The effect of seasonal and meteorological factors on suicide rates in the Lithuanian population in 2001–2021: a case‑crossover study

Abstract

Background

The creation and implementation of suicide prevention measures requires an evaluation of the full range of potential risk factors. This study examined the associations between the total number of suicides in the season-specific context and various meteorological parameters in Lithuania as a country of the Boreal region.

Methods

This study included data on 21,487 cases of suicide in Lithuania from 2001 to 2021. The data were collected from the Register of the Institute of Hygiene. The associations between weather variables and the daily number of suicides were evaluated using a multivariate Poisson regression.

Results

The rate ratio (RR) of suicide was observed to be highest in two summer months, June and July (1.35 and 1.36, respectively), compared to January, with the same tendency in both sexes and different ages. Elevated temperature was associated with a significantly higher RR of suicide in the general population, in men, and across all ages. Moreover, a higher number of sunny hours per day was also associated with higher RR in the whole sample, in both sexes and at the youngest and the oldest ages. Higher atmospheric pressure was found to be associated with lower suicide RR in the total sample, in men, and in the youngest and the middle age groups. An average relative humidity (RH) of 50% or more, compared to an average relative humidity < 50%, was associated with a significantly higher RR of suicide in the total sample and in the men’s group.

Conclusions

There is significant evidence of seasonal effects on suicide. Clinical services in Lithuania should be aware of the risk of suicide on the hot, long summer days, especially among men.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-25586-9.

Background

Suicides are an avoidable societal issue causing more than 800,000 annual deaths around the world [1]. Lithuania has long been known for having exceptionally high suicide rates, both within Northern (Boreal) Europe and worldwide (Lithuania is among the top 10 countries with the highest suicide rates [2]). Although there have been some positive changes in recent years, suicide remains a major public health concern, especially in comparison to other Boreal region countries [3]. Thus, a search for new risk factors for suicide, including those linked to the environment, remains a relevant task. There are many countries where the pattern of seasonal suicides has been confirmed, but the mechanisms remain unclear. However, the most likely main mechanisms for seasonal suicide patterns include seasonal fluctuations in social activities and meteorological factors, such as temperature, the amount and duration of sunshine, air pressure, precipitation, and humidity [4–11]. However, in Lithuania and surrounding countries, there are no studies about seasonal changes, meteorological parameters, or their associations with suicides.

Also, it has been broadly and frequently observed that the suicide rates peak in spring and summer [4, 8, 11–14]. Another theory suggests that greater exposure to sunlight may have an impact on biological processes, potentially by disrupting the production or metabolism of serotonin and melatonin [8, 11–13, 15, 16]. However, most earlier studies examining the effect of sunlight on suicide either overlooked seasonal variations or interpreted them as supporting a positive link between sunlight and suicide [11–13].

A peak of suicides in spring has been observed in countries of the northern and southern hemispheres [17–19]. Large population-based studies have found that ambient temperature, relative humidity [19], daylight duration, and sunlight intensity [17–19] can affect suicidal behavior. Elevated ambient temperatures are also believed to impact serotonin-related neurotransmitter systems, which may increase impulsivity and aggression, potentially contributing to the risk of suicide [20–22]. Earlier studies have identified a positive link between higher temperatures and suicide rates, with this relationship appearing more pronounced among men and older individuals [20, 23].

However, more recent studies have shown that in spring, the number of suicides is on the decline, while new small peaks are occurring in other times of the year, especially in industrialized Western countries [4, 24].

The aim of our study was to evaluate the associations between the total number of suicides in the season-specific context and various meteorological parameters in Lithuania and to formally assess effect modification by age and sex.

We hypothesized a priori that (a) there is an association between meteorological parameters and suicides (the risk of suicide is significantly higher on days with an increasing temperature, low relative humidity, low air pressure, and low number of sunny hours [7]); (b) the association is not limited to the winter and fall months; (c) age and sex modify the effect of season and meteorological parameters on suicide (the elderly and men are more affected by meteorological and seasonal variables).

Methods

Study population

We conducted a population-based case-crossover study of the associations between seasons, meteorological variables, and the risk of suicide in Lithuania from 2001 to 2021. For this study, the population-based register of the Institute of Hygiene under the Ministry of Health was used to identify cases of suicide. All suicide cases of people who were permanent residents of Lithuania and who died by suicide in Lithuania from 2001 to 2021 were included in this study. Each suicide case was assessed and recorded according to the clinical diagnosis set by the physician using the International Classification of Diseases, 10th Revision (ICD-10), codes X60-X84.

Exposure assessment

A continuous time series of daily meteorological factors (mean, minimum, and maximum air temperature, atmospheric pressure, average relative humidity, and sunshine duration) in the study was obtained from the Lithuanian Hydrometeorological Service under the Ministry of Environment of the Republic of Lithuania. Daily meteorological data of the study are based on values measured at the Kaunas monitoring station, Vilnius International Airport station and Vilnius monitoring stations. These data represent weather patterns of a large part of Lithuania, namely, Middle Lowland and South-Eastern Upland climatic regions. About 83.7% of suicides fall within this area. Besides, the seasonal variations in weather variables in the whole territory of Lithuania are similar. These monitoring stations meet the quality control standard LST EN ISO 9001: 2015 (Certificate No. 9000–560 A). After manually checking the data for outliers, we formed a frequency distribution of daily meteorological factors over all seasons during the 21 years from 01/01/2001 to 31/12/2021. The range, the mean value, and the standard deviation (SD) of the daily weather variables were, respectively, −25.7–29.1, 7.3 (11.1) °C for air temperature, 26–100, 81.0 (14.7)% for relative humidity, 953–1021, 987.6 (18.5) hPa for atmospheric pressure, and 0–16.7.7, 5.2 (4.9) hours for sunshine duration.

Statistical analysis

The possible associations between weather variables and the daily number of suicides were evaluated by using a generalized linear model with a Poisson distribution. As the daily numbers of suicides Y_t are a count variable and the mean and variance for suicides were similar size, we assumed that Y_t followed a Poisson distribution with mean λ_t, depending on predictor variables X⁽¹⁾, X⁽²⁾,…, X^(k):

,

where β₁, β₂, _…, β_k are regression coefficients. In our study, the mean and variance for suicides were similar size. For categorical predictors, the exp(β_i) was defined as the adjusted (for the remaining predictors) RR and assesses the change in the rate of suicide as compared to the reference category. For continuous X⁽ⁱ⁾, such as air temperature, the RR shows how time changed the rate of suicide for every unit increase in X⁽ⁱ⁾.

As the annual trend of suicide rate was changing during the study period due to changes in population size and prevention measures, we included the temporal function of the annual trend of suicide rate in the regression models [25].

First, we assessed the associations between the risk of suicide and the seasonality and the day of the week. For this purpose, the day the week and the calendar month were included in the Poisson regression model with the annual trend of suicide rate as categorical predictors. The association between weather variables and suicides was evaluated by including them in the Poisson regression model with covariates: the temporal trend and categorical variables of the calendar month (reference category January) and the day of the week (reference category Monday).

Second, we analyzed the effect of air temperature adjusting for the month and the day of the week.

Third, we assessed the effect of the type of holiday, the day length, and other weather variables on the risk of suicide. These variables were included one by one in the model with predictors such as the month and the day of the week and the variable of air temperature. The day length and weather variables were used as continuous and categorical predictors. The cut-offs of categorical variables were detected by using the regression tree method [26]. The aim of the regression trees is to predict the continuous outcome by dividing the predictor space into high-dimensional rectangles. We divided the set of values of each weather variable into distinct and non-overlapping intervals in which the mean values of the daily suicide rates were the most different. These intervals can help to detect the areas of higher risk of suicide. If the categorized variable was binary, the reference category was chosen based on the lower suicide rate. Otherwise, the first category of the continuous variable (with the lower values) was chosen as the reference category. In the analysis, we used weather variables on the day of the suicide and the previous 1–2 days (with a lag of 0, 1, and 2 days, respectively). The choice of the weather variable (categorical or continuous) and the optimal lag were selected using the Akaike information criterion.

Fourth, we created a multivariate model by including statistically significant weather variables. We checked the autocorrelations of the residuals using partial autocorrelation functions for the created model. If the residuals were correlated, we included the lagged variable of daily suicide in the model.

The analyses were performed separately for men and women and in age groups of ≤ 40 years, 41–55 years, and ≥ 56 years. We used these age groups with an equal proportion, as determined by the tercile method. There was a similar number of suicide cases in each age group. We also tried to analyze other age groups, but we found that the case disparities were too large, and the results might be inaccurate. For the assessment of the impact of weather variables, we used the adjusted rate ratios (RRs) in the multivariate Poisson regression model with a 95% confidence interval (95% CI) and the p-value. For air temperature, RR per a 10 °C increase was used. Statistical analysis was performed using SPSS 20 software (IBM Corp. 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.).

Results

A total of 21,487 suicide cases were included in our dataset, from which 17,588 (81.9%) of these cases were men, while 3,899 cases (18.1%) were women. The main suicide rates by sociodemographic factors in Lithuania by year of investigation (2001–2021) are shown in Table S1. All suicide cases were aged 9–103 (mean 49.0, standard deviation (SD) 17.7). Most suicides occurred in spring and summer. Table 1 shows the characteristics of the study population, with differences in the incidence of suicide by season in sex and age groups.

Table 1.

Characteristics of the cases of suicide in Lithuania by season in sex and age groups

Characteristic	Autumn, n (%)	Winter, n (%)	Spring, n (%)	Summer, n (%)	All, n (%)
All cases	4910 (22.8)	4445 (20.7)	5864 (27.3)	6268 (29.2)	21,487 (100)
Men	3984 (22.6)	3676 (20.9)	4796 (27.3)	5132 (29.2)	17,588 (100)
women	926 (23.8)	769 (19.7)	1068 (27.4)	1136 (29.1)	3899 (100)
≤ 40 year.	1642 (23.0)	1682 (23.5)	1848 (25.9)	1972 (27.6)	7144 (100)
41–55 year.	1491 (21.8)	1374 (20.1)	1932 (28.2)	2047 (29.9)	6844 (100)
≥ 56 year.	1777 (23.7)	1389 (18.5)	2084 (27.8)	2249 (30.0)	7499 (100)

Subsequently, we assessed the associations between the risk of suicide by the calendar month (Table 2) and the day of the week (Table S2). According to the model, the risk of suicide was higher on Mondays as compared to other days (Table S2).

Table 2.

Associations between months and daily number of suicides by sex and age in Lithuania, 2001–2021, expressed as rate ratios and 95% confidence intervals

Months	RR (95% CI) All	p	RR (95% CI) Men	p	RR (95% CI) Women	p	RR (95% CI) ≤ 40 year.	p	RR (95% CI) 41–55 year.	p	RR (95% CI) ≥56 year.	p
December	0.88 (0.82–0.95)	0.001	0.88 (0.81–0.95)	0.001	0.91 (0.76–1.08)	0.257	0.86 (0.76–0.96)	0.009	0.92 (0.81–1.05)	0.220	0.87 (0.77–0.99)	0.039
November	0.98 (0.91–1.05)	0.529	0.96 (0.88–1.03)	0.242	1.09 (0.92–1.29)	0.308	0.89 (0.79–1.00.79.00)	0.051	1.01 (0.89–1.15)	0.845	1.05 (0.93–1.19)	0.439
October	1.04 (0.97–1.11)	0.279	1.02 (0.95–1.10)	0.574	1.12 (0.95–1.32)	0.173	0.94 (0.84–1.06)	0.306	1.00 (0.89–1.14)	0.948	1.19 (1.06–1.34)	0.004
September	1.10 (1.02–1.17)	0.008	1.06 (0.99–1.15)	0.106	1.26 (1.07–1.47)	0.005	0.90 (0.81–1.02)	0.087	1.11 (0.98–1.25)	0.113	1.32 (1.17–1.48)	< 0.001
August	1.23 (1.15–1.31)	< 0.001	1.21 (1.13–1.30)	< 0.001	1.31 (1.12–1.53)	0.001	1.03 (0.93–1.16)	0.556	1.30 (1.16–1.46)	< 0.001	1.38 (1.24–1.55)	< 0.001
July	1.36 (1.28–1.45)	< 0.001	1.34 (1.25–1.44)	< 0.001	1.44 (1.23–1.68)	< 0.001	1.11 (1.00–1.24.00.24)	0.052	1.48 (1.32–1.66)	< 0.001	1.54 (1.38–1.72)	< 0.001
June	1.35 (1.26–1.44)	< 0.001	1.32 (1.23–1.42)	< 0.001	1.47 (1.25–1.71)	< 0.001	1.10 (0.99–1.23)	0.088	1.46 (1.30–1.64)	< 0.001	1.53 (1.37–1.71)	< 0.001
May	1.33 (1.25–1.42)	< 0.001	1.32 (1.23–1.41)	< 0.001	1.39 (1.19–1.63)	< 0.001	1.08 (0.96–1.20)	0.200	1.42 (1.26–1.59)	< 0.001	1.55 (1.39–1.74)	< 0.001
April	1.25 (1.17–1.34)	< 0.001	1.23 (1.14–1.32)	< 0.001	1.40 (1.19–1.63)	< 0.001	1.01 (0.90–1.13)	0.905	1.37 (1.22–1.54)	< 0.001	1.44 (1.29–1.61)	< 0.001
March	1.09 (1.02–1.17)	0.010	1.08 (1.00–1.16.00.16)	0.053	1.18 (1.00–1.38.00.38)	0.049	0.96 (0.86–1.08)	0.517	1.21 (1.07–1.36)	0.002	1.14 (1.01–1.28)	0.033
February	0.96 (0.89–1.03)	0.236	0.94 (0.87–1.02)	0.145	1.03 (0.87–1.22)	0.737	0.97 (0.86–1.08)	0.548	0.97 (0.85–1.10)	0.641	0.94 (0.83–1.07)	0.320
January (ref.)	1		1		1		1		1		1

A higher risk of suicide was associated with all months in the March-September period compared to January. In July, the risk of the daily number of suicides was the highest − 36% (RR 1.36 95% CI 1.28–1.45) among all cases (Table 2). The results of the subgroup analyses by sex and age are also presented in Table 2. The calendar months of July and June were associated with the highest risk of daily number of suicides for both men (34%; RR 1.34 95% CI 1.25–1.44) and women (47%; RR 1.36 95% CI 1.25–1.71) in the older age groups (41–55 years and > 56 years, respectively). This suggests that the younger the age, the lesser the role that seasonality plays in suicide risk (Table 2).

We created a multivariate model with predictors: air temperature, the number of sunny hours, atmospheric pressure, day length, low relative humidity, January 1 st (New Year’s Day) (Table S3), the month, and the day of the week. The autocorrelations of the residuals of the models, except for all suicides, were non-significant. For the model for all suicides, autocorrelations of the 4th and the 5th order were statistically significant. In this model, we included the daily number of suicides on the third and the fifth day before. The autocorrelations of this model were non-significant. The results of the complex models are presented in Table 3. The data presented in Table 3 show a positive association between the risk of suicide and air temperature (except for women) and a higher number of sunny days (except for those aged 41–55 years). A higher risk of suicide was observed during days with lower air pressure, higher than 50% RH, and longer day length, except in the age group of 56 + years and in women.

Table 3.

Complex model of the associations of the risk of suicide with weather variables (adjusting for the month, years, the January 1 st, and the day of the week) by sex and age in Lithuania, 2001–2021, expressed as rate ratios and 95% confidence intervals

Meteorological parameters	RR* (95% CI) All	p	RR (95% CI) Men	p	RR (95% CI) Women	p	RR (95% CI) ≤ 40 year.	p	RR (95% CI) 41–55 year.	p	RR (95% CI) ≥ 56 year.	p
Air temperature (every 10 °C)	1.12 (1.09–1.16)	< 0.001	1.14 (1.10–1.18)	< 0.001	1.06 (0.98–1.15)	0.163	1.12 (1.06–1.19)	< 0.001	1.11 (1.05–1.18)	0.001	1.14 (1.08–1.21)	< 0.001
Number of sunny hours ≥ 9,26 h/per day	1.06 (1.01–1.11)	0.011	1.04 (0.98–1.09)	0.183	1.19 (1.07–1.33)	0.002	1.07 (0.99–1.16)	0.111	0.99 (0.92–1.08)	0.867	1.13 (1.04–1.22)	0.003
Number of sunny hours 0,76 − 9,25 h/per day	1.05 (1.01–1.09)	0.009	1.05 (1.01–1.09)	0.022	1.06 (0.97–1.16)	0.215	1.05 (0.99–1.12)	0.111	0.98 (0.92–1.04)	0.498	1.12 (1.05–1.19)	0.001
Number of sunny hours ≤ 0,75 h/per day (ref.)	1		1		1		1		1		1
Air pressure ≥ 1006 hPa	0.85 (0.78–0.92)	< 0.001	0.86 (0.79–0.94)	0.001	0.81 (0.67–0.98)	0.026	0.84 (0.74–0.95)	0.008	0.82 (0.71–0.94)	0.005	0.90 (0.79–1.03)	0.118
Air pressure 980–1005 hPa	0.92 (0.88–0.97)	0.001	0.92 (0.87–0.97)	0.003	0.94 (0.83–1.05)	0.272	0.91 (0.84–0.99)	0.031	0.94 (0.86–1.03)	0.197	0.92 (0.84–1.00.84.00)	0.040
Air pressure ≤ 979 hPa (ref.)	1		1		1		1		1		1
Relative humidity < 50%	1.09 (1.01–1.17)	0.018	1.11 (1.02–1.20)	0.011	1.00 (0.85–1.19)	0.972	1.12 (0.99–1.27)	0.083	1.08 (0.95–1.22)	0.232	1.07 (0.95–1.20)	0.286
Relative humidity ≥ 50% (ref.)	1		1		1		1		1		1
Day duration ≥ 14.01 h.	1.44 (1.25–1.67)	< 0.001	1.53 (1.30–1.80)	< 0.001	1.14 (0.82–1.58)	0.450	1.59 (1.23–2.05)	< 0.001	1.62 (1.25–2.11)	< 0.001	1.15 (0.90–1.47)	0.261
Day duration 13.426–14 h.	1.35 (1.18–1.55)	< 0.001	1.44 (1.24–1.68)	< 0.001	1.03 (0.76–1.41)	0.847	1.55 (1.22–1.97)	< 0.001	1.45 (1.14–1.86)	0.003	1.08 (0.86–1.37)	0.500
Day duration 9.506–13.425 h.	1.27 (1.15–1.40)	< 0.001	1.35 (1.20–1.51)	< 0.001	0.98 (0.78–1.23)	0.857	1.47 (1.24–1.75)	< 0.001	1.24 (1.03–1.48)	0.023	1.10 (0.92–1.30)	0.302
Day duration ≤ 9.505 h. (ref.)	1		1		1		1		1		1

* in the model the daily number of suicides third and fourth day before were included

Discussion

This population-based time series study provided evidence that different weekdays, seasonality, and meteorological factors could have a protective effect or could be associated with an increased risk of suicide. During the summer months, the risk of suicide was higher: on the average, from 35% (RR 1.35 95% CI 1.26–1.44) in June to 23% (RR 1.23 95% CI 1.15–1.31) in August. The effect estimates in the three age groups (≤ 40, 41–55, and ≥ 56 years) and both sexes were not the same. Positive associations were found between the risk of suicide and air temperature (except for women) and a higher number of sunny days (except for those aged 41–55 years). A higher risk of suicide was observed during days with lower air pressure, lower than 50% RH, and greater day length, except for those aged ≥ 56 years and women. All findings agreed with our hypotheses, except for hypothesis “c”, where our findings only partially agreed with it, as they showed that that men were not more significantly than women affected by the season but instead were more affected by meteorological variables (see Tables 2 and 3).

Previous evidence shows that seasonal and meteorological variables can increase the risk of suicides [27, 28], with studies conducted in different climate regions such as various places in Europe, Asia, and South and North Americas [14, 27–43]. Studies using different methodologies and conducted in countries such as Brazil, China, Mexico, USA, Turkey, Canada, and Germany showed that higher temperatures had a greater effect on suicide rates in the total population [27, 28, 30–32, 35]. Meanwhile, a study conducted in Hungary, a country with a high suicide rate, surveying a period from 1971 to 2013, found that the daily hours of sunshine had an immediate and significant impact on suicides (leading to an increase in suicides) [36]. Similar results were found in Germany where the risk of suicide was by 6.5% higher on days with low/medium cloud cover compared to days with high cloud cover [35]. The opposite results were found in Spain, where an increase of one eighth of the cloud cover in the sky correlated with a 7% increase in the total number of deaths by suicide [38]. The study in Turkey found a weak negative statistically significant correlation between the number of cases and the average humidity and actual atmospheric pressure levels (r = 0.092 and − 0.104, respectively) [31]. A study in Finland showed that men seem to be more prone to suicide attempts under low atmospheric pressure, while women – under high atmospheric pressure [41].

Meanwhile, the same study in Finland taking into account the seasonal normal value during the period of 1971–2000, daily temperature, global solar radiation, and precipitation did not associate them with the number of suicide attempts on a statistically significant level [41].

Our findings agree with this evidence and provide additional insights into a similar association in Lithuania, where only scarce previous evidence exists on associations between seasons and suicides [44]. Interestingly, in a study conducted 20 years ago [44], suicide rates in Lithuania also showed a clear seasonal trend, with the highest numbers in summer and the lowest in December. the same trends were found among men and women: in December, suicide frequencies declined by over 23% among men and by 30% among women, whereas the June peak among men approached 23%, and the July peak among women exceeded 29%, compared to the annual average levels (p < 0.05) [44]. A large-scale study encompassing 12 countries (Brazil, Canada, Japan, Mexico, Romania, South Africa, South Korea, Spain, Switzerland, Taiwan, the United Kingdom, and the United States) also identified seasonal trends in suicide rates, with most countries experiencing a peak in spring and a low point in winter. However, the nature of these seasonal patterns differed across countries, ranging from bimodal to unimodal distributions.

The magnitude of the seasonal effect – measured as the peak-to-trough relative risk – also varied, from as high as 1.47 (95% CI: 1.33–1.62) to as low as 1.05 (95% CI: 1.01–1.10).

Variation in seasonal patterns was also evident across subgroups. In several countries, the amplitude was greater among women and individuals aged 65 and older, compared to men and younger populations, respectively. Additionally, temporal trends in seasonality differed: while some countries showed increasing seasonal variation over time, others exhibited a decline or minimal change. Notably, the seasonal amplitude was larger in communities with colder climates, a higher proportion of elderly residents, and lower unemployment rates (p < 0.05) [29]. However, no studies have comprehensively analyzed the impact of meteorological variables on suicides either in Lithuania or in the Baltic States, and few studies have been conducted in the Boreal region [45–47].

There is not much previous evidence about how age modifies the effect of weather and season on suicide. Studies conducted in Canada, China, and Germany have shown that the association between ambient temperature and suicides is stronger in the elderly than it is in the younger population [28, 32, 35]. However, another study in China found that the association between ambient temperature and suicides is stronger in younger people than it is in the elderly [37].

The seasonal pattern also differed by age group (≥ 65 and < 65, respectively); in several countries (e.g., in Canada, Japan, South Korea, Spain, Switzerland, and the USA), the amplitude was significantly larger for older than for younger people. However, in such countries as Brazil, Mexico, Romania, South Africa, and Taiwan where seasonality is not so pronounced, the effect of seasonal patterns did not differ significantly between the aforementioned age groups [29].

There are only a handful of studies on effect modification by sex on the topic of weather variables, seasons, and suicide.

Several previous studies conducted in Brazil, Germany, and China showed that the association between weather variables and suicide was stronger among men than among women when air temperature increased [27, 35]. Relative humidity also was associated with suicide, but only at lag 3 days and for men. Meanwhile, suicide counts for women increased only with rainfall at lag 0 [27]. However, no effects of cloud cover were found in either men or women [35]. A study in Hungary showed that there were no relevant associations between sunshine duration and suicide risk among women. However, among men, sunshine had a prompt, but very weak increasing effect on the risk of suicide [40].

Men and women often have different stress responses. Men might be more prone to externalizing behaviors (e.g., aggression, defiance, and impulsivity) [48, 49], which extreme weather conditions might exacerbate. Some studies in the literature which investigated the relationship between impulsivity and temperature found a positive relationship [50, 51].

Although seasonality of suicide is seen both in men and women, the seasonal patterns differ between the two sexes. For example, only a single spring peak has been found in men, while two peaks in spring and fall have been reported in women [14]. In Japan, the number of suicides for men peaked 3 weeks earlier than it did for women. In the UK and the USA, women displayed a seasonal pattern with a larger amplitude than men did [29]. However, several previous studies in Scotland, Oxford [14], and Germany [52] showed no such seasonality for men.

It is difficult to speculate on the potential reasons from the biological side for the associations between seasons, weather, and suicide because the mechanisms remain unclear. However, it is believed that the increase in the number of suicides in summer is due to changes in the serotonin system which is responsible for depression, impulsivity, and aggression. It has also been suggested that high ambient temperatures may lead to reduced levels of l-tryptophan, a serotonin precursor, which could be associated with an increase in violent suicides [31].

Some evidence also suggests that seasonal variations in social activity may play a role in seasonal patterns of suicide [29].

In many parts of the world, spring is associated with renewal, which can bring about greater social demands and stress, potentially contributing to a higher risk of suicide. People may also experience disappointment with excessive expectations for a new start, which is a plausible psychiatric mechanism to explain the spring peak in suicides [29]. A very similar phenomenon is called the “Broken Promise Effect”. This effect is described in Howard Gabennesch’s article published in 1988 and refers to the phenomenon where suicide rates temporarily increase after periods of heightened expectations that are later unfulfilled. The author of this Effect used this theory to explain temporal fluctuations in suicide rates, showing that certain events (e.g., holidays, political transitions, or personal milestones) can create false hope, and when reality fails to meet expectations, it may lead to an increase in suicidal behavior [53].

Our research revealed an increased risk of suicide on Mondays and January 1 st, which may be attributed to alcohol consumption over the weekend or during New Year celebrations. This is partially confirmed by one study from 1997 to 2002 conducted in the Chuvash Republic. The peak suicide for men and women regardless of previous alcohol consumption was on Wednesdays and Mondays, respectively. The overall suicide rate was the highest on Mondays and the lowest on Thursdays. Both sexes were less likely to suicide during holidays than on weekends or workdays while intoxicated with alcohol. Most men had consumed alcohol prior to suicide, irrespective of the day of the week, whereas this pattern was observed in women only on Fridays and Sundays [54]. Other researchers in their studies did not evaluate the effect of the day of the week on the risk of suicide.

The study has several limitations. The study could be criticized for not analyzing the role of factors such as air pollution. We made a conscious decision not to control for the effects of air pollution on suicide. Air pollutants can be treated as intermediate variables in the pathway from meteorological factors to suicide, and adjusting for the intermediate variables would lead to an underestimation of the true overall effect. Air pollution levels in Lithuania are also relatively low, thus their role in mediating the effects of weather and seasonally was likely small in this study [55].

One more limitation of the study is that we did not have any information on other medical or harmful factors that could have influenced the decision to suicide. These factors include alcohol, tobacco, and other drug use, which are also associated with depression and an increased risk of suicide, yet were not assessed in this study. According to one meta-analysis, alcohol use was associated with a 94% increase in the risk of death by suicide. Meta-regression indicated larger effect sizes for studies with a higher percentage of women, younger age, unadjusted estimates, longer follow-up periods, military samples, and higher frequencies and quantities of alcohol use (relative to the drinker/non-drinker status) [56]. In this aspect, the post-Soviet region still remains sensitive. A study on the relationship between alcohol consumption and suicide mortality conducted in the biggest country in Central Asia, Kazakhstan, revealed that alcohol consumption was a common factor contributing to suicide, especially among young and middle-aged adult men [57].

Moreover, among men, any psychiatric disorder (depression, anxiety, and bipolar disorders), poor medical condition, stressors/bereavement, and living alone appeared to be more significant for predicting completed suicides in late life [58].

Another limitation of our study was that even though the suicide register is prospective in nature, it is possible that some cases which occurred in 2001–2021 were identified later than the real suicide time. However, given the extensive data collection protocol, this number is likely to be so small that it did not influence our results.

One more limitation is that we have not taken into account the COVID-19 pandemic in this study, although hypothetically it could have also affected the distribution of suicides by the day of the week. To test this assumption, we made separate calculations for the period 2020–2021 and did not observe any change in the temporal trend of suicides. Apart from this, the association between weather variables and the daily suicide rate was the same during the period of 2001–2019. We did not find any changes in the statistical significance, and the regression coefficients of weather variables were almost the same.

The final limitation of our study was that the study did not include certain codes, such as Y10-Y34, which were related to insufficient clarification of the circumstances of suicide, or code, such as Y87, which was related to the consequences of suicide that occurred at another time and which could potentially distort the results obtained.

The study has several strengths. First of all, the study covers a relatively long period of 21 years, which allowed us not to consider the effects of local meteorological conditions. One more strength is that we used a regression model, which allowed for eliminating the impact of the time trend. The prospective study protocol ensured that the case selection procedures and criteria remained constant over time.

A formal assessment of heterogeneity between the effect estimates of different subgroups is another strength. This approach should ideally become standard when assessing and interpreting effect modification in epidemiological research. Time-series designs are well-suited for examining short-term risk factors across different periods and locations, as they inherently control for many confounding variables that are constant or change over time [59].

Conclusion

This population-based case-crossover study in Lithuania provided evidence that there is an association between seasons, meteorological factors, and suicides. The rate ratio (RR) of suicide was observed to be the highest over two summer months, June and July, compared to January, with the same tendency in both sexes and different ages.

Higher temperatures and a higher number of sunny hours per day were also significantly associated with a higher RR of suicide. However, higher atmospheric pressure was found to be a protective factor for suicide and reduced the RR in the whole sample, in men, and in the youngest and the middle-age groups. More evidence is needed on the associations between weather and suicide in the season-specific context to guide decision-making, policy, and clinical practice. The results of this study will help social workers who work with high-risk individuals to identify the periods of the highest suicide risk based on weather forecasts.

Abbreviations

CI

Confidence interval

hPa

Hectopascal

ICD

International Classification of Diseases

LST EN ISO

Lithuanian Standards adopted European (EN) and International (ISO) standards

r

Correlation coefficient

RH

Relative humidity

RR

Rate ratio

SD

Standard deviation

Authors’ contributions

V.V. initiated, planned, and designed the study, and also drafted the manuscript. J.V. contributed to the concept of the work and the analysis and interpretation of data, and also drafted the manuscript. R.R. Critically reviewed the work for important intellectual content, participated in drafting the manuscript, and reviewed and edited the text. O.M. conducted suicide data acquisition, analysis, and interpretation for the work and provided important comments on the data analysis. K.G. substantially contributed to the analysis, drafting of the work, material preparation, and data interpretation. V.D. collected the weather data from the Lithuanian Hydrometeorological Service, conducted quality control measures and statistical preparation, and critically reviewed the Methods section. A.A. and S.L. participated in drafting the discussion part, reviewed and edited other parts of the text, and provided recommendations. All authors read and approved the final manuscript.

Funding

Funded by the European Union (AURORA under the Grant Agreement n° 101157643). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Climate, Infrastructure and Environment Executive Agency (CINEA). Neither the European Union nor the granting authority can be held responsible for them.

Data availability

The data that support the findings of this study are available from the Institute of Hygiene under the Ministry of Health and Lithuanian Hydrometeorological Service, but restrictions apply to the availability of these data, as they were used under license for the current study and thus are not publicly available. The data are, however, available from the authors upon reasonable request and with permission of the Institute of Hygiene under the Ministry of Health and Lithuanian Hydrometeorological Service.

Declarations

Ethics approval and consent to participate

The Lithuanian Bioethics Committee has confirmed that mortality data is not subject to the requirements of the Law on the Ethics of Biomedical Research in Lithuania (certificate No. 6B-20-241; date 2020-10-02); also, the same Committee waived the informed consent requirement for this study according to the Law on the Ethics of Biomedical Research in Lithuania (No. VIII-1679; date 2000-05-11 d.). The study complies with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.