Investigating the Clinical Profile of Suicide Attempters Who Used a Violent Suicidal Means

Round 1

Reviewer 1 Report

1. The author mentioned, "There are no studies up to now investigating personality traits in VSA compared to NVSA. There is also no consensus for putative associations between specific personality traits and the choice of a suicide method based on concepts of lethality and seriousness”. And “In the present study, we explored putative differences in demographic, clinical and personality features between VSA and NVSA”. However, except for Impulsive, Aggression, and Anger, no other personality measures were used, nor any specific tool that may comprehensively assess the personality traits were used. This needs to be rationalized in the manuscript.

2. Please mention something about sample size calculation or approximate size needed or fixed on some basis. Also, mention the sampling method employed.

 

3. Regression analysis could have been better. Please mention why the given method of statistical analysis was used.

Author Response

Reviewer #1:
1. The author mentioned, "There are no studies up to now investigating personality traits in VSA compared to NVSA. There is also no consensus for putative associations between specific personality traits and the choice of a suicide method based on concepts of lethality and seriousness”. And “In the present study, we explored putative differences in demographic, clinical and personality features between VSA and NVSA”. However, except for Impulsive, Aggression, and Anger, no other personality measures were used, nor any specific tool that may comprehensively assess the personality traits were used. This needs to be rationalized in the manuscript.
Reply: We thank reviewer #1 for this comment. We focused on these specific personality characteristics because abnormal impulsive-aggressive traits and anger expression has been consistently associated with suicidal behavior in previous studies [6, 7].
We stated this the Introduction section of the revised manuscript.

2. Please mention something about sample size calculation or approximate size needed or fixed on some basis. Also, mention the sampling method employed.

Reply: We did not perform a power analysis before conducting this study. In the absence of previous studies which examined differences in the personality features, we did not have effect sizes to calculate the desired sample size based on a power analysis. As stated in the introduction most of the previous studies investigated differences in the demographic characteristics like gender, age or employment status. Thus, the study design is original, as no previous prospective studies comparing the level of aggression or anger traits and states as well as impulsivity dimensions between violent and non-violent suicide attempters were found.
We have added a sentence in the Methods section about the lack of a preliminary power analysis. We also stated the potentially low sample size and a potentially limited statistical power in the limitation section.

3. Regression analysis could have been better. Please mention why the given method of statistical analysis was used.
Reply: We expanded the description and the rationale of the discriminant analysis in the revised version of the manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

The study aimed to investigate the sociodemographic and clinical profile of a sample (N=100) of people who had been hospitalized after a suicide attempt. Participants filled out a series of instruments. Researchers analyzed the data to explore if there were differences between those who had made a ‘violent’ suicide attempt versus those who had made a non-violent suicide attempt. Apart from one variable, there were no differences between the two groups.

 

Overall, I agree with the authors that the topic of the study is very important to increase our knowledge regarding who might be at risk of further suicidal behaviour. The fact that the study did not find significant differences between the two groups may also be an important finding in itself. Nonetheless, I have a few comments and considerations that may contribute to improving the manuscript. I detail my comments below.

 

Abstract and manuscript

Authors often state that people ‘chose’ a suicide method. (for example lines 24, 63,…). I suggest changing this in ‘use’. A choice implies that different options are available, and that people make a selection. However, some people just use a means that is available to them.

 

The abstract and manuscript includes several abbreviations (VSA, NVSA, SB). Please write in full throughout the manuscript.

 

Line 99: delete ‘current’

 

Methods section

Can authors clarify the sample size? Why N=100? Sample size/power calculation, or coincidence?

 

When was the study conducted?

 

Clinical assessment tools: Please include the Cronbach’s alpha coefficient for this sample.

 

Statistical analysis

Analysis involved mostly chi-square and t-tests. I wonder if this is the most appropriate method of analysis as it does not consider confounding variables. Have authors considered more sophisticated methods of analysis to compare the two groups?

 

Results

Multivariate analysis: Please report which variables were included in this analysis.

 

Discussion

Line 213: change ‘suicide survivors’ in ‘suicide attempt survivors’ to avoid confusion with people bereaved by suicide (who are also known as ‘suicide survivors’)

 

The most striking finding of the study is the absence of differences between the two groups, which receives little attention in the discussion. Let me suggest two plausible explanations:

 

According to the study methodology, almost all suicide methods were considered ‘violent’ methods except for attempted suicide by overdose of drugs or medication (line 78). Medication was not explicitly mentioned, but I assume it was included. Is it possible that the definition of ‘violent’ method was not adequate and alternative definitions should be considered.

 

A second explanation could be related to sampling bias. The study was conducted in hospital, thus the suicide attempt of each subject had required medical assistance in hospital. Suicide attempts that did not require medical assistance, maybe because they were less lethal, less violent, could not be recruited for the study. Thus, a study that would recruit community samples might be a more adequate design for detecting differences between groups of suicide attempters.

 

References

It seems that many references are between 5 and 40 years old. I do not mind authors using ‘old’ references. However, especially in the field of the bio-psycho-social understanding of attempted suicide and risk of suicidal behaviour, there are more recent references available.

 

Wishing you good luck with the revision.

Author Response

Reviewer #2:
The study aimed to investigate the sociodemographic and clinical profile of a sample (N=100) of people who had been hospitalized after a suicide attempt. Participants filled out a series of instruments. Researchers analyzed the data to explore if there were differences between those who had made a ‘violent’ suicide attempt versus those who had made a non-violent suicide attempt. Apart from one variable, there were no differences between the two groups.

Overall, I agree with the authors that the topic of the study is very important to increase our knowledge regarding who might be at risk of further suicidal behaviour. The fact that the study did not find significant differences between the two groups may also be an important finding in itself. Nonetheless, I have a few comments and considerations that may contribute to improving the manuscript. I detail my comments below.

Abstract and manuscript
Authors often state that people ‘chose’ a suicide method. (for example lines 24, 63,…). I suggest changing this in ‘use’. A choice implies that different options are available, and that people make a selection. However, some people just use a means that is available to them.

The abstract and manuscript includes several abbreviations (VSA, NVSA, SB). Please write in full throughout the manuscript.
Reply: We thank reviewer #2 for this suggestion, which we have implemented in the revised version of the manuscript.

Line 99: delete ‘current’
Reply: We thank reviewer #2 for carefully reading the manuscript and have deleted the word “current”.

Methods section
Can authors clarify the sample size? Why N=100? Sample size/power calculation, or coincidence?
Reply: This is actually coincidence. Regarding the power calculation, please see response to the editor and to reviewer #1.

When was the study conducted?
Reply: The study was conducted from 2018 until 2020. We stated this in the revised version of the manuscript.

Clinical assessment tools: Please include the Cronbach’s alpha coefficient for this sample.

Reply: Cronbach’s alpha is a measure of internal consistency, indicating how closely related a set of items of a questionnaire are as a group. It is thus considered to be a measure of the reliability of a scale. We used in the present study psychometric instruments which were validated and have been extensively used. In the cited references to these questionnaires, the Cronbach's alpha coefficient was given.
We stated in the Methods section: “All psychometric scales utilized were validated and have been extensively used.”

Statistical analysis:
Analysis involved mostly chi-square and t-tests. I wonder if this is the most appropriate method of analysis as it does not consider confounding variables. Have authors considered more sophisticated methods of analysis to compare the two groups?

Reply: We did not observe any differences in potentially confounding variables, like age, gender, education as well as other demographic variables. Therefore, we did not use a statistical analysis model, like ANCOVA to control for confounders. We stated this in the Methods section of the revised manuscript. As a sophisticated analysis method, discriminant analysis was performed to separate the two groups of suicide attempters based on the continuous clinical and personality variables to find a linear combination of the separating characteristics.

Results
Multivariate analysis: Please report which variables were included in this analysis.
Reply: We included the scores of the following questionnaires with the corresponding subscales in the discriminant analysis: SIS, BSS, MADRS, BDI-II, H-R-Scale, UPPS, K-FAF, STAXI-2, CTQ.

Discussion
Line 213: change ‘suicide survivors’ in ‘suicide attempt survivors’ to avoid confusion with people bereaved by suicide (who are also known as ‘suicide survivors’)

Reply: We thank reviewer #2 for carefully reading the manuscript and have corrected the misleading expression.

The most striking finding of the study is the absence of differences between the two groups, which receives little attention in the discussion. Let me suggest two plausible explanations:
According to the study methodology, almost all suicide methods were considered ‘violent’ methods except for attempted suicide by overdose of drugs or medication (line 78). Medication was not explicitly mentioned, but I assume it was included. Is it possible that the definition of ‘violent’ method was not adequate and alternative definitions should be considered.
A second explanation could be related to sampling bias. The study was conducted in hospital, thus the suicide attempt of each subject had required medical assistance in hospital. Suicide attempts that did not require medical assistance, maybe because they were less lethal, less violent, could not be recruited for the study. Thus, a study that would recruit community samples might be a more adequate design for detecting differences between groups of suicide attempters.

Reply: Thank you for these suggestions. We have added these explanations in the limitation section of the revised manuscript.
“We did not find differences in most of the demographic and clinical variables as well as personality characteristic studied. Beside the true effect, this could be also related to the sampling bias. The inpatient admission numbers of subjects after a suicide attempt in a psychiatric hospital alone cannot represent the entire sample of suicide attempters. We still know little about the extent to which outpatient support systems or counseling centers instead cared for patients after a non-violent suicide attempt or whether patients received initial care in an emergency department after a suicide attempt without being referred to psychiatry for further treatment. However, it is common for almost every patient to be referred to a psychiatric hospital participating in the project “Network for suicide prevention in Thuringia” for further evaluation of suicide risk. Furthermore, violent means were defined based on Åsberg’s list. This definition of VSAs remains controversial, perhaps leading to different results in studies based on other criteria. However, when the comparison was reanalyzed with respect to the UPPS “sensation seeking” and when the discriminant analysis was performed with the regrouped samples that included gas poisoning as a non-violent means [5], similar results were found.”

References:
It seems that many references are between 5 and 40 years old. I do not mind authors using ‘old’ references. However, especially in the field of the bio-psycho-social understanding of attempted suicide and risk of suicidal behaviour, there are more recent references available.

Reply: We have updated the reference list to include recent publications.
Wishing you good luck with the revision.

Author Response File: Author Response.pdf