The benefits of making peace with pain: chronic pain acceptance moderates the indirect effect of perceived burdensomeness between pain severity and suicidal cognitions

Abstract

Objective

The purpose of this study was (1) to examine the degree to which perceived burdensomeness mediates the relationship between pain severity and suicidal cognitions and (2) to determine whether this mediated relationship was moderated by pain acceptance. We predicted that high levels of pain acceptance would buffer relationships on both paths of the indirect effect.

Methods

Two-hundred seven patients with chronic pain completed an anonymous self-report battery of measures, including the Chronic Pain Acceptance Questionnaire, the Interpersonal Needs Questionnaire, the Suicidal Cognitions Scale, and the pain severity subscale of the West Haven–Yale Multidimensional Pain Inventory. Conditional process models were examined with Mplus.

Results

Chronic pain acceptance significantly moderated both paths of the mediation model. Results from the conditional indirect effect model indicated that the indirect effect was significant for those with low (b = 2.50, P = .004) and medium (b = 0.99, P = .01) but not high (b = 0.08, P = .68) levels of pain acceptance and became progressively stronger as pain acceptance scores decreased. The nonlinear indirect effect became nonsignificant at acceptance scores 0.38 standard deviation above the mean—a clinically attainable treatment target.

Conclusions

Higher acceptance mitigated the relationship between pain severity and perceived burdensomeness and the relationship between perceived burdensomeness and suicidal cognitions in this clinical sample of patients experiencing chronic pain. Findings indicate that any improvement in pain acceptance can be beneficial, and they provide clinicians with a clinical cut-point that might indicate lower vs higher suicide risk.

Suicide is the tenth leading cause of death for adults in the United States,¹ and as such, decreasing suicide risk is a prominent public health issue. Several competing theoretical models of suicide have articulated antecedents, correlates, and underlying mechanisms of suicide ideation and attempts.^2–7 These models emphasize that suicidal thoughts and actions emerge from a complex network of demographic, psychological, social, and neurobiological factors. Despite the recent proliferation of suicide-related research, rates of suicide ideation and attempts continue to rise, particularly in vulnerable populations.⁸ Thus, identifying psychosocial variables that act as risk and protective factors in these populations remains imperative.

Relative to civilians, US service members and veterans exhibit especially high rates of suicide ideation, attempts, and deaths, even after adjustment for related demographic and psychological variables.⁹ For example, the average number of veteran deaths by suicide per day was between 17 and 18 from 2005 to 2018, despite decreases in the total number of military personnel. In response to this crisis, the US Department of Defense and the US Department of Veteran Affairs continue to prioritize intervention for suicidal thoughts and behaviors as a top clinical priority. Military members and veterans who experience chronic pain are especially vulnerable to suicide risk; in the 2016 National Health Interview Survey, the frequency of chronic pain was 25% higher in veterans than in the general US population.¹⁰ Previous research has shown that both the experience of chronic pain and pain catastrophizing are associated with higher levels of suicide ideation^11,12 and suicidal behaviors.^13–15 That said, suicide risk is not a foregone conclusion for individuals with chronic pain, which makes it difficult to predict with certainty who might be at risk in this vulnerable population and highlights the need to better understand mediators and moderators of this relationship. Toward that end, the goals of the present study are (1) to examine perceived burdensomeness as a mechanism underlying the relationship between chronic pain and suicide cognitions and (2) to determine the degree to which chronic pain acceptance moderates this mechanism. Because these constructs are clinically malleable, addressing these questions might influence the development of intervention programs that could be used to decrease suicide risk in this population.

Perceived burdensomeness, one of 3 facets in the Interpersonal–Psychological Theory of Suicide, has been defined as “an individual’s belief that they are a burden to themselves and/or to others, and that other people would benefit more from the death, versus the continued life, of the individual.”¹⁶ Perceived burdensomeness is linked to higher suicide risk in military personnel.¹⁷ Perceptions of burdensomeness predict suicide ideation above and beyond other theoretically related factors, such as depressive symptomology, feelings of hopelessness, and functional impairments,¹⁸ and have been linked to undesirable or poor health issues. Perceived burdensomeness is also associated with reports of pain intensity in patients who suffer from chronic pain.¹⁹ This relationship could be attributable to feelings of shame associated with perceptions that individuals with chronic pain have created hardship for those who are close to them by requiring the daily direct assistance of others in mitigating their pain symptoms or by having their loved ones watch them suffer.²⁰ This might be particularly true when the pain condition is viewed as permanent and stable.¹⁸ Additionally, perceived burdensomeness has been associated with suicidal cognitions in both civilian and military samples.^6,21–23 Thus, perceived burdensomeness might mediate the relationship between pain severity and suicide cognitions.²⁴

To the extent that perceived burdensomeness mediates the relationship between pain severity and suicidal cognitions, it is imperative to identify modifiable factors that might serve as moderators at each stage of the indirect effect. One possible candidate is chronic pain acceptance, which is defined as “the ability to engage in activities that are meaningful towards an individual’s life goals, even if they include experiencing pain, as well as the ability to refrain from attempts to reduce that pain.”^25–27 Notably, pain acceptance is negatively associated with suicidal cognitions in populations with chronic pain, with pain acceptance accounting for a significant portion of the variance in suicidal cognitions, over and above relevant covariates.^28,29 Chronic pain acceptance could act as a buffer against suicidal cognitions in patients with chronic pain, as high levels of pain acceptance might counteract the negative effects of phenomena related to experiencing chronic pain.³⁰

To our knowledge, no studies have examined the effects of pain severity, perceived burdensomeness, and chronic pain acceptance on suicide-related outcomes comprehensively within a single conditional process model. Examining these variables in such a framework is crucial to improving our understanding of the complex relationship between pain severity and suicidal cognitions, which could have important clinical implications. We predicted that the data would show a significant indirect effect in which pain severity would predict perceived burdensomeness (a-path), which, in turn, would predict suicide cognitions (b-path). We also predicted that pain acceptance would moderate the a-path and b-path in this model. Specifically, with respect to the a-path, we predicted that the relationship between pain severity and perceived burdensomeness would be stronger at lower levels of pain acceptance and that the relationship would be fully buffered at very high levels of pain acceptance. The relationship between perceived burdensomeness and suicidal cognitions was also predicted to be similarly buffered by high levels of pain acceptance. We hypothesized that the relationship between perceived burdensomeness and suicide ideation would be stronger at lower levels of pain acceptance and that said relationship would be fully buffered at very high levels of pain acceptance. As a function of these moderated relationships, we also predicted that the conditional indirect effect would be significant only for individuals with low to moderate levels of pain acceptance. The hypothesized model is presented in Figure 1.

Figure 1.

Hypothesized conditional process model.

Methods

Participants

Participants were recruited to participate in a pain study via advertisements and clinician referral from 3 different clinics on a military base in the southwestern United States. Participants were at least 18 years of age and were required to be seeking treatment for a chronic pain condition lasting 4 months or longer. Participants completed a battery of standardized measures that were administered in electronic format at a computer kiosk or via pen-and-paper surveys in the waiting area.

Data were collected from N = 207 participants (56.1% male, 39.5% female, 4.4% unknown). The majority (64.5%) of participants were active-duty military (92.1% Army, 7.2% Air Force, and 0.7% Navy); the remainder were military retirees or family members. With regard to current marital status, 68% were married, 7.5% single, 6.6% dating or engaged, 12.7% separated or divorced, and 0.9% widowed. More than 40% of the participants’ pain conditions were secondary to an injury (24.9% work related, 20% not work related); the other half were related to a medical condition (eg, having undergone surgery) or other cause. Low back pain (25.9%), lower-extremity pain (15.4%), and multisite pain (14.0%) were the most commonly reported ailments. Surgery to address chronic pain issues was reported by 32.7% of the sample, and 45.3% of the sample reported work limitations due to chronic pain that forced them to seek qualitatively different jobs, as compared with 52.4% of the sample who reported doing work similar to what they had done before the onset of their chronic pain. The average chronic pain duration per patient was 3.84 years.

Measures

Pain severity

The West Haven–Yale Multidimensional Pain Inventory (MPI), Version 2, is a 52-item self-report scale that assesses 12 different dimensions of pain.³¹ The MPI has been widely used in studies of chronic pain. For the present study, the Pain Severity scale of the MPI was used to assess the participants’ subjective pain ratings. Its 3 items are scored on a 0-to-6 Likert scale, with higher mean scores indicating more severe chronic pain (mean = 3.47, SD = 1.43, range = 0–6). Items include “Rate the level of your pain at the present moment,” “On the average, how severe has your pain been during the last week?,” and “How much suffering do you experience because of your pain?” The Pain Severity scale exhibited good internal consistency in the present study (α = 0.76).

Perceived burdensomeness

The Interpersonal Needs Questionnaire (INQ-PB) is an 18-item measure with 2 subscales.³² In the present study, the Perceived Burdensomeness scale was used to evaluate the extent of the participant’s perceptions that he or she is a burden on others. Items are scored on a 7-point Likert scale ranging from 1 to 7, with higher scores indicating higher levels of perceived burdensomeness (mean = 16.05, SD = 9.84, range = 9–57). The INQ-PB scale exhibited satisfactory reliability (α = 0.87).

Suicidal beliefs

The Suicide Cognitions Scale (SCS) is an 18-item self-report scale that assesses cognitions and beliefs commonly expressed by suicidal individuals.³³ Items are scored on a 5-point scale ranging from 1 (disagree strongly) to 5 (agree strongly). The SCS was used as our measure of current suicide risk (mean = 24.68, SD = 11.71, range = 17–76) because it has been found to accurately differentiate patients who have attempted suicide from those who have not across a wide range of psychiatric and medical settings, and it also predicts future behaviors even when accounting for suicidal ideation. The SCS exhibited excellent internal consistency in the present sample (α = 0.98).

Chronic pain acceptance

The 20-item Chronic Pain Acceptance Questionnaire (CPAQ) assesses the degree to which persistent pain and related experiences influence behaviors and the degree of effort put into controlling pain.³⁴ The CPAQ has acceptable psychometrics and uses a 7-point Likert scale ranging from 0 (never true) to 6 (always true), with higher scores indicating higher levels of pain acceptance (mean = 64.49, SD = 20.34, range = 10–115). The CPAQ exhibited good internal consistency in the present study (α = 0.86).

Data analytic plan

All analyses reported here were conducted in Mplus version 8.³⁵ To test our first hypothesis, bias-corrected bootstrapping was used to conduct a test of simple mediation. To examine the second and third hypotheses, we examined conditional main effect models of (1) pain severity on perceived burdensomeness and (2) perceived burdensomeness on suicide cognitions (controlling for pain severity) across the full range of pain acceptance scores with the Johnson–Neyman regions-of-significance technique. Doing so allowed us to determine (1) the exact values of pain acceptance at which pain severity and perceived burdensomeness predict their respective outcomes, (2) the size of the relationship between pain severity and perceived burdensomeness at all possible values of pain acceptance, and (3) the proportion of individuals in the sample whose pain acceptance score falls into the region of significance. The key advantage of this method is that it examines conditional main effects at all levels of the moderator, as opposed to arbitrarily chosen values that can sometimes produce misleading inferences if they are just outside the region of significance. Additionally, this method provides key information about the proportion of individuals for whom the relationship between the variables under consideration is significant by identifying the tipping point at which the relationship becomes significant. This information is important, as it provides a target for clinicians to aim for when working with patients. Finally, we examined the full conditional process model illustrated in Figure 1, and we examined the conditional indirect effect at point estimates representing low, medium, and high levels of pain acceptance, consistent with values at the mean of the pain acceptance variable for medium levels and ±1 standard deviation around the mean for low and high levels. Planned contrasts were used to conduct formal tests of the difference between the indirect effects at these levels.

Given that many of our variables were significantly skewed, and to account for any possible missing data, we ran all models with the both the Maximum Likelihood (ML) and Maximum Likelihood Robust (MLR) estimators, using bootstrapped confidence intervals for all effects. MLR is a robust estimator that adjusts standard errors to correct for non-normality but does not allow for bootstrapping. Bootstrapping allows for nonsymmetric confidence intervals, which could be important for parameter estimates such as indirect effects, which can often have non-normal sampling distributions.³⁶ The parameter estimates and associated substantive conclusions were identical for both sets of analyses. Given that our focus was on estimation of the conditional indirect effects, we report the results of the ML with bootstrapping results.

Results

Preliminary analyses

Before we examined models, it was important to determine the degree to which our sample was similar to other samples in the extant literature. Our sample’s overall mean SCS score was 24.68. In the psychometric article describing the SCS,³³ they reported on 2 samples. The first sample, consisting of Army patients who were discharged from inpatient psychiatric hospitalization for acute suicidal ideation and/or a suicide attempt, had an extrapolated mean of 50.20; given the nature of this sample, we were not surprised at the difference between their mean and ours. The second sample, consisting of US Air Force personnel in an outpatient mental health clinic, had an extrapolated mean of 26.06, which is much more like our sample’s mean. For the second sample, they reported means for 4 groups within that sample based on their reported history: no ideation (68.46%), non-suicidal self-injury (8.05%), suicide ideation (16.11%), and suicide attempter (7.38%) groups, with extrapolated means of 22.78, 23.91, 32.70, and 45.19 for each group, respectively. In our sample, 14 people (6.76%) had scores of 46 or higher, and 11 people (5.32%) had scores between 33 and 46. Overall, our sample appears comparable to Sample 2 from Bryan et al.³³ in terms of the proportion of at-risk-for-suicide patients based on a comparison of mean scores.

Tests of mediation

Hypothesis 1 predicted that perceived burdensomeness would mediate the relationship between pain severity and suicide cognitions. This prediction was supported. There was a significant total effect of pain severity on suicide cognitions (b = 3.41, t(205) = 5.30, P < .001). However, this effect was largely mediated by perceived burdensomeness; in the mediation model, pain severity significantly predicted perceived burdensomeness (b = 3.04, t(205) = 6.17, P < .001), and perceived burdensomeness predicted suicide cognitions (b = 1.01, t(204) = 12.95, P < .001). Accordingly, the overall indirect effect was significant (b = 3.05, P < .001, 95% CI = 1.98–4.24), which indicates that the indirect effect mediated 89.31% of the total effect for the full sample. After accounting for the indirect effect, the direct effect of pain severity on suicide cognitions was not significant (b = 0.37, t(204) = 0.95, P = .34).

Tests of moderation

We next separately examined whether the a-path (Hypothesis 2) and b-path (Hypothesis 3) in the mediation model were moderated by pain acceptance. Results indicated that both hypotheses were supported, as the interaction effect was significant for both paths. With respect to the a-path, results indicated that pain acceptance significantly moderated the relationship between pain severity and perceived burdensomeness (b = –0.59, t(203) = –2.64, P = .008), with the interaction model accounting for an additional 4.4% more variance than a model with both main effects alone (R² = 0.32). We applied the Johnson–Neyman regions-of-significance method to determine values for which the relationships were significant in their uncentered metrics (see Figure 2, top panel). For the interaction between pain severity and chronic pain acceptance, the relationship between pain severity and perceived burdensomeness became significant at chronic pain acceptance values below 75.40; 68.32% of participants scored below this threshold, which was just slightly (0.53 standard deviations) above the average for the sample (mean = 64.49). These results indicate that pain severity has no association with perceived burdensomeness at higher levels of pain acceptance; however, at low to moderate levels of pain acceptance, perceived burdensomeness is significantly correlated with pain severity, and the relationship grows progressively stronger as pain acceptance decreases, consistent with our hypothesis.

Figure 2.

Johnson–Neyman plots of the conditional main effects for pain severity (top panel) and perceived burdensomeness (bottom panel) at varying levels of pain acceptance. Conditional main effects are significant for areas where 95% confidence bands do not include zero. CPAQ= Chronic Pain Acceptance Questionnaire.

The b-path of the model was also significantly moderated by pain acceptance (b = –0.42, t(202) = –2.34, P = .02), with the interaction model accounting for an additional 2.5% more variance than did a model with both main effects alone (R² = 0.77). When the Johnson–Neyman technique was applied, the relationship between perceived burdensomeness and suicide cognitions is significant at pain acceptance values below 86.50; 84.05% of participants scored below this threshold, which is (1.07 standard deviations) above the sample mean (see Figure 2, bottom panel). As with the a-path, the conditional main effect grows stronger as pain acceptance decreases, consistent with our predictions.

Conditional process modeling

Finally, we ran the full conditional process model illustrated in Figure 1 to address Hypothesis 4, which predicted that conditional indirect effect would be significant only for individuals with low to moderate levels of pain acceptance, consistent with our expectation of a buffering effect at higher levels of pain acceptance for both the a-path and b-path of the model. This hypothesis was also supported. Results of the model are summarized in Table 1. We examined the conditional indirect effect at the mean value of chronic pain acceptance, as well as values 1 standard deviation above and below the mean. The latter value was selected given that it was within the range of pain acceptance values where both interaction effects in Hypotheses 2 and 3 were fully buffered.

Table 1.

Conditional indirect effects at varying levels pain acceptance and planned contrasts

Indirect effect or contrast	Level(s) of pain acceptance	B	t	P	95% CI Lower	95% CI Upper
Indirect effect	Low	2.50	2.86	.004	0.77	4.25
Indirect effect	Medium	0.99	2.59	.010	0.29	1.81
Indirect effect	High	0.08	0.41	.679	–0.39	0.44
Contrast	Low vs medium	1.52	2.69	.007	0.45	2.62
Contrast	Low vs high	2.42	2.67	.008	0.69	4.33
Contrast	Medium vs high	0.91	2.44	.015	0.25	1.76

N = 207. Because the direct effect was the same for all 3 levels of pain acceptance, the contrasts are inferential tests of the differences between the respective groups that are accurate for both the indirect and total effects.

In the conditional process model, the direct effect was not significant (b = –0.237, t(202) = –0.64, P = .52). However, as can be seen in Table 1, the indirect effect was significant for those with low and medium, but not high, levels of pain acceptance, as expected. Planned contrasts showed large and significant differences in the indirect effect between those with estimated low pain acceptance and those with medium and high levels (Cohen’s d = 0.42 and 0.64, respectively). Additionally, the indirect effect was also significantly stronger for those with estimated medium levels than for those with higher levels (Cohen’s d = 0.38). It is important to note that high levels of pain acceptance seem to be fully protective; neither the direct nor indirect effect of pain severity on suicide cognitions was significant for those with high pain acceptance. For those with average and low levels of pain acceptance, perceived burdensomeness fully mediates the relationship between pain severity and suicide cognitions. We then graphed the conditional indirect effect across the entire range of CPAQ scores (see Figure 3) to determine exactly where the indirect effect becomes significant. Results showed that the 95% confidence interval for the indirect effect includes zero for pain acceptance scores above 72.50 (0.38 standard deviations above the mean); accordingly, the indirect effect is significant at values below this threshold, which 61.83% of participants scored equal to or below. As indicated in the results presented in Table 1, the indirect effect is substantially larger at lower than at increasingly higher values of pain acceptance.

Figure 3.

Plot of the conditional indirect effect across all values of the moderator. The conditional indirect effect is significant for areas where the 95% confidence bands do not include zero. CPAQ= Chronic Pain Acceptance Questionnaire.

The indirect effect moderation results show significant differences in the indirect effect for the individuals at specific point estimates representing the mean and at values ±1 standard deviation from the mean. To further probe these results to better understand and contextualize the relationship among these variables across the entire sample, we trichotomized CPAQ scores and conducted a series of ANOVAs on each variable in the mediation model, specifying a priori repeated contrasts for each variable (ie, low vs medium and medium vs high). In terms of the percentage of the sample that met these thresholds, 17.30% of the sample could be categorized as either low or high on pain acceptance (based on being below or above 1 standard deviation away from the mean, respectively), and 65.30% of the sample could be categorized as having medium levels of pain acceptance (based on being between ±1 standard deviation away from the mean). Results are summarized in Table 2. As can be seen, there were significant differences between the 3 groups on all 3 variables, with large effect sizes (F for all significant at P < .001, and η² values for all greater than 0.24), and for all contrasts, medium to large effect sizes were observed (Cohen’s d ranging from 0.62 to 1.49), with the largest differences being observed between the low and medium groups.

Table 2.

ANOVAs and contrast tests on mediation model variables between trichotomized CPAQ groups

Outcome or contrast	Low pain acceptance, mean ± SD	Medium pain acceptance, mean ± SD		High pain acceptance, mean ± SD	F (for ANOVAs) or t (for contrasts)	P	η2(for ANOVAs) or d (for contrasts)
Pain severity	4.87 ± 0.78	3.43 ± 1.28		2.17 ± 1.19	44.61	<.001	0.31
C1: Low vs medium	Diff = 1.44				6.30	<.001	1.19
C2: Medium vs high			Diff = 1.26		5.56	<.001	1.27
Perceived burdensomeness	26.15 ± 14.04	14.87 ± 7.78		10.47 ± 2.85	31.48	<.001	0.24
C1: Low vs medium	Diff = 11.28				6.77	<.001	1.20
C2: Medium vs high			Diff = 4.40		2.64	.009	0.99
Suicide cognitions	38.29 ± 17.21	22.48 ± 7.53		18.22 ± 1.87	41.29	<.001	0.34
C1: Low vs medium	Diff = 15.81				8.03	<.001	1.49
C2: Medium vs high			Diff = 4.26		2.19	.03	0.62

F values are based on robust estimates given differences in group variability where appropriate. Cohen’s d values use the pool standard deviation for the groups involved in the contrast.

Abbreviations: C= contrast; Diff= estimated difference between groups in noted contrast (and estimated differences are centered under the 2 groups in the comparison); SD= standard deviation.

Discussion

Given the rates of suicide ideation and attempts in both the general population and the military, it is becoming increasingly important to understand the various demographic, psychological, social, and neurobiological factors that predict suicidal thoughts and behaviors. These relationships among variables that give rise to such thoughts and behavior are complex, and unraveling them requires researchers to use increasingly complex and creative modeling strategies. This study is the first to examine how pain acceptance moderates the indirect effect of perceived burdensomeness on the relationship between pain severity and suicide cognitions in a conditional process model, and our results have important and clear clinical implications. Given that perceived burdensomeness and pain acceptance are clinically malleable variables, these findings add valuable insight on how to structure cognitive–behavioral interventions for individuals suffering from chronic pain.

Prior research has shown that both the experience of chronic pain and pain catastrophizing are associated with higher levels of suicide ideation^11,12 and behaviors,^13–15 though there are some inconsistent findings in the extant literature.^37,38 The total-effects model we examined showed that pain severity positively predicted suicide cognitions. However, the overwhelming majority (nearly 90%) of this effect was mediated by perceptions of burdensomeness, consistent with earlier findings linking pain severity to perceived burdensomeness¹⁹ and perceived burdensomeness to suicidal cognitions.^16,21–23 Given that perceptions of burdensomeness have been shown to be clinically malleable,³⁸ routes to reducing suicidal cognitions and downstream suicidal attempts and completions in patients with chronic pain could include (1) addressing suicidal cognitions, (2) employing interventions to bolster social support and social connectedness,³⁹ and (3) increasing pain acceptance.

Although we found the hypothesized mediation effect, our findings reveal that this was not a one-size-fits-all phenomenon. Chronic pain acceptance moderated both paths of the mediation model, buffering the relationships we predicted it would. First, pain acceptance moderated the relationship between pain severity and perceived burdensomeness such that this relationship was not significant for values of pain acceptance that were more than one-half a standard deviation above the mean of pain acceptance, a value at which a substantial proportion of the sample scored. Similarly, pain acceptance buffered the relationship between perceived burdensomeness and suicide cognitions such that this relationship was significant only at moderate to low values of pain acceptance (at about 1 standard deviation above the mean). These findings highlight the value of examining theoretically related moderators of mediated relationships. More importantly, they provide clinicians with a clinical cut-point that could indicate lower vs higher suicide risk. Well-validated interventions, such as Acceptance and Commitment Therapy, directly target chronic pain acceptance and thus could be exceptionally powerful in a high-risk population with chronic pain.

Formal tests of the conditional indirect effect mirrored the moderation of the individual paths and are even more interesting. Our results clearly indicate that the mediated relationship of pain severity and suicidal cognitions by perceived burdensomeness is dependent upon the degree to which the participant has made peace with pain. Looking across the full range of the construct, the mediated relationship is significant only at values of pain acceptance less than 0.38 standard deviation above the mean—again, a relatively attainable treatment target. Moreover, it is important to note that practically any movement on pain acceptance matters clinically; although the indirect effect was fully buffered at point estimates representing high levels of pain acceptance, the indirect effect was significantly lower for individuals at point estimates representing moderate, relative to low, levels of pain acceptance. This assertion is bolstered by the results showing that the trichotomized pain acceptance groups differed substantially across all 3 variables in the mediation model, with the largest differences being observed between individuals in the low vs medium groupings. Together, these findings suggest that clinicians should use therapeutic approaches that facilitate increased pain acceptance, particularly for individuals with high levels of chronic pain or perceived burdensomeness, as pain acceptance can potentially reduce perceptions of burdensomeness, which in turn, can reduce the degree to which perceptions of burdensomeness affect suicidal cognitions. Even modest gains in pain acceptance could result in less suicide ideation.

This study has limitations that impact the interpretation of our findings. First, the sample was drawn from a military population and was predominantly male, which limits the overall generalizability of the findings. Second, the data were cross-sectional, which limits our ability to draw causal conclusions. Future studies should use longitudinal data to investigate these questions, which might reveal different relationships from those we observed. Third, our data were all self-report measures, which might have inflated the observed associations because of common method variance; investigators seeking to further this work not only could examine these questions longitudinally but could also use experimental paradigms that would allow for true causal conclusions. Fourth, we examined only suicide cognitions; the degree to which these findings are related to suicide attempts and completions remains an open question, though it is reasonable to assert that reducing suicide cognitions should result in fewer suicide attempts and deaths by suicide. Additionally, the relationships examined reflect only the variables included in the models presented, which do not include other possible mediators, moderators, and confounding variables, which could affect our hypothesized indirect effects. Finally, given that everyone in this sample experienced chronic pain, this limits the generalizability of the conclusion, and future studies should seek to replicate and extend these findings in other samples.

Despite these limitations, our findings are extremely promising and provide clinicians with specific guidance to consider when treating patients with chronic pain. The theoretical applications are important, as well. Predominant models of suicide ideation, attempts, and deaths, such as the Interpersonal–Psychological Theory of Suicide, posit that suicidal behaviors require the coexistence of intractable feelings of perceived burdensomeness and thwarted belongingness. Our results highlight how pain acceptance could short-circuit the relationship between pain severity and perceived burdensomeness. Given that pain acceptance reflects the degree to which individuals engage in activities that are meaningful even if they are experiencing pain,^25–27 such acceptance might facilitate successful social connectivity and reduced thwarted belongingness simply because individuals with high pain acceptance are more likely to engage in activities with others, thereby making both thwarted belongingness and perceived burdensomeness seem less intractable. Bolstering pain acceptance would likely reduce the impact of both proximal and sufficient causes of passive suicidal ideation, as clearly indicated in our findings. Moreover, although theories such as the Fluid Vulnerability Theory of Suicide highlight the highly temporal nature of suicide risk, as factors such as the experience of pain wax and wane over time, we propose that providing this population with tools that help elevated pain severity seem less intractable could substantially decrease associated suicide risks.

Funding

This work was supported by the Department of Defense/Defense Health Program (D61_I_10_J5_148) and the National Center for Advancing Translational Sciences, National Institutes of Health (grant KL2 TR001118).

Conflicts of interest: WH, CJB, DDM, and KEK have received government grants / research support from the National Institutes of Health, Department of Defense, and Veterans Affairs. None of the authors have any conflicts of interest to report.