The main findings of our study are that (i) there is an influence of sex on the frequency, mode, and severity of suicidal self-poisonings and (ii) age associates with the patterns of substance choice in suicidal self-poisonings. These findings may be useful for targeted programs of suicide prevention.

The average age of 40.5 years with a 1:2 male: female ratio in this work confirms the results of comparable studies [17, 20, 28, 29]. Similarly as in the UK study [30], the sex ratio converged in the older age groups. 34% of the cohort had a citizenship other than German. Compared to around 26% of the German population with a migration background or with 12% foreigners, the proportion in this study is significantly higher [31]. This may be due to a bias towards a major city in southern Germany with around 1.4 million inhabitants (the predominant enrollment area), with a higher proportion of non-German citizens.

Proportion of jobless patients was significantly higher than the average jobless rate in Germany in the study period (14% vs 6.7%) [32] thus supporting the notion that individuals searching for job are at a higher risk for (para-)suicidal behavior [33, 34].

As in comparable studies [35, 36], most patients presented to hospital within 3 h of substance use.

The substance pattern, with 55% of mono-intoxications and a preferred choice of antidepressants (17%), followed by benzodiazepines (15%) and non-opioid analgesics (15%), is similar to the results of recent studies [20, 37], in which antidepressants were used more frequently than benzodiazepines than in older studies [35, 38]. The tendency towards increased antidepressant vs benzodiazepines use in the context of (para-)suicidal poisoning has been observed in other studies, and it was associated with the increased frequency of antidepressant prescribing [39, 40]. This is supported by the fact that almost 60% of our patients ingested their own long-term medication. Analysis of substance choice revealed a sex-specific pattern with preference for antidepressants and benzodiazepines among females and use of carbon monoxide, chemicals, insecticides, drugs/intoxicants, anticoagulants, cardiovascular drugs, and other sedatives among males. The preferred choice of a pesticide/insecticide [27, 41] or gases, and chemicals [42] by males has already been described and was reconfirmed in this study, in which males were about three times more likely to use non-medical substances for SRB than females (17% vs 5%). Similar to the results reported here, Muheim et al. demonstrated that females more often chose non-opioid-analgesics and antidepressants, while males preferred sedatives. In contrast, a significantly increased use of opioids in males could not be confirmed in our work [43].

In terms of age-specific differences, younger patients tended to use antibiotics, antihistamines, drugs/narcotics, and non-opioid analgesics more frequently than older patients. Preferential use of non-opioid analgesics in the youngest patients was also found in a study by Froberg et al. [44] and could be due to a weaker intention to die, unawareness of the generally wider therapeutic range of this class of drugs (with the exception of acetaminophen), or ease of availability. The elderly were more likely to choose benzodiazepines, Z-drugs, anticoagulants, and cardiovascular drugs, i.e. long-term medications in this age group. Accordingly, we observed the highest frequency of own medication as a source in patients aged > 64 years. Of note, we did not observe differences in opioids use between the age groups, however, prior analysis found that individuals aged 45 to 54 years had a highest rates of suicide deaths with opioid poisoning in US [45].

9% of cases underwent therapeutic interventions, such as the administration of activated charcoal. In other studies, this frequency ranged from 1.5% to about two-thirds of patients [20, 46, 47]. This considerable variability may be a consequence of a limited preclinical availability of activated charcoal, variation in patient compliance, and a limited appropriate time window. Hemodialysis in 2% of cases and antidote therapy in 16% was comparable to a previous report (2% and 18%, respectively) [17].

In a study by Cook et al., the majority of patients were discharged within 24 h [20]. In our study, the length of inpatient stay was longer, as most patients stayed over 24 h. This may reflect a relatively high proportion of psychiatric follow-up therapy in this study, as a seamless transfer may result in a longer stay due to time bridging. Alternative explanation could be that our patients were more severely poisoned with 33% treated in the ICU compared to 1.5% in UK [20], 16% in Israel [28], and 11% in Greece [17]. The rate of ICU treatment in another German study was 61%, almost twice as high as in our study [19]. These different rates of ICU stays may be influenced by fluctuating ICU capacities or internal hospital monitoring strategies. For example, in most hospitals, many patients must be monitored in an ICU after a suicide attempt, even in mild poisoning cases, because adequate monitoring and self-protection cannot be guaranteed on a normal ward. Our department with an ICU and protected IMC, offers the possibility of a seamless “step-up/step-down concept”, which allows monitoring of severely poisoned patients in the ICU and less severe cases in the IMC, thus providing a better individualized monitoring of patients. Given the current shortage of intensive care beds, studies showed that from an internal medicine perspective, most self-poisonings can be adequately and sufficiently treated on an IMC or with supportive therapy by means of monitoring alone [20, 46, 48].

Co-ingestion of alcohol in about one in three patients is in line with the results of comparable studies [20, 38, 46]. Co-ingestion of alcohol predominated among males and in the two middle age groups, which corroborates the results of a similar study [42]. Also, a significantly more frequent co-ingestion of drugs was found in males. Regarding the age group-specific differences, co-ingestion of alcohol among adolescents (< 18 years) was significantly lower in this work (12%) than in a Finnish study, which detected alcohol in about half of the adolescents who committed suicide [49]. Thus, adolescents with a completed suicide appear to differ from the group of young people who attempted suicide in terms of co-ingestion of alcohol.

Alcohol and nicotine abuse predominated among males. Comparably, Mauri et al. found a significantly higher proportion of alcohol dependence (“alcoholism”) in males with (para-)suicidal intoxication [35]. According to a 2021 study, 17.6% of the German population had “harmful” alcohol consumption [50]. In contrast, only 9.3% of patients self-reported the alcohol abuse. Nicotine abuse in 30% of our cohort corresponded to the daily or occasional smoking behavior of 18- to 64-year-old German adults [50]. Illicit substance abuse (6.4%) was higher compared to the general German adult population (2.5%, [50]). However, above-average substance abuse rate is not surprising, as addictive disorders are associated with an increased risk of SRB [51]. The tendency of young males to abuse illicit substances more frequently [50] was hinted in this study (8% males versus 6% females), but without statistical significance.

Considering possible reasons for SRB, partner and family conflicts were common trigger factors, comparable to the findings of Hatzitolios et al. [17]. Males more often reported problems with job, finances, or law or justice as trigger factors, while females more often suffered from the loss of a caregiver or pet. This corresponds to the observations of a study from Hong Kong, which found a predominantly interpersonally triggered parasuicidal behavior in females compared to a personally triggered behavior in males [52]. Interestingly, there were no statistically significant differences in problems in relationship with partner between the sex groups in your study. In contrast, a Turkish study recorded relationship problems more frequently in female than in male suicides [53]. Furthermore, a study investigating psychosocial factors associated with suicidal thoughts in comparison to suicide attempts revealed that relational trauma was a significant risk factor distinguishing males from females [54]. Nevertheless, that study included not only subjects who attempted suicide but also those who reported suicidal ideations, and furthermore, did not distinguish between the means of suicide. In contrast, our study focused specifically on (para-)suicidal poisoning, which could potentially account for a lack relationship problems among triggers differentially reported by females and males. Furthermore, family problems and problems with the social environment were mentioned most frequently by the youngest patients and partner conflicts by those aged 18 to 44, while health problems and “loss of caregiver/pet” were predominantly documented for those aged ≥ 64 years. Similarly, loss of caregivers and associated isolation have been discussed as risk factors for suicide attempts in the elderly [55]. Finally, males had a significantly higher percentage of intensive care treatment. This could be explained by fact that, compared to females, males were more likely to develop a higher severity level of poisoning. In our prior analysis of the present cohort, we demonstrated a lower frequency of parasuicides in men, which is presumably associated with a stronger intention to die [25]. Furthermore, compared to females, males more often used non-medical substances which we found previously to predict severe or fatal course of suicide [24]. Higher lethality of suicidal behavior in males has been demonstrated in past studies not only by choice of method but also within the same method [56, 57].

Our study has several limitations including the retrospective nature, the monocentric setting, and a suspected sampling bias. Patients aged < 18 years were the smallest age group because our center mainly treats adults and only in exceptional cases children. This could be perceived as a bias since distribution of patients into age groups was done according to clinical considerations, but not to ensure a uniform group size. Furthermore, most of patients came from the area of a major German city, which could not be representative for a broader population. Therefore, it can be supposed that frequency of singles living alone as well as use of drugs could be higher than in rural regions.

Since this was a single-center survey of a pre-selected patient population, it can be assumed that many intentional overdoses, also with (para-)suicidal intent, did not lead to a medical presentation due to a barely or slightly symptomatic course. For example, some patients reported having taken drugs in the past with suicidal intent but resuming their usual daily routine a few hours later after a symptom-free interval and without this incident being noticed by anyone. On the other side of the clinical spectrum, severely intoxicated persons who are found dead or do not reach a clinic alive are not included in this work. Overall, a certain selection bias of patients must be assumed.

Information on the ingested substances (in cases without toxicological analysis), trigger factors, and suicide circumstances was often based on incomplete information provided by the patients or obtained during external anamneses. In some cases, the information on the dose of the substances could only be estimated by the paramedics based on empty blisters or medicine boxes found in the rubbish bins. However, there remains considerable uncertainty as even in cases when date of purchase or dispensing was known, this was not necessarily the time of first substance use and thus could lead to dose underestimation. In contrast, in cases when empty blisters were found, it was assumed that all tablets were ingested which could lead to dose overestimation. In certain instances, an indication of quantity was only possible by means of the number of tablets taken without knowledge of the dosage of a drug. In these cases, the lowest commercially available dosage strength of a tablet was considered to estimate the minimum amount taken to ensure that an overdose actually occurred and not “just” several tablets of the lowest strength were taken within the MDD.

MDD was determined on the basis of the prescribing information, whereby a high MDD was presumed for the clear detection of actual intoxication. However, this procedure allows a certain degree of discretion on the part of those conducting the study. In addition, substances such as benzodiazepines or opioids are subject to a strong habituation effect, so that no standardized MDD can be determined.

Furthermore, as our survey was limited to the inpatient-somatic stay cases, a long-term follow-up of patients is lacking.

Finally, since the data collection was not hypothesis-driven, but it was rather aimed at capturing the interrelationships of a variety of parameters, random findings may have arisen due to multiple testing. Therefore, statistically significant results should be interpreted as trends until they will be validated in separate confirmatory studies.