Journal of Affective Disorders 83 (2004) 1 – 9 www.elsevier.com/locate/jad Research report Sex-specific time patterns of suicidal acts on the German railway system. An analysis of 4003 cases N. Erazoa, J. Baumerta,b, K.-H. Ladwiga,b,* a Institut und Poliklinik für Psychosomatische Medizin, Psychotherapie und Med. Psychologie des Klinikums rechts der Isar der Technischen Universität München, Germany b Institute of Epidemiology-GSF National Research Centre for Environment and Health, Germany Received 9 January 2004; received in revised form 26 April 2004; accepted 26 April 2004 Abstract Objective: To examine sex-specific time patterns of suicidal behaviour on railway tracks as basis of suicide preventive strategies. Method: Cases were derived from the National Central Registry of all person accidents on the German railway net (STABAG) between 1997 and 2002 satisfying the operational definition of suicidal behaviour and included sex, age, date, clock time and outcome of the incidence. Results: Over the 6-year observation period, 4003 fatal and non-fatal suicidal incidences were documented. Male to female ratio was 2.70:1 ( pb0.0001). The female subgroup was significantly older than the male subgroup ( pb0.0001). The monthly distribution revealed a bimodal pattern ( p=0.01), particularly in men younger than 65 years, with an excess risk in April and September. This circannual pattern attenuated in the second half of the observation period as shown by adjusted Poisson regression. Monday and Tuesday proved to be high risk days for both sexes. For males and females, a bimodal diurnal distribution pattern with a morning and an evening peak was observed. While both sexes followed the same pattern in the winter half year, in summer females showed a pronounced excess risk in the morning hours while for men the evening peak was substantially amplified ( pb0.0001). Risk assessment revealed a marked broadening by approximately 6 h of the time window in the summer half year compared to the winter half year. Limitations: About 30% of cases were excluded because of missing data on sex specification. However, cases with and without sex information did not differ significantly in the variables under investigation. Conclusion: The analysis revealed marked seasonal, weekly and diurnal peaks of railway suicide intensity. Differences between men and women indicate sex-specific processes underlying their suicidal behaviour. The findings may increase alertness of railway and security personal for particular vulnerable time windows of excess risk for railway suicides. D 2004 Elsevier B.V. All rights reserved. Keywords: Railway suicide; Time pattern; Seasonal excess risk; Sex differences * Corresponding author. Institut und Poliklinik fqr Psychosomatische Medizin, Psychotherapie und Med. Psychologie, Klinikum r.d.I. der TUM, Langerstrage 3, 81675 Mqnchen, Germany. Tel.: +49 89 4140 4399x4311; fax: +49 89 4140 4845. E-mail address: kh.ladwig@lrz.tu-muenchen.de (K.-H. Ladwig). 0165-0327/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2004.04.012 2 N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 1. Introduction Seasonal patterns and diurnal variations of suicidal behaviour are among the most widely established findings in suicide research since its beginning in the 19th century (Morselli, 1981). An asymmetrical distribution of deaths by suicide with a seasonal excess risk in spring/early summer and a diurnal peak between the morning and the early afternoon was consistently confirmed in particular among the elder population and for violent means of suicides (Eastwood and Peacocke, 1976; Maldonado and Kraus, 1991; Maes et al., 1993; Chew and McCleary, 1995; Yip et al., 1998; Hakko et al., 1998a; Altamura et al., 1999; Preti and Miotto, 2001; Partonen et al., in press). Apart from age and degree of violence of the suicide means, temporal variations may be influenced by sex. There is evidence indicating different seasonal asymmetries of suicides for males and females, with a pronounced single spring peak for males, and a bimodal distribution pattern with peaks in spring and autumn for females (Meares et al., 1981; Näyhä, 1983; Lester and Frank, 1988; Micciolo et al., 1991; Preti and Miotto, 1998). Regarding diurnal variations, the day peak was reported to occur later in time for females compared to males (Gallerani et al., 1996). Photoperiod variations leading to subsequent changes of mood and impulsivity induced by the serotonergic pathway is among the best confirmed explanations for the distinct time patterns of suicidal behaviour (Preti, 1997; Preti and Miotto, 1998, 2001). Recent studies have challenged the widely held notion of temporal windows for higher suicide risks. Rihmer et al. (1998), for example, confirmed a marked and significant seasonality with a spring and summer peak between 1981 and 1989, but this seasonality disappeared in the later period between 1990 and 1996. The observed decline in seasonality may be due to changes in socio-psychological variables in Western societies (Yip et al., 2000) and a more effective psychopharmacological treatment of depressive disorders (Rihmer et al., 1998). Without doubt, railway suicides belong to the most violent means of self-destruction. Nevertheless, temporal patterns of railway suicides seem to be distinct from time patterns of violent suicide methods observed in previous studies as outlined before. According to the evidence provided by five population derived railway suicide studies (Veress and Szabó, 1980; Emmerson and Cantor, 1993; Schmidtke, 1994; Deisenhammer et Table 1 Studies on temporal patterns of railway suicides Study Population Sample size Study period Seasonal distribution Daily distribution Circadian distribution Veress and Szabó, 1980 Hungary Suicides: n=436 Attempts: n=50 1965–1969 – – Emmerson and Cantor, 1993 Schmidtke, 1994 Brisbane/Australia Suicides: n=23 1980–1986 – Daylight Federal Republic of Germany Suicides: n=6090 1976–1984 High: April, September/ October No pronounced pattern Females: no pronounced pattern found Males: high: October, low: December Attempts: high: August High: August/ September, low: February Females: high: Monday, Tuesday, Friday Males: high: Monday, Tuesday Females: high: 15–21 h, low: 0–3 h Males: high: 18–21 h, low: 0–3 h With circannual variations Females: high: 15–18 h Males: high: 21–24 h High: 10–24 h, low: 24–6 h With sex-specific circannual variations Attempts: n=391 Deisenhammer et al., 1997 Austria Suicides: n=552 Attempts: n=65 1990–1994 van Houwelingen and Beersma, 2001 The Netherlands Suicides: n=2830 1980–1994 No pronounced pattern High: Monday, low: Thursday High: Monday, Tuesday N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 al., 1997; van Houwelingen and Beersma, 2001) seasonality seems to peak in late summer/autumn and the circadian distribution is markedly shifted to later afternoon and evening hours. Schmidtke (1994) and, recently, van Houwelingen and Beersma (2001) investigated possible influences of season on daily railway suicide variations and found a seasonal modulation of circadian rhythms, which seems to be influenced by the sex factor. In all, however, evidence still is inconsistent and partly relies on studies with small sample sizes, a wide range of different study periods or may also be outdated (Table 1). 1.1. Objectives As a basis for preventive strategies to reduce suicides on railway tracks, the identification of distinct risk patterns of suicidal behaviour is crucial. Furthermore, as railway suicides are the best documented suicides in terms of time, studies in this area can contribute to a more sustained knowledge about temporal patterns. The aim of the present study was, therefore, to determine in an updated and complete data set of suicide events on the entire German railway net whether distinct seasonal, weekly and diurnal rhythms of suicide events could be established. A particular interest is given to possible sex-specific differences in temporal asymmetries. This knowledge may offer useful information for planning railway suicide prevention strategies. 2. Materials and methods The present analysis covers the 6-year observation period of 1997–2002. The database is derived from the national central registry of all person accidents of the German railway company (Deutsche Bahn) which covers the entire German railway track system except municipal subway providers. In the period studied, the mean track length of the German railway system was 37,079.86 km. About 77% of all railway traffic occurs between 06:00 and 21:00 h, 23% during the remaining night-time. Cases of the central registry were included into the data set which satisfied the operational definition of an act of suicidal behaviour according to the ICD-10 3 (International Classification of Diseases, 10th revision; WHO, 1992) category bintentional self-harm by jumping or lying in front of a moving objectQ (code X81). Misclassification and missing records were unlikely as the local police and the local coroner investigate every unnatural death by law. The register contains 5731 suicidal acts during the 6-year study period. For them, an age distribution with two categories (subjects aged V65 and N65 years) was provided. Sex was documented for 4003 suicide victims which were considered as study population of the present analysis. The central registry contains information about the exact time and outcome of the suicidal event. In 90.5% (n=3622) of all cases of the study population outcome was fatal with bdeath within 30 daysQ. Non-fatal events (n=381), which led to a measurable temporary intermission of the railway system (e.g. emergency halt of a train on the open track), were included into the analyses. To analyse the circadian distribution patterns, the day was divided into (i) the 24 h of a day and (ii) eight periods of 3 h, also beginning N00:00 h. To analyse the seasonal distribution patterns, the year was divided into (i) the 12 months of a year and (ii) a summer half year (April–September) and a winter half year (October–March). To analyse a possible time trend of seasonality, the first 3 years were compared with the second 3 years of the 6-year observation period. Statistical associations between categorized variables were assessed by the v 2 test for association. Monthly and diurnal variations were carried out with the v 2 test for equal proportions. Amplitude of the monthly variations in suicides is given as the peak’s percentage above the mean (Preti and Miotto, 2001). A time trend of the monthly distribution stratified for suicides between 1997 and 1999 or 2000 and 2002 was adjusted for age group, track length and year of suicide by Poisson regression (McCullagh and Nelder, 1989). In case of over-dispersion of the Poisson model, the dispersion parameter was estimated by the ratio of the deviance to its associated degrees of freedom (McCullagh and Nelder, 1989). For all statistical analyses a p-value less than 0.05 was considered to be statistically significant. Analyses were performed with SAS, Version 8.2, and, additionally, with the statistical software package SPSS for Windows, Version 11.0. 4 N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 3. Results 3.1. Sex differences A total of 2922 railway suicide victims were male and 1081 female leading to a male/female ratio of 2.70:1 ( pb0.0001). The percentage of the total number of railway suicides accounted for by females did not vary significantly over the period under investigation with an average of 27%. In the female subgroup, 10.1% of the victims were older than 65 years, whereas only 5.8% of the male victims fell into this age category ( pb0.0001). 3.2. Seasonal patterns The analysis of the monthly distribution of railway suicide events revealed a significant bimodal distribution for male suicide victims with a spring peak in April and an autumn peak in September followed by a winter low in December (v 2 test for equal proportions; p=0.01). Females, in principal, followed a similar seasonal pattern. They showed, however, a less evident asymmetric distribution, which did not reach significance (Fig. 1). Additionally, an age-effect on the asymmetrical monthly distribution was observed: against expectation, male subjects younger than 65 years exhibited a more pronounced seasonality with a peak in April (v 2 test for equal proportions; p=0.03) than the elderly with no significant monthly variations ( p=0.21). 3.3. Seasonality over the study period To test a possible time trend in seasonality, monthly variations of the years 1997–1999 were compared with those of the years 2000–2002. For the first time period, a pronounced asymmetrical distribution was found with peaks in April and September (v 2 test for equal proportions; p=0.008). No significant evidence of seasonality, however, was found for the second period ( p=0.178). In Fig. 2, the summarized April and September amplitudes for the years 1997–1999 contrasted to the years 2000–2002 are displayed, stratified by sex: Whereas in the first observation period, among male Fig. 1. Monthly distribution of railway suicides over a 6-year period, stratified by sex (German railway net, 1997–2002). N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 5 3.4. Daily distribution Fig. 2. Summarized amplitudes of April and September for the two time periods 1997–1999 and 2000–2002 stratified by sex (German railway net, 1997–2002). There were more railway suicides for both sexes at the beginning of the week: Monday and Tuesday proved to be high risk days while on the weekend, in particular on Sunday, the lowest suicide density was found (v 2 test for equal proportions; pb0.0001). Most males committed suicide on Mondays, while females rather peaked Tuesdays and had an additional peak on Fridays (Fig. 3). 3.5. Time of the day suicide victims, the summarized spring and autumn percentage above the mean reached more than 30%, this peak disappeared in the second observation period, indicating a reduced seasonality. Adjusted for age group, year of suicide and track length by Poisson regression, a significant monthly variation was estimated in the first 3 years of the study period 1997–1999 ( pb0.01). In contrast, the monthly variation disappeared in the time period 2000–2002 ( p=0.347). In all, regarding 3-h periods, a significant variation in the circadian distribution was observed (v 2 test for equal proportions; pb0.0001): For both sexes, a bimodal pattern with a morning peak between N09:00 and 12:00 h (n=580 suicides, 14.5%) and an evening peak between N18:00 and 21:00 h (n=660 suicides, 16.5%) was found. This diurnal distribution was significantly modified by the winter and the summer half year (v 2 test for association; pb0.0001) as can be seen in Fig. 4. While in the winter half year Fig. 3. Daily distribution pattern of railway suicides, stratified by sex (German railway net, 1997–2002). 6 N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 Fig. 4. Circadian distribution pattern of railway suicides in summer and winter time, stratified by sex (German railway net, 1997–2002). the distribution patterns for both sexes were nearly similar, stratification by sex revealed pronounced differences for the summer half year: Although females and males still exhibited a bimodal distribution pattern, in particular for men, the time window in the summer half year clearly broadened into the late evening hours between N21:00 and 24:00 h (n=271 suicides, 17.4%), whereas women showed an excess risk in the morning time between N09:00 and 12:00 h (n=105 suicides, 19%). summer-to-winter ratio revealed to be 0.74 (95% CI: 0.66–0.82) indicating that in the summer half year the evening peak was about 3 h later than in the winter half year (v 2 test for association; df=1; pb0.006). 3.6. Morning and evening shift A particular interesting issue was the morning and evening shift of suicide peaks in the summer time compared to the winter time. As is shown in Fig. 5, risk assessment revealed a summer-to-winter ratio of 1.45 (95% CI: 1.21–1.74) for the time between N03:00 and 6:00 h contrasting the summer-to-winter ratio of 0.82 (95% CI: 0.74–0.90) for the time between N06:00 and 9:00 h. This result indicates that in the summer half year the morning peak was about 3 h earlier than in the winter half year (v 2 test for association; df=1; pb0.0001). For the evening, the summer-to-winter ratio was 1.44 (95% CI: 1.27–1.64) for the time between N21:00 and 24:00 h, whereas between N18:00 and 21:00 h the Fig. 5. Shifts in circadian distribution patterns of railway suicides by seasonal modulation (German railway net, 1997–2002). N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 4. Discussion The results of the present study on railway track suicides suggest pronounced temporal fluctuations with marked sex differences in suicide rates regarding months, days and hours with interaction patterns between clock time of occurrence, time of the year and sex of the victim. The more violent the method of deliberating self harm, the greater the distribution shifts towards a statistically significant seasonality (Maes et al., 1993; Hakko et al., 1998a,b). This asymmetric distribution was confirmed by our data revealing a suicide peak in April and September followed by a low in December for men. For the females, no significant seasonality was found. This observation is quite similar to findings of other studies on suicidal death on either tracks (Schmidtke, 1994; Deisenhammer et al., 1997) or by all means corroborating a greater seasonality in males (Meares et al., 1981; Micciolo et al., 1991). Interestingly, the seasonality was particularly pronounced for the years 1997–1999 compared to the years 2000–2002. This finding may point to a decreasing seasonality which corresponds with recent suicide studies reporting a complete disappearance of seasonal asymmetries in the last decade of the 20th century (Rihmer et al., 1998; Yip et al., 2000; Parker et al., 2001). Rihmer et al. (1998) explained this observation for Gotland/Sweden as the result of an increasing description of antidepressants. In so far, seasonality of suicides might indicate the rate of depressive suicides in the given population and a decreasing tendency of seasonality might be a sign of the continuous improvement of diagnosis and treatment of depressive disorders. This finding, however, is challenged by recent findings, which again revealed a clear seasonality of suicides in men and women (Partonen et al., 2004, in press). Concerning the weekly distribution, previous studies indicated a peak at the beginning of the week (Angermeyer and Massing, 1985; Schmidtke, 1994; Deisenhammer et al., 1997; van Houwelingen and Beersma, 2001), which is fully confirmed by our data. A rise at the beginning of the week seems to be explained best in terms of socio-psychological variables: For a depressive person, the beginning of the working-week triggers feeling of personal failure and 7 isolation when the surrounding is occupied by their duties. Agreement exists that suicide by all means is largely restricted to daytime (Barraclough, 1976). According to our data, in particular the female subgroup followed this pattern revealing a bimodal course with excess risks in the morning and afternoon hours. The male subgroup, was rather characterised by only one pronounced peak in the late evening. These sex-specific differences were also observed by other railway studies (Schmidtke, 1994; Deisenhammer et al., 1997; van Houwelingen and Beersma, 2001) agreeing that only female suicides are restricted to daytime whereas men tend to commit suicide in the evening or night-time. Reasons for this particular time trend are unknown; however, men may be concerned in limiting the possibility of being seen and eventually hindered in their suicidal plan. Looking two-dimensionally at clock time and time of the year, in the present analysis, marked differences between the winter and summer distribution regarding time of the day were found: Peaks of the summer half year compared to those in the winter half year were clearly shifting to earlier and later clock times, respectively, corresponding to time of sunrise and sunset. Interestingly, a separate analysis for both sexes brought about different peaks for males and females particularly for the summer half year: In the male subgroup, a pronounced late evening peak after sunset was observed whereas the female subgroup was characterised by a single morning peak and no evening peak. These findings support previous analyses on railway suicides revealing pronounced interaction patterns of season, clock time of excess suicide risk and sex of the suicide victims on railway tracks (Schmidtke, 1994; van Houwelingen and Beersma, 2001). Hence, despite our and previous data demonstrating a decline in the amplitude of seasonal variation in suicide rates (Rihmer et al., 1998; Yip et al., 2000), in railway suicide influences on the 24-h patterns are still clearly present. This more subtle influence, however, of time of the year on suicide rates adds a different dimension to seasonality in suicidal behaviour. A peak around sunset and marked differences between men and women strengthen the supposition of biological factors in connection with the light–dark circle being involved in suicidal behaviour particularly for men. 8 N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 The results of the present analyses contribute to suicide research in general by updating scientific knowledge on systematic patterns of suicidal behaviour. Besides this, the results are of practical significance for suicide preventive strategies. For example, the findings on peaks of railway suicide intensity will be broad to knowledge in annual educational programs to all security personal of the Deutsche Bahn to increase alertness for particular vulnerable time windows of excess risk for railway suicides. Maintaining a sense of realism, prevention will never reach all suicidal persons but extensive knowledge on suicidal behavioural patterns may contribute to tailor strategies to reverse the increasing trend to commit suicide on the railway track. 5. Limitations of the study The study population contained fatal and non-fatal suicides. The decision to include non-fatal suicides based on the high degree of violence of railway suicides with survival only being due to unpredictable chance which only permits a true wish to die (cf. Veress and Szabó, 1980; Deisenhammer et al., 1997). An investigation is currently underway to examine whether fatal outcome is influenced by variables as sex, age, date and clock time. The exclusion of 30% of all cases for missing data on sex specification was a substantial figure which, however, were due to administrative reasons as the data set was established purely because of technical interests. A drop-out analysis revealed that cases with and without sex information did not differ significantly in the variables under investigation in the present study. Also the rough age information with only two age categories was solely due to above mentioned administrative reasons of the central data registry of the German railway company. Acknowledgements This study was supported by a grant from the Deutsche Bahn (to Prof. Ladwig). We are indebted to Fritz Schrfder and Jqrgen Bedau for initiating the research project and to Holger Senzel (Deutsche Bahn) for providing the data set. References Altamura, C., VanGastel, A., Pioli, R., Mannu, P., Maes, M., 1999. Seasonal and circadian rhythms in suicide in Cagliari, Italy. J. Affect. Disord. 53, 77 – 85. Angermeyer, M.C., Massing, W., 1985. The monthly and weekly distribution of suicide. Soc. Sci. Med. 21, 433 – 441. Barraclough, B.M., 1976. Time of day chosen for suicide. Psychol. Med. 6, 303 – 305. Chew, K.S.Y., McCleary, R., 1995. The spring peak in suicides: a cross-national analysis. Soc. Sci. Med. 40, 223 – 230. Deisenhammer, E.A., Kemmler, G., De Col, C., Fleischhacker, W.W., Hinterhuber, H., 1997. Eisenbahnsuizide und-suizidversuche in Österreich von 1990–1994. Nervenarzt 68, 67 – 73. Eastwood, M.T., Peacocke, J., 1976. Seasonal patterns of suicide, depression and electroconvulsive therapy. Br. J. Psychiatry 129, 472 – 475. Emmerson, B., Cantor, C., 1993. Train suicides in Brisbane, Australia, 1980–1986. Crisis 14, 90 – 94. Gallerani, M., Avato, F.M., Dal Monte, D., Caracciolo, S., Fersini, C., Manfredini, R., 1996. The time for suicide. Psychol. Med. 26, 867 – 870. Hakko, H., R7s7nen, P., Tiihonen, J., 1998a. Seasonal variation in suicide occurrence in Finland. Acta Psychiatr. Scand. 98, 92 – 97. Hakko, H., R7s7nen, P., Tiihonen, J., 1998b. Secular trends in the rates and seasonality of violent and non-violent suicide occurrences in Finland during 1980–95. J. Affect. Disord. 50, 49 – 54. Lester, D., Frank, M.L., 1988. Sex differences in the seasonal distribution of suicides. Br. J. Psychiatry 153, 115 – 117. Maldonado, G., Kraus, J.F., 1991. Variation in suicide occurrence by time of day, day of the week, month, and lunar phase. Suicide Life-Threat. Behav. 21, 174 – 187. Meares, R., Mendelsohn, F.A.O., Milgrom-Friedman, J., 1981. A sex difference in the seasonal variation of suicide rate: a single cycle for men, two cycles for women. Br. J. Psychiatry 138, 321 – 325. Maes, M., Cosyns, P., Meltzer, H.Y., De Meyer, F., Peeters, D., 1993. Seasonality in violent suicide but not in non-violent suicide or homicide. Am. J. Psychiatry 150, 1380 – 1385. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, 2nd edition. University Press, Cambridge. Micciolo, R., Williams, T., Zimmermann-Tansella, C.H., Tansella, M., 1991. Geographical and urban–rural variation in the seasonality of suicide: some further evidence. J. Affect. Disord. 21, 39 – 43. Morselli, E., 1981. Suicide: An Essay on Comparative Statistics. Kegan Paul, London. N7yh7, S., 1983. Autumn incidence of suicides re-examined: data from Finland by sex, age and occupation. Br. J. Psychiatry 141, 512 – 514. Parker, G., Gao, F., Machin, D., 2001. Seasonality of suicide in Singapore: data from the equator. Psychol. Med. 31, 540 – 553. Partonen, T., Haukka, J., Viilo, K., Hakko, H., Pirkola, S., Isomets7, E., Lfnnqvist, J., S7rkioja, T., V7is7nen, E., R7s7nen, P., 2004. Cyclic time patterns of death from suicide in northern Finland. J. Affect. Disord. 78, 11 – 19. N. Erazo et al. / Journal of Affective Disorders 83 (2004) 1–9 Partonen, T., Haukka, J., Nevanlinna, H., Lfnnqvist, J., in press. Analysis of the seasonal pattern in suicide. J. Affect. Disord. Preti, A., 1997. The influence of seasonal change on suicidal behaviour in Italy. J. Affect. Disord. 44, 123 – 130. Preti, A., Miotto, P., 1998. Seasonality in suicides: the influence of suicide method, gender and age on suicide distribution in Italy. Psychiatry Res. 81, 219 – 231. Preti, A., Miotto, P., 2001. Diurnal variations in suicide by age and gender in Italy. J. Affect. Disord. 65, 253 – 261. Rihmer, Z., Rutz, W., Pihlgren, H., Pestality, P., 1998. Decreasing tendency of seasonality in suicide may indicate lowering rate of depressive suicides in the population. Psychiatry Res. 81, 233 – 240. Schmidtke, A., 1994. Suicidal behaviour on railways in the FRG. Soc. Sci. Med. 38, 419 – 426. 9 van Houwelingen, C.A., Beersma, D.G., 2001. Seasonal changes in 24-h patterns of suicide rates: a study on train suicides in The Netherlands. J. Affect. Disord. 66, 215 – 223. Veress, L., Szabó, T., 1980. Die Besonderheiten der am Eisenbahnkfrper verqbten Selbstmorde. Suizidprophylaxe 7, 401 – 426. World Health Organization, 1992. The ICD-10 classification of mental and behavioural disorders. Clinical Descriptions and Diagnostic Guidelines. WHO, Geneva. Yip, P.S.F., Chao, A., Ho, T.P., 1998. A re-examination of seasonal variation in suicides in Australia and New Zealand. J. Affect. Disord. 47, 141 – 150. Yip, P.S.F., Chao, A., Chiu, C.W.F., 2000. Seasonal variation in suicides: diminished or vanished. Experience from England and Wales, 1982–1996. Br. J. Psychiatry 177, 366 – 369.