Full Text

Introduction Annually, more people die from suicide than from HIV, malaria, breast cancer, or from war and murder. In 2019, more than 700,000 people died this way, that is, one out of every 100 deaths is suicide [1]. McDowell AK et al. indicate the prevalence of suicidal behavior in the population based on data from the Centers for Disease Control and Prevention, USA. Thus, serious suicidal thoughts were noted in 3.7% of the population, 1.0% had a plan of suicide at the time of the study, 0.5% had a history of attempts, and 0.01% ended their lives by suicide [2]. According to Zubrick SR et al., in the Australian population in the age interval of 12-17 years, the prevalence of suicidal ideation was 7.5% over the past 12 months, suicidal attempts - 2.4%, while only 0.6% of adolescents sought medical help due to attempts [3]. It is assumed that against the background of the spread of coronavirus infection, suicide rates will increase from 1% to 145% [4, 5]. The need for a more sparing attitude to financial and labor resources with the continuing difficult epidemiological situation highlights the relevance of research aimed at finding risk factors and protective factors [6, 7]. Adequate target audience research has the potential to help segment vulnerable populations for appropriate targeted interventions. This resource-saving approach optimizes the activities of crisis services and increases the prevention effectiveness. The aim of this work is to form a general idea of the demographic characteristics of persons hospitalized in a psychiatric hospital with suicidal behavior, as well as to identify the age intervals of the risk of specific types of suicidal behavior. Materials and methods As part of the study, a retrospective analysis of the archival medical records of the Kaluga Regional Psychiatric Hospital (hereinafter: KRPH) for the period of 2017-2018 was conducted. In total, 435 Medical Records of an Inpatient were processed. The sample size was calculated using the method of K. A. Otdel'nova at t=2.0; p=0.95 [8]. All admissions were divided into two groups: admissions with suicidal ideation (hereinafter referred to as SI) and admissions with suicidal attempts, including self-harm with specified suicidal intentions (hereinafter referred to as SA). Results Of the total number of hospitalizations for 2017-2018, manifestations of suicidal behavior and self-harm were found in 435 cases. The ratio of patients with SA and SI is practically 1:1, 218 and 217 cases, respectively. ps202204.4htm00192.jpg When analyzing the gender composition of patients who were treated at the KRPH for the specified period, it was found that 46% of those who were admitted with attempts were males and 54.5% were females. On the opposite, in the group of patients with SI, 54% were males and 45.5% were females (Fig. 1). However, this ratio was not stationary during the period under study. Thus, a natural assumption arises: to put forward a hypothesis of hospitalizations with one form or another of suicidal behavior being independent of the gender of the suicidal person, that is, there is no quantitative advantage in one group or another (Hypothesis - H0) against the alternative H1 (Table 1). We shall determine the Yule's coefficient of association. ps202204.4htm00194.jpg ps202204.4htm00196.jpg On the Chaddock scale, the connection is characterized as very weak. We shall determine the Pearson's contingency coefficient. ps202204.4htm00198.jpg On the Chaddock scale, the connection is characterized as absent. ps202204.4htm00200.jpg Based on the analysis of this table, it can be concluded that there is no influence of gender on the form of suicidal behavior. However, in order to be completely satisfied, we shall calculate the value of the Pearson's chi-squared test χ2 (Table 2). Calculating the observed value ps202204.4htm00202.jpg or ps202204.4htm00204.jpg The critical value of the Pearson's chi-squared test is 3.841 at a significance level of p=0.05 and the number of degrees of freedom f = (r - 1) × (c - 1) = (2 - 1)(2 - 1) = 1. We compare the value of the χ2 test with the critical value for the number of degrees of freedom f. The obtained χ2 test value is lower than critical; thus, the gender does not affect the type of suicidal behavior with a probability of 0.95 and there are no quantitative advantages in one group or another. When using the built-in EXCEL function, CHITEST, to calculate the probability of a competing hypothesis, we obtained the following: CHITEST=0.0758>0.05. Preliminary study of the age composition of suicides ps202204.4htm00206.jpg For a preliminary analysis, we shall calculate the statistical characteristics of the samples of SA and SI (Table 3). Finding the characteristics of the samples: Mean (mean age of suicide for samples SA (X) and SI (Y)) ps202204.4htm00208.jpg Unbiased sample variances ps202204.4htm00210.jpg The marginal errors for the confidence intervals of the general mean will be the same. Both samples are not normally distributed (the hypothesis is verified by the Pearson's test χ2); however, they do not contradict the norm by the conditions of the entry of a sample part into the corresponding intervals. Analysis and testing of hypotheses about the means in the SA and SI groups Assuming that there is a mean age in the SA group, we will test the hypothesis of equality of the means. Н0: M (Х) = M (Y), against the alternative hypothesis: Н1: М (Х) > М (Y) The statistical criterion for testing this hypothesis is the normalized random variable ps202204.4htm00212.jpg Calculating the observed value of the criterion ps202204.4htm00214.jpg. The critical region is right-sided, respectively, zcr is defined as the argument of the Laplace's function, at which ps202204.4htm00216.jpg and the critical region is determined by the inequality Z > zcr. ps202204.4htm00218.jpg. According to the Laplace's function tables ps202204.4htm00220.jpg, we find zcr=1.645. Since 3.24 = zobs > zcr = 1.65, we reject the hypothesis H0 and accept H1: the general means differ significantly, i.e. there is a significant difference in admissions in the SA and SI groups depending on the age cohort. Thus, there are significant differences in the admissions of patients with suicidal attempts and ideations in different age cohorts. For both features (SI and SA), we shall perform an interval estimation of the mean values with a reliability of 0.95. We shall estimate the general mean M(X) = a according to sample statistics x and S using the theory of confidence intervals for the normal distribution. Since σ is unknown, we use the Student's distribution. The confidence interval for the true value M (X)=a with a reliability of γ=0.95 is found by the formula: ps202204.4htm00222.jpg. Where tγ = t(1 - γ, n - 1), quantile of the Student's distribution with n=218 is found by the table or by the built-in EXCEL function TINV (0.05; 217)=1.971. Confidence interval: ps202204.4htm00224.jpg Thus, the mean age in the SA group will be within the range (36.048; 41.393). The confidence interval is more reliable according to the t-test (the marginal error is 2.672 versus 2.658 according to the CONFIDENCE function (0.05; 20.02; 218)=2.658) than according to the Laplace's normal distribution criterion, although it is less accurate. The same works for Y. The confidence interval for the true value M(Y)=b with a reliability of γ=0.95 is found by the formula: ps202204.4htm00226.jpg. Confidence interval: ps202204.4htm00228.jpg. Thus, the mean age in the SI group will be within the range (29.828; 35.186). The confidence intervals for the SI and SA groups do not overlap; this implies that the age samples do not belong to the same general population, therefore, they cannot be considered as rigidly related. Analysis and testing of hypotheses about the probability of admission with SI and SA for certain age cohorts For both features (SI and SA), we shall perform an interval estimation of the proportion of observation units with the values of the feature in different cohorts, assuming the estimation reliability is 0.95. Now we consider the first three cohorts of 5-9, 10-14, and 15-19 years in the SA sample. Since the proportion of observation units with the values of the feature in these cohorts for the sample w=(2+27+22)/218=0.2339 is the relative frequency (Table 3), the error of mean of the proportion for the sample is found by the formula: ps202204.4htm00230.jpg Marginal error of the proportion: ps202204.4htm00232.jpg Where tγ = t(1 - γ, n - 1), quantile of the Student's distribution with n = 218 is found by the table or by the built-in EXCEL function TINV (0.05;217)=1.971. Confidence interval: ps202204.4htm00234.jpg or 0.1774 < px< 0.2905 Accordingly, the proportion of attempts in the age range of 5-19 years, with a reliability of 0.95, should be in the interval (0.1774 < px< 0.2905) for the entire population of 5-19 years. Now we consider the first three cohorts of 5-9, 10-14, and 15-19 years in the SI sample. Since the proportion of observation units with the values of the feature is higher than the obtained mean for the sample wу=(16+49+20)/217=0.3917 - the relative frequency (Table 3), the error of mean of the proportion for the sample is found by the formula: ps202204.4htm00236.jpg. Marginal error of the proportion: ps202204.4htm00238.jpg Where tγ = t(1 - γ, n - 1), quantile of the Student's distribution with n = 10 is found by the table or by the built-in EXCEL function TINV (0.046; 9)=2.313. Confidence interval: ps202204.4htm00240.jpg or 0.3264 < py < 0.4570 The proportion of admissions with SI in the age range of 5-19 years, with a reliability of 0.95, should be in the interval (0.3264; 0.4570) for the entire population of 5-19 years. Confidence intervals do not overlap, which means that the probability of admission with suicidal ideation is much higher for these cohorts than the probability of hospitalization with suicidal attempts and self-harm. To substantiate this assumption, we will test the hypothesis Н0: рy =рх with a competing hypothesis Н1: рy > рх The observed value of the criterion is calculated by the formula: ps202204.4htm00242.jpg Where ps202204.4htm00244.jpg is the best point estimation of the probability when the null hypothesis is valid. Plotting the critical region: with a competing hypothesis Н1: р1 > р2, ucr for the right-sided critical region is found from the condition ps202204.4htm00246.jpg, and the critical region is as follows: U > ucr, ps202204.4htm00248.jpg consequently, ucr = 1.645. U = 3.55 > ucr = 1.645. Therefore, the hypothesis is not accepted, and it can be assumed that in the entire general population for these cohorts, the probability of hospitalization with suicidal ideation is greater than the probability of hospitalization with suicidal attempts and self-harm. The rest of cohorts in various combinations are considered in the same way. For the age ranges of 25-29, 30-34 and 35-39 years, the probabilities, in general, are p=0.275 in the SA group and p=0.180 in the SI group. For 30-34, 35-39 and 40-44 years, the probability was p=0.271 in the SA group and p=0.188 in the SI group. Thus, admissions with SA in the named cohorts prevail over admissions with SI; hypotheses are tested in the same way (Fig. 2, Table 4). ps202204.4htm00250.jpg For cohorts 65-69, 70-74, 75-79, 80+, the probability of hospitalization is lower both in the SI group and in the SA group in general, while maintaining the ratio within the cohorts: SA prevail over SI and hypotheses are tested in the same way. Discussion The lack of clear prevalence of the male or female gender in the sample presented requires further study with the expansion of the cohort of respondents and the use of materials from emergency care facilities. Significant differences in the structure of psychiatric diagnoses in hospitalized patients depending on the profile of the hospital have been previously described [9, 10]. The incompleteness of the “gender paradox” highlights the relevance of nonselective prevention, which includes community outreach and more open dialogue on suicide and stigmatization reduction. The most important component of suicide prevention is the availability and accessibility of crisis assistance, which implies increased funding for services not only in the field of prevention and health promotion, but also treatment of mental disorders [11]. Resources should be aimed both at training practitioners and increasing their competence in assessing suicidal risk and at increasing public awareness of the warning signs of suicide and possible ways of providing help [11, 12]. Support groups of depression and suicide survivors can be helpful [13]. The age intervals of risk determined in this study indicate the sequence of changes in the stages of the suicidal continuum: from hospitalizations with SI in the range of 29.8-35.2 years to admissions with SA in a cohort of 36.0-41.4 years. The maximum number of cases with SI in a cohort of adolescents allows us to draw indirect conclusions about the prevalence of this phenomenon in the population. It is appropriate to indicate the data of M. I. Cherepanova obtained in the course of studying suicidal behavior of young people in Altai Krai: 10% of schoolchildren and 14% of students expressed the opinion that by depriving oneself of life it is possible to “solve important problems for oneself”. Besides, according to the author, 7% of schoolchildren and 13% of students who answered vaguely can also be attributed to the suicidal risk group [14]. Suicide prevention among youth may include student conferences/summits on suicide prevention, training for law enforcement officials, teachers, mental health caregivers, countering child abuse [13] and bullying, funding for fight against substance abuse, and strengthening of mental health [15]. Previously, a number of initiatives were highlighted that have proven their effectiveness along with existing activities: the introduction of 24-hour crisis teams, observation/supervision of patients for the next seven days after discharge, support for patients with severe mental disorders after discharge with special attention to non-compliant [16], informational support for “survivors” [17]. Conclusions With economic pressure on the general population, the pandemic is making the development of crisis services more urgent than ever [18]. The resolution of the issues on the implementation of activities that have demonstrated obvious progress within the framework of well-known national strategies can be called timely. Broad communication, public awareness and educational strategies are needed to improve understanding of suicidal behavior and remove barriers to requesting help. The cornerstone in initiating interagency cooperation is raising the awareness of officials of the need for strategies to prevent suicide, their level of knowledge, and clear indicators of the evidence-based effectiveness of preventive measures [19]. Авторы заявляют об отсутствии спонсорской поддержки и конфликта интересов. The authors declare that there is no conflict of interests and sponsorship support

About the authors

E. A. Nosova

N. A. Semashko National Research Institute of Public Health

Email: nosova.evgenya@lenta.ru

B. A. Spasennikov

N. A. Semashko National Research Institute of Public Health

O. Yu. Aleksandrova

N. A. Semashko National Research Institute of Public Health

References

Statistics