Introduction

Suicidality is a complex phenomenon. It can be seen as a symptom of psychiatric disease. This corresponds to a reported high (up to 98%) prevalence of psychiatric disease in people who died by suicide (Bertolote and Fleischmann 2002). However, suicidality also has characteristic dynamics on its own, across different psychiatric diagnoses and different persons. In a complex interplay of factors suicidality develops, waxes, wanes and can lead to suicide. The following stages of the suicidal continuum are usually discerned: prior suicide acceptability, life-weariness, death-wishes, suicidal ideation, suicidal intent or plans, minor suicide attempts, major suicide attempts and completed suicide (Paykel et al. 1974; Salander Renberg 2001). One of the most important models to describe this process and the factors influencing suicidality is the entrapment model. This model gives a central role to the perception of having no escape potential from a situation in which an individual is placed as a consequence of triggering events. This perceived entrapment induces hopelessness and the risk of suicidal behavior (van Heeringen et al. 2008).

Religion and religiosity may influence entrapment in different ways. Religious beliefs and behaviors offer another perspective to reality and offer hope, which both could lower entrapment. They are also related to a lower risk for psychiatric disorders (Koenig et al. 2012, p. 179; Verhagen et al. 2012). Furthermore, all major religions, and especially the three monotheistic religions, reject suicide, in particular when committed for emotional or personal reasons (Koenig et al. 2012, p. 178). Hence, the strength of religious belief and suicide tolerance are negatively associated (Neeleman et al. 1997) and a higher measure of religiosity is correlated in general with a lower probability of suicide or suicidality (Colucci and Martin 2008; Koenig et al. 2012; Lizardi and Gearing 2010; Stack and Kposowa 2011). A recent meta-analysis evaluating 2339 completed suicide cases suggested an overall protective effect of religiosity from completed suicide with a pooled odds ratio of 0.38. This figure was even lower for western and religious homogenous areas (Wu et al. 2015).

One of the possible ways in which religion lowers the risk of suicide is by providing moral objections to suicide. The term ‘Moral objections to suicide’ (MOS) is being used regularly since Linehan included it her reasons for living inventory (RFLI) (Linehan et al. 1983). Moral objections comprise a set of various beliefs. One of these beliefs is the conviction that people who commit suicide will go to hell in afterlife. Many Christians for instance, especially orthodox protestants, perceive hell as a reality (Exline 2003). One can hypothesize that this conviction holds back people from committing suicide.

Clinical impressions support this hypothesis. The effect seems to be nonlinear and subject to influences from the psychopathological and suicidal state. For example: people who are severely depressed sometimes report no restrain by anxiety for hell anymore. In their experience nothing can be worse than this life.

This article reviews the literature to explore two hypotheses:

  1. (1)

    ‘Moral objections to suicide, especially the conviction of going to hell after committing suicide, exert a restraining effect on suicide and suicidality.’

  2. (2)

    ‘There is a differential effect of moral objections to suicide on the different stages of the suicidal continuum.’

Under the heading ‘Moral Objections’ Linehan et al. (1983) combine four reasons for living: ‘My religious beliefs forbid it.’, ‘I believe only God has the right to end a life.’, ‘I consider it morally wrong.’, and ‘I am afraid of going to hell.’ Studies on this inventory repeatedly pointed out a high score on internal consistency and correlation for the subscales, including the MOS-subscale (e.g., Garza and Cramer 2011; Linehan et al. 1983).

Previous reviews on how religion relates to suicidality pay little attention to the subject of moral objections. Some reviews mention it or sum up some articles about this subject (Colucci and Martin 2008; Lizardi and Gearing 2010), but none except one provides a more profound analysis or hypothesizes a model for the specific relation of MOS to suicidality. Bakhiyi et al. (2016) recently meta-analyzed 39 studies on the protective effect of the RFLI and concluded that reasons for living may protect against suicidal ideations and attempts and even show a predictive value. Moral Objections to Suicide and Survival and Coping Beliefs had a particular inhibiting effect on suicidality. No study investigating suicide death was found.

Given this background a specific review of the effect of moral objections to suicide on suicidality has to be provided. If MOS has a particular inhibiting effect on suicidality, it is clinical relevant to know how strong this effect is, how it functions and in what way it can be influenced in clinical practice. This article provides for a review and a model. First, the available literature will be reviewed. Next, based on the findings from the literature, a model of differential inhibiting influence of moral objections to suicide, especially fear of hell, on the suicidal continuum will be proposed.

Method

PubMed Medline and PsycInfo were searched on April 15, 2016, using the search terms: ‘hell’ and ‘suicide’/‘suicidality’; ‘afterlife’ and ‘suicide’/‘suicidality’; ‘moral objections’ and ‘suicide’/‘suicidality’; ‘reasons for living’ and ‘suicide’/‘suicidality’. No publication year limit was imposed. Articles concerning other subjects than suicide (like assisted suicide or suicide bombers), not dealing with moral objections or religious beliefs or feelings against suicide, or not available in Western languages were excluded (step 1).

Articles relevant to the proposed hypotheses were selected and reviewed. The analysis regarding the first hypothesis was done on all results specifically dealing with suicidal behavior as a variable dependent of or correlated with the score on the MOS-subscale, analogous to the hypothesis under investigation (step 2). These results were summarized, and the quality of each study was rated by the first author, using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies by the National Institute of Health (NIH) (“Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies—NHLBI, NIH”, w.d.). The contribution of each result with respect to the first hypothesis was rated. This rating was done by the first author, based on study quality, strength of effect and relevance to the central hypothesis. Four categories were discerned: (1) fair to good quality rating and support of hypothesis; (2) poor quality rating and support of hypothesis; (3) no support or disapproval of hypothesis; (4) poor quality rating and disapproval of hypothesis.

The analysis regarding the second hypothesis aimed to review all results of step 1 regarding the possible differential effect of MOS on the different stages of the suicidal continuum. Except for prior suicide acceptability, which for a great deal overlaps with or reflects MOS, evidence regarding positive or negative influence of MOS on transition to every stage in the suicidal continuum was gathered from these results.

If possible, longitudinal or prospective data on the effect of MOS on different stages were compared. When only cross-sectional studies were found, the studies reporting an association of MOS to suicidality in more than one stage were selected. The significance and direction of these associations within these studies were compared. When a next stage in the suicidal continuum showed a (more) significant association than a prior stage, a presumable differential effect was supposed. Studies using a continuous scale of suicidality, and not staging or grouping results on the suicidal continuum were excluded. This evaluation was followed by the proposal of a model to represent the effect of MOS on the suicidal process.

Results

The search combining ‘moral objections’ and ‘suicide’/‘suicidality’ brought forward 45 articles in PubMed Medline. Nineteen of them matched the selection criteria (step 1). In 11 studies suicidal behavior was examined as a dependent variable (or correlated one) in relation to the score on the MOS-subscale, analogous to the first hypothesis (step 2) (Dervic et al. 2004, 2006a, b, 2011; Garza and Cramer 2011; Lee and Oh 2012; Lizardi et al. 2008a, b; Malone et al. 2000; Oquendo et al. 2005; Richardson-Vejlgaard et al. 2009a, b). The same search in PsycInfo produced 73 results, 47 of them matched the selection criteria (step 1). Two articles written in non-Western languages were excluded (Deng et al. 2012; Mahmoudi et al. 2010). All 11 articles selected from PubMed Medline were found again, and four additional articles (Chatterjee and Basu 2010; Connell and Meyer 1991; Flowers et al. 2014; Valencia et al. 2009) were selected in step 2.

The search combining ‘afterlife’/‘after-life’ and ‘suicide’/‘suicidality’ generated 11 results in PubMed Medline, with 1 relevant article (step 1) that did not meet the criteria for the analysis regarding the first hypothesis (step 2) (Feigelman et al. 2014). In PsycInfo this search generated 364 results, with 19 relevant articles (step 1), but no article eligible for the analysis regarding the first hypothesis (step 2).

The search combining ‘hell’ and ‘suicide’/‘suicidality’ generated eight results in PubMed Medline, but no relevant articles apart from a single column in a journal on psychosocial nursing (Rix 1994) In PsycInfo this search generated 338 results, and 1 relevant article (step 1) (Echávarri et al. 2015). This article was not eligible for the analysis regarding the first hypothesis (step 2). Replacing ‘suicide’/‘suicidality’ in all those searches with ‘self-injury’ generated no additional results.

All 15 articles of step 2 that specifically dealt with suicidal behavior as a dependent or correlated variable of the score on the MOS-subscale are summarized in Table 1. This table shows for each article the population, outcome, a quality rating conform the NIH rating and an overall rating of support for the central hypothesis of this review.

Table 1 Overview of articles in primary analysis
Full size table

Four of these 15 articles were authored by Dervic. She was also co-author of the study by Lizardi. Three of these four articles reported for different groups (patients with a bipolar disorder, a depressive disorder, with a self-reported history of abuse and a cluster B personality disorder; resp. n = 149, 371, 119, 147) a significant association between MOS and suicidality: the higher the score on the MOS-subscale, the lower the prevalence of suicidal behavior (Dervic et al. 2006a, b, 2011). These studies did not evaluate the effect of the four moral objections from the RFLI separately.

The fourth study by Dervic et al. (2004) proved a mediating role for MOS in the inhibiting effect of religious affiliation on suicidality among people with a depressive episode. A higher MOS-score was found to be significantly associated with lower suicidal behavior. Mainly MOS accounted for the effect of religious affiliation on suicidality: the significant association between religious affiliation and suicide attempt did not remain significant when controlled for the intermediating effect of MOS.

Lizardi et al. (2008a, b) evaluated the influence of MOS on suicidal behavior in a cross-sectional study with depressed patients (n = 265). Those with a lower MOS-score showed a significantly higher number of suicide attempts, a higher percentage of being not religious affiliated, higher scores on depression, hopelessness and impulsivity, and lower scores on anxiety and reasons for living. Logistic regression, excluding possible mediating factors, pointed out an independent relation between a lower MOS-score and suicide attempts. Additionally, Oquendo et al. (2005) demonstrated that only the MOS-score and age were significantly related to lethality of attempts (n = 460). The MOS-score was higher in patients with low-lethality suicide attempts than in those with high-lethality attempts. For all other outcomes regarding suicidality MOS did not make a significant difference. A study by Malone et al. (2000) showed a higher MOS-score in depressed inpatients without a history of suicide attempts (n = 84). Richardson et al. (2009b) examined the suicidality of depressed patients (n = 521; with and without an alcohol use disorder). A higher MOS-score was significantly associated with lower suicidal ideation. Another study by Richardson et al. (2009a) (n = 804; persons with a depression or bipolar disorder) showed that higher suicidal ideation was modestly correlated with a lower MOS-score.

Two studies reported on suicide attempters. Valencia et al. (2009) found no difference in MOS-scores between attempters with violent and nonviolent suicide methods, and only little difference in MOS-scores between attempters rated ‘without suicidal intent or low risk’ versus ‘with suicidal intent or high risk’ (n = 102). Flowers et al. (2014) (n = 150, African-American female suicide attempters) found a significant correlation between a lower MOS-score and suicidal intent as measured in the Suicidal Intent Scale (SIS).

Four studies of those 15 described also a positive relation between higher moral objections and lower indicators of suicidality in other and broader defined populations. Research by (Garza and Cramer 2011) among Hispanics (n = 168; partially outpatient population) and by Lee and Oh (2012) in Korean students (n = 277; no specific psychiatric population) also showed that a lower MOS-score was significantly related to higher suicidality and/or suicidal behavior, even after controlling for interacting variables. Connell and Meyer (1991) showed a negative correlation between the MOS-score and suicidal intent (n = 205; college students). Chatterjee and Basu (2010) found the same correlation between MOS and a history of suicidal ideation and attempt (n = 120; female college students).

Meta-analyzing these results turned out to be unfeasible, mainly because of too much difference in (statistical) methods and partly because of study-overlap (see Discussion).

Differential Restraining Effect

In the analysis regarding the second hypothesis all results of step 1 were reviewed for evidence concerning a differential restraining effect on the suicidal continuum. No longitudinal or prospective data were found.

Only one study provided data on the influence of MOS on the transitions from no suicidality at all to life-weariness or to death-wishes, suicidal ideations, intent or plans. This study by Connell and Meyer (1991) reported a significantly lower MOS-score in non-suicidal persons when compared with persons with any suicidality, and in persons with a history of brief suicidal ideations when compared with persons with more advanced suicidality. So MOS in this study is associated with less progression to death-wishes and next to suicidal ideations and intent.

The influence of MOS on transition from no attempts to (minor) suicide attempts brought up more results. Dervic et al. (2011) found that of 7 suicide-related variables only the number of past suicide attempts was significantly negatively related to moral or religious objections to suicide. In line with this result another study by Dervic et al. (2006) found no significant association between MOS and suicidal ideations, but only a significant lowering effect on suicidal attempt. In addition another study by Dervic, Grunebaum, et al. (2006) found a significant higher MOS-score in attempters compared with non-attempters. Findings by Oquendo et al. (2005) also supported the suggestion of an inhibiting effect of MOS on the development of suicidal attempts. In this study MOS was not significantly associated with suicidal ideations or intent, but significant negatively related to both attempt status and potential lethality.

On the transition to major suicide attempts three studies were reported. Malone et al. (2000) made a distinction between suicide attempts with a low and a high potential lethality. MOS were negatively associated to both, but was clearly stronger reported by subjects with low-lethality suicide attempts than by those with high-lethality attempts. In line with this, as mentioned already, Oquendo et al. (2005) demonstrated that only the MOS-score and age are significant related to potential lethality of attempts. Valencia et al. (2009) also report an (not so strong) association between MOS and the suicidal intent of an attempt, with lower MOS in higher suicidal intent versus those with lower suicidal intent.

On the relation of MOS to completed suicide no direct data exist. Only one study prospectively examined the relation between suicide acceptability (a factor related to MOS) and completed suicide. In a large cohort (n = 9115), after controlling for both age and gender Feigelman et al. (2014) found a significant negative association between belief in an afterlife, suicide acceptability and support for euthanasia and later completed suicide (n = 141).

Discussion

Limited Evidence

All analyzed studies support the hypothesis (1) that moral objections to suicide, especially the conviction of going to hell after committing suicide, exert a restraining effect on suicide and suicidality. No study reports an inverse relation. However, one can make several remarks regarding the quality of this small body of evidence.

First, as already mentioned, no prospective or longitudinal studies on this subject are published yet. This hampers the quality of evidence and makes it practically impossible to get conclusive evidence for a possible causal relation.

Second, five of the 15 available studies are (co)authored by Dervic. These five studies show an overlap in studied population and even data used. All patients in these studies have been recruited from the New York State Psychiatric Institute and of the Western Psychiatric Institute and Clinic (Pittsburgh, Pennsylvania). Three studies report on data of subjects also studied in the first publication of Dervic on this subject (2004), varying from 23% up to 100% (Dervic et al. 2006, 2011; Lizardi et al. 2008a, b).

Third, adjacent to this: the outcomes and effect of the RFLI and of religiosity in general differ from culture to culture (Lee and Oh 2012; Wu et al. 2015) Twelve of the 15 analyzed studies report on a US population. Although two studies pay special attention to the Spanish-speaking and Latino population in the USA, there are only three reports on populations in other countries.

Last, there is a possible social desirability bias: use of self-report to evaluate suicidality can generate an overestimation of the effect of the RFLI and its MOS-subscale. Morrison for example showed for African Americans a higher amount of ‘hidden ideators’ than for Caucasian Americans: 35 out of 36 non-Caucasian patients who had checked the box ‘no suicidal thoughts’ in a prior self-report inventory, were found to have suicidal thoughts at further face-to-face evaluation (Morrison and Downey 2000). Most data used in this review report on self-rated questionnaires with Likert items. The social and psychological barrier to report suicidality can be high and differs from culture to culture, and self-report, especially using single check-boxes, but also more differentiated scoring, possibly induces a negative reporting bias on suicidality.

In sum, the hypothesis that moral objections to suicide, especially the conviction of going to hell after committing suicide, exert a restraining effect on suicide and suicidality is supported by 15 cross-sectional studies of fair to poor quality, mainly clinical samples from the USA and some non-clinical samples worldwide.

The findings regarding this hypothesis accord with results from other studies, not included in the primary analysis. Joe et al. (2007) showed for young persons (14–22 y; n = 3310) that the acceptability of suicide (not specifically operationalized in MOS) to an individual enlarges the prevalence of suicidal plans: young persons with the greatest acceptance of suicide were more than fourteen times more likely to plan their suicide as those with the least acceptance of suicide (26.3 vs 1.8%).

RFLI, MOS and Fear of Hell

Does the score on the MOS-subscale directly correlate to the presence of its subfactor fear of hell? Or are the other three factors of the MOS-scale not interlinked with fear of hell? This question cannot be answered from the literature. Richardson states, without showing any data, that principal components analysis shows that the four items of the MOS tap on a unitary factor (Richardson-Vejlgaard et al. 2009a) conform the previous mentioned high score on internal consistency and correlation for the subscales.

Only Echávarri et al. (2015) reported separately on the question about fear of being sent to hell, and not on the combined MOS-subscale (n = 193). The group with only suicidal ideations had higher scores on fear of being sent to hell than the group with suicidal attempts (95% CI [2.45, 3.55] vs [1, 77, 2,52]; p < 0,050). These few findings point to a congruent effect of fear of hell to other moral objections to suicide, both being negatively associated to suicidality.

Relation to Psychopathology

Other results of the search (step 1) help to create a more differentiated perspective with regard to interactive effects with psychopathology. Does (the course of) psychiatric disease alter the effect of MOS on suicidality?

One can expect MOS to be less effective in reducing suicidal behavior with a more serious psychopathology. Only Garza and Cramer (2011) address this with regard to depression in a subanalysis and reports ambivalent results. Dervic et al. (2006) did not find a significant association of MOS to cluster B personality disorder (CBPD), which suggests a protective effect of MOS on suicide attempt to the same extent in those with or without comorbid CBPD. The presence of an anxiety disorder raises the risk of developing suicidal ideations (Sareen et al. 2005) Anxiety, both in trait or state, could thus be a confounding factor in the effect of MOS, specifically fear of hell. Very little research has been done on this relation. Lizardi et al. (2008a) found no significant association of psychic anxiety with MOS or a history of suicide attempts.

After all, no robust results are available about the interactive effect of psychopathology on the relation between MOS and suicidality.

Differential Effect of MOS

From the results of this review a differential pattern of influence of MOS on the different stages of the suicidal continuum can be suggested (according to hypothesis 2). This differential effect is represented in a schematic way in Fig. 1. The upper bar represents the different stages of the suicidal expression process. The lower bar represents the moral objections. The arrows show the extent to which MOS have a restraining effect on the progression in different stages suicidal expression process: the broader the arrow, the more effect.

Fig. 1
figure 1

Inhibitory influence of MOS on the suicidal continuum

Full size image

Conclusion and Implications

This review explored the literature to test the hypothesis that ‘moral objections to suicide, especially the conviction of going to hell after committing suicide, exert a restraining effect on suicide and suicidality.’ The hypothesis was supported by all 15 cross-sectional studies on this subject.

The reviewed literature raised sufficient evidence to confirm a restraining effect of moral objections on suicide and suicidal behavior in a psychiatric population, with an indication for the same effect in the general population. On the effect of fear of hell sparse but congruent data were found.

Moral objections to suicide seem to counteract the development of suicidal ideations and attempts, and possibly, when an attempt is done, the lethality of suicidal attempts. A differential pattern of influence of MOS on the different stages of the suicidal continuum is suggested. The restraining effect of moral objections seems to be greater on the severity of suicide attempts than on the measure of suicidal ideations.

The presence of moral objections to suicide is a relevant and an important factor in the process of suicidality, of course cooperating with other restraining and promoting factors. Therefore, prospective and sufficiently detailed research is needed to clarify the role of MOS in this process and to develop an interactive model of suicidality and religion. Such a model has to encompass a broader and more detailed elaboration of religiosity, to include social and cultural aspects, and to account for the role of psychopathology and psychological factors like anxiety and impulsivity (Turecki 2005).

The presence of moral objections to suicide is a relevant factor in the exploration of suicidality in every individual patient. Suicidal patients with moral objections can benefit from the inclusion of MOS and life-maintaining beliefs from religion in the treatment of suicidality and of its underlying psychopathology. This has to be done in a sensitive and supportive way, not just reinforcing fear, but instilling hope. Future research can reveal methods and ways of doing this, reinforcing this important reason for living.