Few Sex Differences in Hospitalized Suicide Attempters Aged 70 and Above

Relatively little research attention has been paid to sex issues in late life suicidal behaviour. The aim was to compare clinical characteristics of women and men aged 70+ who were hospitalized after a suicide attempt. We hypothesized higher depression and anxiety scores in women, and we expected to find that men would more often attribute the attempt to health problems and compromised autonomy. Participants (56 women and 47 men, mean age 80) were interviewed by a psychologist. In addition to psychiatric and somatic health assessments, participants responded to an open-ended question concerning attributions of the attempt. There were no sex differences in depression and anxiety. Forty-five percent of the men and 14% of the women had a history of substance use disorder (p = 0.02). At least one serious physical disability was noted in 60.7% of the women and 53.2% of the men (p = 0.55). Proportions attributing their attempt to somatic illness did not differ (women, 14.5% vs. men 17.4%, p = 0.79), and similar proportions attributed the attempt to reduced autonomy (women, 21.8% vs. men, 26.1%, p = 0.64). We found strikingly similar figures for depression scores, functional disability and attributions for attempting suicide in older men and women. Larger studies are needed in diverse settings as sex differences might be influenced by cultural context.


Introduction
Suicide prevention interventions that target older people appear to be more successful in women than in men [1]. This presents a public health challenge, as suicide rates are particularly high among older men in most countries worldwide. It has been suggested that those who survive suicide attempts in later life constitute a group that could provide keys to the prevention of suicide in later life [2]. It is therefore surprising that relatively little research attention has been paid to sex and gender issues in clinical presentations of older people with suicidal behaviour.
We could identify no study that focused on sex differences in psychopathology in older persons who presented at hospital after a suicide attempt. Regarding suicidal behaviour with fatal outcome, there is some evidence from both mixed age [3] and older adult [4,5] cohorts that symptoms of depression and anxiety are more common in women who die by suicide compared to their male counterparts. This may in part reflect the sex differential in depression rates in underlying populations. While the sex gap tends to decrease with age, many reports suggest a female preponderance of anxiety and depression in later life [6][7][8][9]. A Canadian study identified mental disorders as the major precipitant stressor in older women who died by suicide, but physical illness was the main stressor in men [4]. Findings from our own psychological autopsy study [10] showed that serious physical illness/disability was associated with a four-fold increase in suicide risk in men aged 65 and above, but no association was found in women. Men may be more vulnerable to react with suicidal behaviour when faced with physical illness and disability. The social roles to which men aspire are often thought to require higher levels of physical function and vigor. Loss of vigor/function may be perceived by many men as having adverse consequences for their identity. The authors of a qualitative study on end of life plans in older adults with chronic health issues noted that men emphasized that physical illness threatened their need to be independent and powerful [11]. Suicide was brought up by these men as a means of regaining dignity and control. Paralleling these results, a study of attitudes toward suicide in community living adults [12] revealed that suicide in connection with ill-health was considered more rational, more courageous and more acceptable than suicide under other conditions, especially among men.
The aim of the current study was to examine potential sex differences in clinical presentations of older women and men who were hospitalized after a suicide attempt. In light of the findings outlined above, we hypothesized that women would exhibit more severe symptoms of depression and anxiety. Further, we anticipated that men would be more likely to spontaneously attribute their attempt to compromised physical health and functional disability when asked the open-ended question "Why did you attempt suicide?".

Participants
Fifty-six women and 47 men aged 70 and above who were admitted to medical emergency departments in connection with a suicide attempt were included in the study [13]. Mean age did not differ between women and men (mean value 80.5, SD, 5.2, vs. 78.7, SD, 5.4, t = -1.641, df = 101, p = 0.104). Participants were recruited from 5 hospitals in western Sweden during 3 years (2003)(2004)(2005)(2006). Persons who suffered from terminal illness (n = 2), those who scored below 15 on the Mini Mental State Examination (MMSE) [14] (n = 2) and those who had insufficient knowledge of the Swedish language (n = 1), were excluded in accordance with pre-specified criteria. Twenty-eight persons declined participation. Seven persons left hospital without study information and two died before the scheduled interview, yielding 103 participants. This corresponds to a participation rate of 77.4% (74.6% for women and 81.0% for men, p = 0.384). Participant flow in its entirety is shown in Figure 1. A suicide attempt was defined as "a situation in which a person has performed an actual or seemingly lifethreatening behavior with the intent of jeopardizing his or her life, or to give the appearance of such an intent but which has not resulted in death" [15].

Major and minor depression
An algorithm based on the Comprehensive Psychopathological Rating Scale (CPRS) [16] and in accordance with the DSM-IV [17] was used to diagnose major depression [18]. Minor depression [13] was diagnosed in accordance with DSM-IV research criteria.

Substance abuse or dependence
A diagnosis of lifetime alcohol use disorder was based on interview data, medical record review or the regional hospital discharge register and 21 men and 6 women were identified by this procedure [19]. Review of medical records showed that seven women and three men had a history of prescription drug dependence, all but two of whom (both women) also had a lifetime alcohol use disorder.

Dementia
A research diagnosis of dementia was made in accordance with DSM-III-R [20] using the algorithm developed by Skoog et al [18] which requires: a. Impairment in short-and longterm memory and b. At least one of: b1/ Impairment in abstract thinking, b2/ Aphasia, b3/ Agnosia and b4/ Personality change.

Neuropsychological examination
The cognitive examination included the MMSE [14] and tests of short-and long-term memory, abstract thinking, aphasia, apraxia and agnosia as described previously [18].

Hopelessness
Hopelessness was examined with a single item from the Geriatric depression scale (GDS) [21] ("Do you think your situation is hopeless?") due to its central relevance in the study of suicidal behaviour.

Medically serious suicide attempt
A suicide attempt was classified as medically serious if it warranted hospitalization for at least 24 hours and treatment in a specialized unit (including intensive care unit, hyperbaric unit, and burn unit), surgery under general anesthesia (e.g., tendon rupture), or extensive medical treatment (beyond gastric lavage, activated charcoal, or routine neurological observations), including antidotes for drug overdose [2,22]. Persons who were hospitalized for more than 24 hours after a suicide attempt by hanging were also considered to have made a medically serious suicide attempt (MSSA), due to the high risk of fatality associated with this method.

Physical disability
A physician who was unfamiliar with the aim of the study and blind to participants' sex applied the World Health Organization description of disability ("an umbrella term, covering one impairment, activity limitation or participation restriction reached a level of 3 (severe/constant disability) or 4 (extremely severe disability) on the CIRS-G.

Social Factors Loneliness
A single question ("Do you feel lonely?") with a yes/no response was used to investigate perceived loneliness.

Family discord
A single item from a revised version [26] of the Recent Life Change Questionnaire [27] was used to examine family discord during the 6 month period prior to the attempt.

Instruments The Comprehensive Psychopathological Rating Scale (CPRS)
The CPRS [16] was used to examine psychiatric symptoms during the month prior to the suicide attempt. This scale comprises 67 psychiatric symptoms, 40 of these are reported and the remaining 27 are observed and rated by the interviewer. Each item is scored from 0-6, with 6 indicating the most severe level of symptoms and 0 the absence of the symptom. For the purpose of this study, we examined 18 of these symptoms. A symptom was defined as present when the score was 2 or more, as a score of 1 was considered to be within the range of normal psychopathology. The Montgomery-Åsberg Depression Rating Scale (MADRS) [28], derived from the CPRS, was used to rate depressive symptoms during the month preceding the index attempt. This scale includes ten items, yielding a maximum score of 60. The Brief Scale for Anxiety (BSA) [29], also derived from the CPRS, provide an overall rating of anxiety burden. We used a modified 9-item version (phobia question excluded), with a maximum score of 54.

Geriatric Depression Scale (GDS)
We also employed a depression scale developed specifically for older adults, and validated in that age group, the Geriatric Depression Scale (GDS) [21]. The 20 item Swedish version was used in the current study [30].

Sense of Coherence (SOC)
The 29-item Swedish version [31] of the SOC scale [32] was used to examine the extent to which participants found their lives meaningful, manageable, and comprehensible. The SOC has a seven-grade scale of response alternatives for each question, yielding a maximum score of 203. A high score indicates strong SOC.

Suicide Intent Scale (SIS)
Suicide intent was measured using the Suicide Intent Scale [33]. This scale comprises 8 objective and 7 subjective items. Items are scored from 0 (low intent) to 2 (high intent) yielding a maximum score of 30.

Personality traits
The Eysenck Personality Inventory (EPI) [34] was administered to measure two personality dimensions: neuroticism -emotional stability and extroversion -introversion. Personalities characterized by emotional reactivity, anxiety and psychosomatic concern, low ego-strength and guilt proneness score high on the neuroticism scale. Persons who score high on the extroversion scale are characterized as sociable, outgoing, impulsive and uninhibited.

Reasons for attempting suicide
An open-ended question, "Why did you attempt suicide?" was used to explore attributions for the attempt. In qualitative analyses, two coders (SW and AIB) identified nine attributions; somatic problems and pain, functioning and autonomy, psychological problems, social problems, lack of meaning, perceived burden and escape, no memory or understanding and wanted to die without a specific reason [35].

Ethics
The Research Ethics Committee at the University of Gothenburg approved the study (S 063-03, approval date 2003-02-18). All the participants provided written consent following oral and written information about the study.

Statistics
Due to small numbers, Fisher's exact test was employed to test for sex differences in proportions. The t-test was used to test for potential sex differences in means. As age and education level strongly influence MMSE score, a linear regression including these variables was performed. Ordinal regression was used to test for sex differences regarding number of depressive, anxiety, cognitive and somatic symptoms related to depression. All statistical tests were two-sided. Statistical significance was determined where p-values were less than 0.05.
The Bonferroni correction method was applied to address the problem of multiple comparisons.

Results
No sex differences were observed regarding sociodemographic characteristics (Table 1). Two thirds of the women fulfilled criteria for major depression, a proportion similar to that observed in men. Rates of minor depression were also similar. A history of alcohol/substance use disorder was significantly more common in men (44.7%) than in women (14.3%). This difference remained after a Bonferroni correction. Table 1 shows further that proportions using violent methods were similar in women (28.6%) and in men ( 27.7%). Six out of ten (60.7%) of the women and slightly more than half (53.2%) of the men had at least one serious physical disability in accordance with the CIRS-G. Slightly over a third reported family discord during the past 6 months in both women and men.  Table 2 shows proportions scoring 2 or above on CPRS depression-related symptoms. None of the specific symptoms were related to sex, nor were the number of symptoms in specified categories. Over two thirds of the women reported that they found their situation hopeless, a proportion that was significantly higher than that for men (43.8%), but the difference was not statistically significant after the Bonferroni correction. MADRS scores were almost identical in women and men and similar depression scores were observed on the GDS (   Table 3. Mean values on rating scales in hospitalized suicide attempters aged 70 and above, by sex.
Self-reported attributions for the attempt are shown in Figure 2. Proportions reporting that the attempt was due to somatic problems and pain did not differ between women and men.

Discussion
Contrary to our hypotheses, clinical characteristics and self-reported reasons for attempting suicide were strikingly similar in women and men in this northern European cohort of older adults with non-fatal suicidal behaviour. A partial explanation for the lack of sex differences in our study might be related to survival bias, as the chance of surviving suicidal acts is smaller in men than in women since men tend to use violent methods. Further, within a given method, greater lethality is observed in men [37]. Women and men who survive suicidal behaviour may share common clinical characteristics and attributions to a greater degree than those who die. Another explanation might be that sex differences may decrease with age [38].
Further, our study may have been underpowered to detect some potential sex differences.
The finding that affective psychopathology did not differ by sex was somewhat surprising.
The interviewer was a male clinician with extensive experience conducting research interviews with older adults, which might have helped men to feel more comfortable when talking about their symptoms in a research interview. Another possible explanation might be that the phenomenon of late life depression has received media attention in recent years, and this would be expected to reduce stigma of mental illness, which might facilitate the reporting of symptoms in men.
There was no sex difference regarding proportions with serious physical disabilities. We recently reviewed the litterature on this topic [39] and could find no clinical reports of physical disabilities in older women and men with non-fatal suicidal behavior for comparison.
Our findings expand on those of a European multicenter study [40] that found an association between functional disability and death wishes, with no sex difference.
Proportions attributing the attempt to physical problems and pain did not differ by sex and we found no differences regarding the attributions functioning and autonomy. While we could find no similar study for comparison, we do note that a US study on community-dwelling older adults (mean age 73.6) [41] found that greater value placed on autonomy among men amplified the relation between depressive symptoms and suicide risk. This was not the case for women in that study.
The women in our study scored slightly lower on the MMSE compared to men, but the association did not survive the Bonferroni correction. In general, research on cognitive aging suggests that women tend to outperform men on many cognitive tasks in late life [42].
However, a recent Japanese study [43] on community-dwelling older adults found a modest effect of sex with lower MMSE scores among women in the oldest age group. The sex difference in MMSE scores in our study is unlikely to be fully explained by differences in education, age, major depression or the number of days between the index attempt and the research interview. It is possible that women's suicide attempts were more likely to have specific cognitive or neurologic sequelae, which could have led to their slightly (nonsignificantly) lower MMSE scores. Our methods were however unable to detect those sequelae.
In the current study we found no sex differences regarding suicide intent as measured by the SIS, and this was the case for the subgroup with major depression as well. A multinational mixed age study found higher suicide intent among male attempters compared to female [44].
In contrast, one US study focusing on suicide attempters with major depression [45] found that women aged 70+ scored numerically higher on the SIS scale than their male peers (17.4  vs. 15.1), while in our study men scored numerically higher (16.4 vs. 15.3). In our study, the medical lethality of the suicide attempt was similar in women and men with major depression.
Our results diverge from those reported in the US-based study where greater medical lethality was found in men compared to women [45]. A number of methodological differences might be at play, including different methods for the evaluation of potential lethality. Our rating was based on the medical interventions required, and there could be a gender bias regarding emergency room physicians' treatment decisions. Another difference is that the US study was set in tertiary care and ours in a general hospital emergency setting. Half of the men with major depression in our study had a lifetime history of alcohol/substance abuse or dependence, a proportion identical to that reported in the above cited US study. Proportions were lower for women in both studies, which may reflect lower rates of substance use issues in the background population. In a study that employed a population-based comparison group [19] we reported that alcohol use disorder was associated with a ten-fold increase in risk for suicide attempt in both men and women.
Participants can be considered representative of the underlying population in terms of sociodemographics [13]. However, it must be stressed that our study design can capture only those who are in contact with hospital services after a suicide attempt. Our findings cannot be extrapolated directly to other cultural settings. Normative and social structures of femininity and masculinity can differ between cultures and change over time [46,47] as do suicide rates [48]. In 2016 suicide rates were at a moderately high level in Sweden, 13.2/100 000 in women and 28.6/100 000 for men aged 70 and above [49]. Corresponding figures for non-fatal suicidal behavior were 51.7/100 000 in women and 52.8/100 000 in men. One important issue is that shooting is a relatively uncommon method in Sweden as it is in the rest of Europe. This may influence patterns of both fatal and non-fatal behaviors. Over three-quarters of the participants in our study used overdose [22], a larger proportion than that (63%) reported in a US study [50], Another difference is that two thirds of the participants in our study lived alone, a larger proportion than the US study involving hospitalized suicide attempters [45].

Methodological considerations
One strength of our study is the high age cut-off for inclusion. Further, the same psychologist carried out all the interviews. While this is one of the largest clinical samples of older individuals who attempt suicide, the study was not powered to detect relatively small effect sizes. Another limitation is that the data were collected more than 10 years ago. Our study lacked a measure of masculinity and femininity, which are important to consider as masculinity, rather than biological sex, was found to be associated with serious suicidal thoughts in older, but not younger cohorts [51].

Implications for the prevention of suicidal behavior
Relatively high ratings on intent and lethality in women demonstrate a level of seriousness that needs to be taken into consideration in clinical management after a suicide attempt. The fact that serious functional disability was present in considerable proportions in both sexes, taken together with the finding that autonomy was one of the most important attributions for attempting suicide in both men and women, highlights the need for interventions that can strengthen the older adult's ability to deal with everyday life issues. Person-centered interventions can help to strengthen the feeling of capability [52] by facilitating opportunities for older men and women to do and be what they value, even in the face of significant health issues.

Conclusion
The unexpected lack of sex differences in the present study might be influenced by cultural context in which gender norms play a part [53]. Studies are needed in varied cultural settings.
These will need to address measures of gender role orientation and attitudes, including also depressed older women and men without suicide attempt history for comparison.

Supplementary material
Supplementary sources had no role in the study design, the data collection, the analyses and interpretation of data, the writing of the report or in the decision of submit the paper for publication.

Declaration of Conflicting Interests
The authors declared no potential conflicts of interests with respect to the research, authorship, and/or publication of this article.