Previous studies consistently showed asymmetrical distributions of suicide deaths. For both sexes, major peaks were found in late spring or summer (“Spring fever”) [7-12], whereas a nadir was noted in December (winter) [13]. Compared with monthly patterns, the daily (referred to as “weekday” or “weekly” in the literature) pattern of suicides is relatively under-researched, particularly in the Hong Kong context. Notwithstanding the limited number of studies, the findings on the daily pattern of suicides were similar to those observed in the related studies, with marked peaks on Mondays and nadirs on weekends [14,15]. Thus, charcoal burning suicides may also have temporal variations if they follow a similar pattern.

However, being a non-violent suicide method, charcoal burning is often assumed to have no obvious temporal patterns. In some seasonality studies, suicides by violent methods (e.g. jumping and hanging) showed more marked temporal variations when compared with nonviolent suicides (e.g. poisoning) [8,16,17]. This may suggest that charcoal burning suicides do not have obvious temporal variations in either monthly or daily distributions. Accordingly, the recent surge in suicides by charcoal burning would lead to a smoothing of the temporal variations in suicide in Hong Kong.

The unique sociodemographic profile of the victims of charcoal burning suicide suggests that they may tend to commit suicide at specific times. Charcoal burning suicide victims were more likely to be economically active and over-indebted than other suicide victims [1]. Although economic activity status had no influence on the monthly pattern of overall suicide deaths [7], its exact impact on the temporal patterns of charcoal burning suicides is unclear. It is reasonable to assume that the employed victims were particularly vulnerable to the fear of job loss, which has been found to significantly predict depressive symptoms and suicide mortality [18,19]. As the weekend represents a potential break from the stresses experienced during the work week, it is possible that suicide rates are lower during the weekend than during the work week. Employed victims may also experience a greater fear of job loss during seasonal layoff periods and thus may be more inclined to commit suicide.

The unemployed may also have a higher suicide risk at specific times as a result of the stress of job searching. Some researchers have noted that during weekends, the unemployed tended to spend more time on socializing activities and significantly less time on job searching [20]. Consequently, unemployed people may be more likely to commit suicide during the week, as they experience higher levels of stress in relation to job searching during this period of time [20]. The unemployed may also be more anxious about finding a job during recruitment seasons. Nonetheless, community-based research in this area is lacking. The influence of economic activity status on the daily and monthly patterns of charcoal burning suicides is yet to be examined.

In addition to economic activity status, sex difference may also influence the temporal pattern of charcoal burning suicides. In previous studies, a bimodal distribution of monthly suicide deaths was observed in women but not in men [21,22]. Sex also interacted with the suicide method on the temporal patterns of suicide [17,23]. As the victims of charcoal burning suicide were predominately economically active men [1,2], an interaction between sex and economic activity status on the temporal pattern of charcoal burning suicides is possible. However, this is yet to be investigated.

To analyze the monthly and daily patterns, the data were divided into 12months of a year and 7days of a week, respectively. To examine the amplitude of temporal variation in suicides and its statistical significance, multinomial Poisson regression analyses were conducted by fitting the number of suicides for each month/day of the week with the corresponding time variable [10,25]. Given that the size of the usual working age population in Hong Kong increased by 10% during the study period, from 3.08 million in 2001 to 3.33 million in 2008, the Poisson regression is a more appropriate method than the Chi-square test, which assumes that there has been no significant change in the size of the population. The Poisson regression can also adjust for changes in regard to sex and economic activity status within the population over the study period.

To adjust for differences in age and sex in the suicide rate and population structure, we included the age, sex, and economic activity status of the suicide victims, the year of incidence and the offset term of population size in the regression equation as the confounding variables. Mathematically, the regression equation is written as follows:

The temporal variation was modeled by comparing the actual frequencies of suicides with the expected values based on the assumption that the suicides are equally distributed. The exponential of the regression coefficient (β) represents the incidence risk ratio (IRR), which describes the multiplicative effect of the corresponding independent variable on the risk [26-28]. A significant IRR indicates that there is a significant difference in the adjusted suicide risk between the corresponding month (or day of the week) and the reference period of time. In the present study, the suicide risks for December and Saturday (which previous studies reported as being the periods of lowest risk [13-15]) were used as references to examine the respective monthly and daily patterns of suicide.

Previous studies have attributed the monthly patterns observed in suicide rates to the influence of seasonality-linked biological, psychosocial and meteorological correlates [12,16,17,31]. As the April peak in the seasonality of suicide has also been observed in other suicide methods, charcoal burning suicides may share some of the common causes of suicides by other methods. In addition to the presence of spring peak in April, it is observed that attempters were more likely to commit suicide by burning charcoal in the colder winter months. From a practical point of view, the sub-tropical climate of Hong Kong would implicitly discourage suicide attempters from burning charcoal in summer to take their own lives. By contrast, the cooler temperatures in winter would make charcoal burning a more acceptable option for suicide. However, the extent to which these explanations are able to account for the monthly pattern of charcoal burning suicides remains unclear. Some studies suggest that the correlation between biological factors and seasonal variation is only observed in violent suicides [32,33]. Furthermore, the findings on the effects of meteorological variables on suicide seasonality are rather inconsistent and have not been replicated in subsequent studies [10,34]. It is worth noting that only employed women tended to commit suicide by burning charcoal in spring (i.e., April). Hence, the seasonal occurrence of particular mental disorders involving suicide ideation fails to explain the temporal variations in charcoal burning suicides. In fact, this is in line with the observation that charcoal burning suicide victims were less likely to be associated with psychiatric illness or substance abuse, compared with those who killed themselves using other methods [1-5]. We expect that more detailed data on the events prior to the death of charcoal burning suicide completers, and on what the suicide attempters were thinking, will help us to explore this knowledge gap in future studies.

The results confirm our hypothesis concerning the daily pattern of charcoal burning suicides. The risk of charcoal burning suicide was found to be higher on weekdays (i.e., Monday, Tuesday, and Thursday) than on weekends. This is consistent with most of the other published findings, which report a peak at the beginning of the week [14,15,35,36]. As suggested by Erazo et al. [36], a significantly higher rate of suicide on Monday seems to be best explained in terms of socio–psychological variables. Relatively speaking, a sense of personal failure and isolation is more likely to be triggered in a depressive person at the beginning of the working week, when their surroundings reflect their duties [36]. A recent study on the relationship between the weekend, work, and well-being also indicated that both men and women tended to experience an enhanced sense of well-being during the weekend compared with working days [37]. Yet, the second peak on Thursday observed in the present study is rather novel in the literature. Further investigation of the relationship between weekdays and psychological stress is warranted.

In line with our expectations, there were pronounced differences in the temporal patterns of charcoal burning suicides in relation to sex. The results identified marked peaks in the rates of charcoal burning suicide in spring and winter (January, February, April, May and November) for working-age women, a less obvious spring peak with a summer nadir was observed for working-age men. This seasonal difference in the magnitude of variation] may be further explained by gender differences in relation to thermal comfort. In an earlier physiological study, Lan et al. [38] found that Chinese women were more sensitive to temperature differences than men and that women preferred a relatively warmer environment because they had a higher comfortable operative temperature. It can thus be expected that women are more inclined to burn charcoal in “cold months” and men are less likely to do so in “hot months”. As our knowledge of the determinants of charcoal burning suicide remains scant, further research is needed to investigate the differential effects of charcoal burning on vulnerable men and women in terms of monthly patterns.

Role of funding source.