Suicidal Ideation in Multiple Sclerosis

Methods 

Results 

Risk Markers of Any Suicidal Ideation 

Bivariate logistic regression analyses revealed significant relations between suicidal ideation and several variables. Suicidal ideation was less likely among people who drove at least occasionally (odds ratio [OR]=.59; 95% CI, .39–.89), had been diagnosed longer (OR=.97; 95% CI, .96–.99) were older (OR=.98; 95% CI, .97–.99), and reported more perceived social support (OR=.99; 95% CI, .98–.99). There was a nonsignificant trend suggesting lower risk of suicide among those who were married (OR=.68; 95% CI, 0.44–1.02). Suicidal ideation was significantly more likely among people who had a low income (OR=2.27; 95% CI, 1.48–3.48), progressive disease (OR=1.67; 95% CI, 1.08–2.57); and had more physical disability in the domains of mobility (OR=1.14; 95% CI, 1.02–1.28), bowel function (OR=1.46; 95% CI, 1.26–1.71), and bladder function (OR=1.35; 95% CI, 1.18–1.54). Finally, suicidal ideation was significantly more likely among people who reported higher depression severity scores (OR=1.26; 95% CI, 1.20–1.33). There were no significant bivariate relations between suicidal ideation and sex, race, living alone, reporting falls, being unemployed, alcohol use, and current use of an interferon-based disease-modifying therapy. Consequently, these latter variables were omitted from the final multivariate model.

In the fully adjusted multivariate model, bowel-functioning disability (OR=1.28; 95% CI, 1.02–1.62) and depression severity (OR=1.23; 95% CI, 1.16–1.30) remained the only variables independently associated with suicidal ideation (see table 1).

Table 1. Logistic Regression Examining the Relation Between Patient Variables and Suicidal Ideation or Persistent Suicidal Ideation During the Past 2 Weeks

	Characteristics	Mean ± SD	%	Suicidal Ideation∥ (OR [95% CI])		Persistent Suicidal Ideation∥ (OR [95% CI])
				Unadjusted Bivariate Analyses	Fully Adjusted for Significant Bivariate Predictors	Unadjusted Bivariate Analyses	Fully Adjusted for Significant Bivariate Predictors
	Age	55.1±11.8		0.98(0.97–0.99)#	NS	NS	NS
	Sex (male)		86.4	NS	—	NS	—
	Race (white)		92.1	NS	—	NS	—
	Live alone		18.5	NS	—	NS	—
	Married		62.7	0.68(0.44–1.02)¶	NS	0.47(0.23–0.94)#	NS
	Income <$22,000		37.6	2.27(1.48–3.48)††	NS	2.37(1.16–4.88)⁎⁎	NS
	Drive (at least occasionally)		64.5	0.59(0.39–0.89)#	NS	0.54(0.27–0.81)⁎⁎	NS
	Fall (at least occasionally)		81.9	NS	—	NS	—
	Unemployed		85.5	NS	—	NS	—
	Any alcohol use		44.0	NS	—	NS	—
	Progressive MS subtype		59.4	1.67(1.08–2.57)#	NS	3.39(1.37–8.38)⁎⁎	NS
	Disease-modifying therapy now⁎		35.3	NS	—	NS	—
	Years with diagnosis	18.1±12.5		0.97(0.96–0.99)⁎⁎	NS	0.97(0.93–0.99)#	NS
	Bowel disability (incontinence)†	1.7±1.4		1.46(1.26–1.71)††	1.28(1.02–1.62)⁎	1.61(1.27–2.05)††	NS
	Bladder disability (incontinence)†	2.4±1.6		1.35(1.18–1.54)††	NS	1.37(1.09–1.72)††	NS
	Mobility disability†	4.0±1.9		1.14(1.02–1.28)#	NS	1.32(1.06–1.65)⁎⁎	NS
	Social support‡	69.6±27.9		0.99(0.98–0.99)††	NS	0.98(0.97–0.99)⁎⁎	NS
	PHQ depression severity§	9.4±5.9		1.26(1.20–1.33)††	1.23(1.16–1.30)††	1.40(1.27–1.53)††	1.27(1.18–1.37)††

NOTE. The n range is 426 to 445 for univariate models. N equals 398 for fully adjusted model. Descriptive variables (means, percentages) are based on the total sample of N equal to 451 where available. Abbreviation: NS, not significant.

⁎ Current use of an interferon-based disease-modifying therapy. † Perceived disability by using adapted performance scales from Patient Determined Disease Steps measure of the NARCOMS Registry Survey. ‡ Modified Social Support Survey total score. § Depression symptom severity score of the PHQ without suicidal ideation. ∥ Suicidal ideation item of the PHQ. ¶ P<.06. # P<.05. ⁎⁎ P<.01. †† P<.001.

Discussion 

Suicidal ideation within the past 2 weeks was endorsed by 29.4% of VHA patients with MS, with 7.9% reporting it persistently on more than half the days. Consistent with our hypotheses and the broader literature on suicide, depression severity was the single most important factor associated with thoughts of suicide because it remained the only measured variable significantly and uniquely associated with both suicidal ideation and persistent suicidal ideation in all multivariate analyses. However, depression screening alone was insufficient to identify 1 out of every 3 patients who endorsed recent suicidal ideation. We conclude that depression screening in similar samples should include an explicit item about suicidal ideation.

Although it is difficult to compare prevalence across different assessment periods, the level of recent suicidal ideation observed in the present sample is similar to the 28.6% lifetime estimate recently reported by Feinstein20 in a predominantly female sample of subjects with MS. The prevalence rate in our sample also substantially exceeds the general population past-year and lifetime estimates (3.3% and 13.5%, respectively) obtained from the U.S. National Comorbidity Survey,19 similar levels found in its subsequent replication,18 and the 6.6%-past 2-week rate obtained from a large recent sample of VHA primary care patients.33

Unadjusted bivariate analyses suggest that suicidal ideation in this sample was more prominent early in the course of the disease among people who were more recently diagnosed. These results are consistent with the epidemiologic literature on completed suicides in MS, which suggests that young, recently diagnosed males may represent a group at particularly high risk.12, 14

We also found suicidal ideation to be more common among people with a progressive MS subtype, a result that has been replicated in at least 1 other sample.34 Given the observation in this study and others that there may be a period of risk early in diagnosis, this correlation may seem contradictory. It is likely, however, that suicidal ideation is complex and subject to multiple influences that may differ over time. For example, hopelessness, a phenomenon strongly associated with suicide risk,35 has been shown to be more common among people with progressive MS and to increase over the course of the disease.36 Surprisingly, suicidal ideation was not more common among men and whites, as frequently seen in other samples.10, 14, 37

Suicidal ideation in this sample was also associated with lower levels of perceived social support and, for general suicidal ideation, not being married. Both markers of social isolation have been shown to be linked with suicidal ideation in the general population.20, 38 Unexpectedly, no relation between living alone and suicidal ideation was observed.

The association between lower income and suicidal ideation found here has not been previously reported in people with MS. It is not surprising, however, given the observation that economic difficulties may represent a particularly salient risk for suicide more broadly, particularly for men.39, 40 Contrary to expectation, no relation between current employment and suicidal ideation was observed, although this variable was unable to capture the disruption associated with recent job loss.

Three specific aspects of physical disability, decreased mobility, decreased bowel function, and decreased bladder function, were all associated with an increased likelihood for suicidal ideation. This finding differs from Feinstein’s20 largely female sample in which there was no relation between global physical disability and suicidal ideation and epidemiologic literature on completed suicides in MS in which subjects were more likely to be mildly or moderately disabled.10, 14 Decreased bowel functioning appears to be a particularly salient problem, possibly because it often appears in the context of broader disability and because of its potential impact on participation in social and recreational activities.

Despite concern that interferon-based disease-modifying agents may increase the risk of suicide,41, 42 we found no relation between current use of interferon (β-1a or interferon β-1b) and suicidal ideation. Similarly, alcohol use (vs abstinence) assessed by a single item was not associated with increased suicidal ideation in MS, although at least one previous study20 suggests that this relation may be observed with more detailed assessment of the quantity, frequency, and consequences of alcohol use.

Although it cannot be definitively established with the current data, it is likely that many disease, disability, and psychosocial variables identified in this study are associated with suicidal ideation via their common link with depression.43

It is noteworthy that only 2 out of every 3 people with suicidal ideation endorsed levels of depression that could be detected with a routine screen, although 88.6% of those with persistent suicidal ideation screened positive on a 2-item depression screen. This screening performance contrasts sharply with a recent survey of VHA primary care patients in which similar screening methods detected 92.5% of people with suicidal ideation in the past 2 weeks.33 The USPSTF concluded that even when screening for suicide can be conducted with reasonable sensitivity and specificity in primary care settings, low prevalence of suicidal ideation and large numbers of false-positives requiring further assessment may not justify routine screening.44 Unlike general medical patients, routine screening may be more appropriate among people with MS who appear in significantly smaller numbers and carry substantially greater risk for suicide.

Identification of suicidal ideation is particularly important considering there are several common and efficacious treatments for depression in MS. In addition to the use of traditional tricyclic, selective serotonin reuptake inhibitors and atypical antidepressant medications,45, 46 cognitive behavioral therapies delivered in group47, 48 and individual formats46, 49 have been shown to reduce symptoms of depression in MS. Recognizing that people with MS often experience significant barriers to care delivery, empirically validated psychotherapy treatments are increasingly available via telemedicine.50

Despite the relatively large sample size and broad community representation, several limitations of the current study are worth noting. All data are cross-sectional precluding causal interpretations. The modest response rate (43.7%) may have introduced bias as people with less social support or greater impairment (eg, cognition, vision, mobility) may have been less likely to complete the survey. Respondents’ demographic characteristics were typical of other samples of VHA patients with MS, but results may not generalize to nonrespondents or to people with MS not receiving care in the VHA. The small size of some subsamples (eg, women, nonwhites) may have provided limited power to detect differences. Also, many demographic variables do not allow for important quantitative work and qualitative differences (eg, employment characteristics like hours or location of employment in or out of the home). Overlapping symptoms (eg, fatigue) may have made it difficult to distinguish some depressive symptoms from symptoms of MS. Additionally, suicidal ideation was assessed with a single item from the PHQ-9, an instrument that has well established psychometric properties and is widely used but has not been specifically validated for use in MS. Persistent suicidal ideation was evaluated in a relatively narrow time frame of the 2 weeks before assessment and was based on a single assessment point. Finally, although suicidal ideation is necessarily based on self-report, all other data on risk factors were collected via self report without corroboration from other sources and may be influenced by factors such as social desirability.

Conclusions 

Suicidal ideation is extremely common among people with MS and is associated most significantly with depression but also with younger age, earlier disease course, progressive disease subtype, lower income, social isolation, not driving, and higher levels of physical disability. Minimal screening for depression alone failed to detect nearly one third of people with suicidal ideation in the present sample. Current findings support the importance of routine assessment by using brief screening measures that explicitly address suicidal ideation, such as the 9-item PHQ. Future research on suicide in MS should clarify factors associated with the progression from suicidal ideation to active suicidal intent and plan and ultimately to suicide attempt. To date, research in MS has focused predominantly on completed suicide as an outcome, with minimal exploration of suicidal ideation and no examination of the temporal, biologic, and psychosocial variables that might represent important opportunities for intervention.