You are on page 1of 4

Comprehensive Psychiatry 48 (2007) 554 557

www.elsevier.com/locate/comppsych

The myth of the birthday blues: a population-based study about the


association between birthday and suicide
Udo Reulbach, Teresa Biermann, Katrin Markovic, Johannes Kornhuber, Stefan Bleich
Department of Psychiatry and Psychotherapy, University of Erlangen-Nuremberg, Schwabachanlage 6, D-91054 Erlangen, Germany

Abstract

Purpose: The problem of time distribution of suicide has aroused the interest of various investigators in the past. A population-based study
on the association of the birthday in relation to the time of suicide taking into account sex and the chosen suicide method was conducted.
Methods: The study included all residents of Bavaria who died between 1998 and 2003 and whose death was classified as suicide by the
death certificate. In sum, 11 378 deaths from suicide were included. The suicide ratios in various time frames around the individuals'
birthdays were compared with expected suicide ratios.
Results: The occurrence of suicide did not vary around the birthday. Different time frames (3, 7, 14, 30, and 90 days) before as well as after
the individuals' birthdays were taken into account. Even after dichotomization regarding the chosen method as well as sex differences, no
significant association between birthdays and dates of suicide could be found.
Conclusion: There is no evidence for a birthday blues effect on suicide.
2007 Elsevier Inc. All rights reserved.

1. Introduction themselves in a 60-day period around their birthday than


chance would predict. In a study on a similar population
Various factors have been identified as being associated composed of people older than 75 years, an excess of
with suicide. Because short- as well as long-term temporal suicides in the birth month and the following 2 months has
factors greatly influence many daily activities [1], it seems been observed [8], whereas others found that death through
reasonable to assume an influence of such factors on the suicide was significantly more likely to occur within a
occurrence of suicide as well. For example, it has been 3-month period after the individual's birthday than 3 months
observed that suicide occurred more frequently during the before [9]. A study on children dying from suicide produced
daytime than during the night and therefore presumed that evidence that suicide was more common in a 4-week period
the duration of daylight was a determining factor, for around their birthdays [10], although this was discussed
example, in yearly distributions of suicides. The time of controversially in another study, which failed to reproduce
death by suicide seems to vary throughout the day [2-7]. such an association [11]. Several reasons for this phenom-
Several investigations studied the timing of deaths through enon have been addressed in the aforementioned studies. An
suicide, taking into account specific dates of supposedly individual's birthday may affect morale in the elderly,
significant meaning to the individual, such as his or her own thereby impairing their resistance to fatal disease and giving
birthday or the birthday of close relatives. Barraclough and in to some kind of birthday stress [8].
Shephard [7] found that suicides committed by elderly
1.1. Aims of the study
people older than 75 years were more likely to kill
Identifying an association between an anniversary and
self-inflicted death may be of practical as well as theoretical
Corresponding author. Tel.: +49 9131 8533001; fax: +49 9131 value because the defining of high-risk periods of the
8534105. individual is important for suicide prevention. Therefore, the
E-mail address: udo.reulbach@uk-erlangen.de (U. Reulbach). aim of the present study was to evaluate the association of
0010-440X/$ see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.comppsych.2007.06.006
U. Reulbach et al. / Comprehensive Psychiatry 48 (2007) 554557 555

the individual's birthday and the chosen time for self- All statistical tests were 2-sided. Calculations were
inflicted death in a population-based study. performed by using the statistical package SPSS 10 for
Windows (SPSS, Chicago, IL).
2. Materials and methods
3. Results
2.1. Samples
The study included all residents of Bavaria who died Concerning the means of suicide, 9679 suicides were
between January 1, 1998, and December 31, 2003, whose committed choosing a violent method (85.1%; 7397 male
deaths were classified as suicide by the death certificate. subjects and 2282 female subjects) and 1699 suicides
Bavaria is a federal state of Germany and has about 12.3 (14.9%; 972 male subjects and 727 female subjects)
million inhabitants. Data from the death certificate concern- through a nonviolent method. The choice of a violent or
ing time of death, sex and, birthday were obtained from the nonviolent method was significantly associated with sex
Bavarian State Office for Statistics and Data Processing (2 = 274.3, P b .001).
(Munich, Germany). Because of the obligations of secrecy of The distribution of suicides within 3, 7, 14, 30, and
the German law, we were not provided with information 90 days preceding as well as after the individuals' birthdays
concerning the individual's age and year of death. Data did not differ significantly from the expected values (as
concerning occupation, marital status, and actual impairment shown in Tables 1 and 2). No sex differences could be
through alcohol or drugs and medical history were therefore observed, either. However, after stratification regarding the
not available, too. chosen method (nonviolent vs violent), some ratios did differ
In total, in the present study, 11 378 suicides (8369 men significantly from 1 in the explorative analysis. The ratio of
and 3009 women; male: female ratio, 2.8:1) were included. nonviolent suicides was significantly lower in the week
From these subjects, detailed data concerning exact time of before the birthday (ratio, 0.73; P = .049). This was the sole
birth and death were available. significant ratio in the time frames before the birthday. The
To obtain the observed number of deaths occurring ratios for nonviolent methods were significantly higher in
either before or after 3 days, 1 week, 2 weeks, 1 month, the time frames 3 days (1.50, P = .015), 2 weeks (1.19,
and 3 months around the individuals' birthdays, we P = .050), and 1 month after the birthday (1.14, P = .041). In
calculated the shortest time interval between the date of contrast, the ratio for a violent method was significantly
each death and birthday. The shortest time was the birthday lower 1 week after the birthday (0.89, P = .05).
itself, the longest time 182 days. Afterwards, the number of
deaths and the suicide ratios for each aforementioned
interval were calculated. This was done under the
Table 1
assumption that the time difference between the day and
The occurrence of suicide within specific time frames before the individual's
month of death due to suicide and the day and month of birthday
birthday has a uniform distribution.
3d 1 wk 2 wk 1 mo 3 mo
Furthermore, suicides were categorized with regard to birthday birthday birthday birthday birthday
sex differences and the chosen suicide method as well,
Violent method
which was classified into violent and nonviolent. n 102 215 399 833 2458
Violent methods included hanging, drowning, shooting, SR 0.96 1.01 1.00 1.01 1.02
vehicle accidents, and stabbing (International Statistical Men 72 160 300 628 1861
Classification of Diseases, 10th Revision, ICD-10 X70-84) SR 0.89 0.99 0.99 1.00 1.01
nonviolent methods were composed of self-poisoning by Women 30 55 99 205 597
SR 1.20 1.10 1.06 1.06 1.05
poison or gas (ICD-10 X60-69). Nonviolent method
n 13 27 64 138 429
2.2. Statistical methods
SR 0.7 0.73 0.92 0.96 1.01
Men 9 16 37 79 239
The univariate analysis was performed by applying the
SR 0.85 0.75 0.93 0.96 0.99
2 test statistic and by calculating suicide ratios in relation to Women 4 11 27 59 190
the length of each considered time frame. Ratios were SR 0.5 0.69 0.9 0.96 1.05
evaluated for statistical difference from 1 (ratio, observed vs Total
expected suicides) by assuming a binomial distribution. Five n 115 242 463 971 2887
SR 0.92 0.97 0.99 1.01 1.02
time frames before and after birthdays were used. In
Men 81 176 337 707 2100
addition, strata for method and sex were computed. There- SR 0.88 0.96 0.98 1.00 1.01
fore, 90 ratios were calculated. For the first step of an Women 34 66 126 264 787
explorative analysis, the significance level was set at = .05. SR 1.03 1.00 1.02 1.03 1.05
In a second step, all P values were corrected for multiple SR indicates suicide ratio.
testing by using the method of Bonferroni. P b.05, statistically significant before correction for multiple testing.
556 U. Reulbach et al. / Comprehensive Psychiatry 48 (2007) 554557

Table 2 Furthermore, 85.1% of suicides included in the present


The occurrence of suicide within specific time frames after the individual's study were committed using a violent method. In contrast to
birthday
former findings [4], violent suicide methods seemed to be
3d+ 1 wk + 2 wk + 1 mo + 3 mo + overrepresented in the present study. This might be explained
birthday birthday birthday birthday birthday
by the fact that suicide committed by means of a nonviolent
Violent method method, for example, self-poisoning, might not have been
n 95 188 371 826 2458
discovered as a suicide [4] and therefore not be mentioned in
SR 0.9 0.89 0.93 1.01 1.02
Men 69 138 281 637 1880 the death certificate.
SR 0.85 0.85 0.93 1.01 1.02 Effects of stressors on suicide and the chosen method can
Women 26 50 90 189 578 be modified by age and sex as well [14]. It is often assumed
SR 1.04 1.00 0.96 0.98 1.02 that women kill themselves because of emotional and men
Nonviolent method
because of achievement problems [15]. It has been well
n 28 41 83 164 436
SR 1.5 1.1 1.19 1.14 1.03 documented that men have higher ratios of externalizing
Men 17 26 48 89 244 behaviors [15]. In this context, men might refer to a certain
SR 1.6 1.22 1.2 1.08 1.01 problem rather in a traditional masculine way, whereas
Women 11 15 35 75 192 women tend to react to special circumstances rather as a
SR 1.38 0.94 1.17 1.22 1.06
relational being [16].
Total
n 123 229 454 990 2894 In addition, the individual's age might play an important
SR 0.99 0.92 0.97 1.03 1.02 role in the timing of suicide. Elderly persons might be
Men 86 164 329 726 2124 affected by their own anniversary, whereas younger people
SR 0.94 0.89 0.96 1.02 1.02 are predominantly affected by another's [17].
Women 37 65 125 264 770
SR 1.12 0.99 1.01 1.03 1.03 4.1. Limitations of the study
P b .05, statistically significant before correction for multiple testing.
There are some limitations to the present study that need
to be addressed.
However, no significant ratios could be observed in all 90 Firstly, the exact time of death of a suicide in retrospect is
ratios, after correction for multiple testing. often difficult to determine [4].
Secondly, it is sometimes difficult to determine if a death
is a suicide or is unintentional. Because the information
4. Discussion concerning time and manner of suicide relied on the
outcomes of the death certificate, the timing and the
According to the findings of the present study, there is classification of suicides can be assumed to be as accurate
no evidence for an alteration of suicide ratios in as possible.
connection to the individual's birthday, which is in line Thirdly, a larger error of the exact timing of each
with previous studies on suicide as well as on suicide individual's death might be involved in using time of death
attempts [11,12]. In some strata of our analyses, significant resulting from the act of suicide [4]. If the ratio of this error
ratios could be observed. Because no pattern is visible, this did not vary by time of death, this nondifferential error might
is probably due to pure chance. Moreover, regardless of have caused the observed variation to be smaller than the true
the use of any correction system for multiple testing, no variation [18]. The timing of suicide can be assumed to be
evident association between suicide and birthday could be most accurate when violent methods are used [19]. Never-
observed. However, the present results are contradictory to theless, such limitations are unlikely to have a major effect
the findings of previous studies, which produced evidence on a measure of association.
for an increase in suicide ratios in a time frame of up to Fourthly, a major limitation is the impossibility of
3 months after the birthday [7,9]. Another study on the obtaining information on age and other sociodemographic
birthday blues in suicide described higher suicide ratios variables.
in the 3 months after birthdays [9]. However, even if there However, the strength of the present study lies in its
would be a slight difference in the chosen time of suicide, population-based design, the large number of cases, the
it might not be useful in suicide prevention because it accuracy of the data, and the completeness of the study
could be applicable to natural deaths or deaths from other population.
causes, too [13]. However, the psychological stresses
associated with birthdays or anniversaries in general may
reinforce feelings of loneliness and hopelessness and might 5. Conclusions
therefore raise suicidal thoughts and actions as well. It is
feasible that both the prospect as well as the retrospect of Psychiatric, social, and environmental predictors of
birthday may be distressing in subjects at risk of suicide have been widely studied to improve prediction
committing suicide [7]. and prevention of suicide. A better understanding of the
U. Reulbach et al. / Comprehensive Psychiatry 48 (2007) 554557 557

timing of suicide might help to detect at-risk populations [8] Alderson M. Relationship between the month of birth and the month of
and to develop better treatment interventions. However, no death in the elderly. Br J Prev Soc Med 1975;29:151-6.
[9] Kunz P. Relationship between suicide and month of birth. Psychol Rep
such effect as a birthday blues could be observed in the 1978;42:794.
present study. [10] Shaffer D. Suicide in childhood and early adolescence. J Child Psychol
Psychiatry 1974;15:275-91.
[11] Lester D. The birthday blues revisited. Acta Psychiatr Scand 1986;73:
References 322-3.
[12] Beck AT, Lester D. Attempted suicide and month of birth. Psychol Rep
[1] Durkheim E. Suicide: a study in sociology. New York: Free Press; 1973;33:506.
1951. [13] Kunz P. And a time to die. Mt West 1977;3:36.
[2] Caracciolo S, Manfredini R, Gallerani M, Tugnoli S. Circadian rhythm [14] Heikkinen M, Isometsa ET, Aro HM, Sarna SJ, Lnnqvist JK. Age-
of parasuicide in relation to violence of method and concomitant related variation in recent life events preceding suicide. J Nerv Ment
mental disorder. Acta Psychiatr Scand 1996;93:252-6. Dis 1995;183:325-6.
[3] Gallerani M, Avato FM, Dal Monte D, Caracciolo S, Fersini C, [15] Mller-Leimkhler AM. The sex gap in suicide and premature death
Manfredini R. The time for suicide. Psychol Med 1996;26:867-70. or: why are men so vulnerable? Eur Arch Psychiatry Clin Neurosci
[4] Maldonado G, Kraus JF. Variation in suicide occurrence by time of day, 2003;253:1-8.
day of the week, month, and lunar phase. Suicide Life Threat Behav [16] Kotila L, Lonnqvist J. Adolescent suicide attempts: sex differences
1991;21:174-87. predicting suicide. Acta Psychiatr Scand 1988;77:264-70.
[5] Motohashi Y. Circadian variation in suicide attempts in Tokyo from [17] Bunch J, Barraclough BM. The influence of parental death
1978 to 1985. Suicide Life Threat Behav 1990;20:352-61. anniversaries upon suicide dates. Br J Psychiatry 1971;118:621-6.
[6] Williams P, Tansella M. The time for suicide. Acta Psychiatr Scand [18] Copeland KT, Checkoway H, McMichael AJ, Holbrook RH. Bias due
1987;75:532-5. to misclassification in the estimation of relative risk. Am J Epidemiol
[7] Barraclough BM, Shepherd DM. Birthday blues: the association of 1977;105:488-95.
birthday with self-inflicted death in the elderly. Acta Psychiatr Scand [19] Barraclough BM. Time of day chosen for suicide. Psychol Med 1976;
1976;54:146-9. 6:303-5.

You might also like