Approximately 10 million people worldwide died from cancer in 2020, making it the top cause of death [1]. Breast cancer is the most common malignancy in terms of incidence [1]. Given that its 5-year survival rate has reached up to 90% [2], patients with breast cancer have a high risk of developing psychiatric disorders including depression [3]. Among various psychiatric conditions, one that must be addressed immediately is suicidal ideation and behavior. About one in every ten patients with breast cancer reports suicidal ideation 1 year after breast cancer surgery [4], and they have a higher risk of suicide than the general population with standardized mortality ratios ranging from 1.25 to 1.53 among cancer registrants based on population [5].

Suicide in cancer patients has been linked to a variety of risk factors including depression, hopelessness, impaired physical performance, cancer type, and time elapsed after diagnosis [6]. For patients with breast cancer in particular, the following factors are known to increase the risk of suicide: younger age, male sex, separated/divorced/widowed marital status, a history of cancer-related surgery, and short time passed since cancer diagnosis [7]. Protective factors, which decrease risk of suicide, are also integral to a comprehensive suicide risk assessment. Social support, family connectedness, pregnancy, and parenthood are known to decrease the risk of suicide in nonmedical populations [8,9]. Contrary to risk factors, however, studies on protective factors on cancer patients are lacking. A study on Chinese cancer patients reported that confrontation coping strategies and family adaptability and cohesion are protective against suicidal ideation [10].

Grit is a relatively new, positive psychological concept and was operationally defined as perseverance and passion for achieving long-term goals [11]. Grit entails working hard toward challenges and staying determined and interested over a long period of time irrespective of failure, hardship, and setbacks [11]. Grit consists of two factors: consistency of interests and perseverance of effort [11]. The former entails consistently maintaining interests and efforts while the latter means exhibiting an increased level of persistence and diligence even in the face of definite setbacks or failures [12]. Higher scores in grit are associated with increased psychological well-being and resilience [13]. Moreover, evidence continues to accumulate indicating that grit could be a protective factor against suicidality [7,14].

By using a 6-month longitudinal study design for establishing causal relationships between predictors and the outcome (suicidality), this study aimed to examine whether each aspect of grit could be a protective predictor against suicidality and identify risk predictors for suicidality among breast cancer patients in a psycho-oncology clinic in a general hospital. We hypothesized that grit could be a predictive protector against suicidality.

A total of 176 patients with breast cancer participated in this study. Complete baseline data was collected from 154 of them, and those with an education status of college or higher, a familial history of psychiatric disorder, and a lower level of pain were more likely to have missing data (Supplementary Table 1). At the 3-month evaluation, there were 88 participants (attrition rate: 42.86%), and missing data was more likely to be found for those who had surgery within the last 2 months, no history of psychiatric disorder, and a familial history of psychiatric disorder (Supplementary Table 2). Lastly, 63 participants (attrition rate: 52.09%) completed the 6-month assessment, for which there was no difference in characteristics between those with and without missing data (Supplementary Table 3).

Sociodemographic and clinical characteristics are presented in Table 1. The group mean of the PHQ-9 falls between mild and moderate depression, and that of the GAD-7 corresponds to mild anxiety, respectively [17,20]. The average baseline score of the MINI suicidality module was 2.73±4.86. Its mean scores for the 3-month and 6-month assessments were 2.45±5.57 and 2.33±5.31, respectively. There was no significant change in the average suicide risk scores over the three assessment points (F= 0.156, p=0.856).

The results of the linear regression models, in which suicide risk at baseline, 3 months, and 6 months assessments was predicted by baseline factors, are shown in Tables 2-4, respectively. The baseline model contained only factors measured at baseline, and 23.6% of the variance in suicide risk was explained by the predictors (Table 2). Significant associations were found between a history of psychiatric admission (β=0.240), the PHQ-9 (β=0.267), and the PPI (β=0.234) and suicidality; previous psychiatric admission, the higher PHQ-9 score, and the greater PPI score were associated with higher suicidality. For the 3 months model, 7.4% of the variance was explained (Table 3). Although GAD-7 was significantly associated with suicidality in the univariable analysis, it was not retained in the final multivariable model following the stepwise selection procedure. The PPI was the only significant contributor to suicide risk at the 3-month assessment (β=0.240); the PPI score was positively associated with suicidality. In the 6 months model, 18.6% of the variance in suicide risk was explained (Table 4). Significant relationships with suicidality included consistency of interests (β=-0.323) and the PHQ-9 (β=0.280). Whereas the PHQ-9 score was positively associated with suicidality, consistency of interests showed a negative association with it. Perseverance of effort did not show any associations with suicide risk at any of the time points.

In the sensitivity analyses with baseline suicidality additionally entered as a covariate, no significant predictors emerged at the 3-month assessment (Supplementary Table 4), whereas consistency of interests remained a significant negative predictor at the 6-month assessment (β=-0.342, p=0.016) (Supplementary Table 5).

This study is the first to examine the role of grit and other predictors of elevated suicidality among psycho-oncology outpatients with breast cancer. The findings indicate that several factors—consistency of interests, the degree of depressive symptoms, and the intensity of present pain—can contribute to suicidality. Consistency of interests decreased suicidality at the 6-month time point. The degree of depressive symptoms and intensity of present pain increased suicidality at the baseline and 6-month time point and at the baseline and 3-month time points, respectively. These results show that a facet of grit—consistency of interests—can be a protective predictor of suicide and different factors contributing to suicidality occur over time.

The protective role of grit against suicidality in the long term (6 months) among patients with breast cancer is in line with the results of the previous study showing that negative life events predict suicidal ideation 4 weeks later only for those college students low in grit [21]. Similarly, higher grit along with greater gratitude at baseline predicts the greatest decrease in suicidal ideation 1 month later in undergraduates [22]. In another study on the moderating effect of grit on the relationship between hopelessness and suicidal ideation and behavior, the authors showed that grit is a potential protective factor against suicidal ideation and resolute plans and preparations, even in those with a high level of hopelessness in military personnel [23]. Likewise, higher grit is associated with decreased severity of suicidal ideation and behavior in general surgery residents and adults among general populations [14,24]. Given that suicidality in this study is a comprehensive construct encompassing suicidal ideation, plan, and a previous attempt, our finding is in close agreement with those of the studies on non-students in particular.

How grit can be protective against suicidality could be explained with Baumeister’s conceptualization of death by suicide [25]. According to the theory, death by suicide is a motivation to free oneself from aversive self-awareness. Recognition of inadequacies of themselves produces negative emotion, and thus they wish to avoid self-awareness and the concomitant emotion. Then they strive to attain a state of cognitive deconstruction, which helps suppress genuine self-awareness. Suicide can be regarded as the last stage in the process of escaping from themselves and the outside world. A diagnosis of cancer implies the beginning of a life-threatening journey. During the cancer journey, their pursuit of life goals, whether they are short term or long term, are likely to be unwillingly postponed while patients undergo cancer examination, treatment, and rehabilitation, often exhausting. Since those with strong grit have attributes to withstand such adversity, they are likely to be resilient to the formation of suicidal ideation. A previous study reported that grit, along with gratitude, makes patients resilient and helps deter suicide by giving more meaning to their lives [24].

More specifically, among two constructs of grit, only consistency of interests turned out to be a protective predictor against suicidality at the 6-month evaluation. There is a paucity of studies on distinct roles between consistency of interests and perseverance of effort regarding psychiatric domains. An increased level of consistency of interests, not perseverance of effort, is associated with lower severity of maladaptive eating behavior in people with an eating disorder [26]. Those with binge-eating symptomatology regard food stimuli as aversive [27] and threatening against achieving or upholding the ideal body image [28], and those with restrictive behavior rate food more negatively, which is followed by an increase in anxiety and fear [29]. In either case, negative affect is involved in the processing of food-related stimuli. In summary, although preliminary, these studies may lead to the implication that only consistency of interests is partly related to the ability to tolerate negative affect. Besides, those with higher consistency of interests are associated with reduced perceived stress [30,31], which can consequently lower the risk of suicidal ideation [32,33]. Moreover, whereas consistency of interests is negatively associated with depression, anxiety, and stress, perseverance of effort has a negative association with depression only [34], implying that the former might be considered as an indicator of the presence of less psychopathology. Perseverance of effort, however, has mainly been associated with academic performance [11,35] and entrepreneurial success [36]. Although consistency of interests and perseverance of effort seem similar at first glance, there is, in reality, a weak correlation between them [30,37]. In the present sample, consistency of interests and perseverance of effort were weakly correlated (r=0.234, p=0.004). In addition, collinearity diagnostics did not indicate problematic multicollinearity, with variance inflation factors ranging from 1.11 to 1.33. Therefore, the differential associations observed for the two grit dimensions are unlikely to be explained solely by multicollinearity. Management of a psychiatric disorder usually involves long-term care. Industry, reflected in the aspect of perseverance of effort, can be important in academic or entrepreneurial achievement [11,35,36]. For our study sample, however, the ability to sustain an interest (i.e., good compliance with treatment for the improvement of psychological health without burnout) seems to matter more than the tendency to show diligence and industry regardless of challenges or obstacles for lowering the severity of a psychiatric symptom, especially suicidality. In addition, consistency of interests may partly reflect the ability to maintain a stable sense of meaning, identity, and valued life goals despite adversity. From the perspective of psychological acceptance frameworks, sustaining long-term interests may help patients tolerate distress without excessive avoidance, which could contribute to lower suicidality. In contrast, perseverance of effort may be more closely related to achievement-oriented persistence and diligence, which may not directly protect against suicidal thoughts or behaviors in the context of chronic psychological distress. Of note, this protective association remained significant even when baseline suicidality was controlled for, suggesting that the effect of consistency of interests on suicidality at 6 months was not merely a reflection of initial suicidality levels.

Grit is a relatively new concept. As not much research on the association between grit and suicidality has been conducted, our study needs to be replicated by future researchers. Grit is affected by life experiences in addition to genes even though it is often considered to be a trait [14,38], as opposed to a state. Thus, increasing grit may become a specific measure to prevent suicide. By giving patients a feeling of hope, especially in challenging conditions, intervention programs that promote resilience and positive psychology may boost grit on an individual level [14,38].

At baseline, the degree of depressive symptoms was a risk factor for suicidality. Besides, depression turned out to be a predictor for suicidality at the 6-month assessment. These findings are in close agreement with the results of a previous study showing that the intensity of depressive symptoms is a significant correlate of suicidal ideation in patients with breast cancer [4], and depression was reported to be associated with death by suicide among older cancer patients [39]. The present finding that the level of depressive symptoms is significantly related to the current suicidality and, moreover, predicts it in the long term, highlights the necessity of effective management of depression. Treating depression leads to the improvement of quality of life in breast cancer patients and may extend lifespan [40]. In the additional analysis adjusting for baseline suicidality, however, depressive symptoms were no longer retained in the 6-month model. This finding suggests that part of the observed association may have reflected baseline suicide risk.

At the baseline, a history of psychiatric admission was associated with a higher suicide risk. Past studies including nonmedical samples revealed that a history of psychiatric hospitalization increases both the risk of suicide attempt [41] and that of death by suicide [42]. Given that it is well-established that a history of psychiatric disorder is positively associated with suicide attempt [43], it is understandable that psychiatric admission elevates the risk of suicide attempt in that admission due to psychiatric disorder can be regarded as an indicator of a mental condition being poorly controlled [44].

Nonpsychological factors appeared to be less influential for suicide risk in our sample. The intensity of present pain was the only nonpsychological contributor among our models (baseline and 3-month assessments). Many studies on patients with breast cancer have linked pain and suicide risk. The intensity of pain is associated with suicidal ideation in the group of cancer patients including those with breast cancer [45,46]. Similarly, for the sample with advanced cancer that also includes patients with breast cancer, a higher degree of pain was associated with a greater suicide risk [47]. Depression appears to account for the association between pain and suicide risk. The degree of pain increases the prevalence of a major depressive episode in cancer patients, playing a causal role [48], and pain is also associated with increased depression in women with metastatic breast cancer [49]. Major depression is a well-established risk factor for suicide [50], which may be involved in mediating pain and suicidal ideation. Given our finding that the PPI is associated with the current suicide risk and even predicts suicide risk in the short term, effective pain management would be beneficial for reducing not only physical distress but also suicide risk. However, the 3-month model explained only a small proportion of the variance in suicidality (R²=0.074). Accordingly, the association between pain intensity and suicidality at this time point should be interpreted with caution and confirmed in future studies. Similarly, the association between pain intensity and suicidality at the 3-month assessment was no longer significant when baseline suicidality was included as a covariate, and should therefore be interpreted with caution.

Many factors known to increase suicide risk (e.g., age, and living status) [43] did not occur as risk factors in this study sample. Since breast cancer per se is known to increase suicide risk [5], we speculate that identifying a risk factor was difficult for the study sample, which was already at an elevated risk of suicide. It should be noted that consistency of interests was not associated with suicidality in the relatively early stages. Until after certain treatment on psychological and physical symptoms was given, the risk factors in the relationship with suicidality, which appears distress-provoking, may have prevailed over the protective factor. Our findings are in line with that of a previous study showing that, beyond the initial contributors of psychological and physical distress, multifaceted factors associated with suicidal ideation emerged 1 year after surgical treatment of breast cancer [4]. However, this interpretation remains speculative. It is also possible that the significant finding at the 6-month assessment was influenced by the relatively small sample size and multiple testing. Therefore, the temporal pattern observed in the present study should be interpreted with caution and confirmed in larger prospective studies.

This study has several limitations. First, the attrition rate was high at the 3-month and 6-month time points. Although the characteristics of the study sample at those points did not differ largely between the participants and the nonparticipants, we cannot exclude that the high attrition rates might have affected our results. Because participants with greater psychological distress or poorer functioning may have been less likely to complete follow-up assessments, the protective effect of consistency of interests may have been underestimated. On the other hand, if participants with higher consistency of interests were more likely to remain in the study, the observed association may have been overestimated. However, because there were no significant differences in baseline characteristics between completers and non-completers at the 6-month assessment, the direction and extent of potential attrition bias remain unclear. Therefore, these findings should be interpreted as preliminary and hypothesis-generating and warrant replication in larger prospective studies. Second, changes in psychiatric symptoms may be related to specific conditions of breast cancer (e.g., cancer-related symptoms, and treatment status); however, these were not considered. Therefore, time passed since breast cancer diagnosis was adjusted instead. In addition, baseline PHQ-9 and PPI scores may not have fully reflected changes in depressive symptoms and pain intensity over the follow-up period, which may have influenced their associations with suicidality at later time points. Third, our results may not represent the entire population of patients with breast cancer. The participants were recruited from a psychooncology outpatient clinic and thus representative of a population experiencing psychological difficulties. Consistent with this recruitment setting, 97.4% of participants had a current psychiatric diagnosis. Although this enhances the clinical relevance of the findings for patients receiving psychiatric care, the highly selected nature of the sample may have limited the variability of both suicidality and grit. This restricted range may have weakened the observed associations and contributed to the relatively low explanatory power of some regression models, particularly at the 3-month follow-up. Fourth, important cancer-related clinical variables, such as cancer stage, recurrence status, metastasis, and treatment intensity, were not included in the analyses. However, we attempted to partly account for cancer-related clinical status by including time passed since index cancer diagnosis and treatment-related variables (surgery, chemotherapy, radiotherapy, and other therapy) in the analyses. Nevertheless, residual confounding by unmeasured oncological factors may still have affected the findings although the direction and magnitude of this bias cannot be determined with certainty from the present data. Fifth, the relatively low explanatory power of the regression models, particularly the 3-month model, suggests that suicidality in patients with breast cancer is likely influenced by multiple biological, psychological, and social factors that were not fully captured in the present study. Therefore, the identified predictors should be interpreted as contributing factors rather than comprehensive determinants of suicidality. Sixth, the stepwise method for variable selection has known limitations, including the risk of overfitting and model instability in relatively small samples. However, the primary predictors of theoretical interest (consistency of interests and perseverance of effort), as well as age and time passed since index cancer diagnosis, were retained in all multivariable models. Stepwise selection was applied only to the remaining covariates, which may partly reduce concerns regarding a purely data-driven model. Replication in larger samples with pre-specified variable selection strategies is warranted. Seventh, no correction for multiple comparisons was applied across the three regression models. As this study was exploratory in nature, the analyses were designed to identify potentially relevant factors associated with suicidality over time. Nevertheless, the absence of multiple comparison correction increases the risk of type I error, and some findings may therefore represent chance associations. Accordingly, the results should be interpreted with caution and considered preliminary until replicated in independent samples using pre-specified hypotheses and appropriate statistical controls. Eighth, grit was measured only at baseline and was not reassessed during follow-up. Although grit is generally considered a relatively stable trait, we could not examine whether it changed over time or whether any changes were associated with changes in suicidality. Future studies with repeated assessments of grit are needed to determine whether changes in grit are associated with subsequent suicide risk and to evaluate the potential role of grit-based interventions. Finally, all the participants were ethnic Koreans, and thus, generalization of the findings to ethnic groups should be performed cautiously. Despite the limitations, the present study also demonstrates certain strengths. First, this is the first study to investigate the protective effect of the aspects of grit on suicide risk in cancer patients. Second, we included a large clinical sample receiving standard clinical care—both oncologic and psychiatric. Third, suicide risk was assessed with a structured interview by psychiatrists during each visit.

This prospective study revealed that consistency of interests, one facet of grit, can be a novel protective factor of suicidality in the long term. Among the risk factors, the intensity of current pain and the severity of depressive symptoms were associated with suicidality over time, in the short term and long term, respectively. Our findings show that suicidality among patients with breast cancer are influenced by predictors on multiple levels, which appear to change over time. Thorough evaluation of patients with breast cancer on not only risk factors but also protective factors will provide a long-term prospect of their suicide risk. Longitudinal studies will be needed to determine whether intervention on the identified predictors can reduce suicidality.