch and October but are significantly different from a uniform distribution formales only (P