| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Psychiatric Bulletin | Advances in Psychiatric Treatment | All RCPsych Journals |
|
|
|
||||||||||
|
||||||||||||
Peasley Cross Hospital, St Helens and Liverpool University
Centre for Paediatric Epidemiology and Biostatistics, Institute of Child Health, University College London, UK
Correspondence: Dr E. Salib, 5 Boroughs Partnership Trust, Stewart Assessment Unit, Peasley Cross Hospital, St Helens, Merseyside WA9 3DA, UK. Fax: +44 (0)174 (0)1744 458461; e-mail: esalib{at}hotmail.com
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
Aims To examine the association between suicide and month of birth using suicide data for a 22-year period in England and Wales. The sample size of 26 915 suicides greatly exceeds all previous studies.
Method We analysed all suicides (ICD–9 codes E950–959) and deaths from undetermined injury (E980–989) reported between 1979 and 2001 in England and Wales for persons born between 1955 and 1966, using Poisson and negative binomial generalised linear models with seasonal components.
Results Birth rates of people who later kill themselves show disproportionate excess for April, May and June compared with the other months. Overall, we found an increase of 17% in the risk of suicide for people born in the peak month (spring–early summer) compared with those born in the trough month (autumn–early winter); this risk increase was larger for women (29.6%) than for men (13.7%).
Conclusions The‘month of birth’factor in suicide can be interpreted in terms of the foetal origins hypothesis.Our findings might have implications for our understanding of the multifaceted aetiology of suicide and may eventually offer new strategies for research and prevention.
|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
This study examines the association between suicide and season of birth using routinely collected suicide data over a 22-year period in England and Wales. The size of the database used in this study provides a powerful opportunity to assess this issue, with nearly 27 000 suicides from over 11 million births, and greatly exceeds the size of all previous studies.
Our hypotheses were that the risk of suicide would vary according to month of birth and that this association would remain after adjusting for the effects of the total number of births per month in the population. We also predicted that this differential in risk of suicide by month of birth would be present when looking at suicides by gender and method of suicide (violent and non-violent).
|
|
METHOD |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
The term ‘suicide’ used in this study includes all cases of suicide (ICD–9 codes E950–959) and undetermined injury deaths (ICD–9 codes E980–989) combined in order to reduce misclassification biases.
Sample selection
National suicide data in England and Wales between 1979 and 2001 recorded
124 661 suicides: 86 566 (69%) verdicts of suicide and 38 095 (31%) open
verdicts. A total of 26 915 suicides completed between 1979 and 2001 by people
born between 1955 and 1966 were selected for our analyses. Data on month and
year of birth and death, method of suicide and gender were available. The
number of births of people who did not die by suicide during the study period
(control group) was obtained by subtracting monthly births of the suicide
sample from monthly population births between 1955 and 1966. There were 11 035
365 births of people who reached at least 16 years of age in England and Wales
in this period.
Data analysis
The total number of suicides of people born in the period 1955–1966
was aggregated according to month of birth into monthly counts. Monthly
frequencies of births and suicides were adjusted to 31-day months. The number
of births of the suicide group in each month was divided by the number of
total births in England and Wales for the same period to obtain birth rates
for the suicide group. For reasons of confidentiality the database only
included information on gender, method, months and years of birth and death,
thus we calculated the age at death as completed months.
In the analyses, ‘violent’ methods of suicide included hanging, drowning, falling from height, shooting, burning, asphyxia and wounding; poison by solids or liquids and poison by gas were regarded as ‘nonviolent’ methods.
Statistical method
Previous studies have used chi-squared tests to analyse seasonality data.
However, since these tests cannot detect complex cyclical trends efficiently
(Edwards, 1961;
Bradbury & Miller, 1985;
Yip et al, 2000), and
suicide is a rare event among the general population, we fitted generalised
linear models (GLMs) with Poisson and negative binomial responses and
logarithmic links. Overdispersion was formally assessed using a likelihood
ratio test comparing Poisson and negative binomial models, taking into
consideration the problems of testing at boundary values
(Cameron & Trivedi, 1998).
Trends by year of birth were adjusted using cubic splines with three effective
degrees of freedom; this functional form was deemed complex enough to model
these trends by minimising Akaike’s information criterion (AIC).
Seasonal variation was modelled using at most three harmonic components in
order to have enough flexibility to detect asymmetrical seasonal variations
which would be lost if only one symmetrical 12-month cycle had been fitted.
The seasonal components used had 12-month, 6-month and 4-month periods: the
first accounts for symmetrical yearly cycles, and the last two were included
in order to allow the detection of asymmetrical yearly patterns. The models
considered also included interaction terms between year of birth, gender and
method of suicide with the seasonal components in order to allow for possible
differences in their shape. The final models presented were selected by
minimising AIC and, as in all stepwise-type model selection procedures,
correlation among the parameter estimates may produce misleading results
– for instance, entered variables can be proxies for others. The final
models have a large number of parameters, and some of the effects that appear
to be significant when comparing the parameter estimates to their standard
errors may be spurious. The analyses of deviance tables for each of the final
models are shown in the Appendices presented as a data supplement to the
online version of this paper. All calculations were carried out using R
version 2.0.0 (R Development Core Team,
2004) in a Windows environment. The negative binomial GLMs were
fitted using the MASS library (Venables
& Ripley, 1998).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
|
Figure 1 shows the monthly suicide rates for people aged over 16 years born between 1955 and 1966 who died by suicide between 1979 and 2001. Figure 2 shows the adjusted monthly number of births for people who did not die by suicide who were born between 1955 and 1966: this is the control population. There is a steady increase in the number of births up to 1962, followed by a period of slower growth. The beginning of the series exhibits the so-called European seasonal pattern (Parkes, 1976) in the distribution of monthly number of births, with a large majority occurring in the spring with a subsidiary peak in September; the latter part of the series shows the start of the shift to the American pattern, with a pronounced peak in September. The transition to this seasonal pattern was complete in England and Wales by the late 1970s. The two series are not stationary and have distinct, albeit different, seasonal patterns with annual peaks and troughs in the monthly birth rates.
|
|
|
Analysis by gender
Since overdispersion was significant in the Poisson regressions, we fitted
negative binomial GLMs. The final model was obtained using the same model
selection strategy as above and included 33 parameters (see Appendix 2 in the
data supplement to the online version of this paper). The term corresponding
to the difference between suicide and non-suicide frequencies had the
following significant interactions:
The strongest main effects were those of the suicide and non-suicide contrast, gender, year of birth trend, and the 12-month and 6-month seasonal components (P < 0.001 in all cases). These interactions establish the different patterns by gender in the asymmetrical cycles determined by month of birth for the suicide cases once the effect of year of birth is adjusted for. Figure 4 shows the detrended seasonal components by gender for male and female suicide and non-suicide cases as a fraction of their means. This model supports the hypothesis of different seasonal phenomena by month of birth for both men and women who die by suicide. Monthly births of men who died by suicide showed a roughly symmetrical pattern, with one trough in autumn and one peak in late spring; for women who died by suicide the seasonal pattern was definitely asymmetrical, with troughs in late winter and early autumn, and a main peak in midsummer taking place slightly later than the peak for men, and a secondary, significant peak in early winter. The average increase of risk between the autumn trough and the summer peak of the detrended seasonal components was 29.6% (95% CI 8.0–50.7) for women and 13.7% (95% CI 5.2–22.2) for men, thus reflecting both the larger number of men dying by suicide and the stronger effect of seasonal variation of month of birth on women completing suicide. The higher birth rates in the spring and early summer for people who die by suicide suggest that some weather variables may be important, and there appears to be a positive deviation from the detrended monthly births of the suicide group coinciding with the annual increase in temperature and sunshine hours during the same months. This was not evident in those who were born during the same period but did not die by suicide.
|
Analysis by method of suicide and gender
Table 2 shows the
frequencies of suicides categorised by method of suicide, gender and month of
birth; it excludes 29 cases for which month of birth was not known.
Men’s suicides involved violent methods significantly more frequently
than women’s (52.4% v. 38%, P < 0.001). There was
no significant difference in the mean ages at suicide by gender. Since
overdispersion was not significant, we fitted Poisson GLMs to explain the
variation of suicide counts by month of birth (using the same seasonal
components and model selection strategy as in the previous sections), gender,
method of suicide (violent and non-violent) and their interactions. In this
analysis we considered only the 26 886 suicides with known month of birth. The
optimal model has 17 parameters (see Appendix 3 in the data supplement to the
online version of this paper) and its predicted detrended values are shown in
Fig. 5. It includes significant
terms for an overall 12-month cycle of month of birth (P < 0.0001)
and significant terms for differences by gender and method and their
interaction (P < 0.0001 in all cases). There is also a significant
interaction term between gender and a 4-month cycle by month of birth
(P=0.011), which establishes the differences in the shape of the
seasonal components in cases of female suicide in a way similar to the one
described in Fig. 4, and a
3-way interaction among gender, method and a 6-month seasonal term
(P=0.028), corresponding to significant differences in the seasonal
patterns of month of birth by method in suicides by women. For suicides by
men, the monthly birth rate peaks were in spring for violent methods and in
summer for non-violent methods; the troughs for both methods occurred in
autumn. For suicides by women, the peaks were in late spring and the troughs
were in autumn for both methods.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
Interpretation of the findings
The effect of season of birth on suicide can be interpreted, on lines
similar to those used by Bradbury & Miller
(1985), as a tendency for
individuals who later die by suicide to be born during certain months or
seasons of the year, in numbers disproportionate to those from the population
at large. Given the multifaceted aetiology of suicide it is important to study
the phenomenon of seasonality of birth in those who die in this way as this
may help to give directions to future research. The large size of the database
used in this study, which greatly exceeds that of all previous studies of this
subject, enhances its power and its inferential value. The database allows us
to establish the presence of significant seasonal patterns, adjusting in
comparison with a large control population. The interaction terms of our
models correspond to joint variation of two or more variables and allow us to
establish the differences of seasonal patterns of month of birth between
suicide cases and the control population, adjusting for year of birth; suicide
cases and the control population by gender, adjusting for year of birth; and
suicide method and gender in suicide cases.
Our results support the hypotheses that there is a seasonal effect in the monthly birth rates of people who kill themselves and that there is a disproportionate excess of such people born between late spring and midsummer compared with the other months. This applied to suicide as a whole and regardless of the methods used, contrary to the findings of previous studies (Chotai & Asberg, 1999; Salib, 2002). Our findings are consistent with the reported higher birth rates in spring and early summer of people with affective disorders and alcoholism (Castrogiovanni et al, 1998), whose deaths by suicide represent about 10% of total expected annual suicides in England and Wales, according to the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness (Department of Health, 1999). However, the study findings, and without taking into account other risk factors, do not appear in keeping with the fact that people with schizophrenia, who represent about 5% of total annual suicides in Britain (Department of Health, 1999), have a well-recognised winter birth association (Bradbury & Miller, 1985; Castrogiovanni et al, 1998; Mortensen et al, 1999). It may be plausible to assume that the small increase in birth rates of some suicide cases noted in December and January (see Fig. 3), particularly in female cases (see Fig. 4), might be related to the suicide of people with schizophrenia born in winter. Perhaps this could be explored in future research of this area.
Biological assumption of suicidal behaviour
Lester (1995) and Asberg
(1997) reported low levels in
cerebrospinal fluid and urine of monoamine metabolites 5-hydroxyindoleacetic
acid (5-HIAA), homovanillic acid (HVA) and methyl-hydroxyphenylglycol (MHPG)
in suicidal behaviour, impulsivity and depression. Low levels of 5-HIAA in
cerebrospinal fluid have been reported in people exhibiting violent suicidal
behaviour, such as hanging, stabbing, using firearms or jumping from heights,
and impulsivity (Asberg, 1997;
Mann & Malone, 1997), as
well as in patients with an increased risk of a future suicide or suicide
attempt (Nordstrom et al,
1994). Low levels of 5-HIAA have also been found in the
cerebrospinal fluid of depressed patients with high-lethality or well-planned
suicide attempts, compared with other suicide attempts
(Nordstrom et al,
1994). Chotai & Asberg
(1999) reported some
significant season-of-birth variations in 5-HIAA and HVA levels in the
cerebrospinal fluid of a sample of drug-free patients, adjusted for gender,
age, height and diagnostic category. Persons born during the winter and spring
months, February to April, had significantly lower values of 5-HIAA. The
values of HVA as well as the ratios of HVA/5HIAA and HVA/MHPG were
significantly higher for those born during the winter months October to
January compared with others. However, winter variations in serotonin reported
by Chotai et al (1999)
are inconsistent with the findings of this study, essentially the opposite of
the Swedish findings.
The biological assumption is possible because factors that may influence the brain growth process after a few months of gestation could affect the sensorimotor, cognitive, affective and behavioural development. It has been suggested that annual rhythm in peripheral and central serotonergic turnover, as well as in L-tryptophan and cortisol, which are observed in normal and depressed people, are related to annual rhythm in suicide (Maes et al, 1994; Petridou et al, 2002). Seasonal biological rhythm, as in serotonergic activity, may be synchronised by seasonal variations in weather conditions (Maes et al, 1994; Salib & Gray, 1997). A positive relationship between relative risk of suicide during the peak month of suicide incidence and the same-month average sunshine duration has been reported by Petridou et al (2002). Melatonin, the secretion of which is suppressed by sunshine, does not appear to have been investigated as a possible contributory factor in relation to suicide.
Seasonality of suicide, whether in relation to the birth or the death of those who kill themselves and the apparent correlation with some weather parameters (Maes et al, 1994; Salib & Gray, 1997; Salib, 2002), is a striking epidemiological characteristic of this phenomenon, which ought to be explored further as it could have some preventive implications. The high degree of sensitivity of the central nervous system to environmental exposures in utero and during infancy (Brenner et al, 2004) and the long latency from initiation to clinical onset of many neurological diseases (Torrey et al, 2000) and several psychiatric disorders (Castrogiovanni et al, 1998) – including suicidal behaviour, as our study has shown – may raise the possibility that seasonally varying exposures acting very early in life might influence the risk in adulthood. The list of candidate exposures includes not only infections, many of which vary seasonally, but also maternal diet, environmental toxins, photoperiod, temperature, weather and hormonal milieu (Brenner et al, 2004).
The ‘month of birth’ factor in suicide can be interpreted in terms of the foetal origins hypothesis (Barker, 1998) and the maternal–foetal origins hypothesis (O’Keane & Scott, 2005). The foetal origins origins hypothesis suggests that exposure of the foetus to an adverse environment in utero leads to permanent programming of tissue functions and subsequent risk of diseases. One system implicated in the putative altered programming in utero is the hypothalamic–pituitary–adrenal axis (O’Keane & Scott, 2005).
On the basis of the foetal origins hypothesis, it may be plausible to suggest that the ‘month of birth’ factor in suicide may reflect the timing of an errant early neural migration or differentiation process due to one or more of these exposures, resulting in subtle histochemical abnormalities underlying the individual’s constitutional vulnerability and affective predisposition. This could act as a latent risk factor that might enhance other suicide risk factors in some predisposed individuals. This assumption gives some hope that developmental approaches might improve our understanding of the psychopathology of depression and suicidal behaviour and could offer new strategies for treatment and prevention.
|
|
Clinical Implications and Limitations |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
LIMITATIONS
|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODRESULTSDISCUSSIONClinical Implications and...ACKNOWLEDGMENTSREFERENCES |
|---|
Barker, D. (1998) Mothers, Babies and Health in Later Life. Edinburgh: Churchill Livingstone.
Bradbury, T. N. & Miller, G. A. (1985) Season of birth in schizophrenia: a review of evidence, methodology, and aetiology. Psychological Bulletin, 98, 569 –594.[CrossRef][Medline]
Brenner, A.V., Linet, M. S., Shapiro, W. R., et al
(2004) Season of birth and risk of brain tumours in adults.
Neurology, 63, 276
–281.
Cameron, A. C. & Trivedi, P. K. (1998) Regression Analysis of Count Data.Cambridge: Cambridge University Press.
Castrogiovanni, P., Iapichino, S., Pacchierotti, C., et al (1998) Season of birth in psychiatry: a review. Neuropsychobiology, 37, 175 –181.[CrossRef][Medline]
Chotai, J. & Asberg, M. (1999) Variations in CSF monoamine metabolites according to the season of birth. Neuropsychobiology, 249, 57 –62.
Chotai, J., Salander, R. & Jacobsson, L. (1999) Season of birth associated with age and method of suicide. Archives of Suicide Research, 5, 245 –254.
Department of Health (1999) Safer Services; National Confidential Inquiry into Suicide and Homicide by People with Mental Illness. London: Department of Health.
Doblhammer, G. & Vaupel, J.W. (2001)
Lifespan depends on month of birth. Proceedings of the National
Academy of Sciences of the USA, 98, 2934
–2939.
Edwards, J. H. (1961) The recognition and estimation of cyclic trends. Annals of Human Genetics, 25, 83 –87.[Medline]
Huntington, E. (1938) Season of Birth: Its Relation to Human Abilities. New York: Wiley.
Kettl, P. A., Collins, T., Sredy, M., et al (1997) Seasonal differences in suicide in birth rate in Alaska Natives compared to other populations. American Indian and Alaska Native Mental Health Research, 8, 1 –10.[Medline]
Kristoffersen, S. & Hartveit, F. (2000) Is a woman’s date of birth related to her risk of developing breast cancer? Oncology Reports, 7, 245 –247.[Medline]
Langagergaard, V., Norgard, B. & Mellernkjaer, L. (2003) Seasonal variations in month of birth and diagnosis in children and adolescents with Hodgkin disease and non-Hodgkin lymphoma. Journal of Pediatric Hematology/Oncology, 25, 534 –538.[CrossRef]
Lawlor, D. A., Davey Smith, G., Mitchell, R., et al
(2004) Temperature at birth, coronary heart disease, and
insulin resistance: cross sectional analysis of the British women’s
heart and health study. Heart,
90, 381
–388.
Lester, D. (1995) The concentration of neurotransmitter metabolites in the cerebrospinal fluid of suicidal individuals: a meta-analysis. Pharmacopsychiatry, 28, 45 –50.[Medline]
Maes, M., De Meyer, F., Thompson, P., et al (1994) Synchronised annual rhythm in violent suicide rate, ambient temperature and light–dark span. Acta Psychiatrica Scandinavica, 90, 191 –196.
Mann, J. J. & Malone, K. M. (1997) Cerebrospinal fluid amines and higher-lethality suicide attempts in depressed in-patients. Biological Psychiatry, 41, 162 –171.[CrossRef][Medline]
Mortensen, P. B., Pedersen, C. B., Westergaard, T., et
al (1999) (1999) Effects of family history and
place and season of birth on the risk of schizophrenia. New England
Journal of Medicine, 340, 603
–608.
Nordstrom, P., Samuelsson, M., Asberg, M., et al (1994) CSF 5-HIAA predicts suicide risk after attempted suicide. Suicide and Life-Threatening Behaviour, 24, 1 –9.
O’Keane, V. & Scott, J. (2005)
From‘obstetric complications’ to a maternal – foetal origin
hypothesis of mood disorder. British Journal of
Psychiatry, 186, 367
–378.
Parkes, A. S. (1976) Patterns of Genderuality and Reproduction.Oxford: Oxford University Press.
Petridou, E., Papadopulos, F., Frangakis, C., et al (2002) Brief report: a role of sunshine in the triggering of suicide. Epidemiology, 13, 106 –109.[CrossRef][Medline]
Pokorny, A. D. (1960) Characteristics of 44 patients who subsequently committed suicide. Archives of General Psychiatry, 3, 314 –323.
Prener, A. & Carstensen, B. (1990) Month of
birth and testicular cancer risk in Denmark. American Journal of
Epidemiology, 131, 15
–19.
R Development Core Team (2004) R: A Language and Environment for Statistical Computing.Vienna: R Foundation for Statistical Computing.
Salib, E. (2002) Month of birth and suicide: an exploratory study. International Journal of Psychiatry in Clinical Practice, 6, 39 –44.[CrossRef]
Salib, E. & Gray, N. (1997) Weather
conditions and fatal self-harm in North Cheshire 1989–1993.
British Journal of Psychiatry,
171, 473
–477.
Sorenson, H. T., Pedersen, L., Olsen, J., et al
(2001a) Seasonal variations in month of birth and
diagnosis of early childhood acute lymphoblastic leukaemia.
JAMA, 285, 168
–169.
Sorensen, H. H., Pedersen, L., Norgard, B., et al
(2001b) Does month of birth affect risk of
Crohn’s disease in childhood and adolescence?
BMJ, 323, 907
.
Torrey, E. F., Miller, J., Rawlings, R., et al (2000) Seasonal birth patterns of neurological disorders. Neuroepidemiology, 86, 1070 –1077.
Venables, W. N. & Ripley, B. D. (1998) Modern Applied Statistics with S-Plus 2nd edn. New York: Springer.
World Health Organization (1977) Manual of the International Statistical Classification of Diseases, Injuries and Causes of Death (9th revision) (ICD–9). Geneva: WHO.
Yip, P., Chao, A. & Chiu, C. (2000)
Seasonal variations in suicide: diminished or vanished: experience from
England and Wales, 1982–1996. British Journal of
Psychiatry, 177, 366
–369.
Received for publication January 20, 2005. Revision received March 31, 2005. Accepted for publication May 6, 2005.
Related articles in BJP:
This article has been cited by other articles:
|
|
 |
|
  J. Chotai Month of birth in relation to suicide The British Journal of Psychiatry, April 1, 2008; 192(4): 313 - 313. [Full Text] [PDF]
|
|
|
|
|
|
E. Salib and M. Cortina-Borja Author's reply The British Journal of Psychiatry, April 1, 2008; 192(4): 314 - 314. [Full Text] [PDF]
|
|
|
|
|
|
P. Jongbloet Month of birth in relation to suicide The British Journal of Psychiatry, April 1, 2008; 192(4): 313 - 313. [Full Text] [PDF]
|
|
|
|
|
|
P. Tyrer From the Editor's desk The British Journal of Psychiatry, July 1, 2007; 191(1): 96 - 96. [Full Text] [PDF]
|
|
|
|
|
|
P. Tyrer The British Journal of Psychiatry, February 1, 2007; 190(2): 188 - 188. [Full Text] [PDF]
|
|
|
|
|
|
G. Cornelissen and F. Halberg Chronomics of suicides and the solar wind The British Journal of Psychiatry, December 1, 2006; 189(6): 567 - 568. [Full Text] [PDF]
|
|
|
|
 |
|
 Minerva BMJ, May 20, 2006; 332(7551): 1222 - 1222. [Full Text] [PDF]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Psychiatric Bulletin | Advances in Psychiatric Treatment | All RCPsych Journals |
