Associations Between Doing Planned Exercise and Probable Major Depressive Disorder in Individuals Following Spinal Cord Injury

Abstract

Background: In neurologically healthy individuals, exercise positively impacts depressive symptoms, but there is limited knowledge regarding the association between exercise behaviors and depression after spinal cord injury (SCI). Objective: To examine associations between doing planned exercise and probable major depressive disorder (PMDD) after SCI. Methods: Community-dwelling adults, who were one or more years post traumatic SCI, completed self-report assessments at baseline (Time 1) and an average of 3.29 years later (Time 2). Patient Health Questionnaire-9 was used to assess depressive symptoms. Participants self-reported frequency of doing planned exercise. There were 1,790 participants who responded at both Time 1 and 2. Associations were analyzed using logistic regression. Results: Prevalence of PMDD was 10% at Time 1 and 12% at Time 2. Only 34% of participants at Time 1 and 29% at Time 2 reported doing planned exercise three or more times per week. The majority of participants (47%) reported no change in frequency of doing planned exercise between Times 1 and 2. Significant risk factors for PMDD at Time 2 included low household income (p = .0085), poor to fair self-perceived health (p < .0001), and doing less planned exercise at Time 2 (p = .0005). Meanwhile, number of years post injury (p = .04), doing planned exercise three or more times per week at Time 1 (p = .0042), and doing more planned exercise at Time 2 (p = .0005) were associated with decreased odds of PMDD at Time 2. Conclusion: These results demonstrate that a negative association exists between doing planned exercise and PMDD post SCI. Future longitudinal studies are needed to further explain these findings.

It is estimated that 276,000 people in the United States currently live with a spinal cord injury (SCI), and 12,500 new cases occur each year.¹ In addition to motor and sensory loss, secondary conditions, including psychological disorders (eg, major depressive disorder [MDD]), are also common. Within the general population, 7.1% of adults reported experiencing a major depressive episode in the past year.² Comparatively, the estimated prevalence of MDD post SCI varies from 15% to 30% of community dwellers to 30% of individuals undergoing rehabilitation.^3–5 Similar to the general population, depressive symptoms predict reductions in leisure activities,⁶ spending more time in bed,⁷ and increased health care utilization.^8,9 Following SCI, MDD is also associated with poorer subjective health, increased difficulty with self-care, and lower satisfaction with life⁴ as well as a number of negative medical sequelae including pressure ulcers, urinary tract infections,¹⁰ and higher mortality.¹¹

According to the American Psychiatric Association, standard treatment options for MDD include psychotherapy, antidepressants, and brain stimulation, but exercise is also strongly encouraged.¹² Some research has suggested exercise has comparable efficacy to psychotherapy and antidepressants in improving depressive symptoms among those with MDD.¹³ Within the neurologically healthy population, meeting the established health guidelines for physical exercise (30 minutes at moderate to vigorous intensity, 3 to 5 days/week) resulted in the greatest improvement of depressive symptoms.^13–16

The mechanism by which exercise improves depressive symptoms remains undetermined, but as with the pathogenesis of MDD, it is assumed to be multifactorial.^17–19 Some biological theories include hormones, neurotrophic factors, inflammatory biomarkers, cerebral vascular changes, increased synaptic plasticity, improved hippocampal function, and promotion of neuroplasticity.^17,18 Psychological theories include increased self-efficacy, feelings of accomplishment, and changes in self-concept.¹⁹ Furthermore, individuals could potentially experience an increase in social interaction and support through exercise.

Although exercise is an effective treatment option for MDD, there is added value in studying its potency as a protective factor for MDD. Physical activity has been shown to reduce the odds of future depression^20–22 among different age groups and across geographic regions.²² In their 11-year prospective study, Harvey et al found that if all participants had completed 1 hour of physical activity each week, regardless of intensity, 12% of future cases of depression could have been prevented.²¹ Conversely, the withdrawal of exercise has been linked to an increase in depressive symptoms and anxiety, and exercise withdrawal periods of 2 weeks or longer were associated with a significant decline in mental health.²³

While the results of these studies are promising regarding the impact of exercise on MDD, the external validity to populations with disabilities is limited, as they are often excluded. One prospective study found that physical activity was protective for both prevalent and incident depression among community dwellers with and without disability,²⁴ which demonstrates exercise might have an impact on depression among those with physical disabilities. To date, few studies have explored the association between physical activity and depression post SCI, and these studies are limited by categorizing exercise frequency as a dichotomous variable,²⁵ analyzing depressive symptoms as a continuous measure,²⁶ and using a cross-sectional design.²⁶

Additionally, the population with SCI is frequently cited as being sedentary.^27–29 The resulting physical impairment(s) from an SCI may significantly restrict the extent to which individuals may exercise, particularly with higher level injuries. This further underscores the need to identify how exercise is associated with MDD post SCI.

The purpose of this study was to evaluate the association between self-reported frequency of doing planned exercise and probable MDD (PMDD) in community-dwelling individuals with chronic SCI. We also sought to determine whether changes in frequency of doing planned exercise over time are associated with PMDD.

Methods

Participants

Data were collected as a part of an ongoing study assessing secondary outcomes post SCI. Participants were recruited from the SCI Model Systems patient database, registry, and outpatient directory. Inclusion criteria were (1) minimum of 18 years old at time of assessment, (2) at least 1 year post injury at enrollment, and (3) residual impairments resulting from traumatic SCI. Participants completed mail-in self-report assessments at two time points separated by an average of just over 3 years (3.29 ± 0.66 years).

Procedures

After receiving approval from the Institutional Review Board, initial letters were sent describing the purpose of the study. Four to 6 weeks after the initial mailing, self-report assessments were mailed. Additional mailings and follow-up phone calls were made to nonresponders as reminders. A third mailing was used for those who consented during a follow-up phone call but misplaced the previous mailing(s). This method was used for both Time 1 and Time 2 data collections. Only individuals who participated at both Time 1 and Time 2 were included.

Measures

Demographics

Demographic and injury characteristics were collected on all participants (Table 1). To account for possible differences in physical impairments that could serve as barriers to physical activity, three categories were established based upon self-reported level of injury and ability to ambulate: (1) cervical (C)1–C4 level, nonambulatory; (2) C5 level and below, nonambulatory; and (3) any level, ambulatory. Ability to ambulate was identified by “yes/no” response to the question, “Are you able to walk?” Annual household income was self-reported by selecting one of 10 categories ranging from “less than $10,000” to “$150,000 or more.” The 10 categories were reduced to two for analysis: (1) low annual household income (less than $20,000) and (2) other ($20,000 to $150,000 or more).

Table 1.

Demographics and characteristics of study participants at baseline (Time 1) and an average of 3.29 years later (Time 2) (N = 1,790)

Demographics	Time 1	Time 2
	Total n (%)	Total n (%)
Sex
Male	1,324 (73.97)	–-
Female	466 (26.03)	–-
Race/ethnicity
Non-Hispanic white	480 (26.82)	–-
Other	1,310 (73.18)	–-
Injury severity
C1–C4, nonambulatory	175 (9.88)	–-
C5 level and below, nonambulatory	1,034 (58.36)	–-
Any level, ambulatory	563 (31.77)	–-
Annual household income
Low income (less than $20,000)	489 (29.84)
Other income ($20,000 to $150,000+)	1,150 (70.16)
Probable major depressive disorder	172 (10.18)	200 (11.72)
	Mean ± SD (range)	Mean ± SD (range)
Age at spinal cord injury	33 ± 14 (1–81)	–-
Years post-spinal cord injury	13 ± 10 (1–70)	–-
PHQ-9 total score	5.6 ± 5.7 (0–27)	6.0 ± 5.6 (0–27)

Note: C = cervical; PHQ-9 = Patient Health Questionnaire-9.

Self-perceived health status

Participants were asked, “In general would you say your health is_____?” and selected “poor,” “fair,” “good,” “very good,” or “excellent.” For the analysis, similar categories were combined: (1) poor to fair, (2) other (good to excellent).

PMDD

Depressive symptoms were assessed using the Patient Health-Questionnaire-9 (PHQ 9), which is a self-report measure based on the Diagnostic and Statistical Manual of Medical Disorders, 4th edition for major depressive disorders.³⁰ The PHQ-9 can detect changes in depressive symptoms in a population of medical outpatients³¹ and has demonstrated strong internal consistency, Cronbach α (0.87), in those with SCI.⁴ The PHQ-9 asks individuals to indicate the frequency at which they have been “bothered by” nine different depressive symptoms “over the last two weeks”: 0 (not at all), 1 (several days), 2 (more than half the days), or 3 (nearly every day). The total score ranges from 0 to 27. For this study, an individual was identified as having PMDD when five of the nine depressive symptoms were scored ≥2 and one of the five symptoms was either anhedonia (question 1) or depressed mood (question 2). Question 9, “Thoughts of being better off dead or of hurting yourself in some way,” was considered one of the five depressive symptoms if it was present at all (scored ≥1). Individuals with missing data on the PHQ-9 at one or both time points were included in the analysis, as they contributed useful information regarding exercise variables.

Doing planned exercise

Participants were asked, “How often do you do planned exercise? (Lifting weights, swimming, or pushing your chair for the sake of exercise. Do not include passive range of motion—when somebody moves your arms or legs for you).” Frequency was self-reported by selecting one of six categories ranging from “rarely” to “five or more times per week.” For the analysis, the six categories were condensed to three: (1) once a month or less, (2) twice a month to twice a week, (3) three times a week or more.

Change in doing planned exercise

To determine whether the self-reported frequency of doing planned exercise changed over time, Time 2 answers were compared to Time 1. Participants were identified as fitting into one of three categories: (1) doing less planned exercise, (2) no change, (3) doing more planned exercise.

Statistical analysis

Combination of frequencies, percentages, and means were used to report demographic information and describe study variables. All statistical analyses were completed using SAS 9.4 (Cary, NC).

To determine risk factors associated with PMDD at Time 2, a logistic regression was run including factors described in the Methods section: annual household income, self-perceived health status, frequency of doing planned exercise at Time 1, and change in reported frequency of doing planned exercise between the two assessments.

Results

Sample characteristics

At Time 1, 2,547 of 3,629 people (70%) responded; of those, 1,790 (70%) responded at Time 2. Therefore, a total of 1,790 participants were included in the sample. The majority were male (74%). The average age at time of injury was 33 years (±14 years), and at Time 1 the average time since SCI was 13 years (±10 years).

Prevalence of PMDD

Within the sample, 10% had PMDD at Time 1 and 12% at Time 2. Among the participants who had PMDD at Time 1 and completed the PHQ-9 at Time 2, 45% still had PMDD at Time 2, indicating a resolution of PMDD for 55% of the participants between Time 1 and Time 2 (Figure 1). Of those who had PMDD at Time 2 and also completed the PHQ-9 at Time 1, 61% appeared to have a new episode of PMDD at Time 2. Among 1,617 participants who had complete PHQ-9 data at both Time 1 and 2, the majority of participants (1,346; 83%) did not have PMDD at either time point.

Figure 1.

Illustration of the determination of the prevalence of probable major depressive disorder (PMDD) at (a) baseline (Time 1) and (b) an average of 3.29 years later (Time 2) in the sample.

Frequency of doing planned exercise at Time 1 and Time 2

At Time 1 and 2, a large proportion of participants reported rarely doing planned exercise, 37% and 44%, respectively (Table 2). At the other extreme, 15% (Time 1) and 11% (Time 2) of participants reported doing planned exercise five or more times per week. At both time points, the categories with the lowest reported frequencies were once a month followed by two to three times per month.

Table 2.

Self-reported frequency of doing planned exercise at baseline (Time 1) and an average of 3.29 years later (Time 2).

	Time 1		Time 2
	n	%	n	%
Rarely	650	36.91	781	44.40
Once a month	58	3.29	43	2.44
2–3 times per month	134	7.61	139	7.90
1–2 times per week	323	18.34	287	16.32
3–4 times per week	331	18.80	309	17.57
5+ times per week	265	15.05	200	11.37

Change in frequency of doing planned exercise

When comparing the frequency of doing planned exercise at Time 2 to Time 1, the largest percentage of participants (47%) demonstrated no change followed by doing less planned exercise (32%) as shown in Table 3. Only 21% of participants reported doing more planned exercise at Time 2.

Table 3.

Change in frequency of doing planned exercise between Time 2 (average of 3.29 years post baseline) and Time 1 (baseline).

	n	%
Doing less planned exercise	557	32.09
No change	809	46.60
Doing more planned exercise	370	21.31

Factors associated with PMDD at Time 2

The two demographic variables significantly associated with PMDD at Time 2 were years post injury (odds ratio [OR] 0.98; 95% CI, 0.96–1.00; p= .04) and low annual household income (OR 1.61; 95% CI, 1.13–2.30; p = .0085) (Table 4). Variables associated with injury severity were not significant. Those with poor to fair self-perceived health status had significantly higher odds of PMDD at Time 2 (OR 3.46; 95% CI, 2.43–4.92; p < .0001). Participants who did planned exercise three times a week or more at Time 1 had significantly lower odds of PMDD at Time 2 (OR 0.47; 95% CI, 0.29–0.77; p = .0042). Also, compared to those who had no change in the amount of planned exercise they did, those who did more planned exercise at Time 2 had significantly lower odds of PMDD (OR 0.54; 95% CI, 0.33–0.88; p = .0005) while those who did less planned exercise had significantly higher odds of PMDD (OR 1.61; 95% CI, 1.03–2.50; p = .0005).

Table 4.

Logistic regression model for probable major depressive disorder at an average of 3.29 years (Time 2) from baseline (Time 1)

	Odds ratio	95% CI	p
Age at spinal cord injury	1.00	0.99–1.02	.75
Years post spinal cord injury	0.98	0.96–1.00	.04*
Male (ref = Female)	1.06	0.73–1.56	.75
Non-Hispanic White (ref = others)	1.06	0.73–1.55	.76
Injury severity (ref = C1–C4 level, nonambulatory)
C5 level and below, nonambulatory	1.22	0.65–2.30	.54
Any level, ambulatory	1.53	0.79–2.97	.21
Low annual household income (ref = others)	1.61	1.13–2.30	.0085*
Poor to fair self-perceived health status (ref = others)	3.46	2.43–4.92	<.0001*
Change in self-reported exercise frequency (ref = no change)
Doing less planned exercise	1.61	1.03–2.50	.0005*
Doing more planned exercise	0.54	0.330.88	.0005*
Frequency of doing planned exercise at Time 1 (ref = once a month or less)
Twice a month to twice a week	0.72	0.45–1.15	.82
Three time a week or more	0.47	0.29–0.77	.0042*

Note: C = cervical; CI = confidence interval.

^*p < .05.

Discussion

This study provides insight into associations between the frequency of participants doing planned exercise and PMDD post SCI. We found that 10% of participants had PMDD at Time 1 and 12% at Time 2. While these values are greater than the prevalence of depression within the general population,² they are lower than what some studies have reported within the SCI population. A systematic review found that an estimated 30% of individuals post SCI undergoing rehabilitation and 27% of community dwellers have depression, but the review included multiple outcome measures for depression with variable cutoff points.³ We used more stringent criteria for identifying PMDD compared to other studies, where a total PHQ-9 score of ≥ 10 is used as a cutoff.^5,26,32 We identified two studies that used the same stringent diagnostic criteria in community dwellers post SCI, and they reported similar MDD prevalence of 9.7% to11.9%.^4,33

With respect to the longitudinal assessments, 83% of our sample did not have PMDD at either time point. While the diagnostic criteria for PMDD was different, our findings are similar to a 5-year study by Hoffman et al that found 70% of the sample did not have depression at either time point.⁵ Hoffman et al also reported that 61.7% had clinically significant improvements in depression from Time 1 to Time 2,⁵ whereas we found that 55% no longer met our criteria for PMDD at Time 2.

Because our focus is exploring the relationship between doing planned exercise and PMDD post SCI, it is important to discuss the reported exercise frequencies. Within the general population, only 22.9% of US adults met guidelines for both aerobic and strengthening exercise, 32.4% met one guideline, and 44.7% did not meet either.³⁴ Although the general population is not physically active, the presence of comorbid condition(s) appears to be associated with even less activity. Among those with MDD, 67.8% did not meet exercise guidelines.^19,35 Within the SCI population, 44% to 50% reported not engaging in any physical activity,^27–29 and our results were similar. When comparing reported exercise frequencies to SCI-specific exercise guidelines, one study found that 36% met guidelines for aerobic exercise, only 19% met guidelines for resistance training, and only 12% met both.²⁸

Specifically, within the SCI population, sex, age, years post injury, injury severity, and primary mode of mobility have each been identified as predictors of physical activity,²⁷ whereas primary mode of mobility and autonomous motivation were associated with meeting SCI specific exercise guidelines.²⁸ Frequently reported barriers or facilitators to exercise post SCI, intrapersonal or intrinsic factors (eg, motivation), access to resources, knowledge, environmental limitations, and even the physical body^36,37 could also impact one's ability to exercise, as can the presence of an additional medical diagnosis.

When considering the nonexercise factors significantly associated with PMDD at Time 2, low annual household income and poor to fair self-perceived health status were associated with increased odds of PMDD while number of years post SCI was associated with decreased odds. The literature is inconclusive with respect to associations between depression and demographics and injury severity post SCI. Previously suggested risk factors associated with depression include older age,^38,39 female gender,⁴⁰ ethnic minority,³⁸ incomplete injury,⁴¹ and tetraplegia⁴⁰; other studies have reported no associations with demographics or injury variables.⁴

Low household income has previously been associated with an increased risk of depression within the general population⁴² and post SCI.³⁸ In addition, unemployment has been associated with depression post SCI.^26,33 Income is entwined in psychosocial and socioeconomic variables, making it difficult explain this association. Within the general population, there are two proposed theories: social causation (ie, stress and adversity associated with low income increases risk for mental illness) and social selection (ie, downward spiral associated with genetic predisposition for mental illness).^42,43 Although the theory of social causation is leading in regard to depression, a definitive causation has not been established.^42,43 When considering these theories with respect to our study, there is the additional confounding factor of SCI that is not included in either theory.

As for poor to fair self-perceived health status, there are known associations between mental and physical health, but interestingly it has been reported that past physical health has an effect on present mental health.⁴⁴ This effect possibly explains our result that poor to fair self-perceived health status at Time 1 (past physical health) was associated with greater odds of PMDD at Time 2 (present mental health). Years post SCI was associated with decreased odds of PMDD at Time 2. Prior studies have reported increased odds of depression among those with recent injury.^26,45 While not evidence based, it is possible that increased time since injury might enable a person to establish coping skills, acceptance, social support, and even roles within and outside of the home compared to an individual with a more acute injury, thereby decreasing odds of PMDD.

When exploring the exercise variables significantly associated with PMDD at Time 2, it is important to remember that even though exercise has been shown to reduce the odds of future depression^20–22 and improve depressive symptoms,^13–16 there is a bidirectional relationship between physical activity and mental health.⁴⁶ For example, the association between decreased odds of PMDD at Time 2 and doing planned exercise three or more times per week at Time 1 could be interpreted a number of ways: (1) doing planned exercise at Time 1 could have offered some protective benefits from PMDD at Time 2; (2) doing planned exercise three or more times per week could have decreased depressive symptoms; or (3) those who did planned exercised three or more times per week at Time 1 had fewer depressive symptoms and therefore frequently did planned exercise. The same holds true for our variable regarding change in frequency of doing planned exercise between Time 2 and Time 1. A change in frequency of doing planned exercise (increase or decrease) could have impacted depressive symptoms as exercise can improve depressive symptoms^13–16 and exercise withdrawal has been associated with increased depressive symptoms.²³ Alternatively, change in depressive symptoms (eg, mood or motivation) could have led an individual to do more or less planned exercise.

It would be naïve to attribute a change in frequency of doing planned exercise to only PMDD. One study that examined leisure time physical activity trajectories over an 18-month period found that injury severity, age, and years post injury significantly impacted activity trajectories, and pressure sores were associated with a decline in reported physical activity.²⁹ Additionally, any of the frequently reported barriers or facilitators to exercise post SCI presented above^36,37 could contribute to a change in doing planned exercise.

Methodological considerations

Even though our study is longitudinal, direct causality of doing planned exercise on the presence of PMDD cannot be assumed. Participants were only assessed at two time points; therefore, it is not possible to account for anything that transpired between these time points, including experiencing depressive episodes that could have resolved or receiving treatment for depression. Participants did not report whether they were prescribed antidepressants or received treatment(s) for depression (eg, psychotherapy or brain stimulation); therefore, it was not possible to control for these variables. Participants could have been receiving effective treatment for depression and not experiencing depressive symptoms at assessment. It was not possible to account for all variables known to be associated with depression post SCI, such as pain,^5,47–49 substance abuse^5,50 and even self-efficacy.²⁶ All data were self-report.

Future research

Our self-report assessments only explored how frequently participants did planned physical exercise. We did not account for the duration or intensity at which participants exercised, the type of exercise, or baseline fitness levels. Although it is not known what exercise parameters most effectively protect against depression, Dunn et al demonstrated a dose-response relationship between exercise and depression in individuals without SCI, where engaging in aerobic exercise at 17.5 kcal/kg/week for 12 weeks significantly reduced depressive symptoms in those with mild to moderate MDD.¹⁶ To develop a more detailed understanding of the association between MDD and physical activity among individuals following SCI, future intervention studies should investigate specific exercise variables, including frequency, duration, and intensity, using different types of physical exercise.

Conclusion

This study explored self-reported frequency of doing planned exercise and its relationship with PMDD post SCI. Our findings demonstrate that the prevalence of PMDD among community-dwelling individuals post SCI is greater than the general population. Furthermore, the frequency of participants doing planned exercise was low; over the study period, the majority either did not change or decreased the amount of planned exercise they did. Risk factors that increased odds of PMDD included low annual household income, poor to fair self-perceived health status, and doing less planned exercise at Time 2, whereas number of years post injury, doing more planned exercise at Time 2, and doing planned exercise at least three times per week at Time 1 decreased the odds of PMDD. These findings suggest that exercise could have an impact on PMDD post SCI.