A Longitudinal Study of Depression From 1 to 5 Years After Spinal Cord Injury

8/13/2019 A Longitudinal Study of Depression From 1 to 5 Years After Spinal Cord Injury

http://slidepdf.com/reader/full/a-longitudinal-study-of-depression-from-1-to-5-years-after-spinal-cord-injury 1/8

ORIGINAL ARTICLE

A Longitudinal Study of Depression From 1 to 5 Years AfterSpinal Cord Injury

Jeanne M. Hoffman, PhD, Charles H. Bombardier, PhD, Daniel E. Graves, PhD,Claire Z. Kalpakjian, PhD, MS, James S. Krause, PhD

ABSTRACT. Hoffman JM, Bombardier CH, Graves DE,Kalpakjian CZ, Krause JS. A longitudinal study of depressionfrom 1 to 5 years after spinal cord injury. Arch Phys MedRehabil 2011;92:411-8.

Objective: To describe rates of probable major depressionand the development and improvement of depression and totest predictors of depression in a cohort of participants withspinal cord injury (SCI) assessed at 1 and 5 years after injury.

Design: Longitudinal cohort study.Setting: SCI Model System.Participants: Participants (N1035) who completed 1- and

5-year postinjury follow-up interviews from 2000 to 2009.Interventions: Not applicable.Main Outcome Measure: Probable major depression, de-

fined as Physician Health Questionnaire-9 score of 10 orhigher.

Results: Probable major depression was found in 21% of partic-ipants at year 1 and 18% at year 5. Similar numbers of participantshad improvement (25%) or worsening (20%) of symptoms overtime, with 8.7% depressed at both 1 and 5 years. Increased pain(odds ratio [OR], 1.10), worsening health status (OR, 1.39),and decreasing unsafe use of alcohol (vs no unsafe use of alcohol; OR, 2.95) are risk factors for the development of depression at 5 years. No predictors of improvement in depres-sion were found.

Conclusion: In this sample, probable major depression wasfound in 18% to 21% of participants 1 to 5 years after injury.To address this high prevalence, clinicians should use theserisk factors and ongoing systematic screening to identify thoseat risk for depression. Worsening health problems and lack of effective depression treatment in participants with SCI maycontribute to high rates of chronic or recurrent depression inthis population.

Key Words: Depression; Rehabilitation; Spinal cord inju-ries.

© 2011 by the American Congress of Rehabilitation Medicine

MAJOR DEPRESSION is a prevalent and highly disablingsecondary condition associated with SCI. The point prev-

alence of major depression typically is estimated to be in therange of 15% to 23%.1 This is substantially higher than thepoint prevalence of major depression in primary care settings(10%)2 or the 1-year prevalence of major depression in the U.S.population (6.6%).3 Depressive symptoms are linked to a hostof negative outcomes, including pressure ulcers and urinarytract infections,4 lower self-appraised health,5 fewer leisureactivities,6 poor community mobility, poor social integration,and fewer meaningful social pursuits.7,8 Depressive symptomseven predict mortality after controlling for other injury severity–and health-related variables.9

Although much is known about the point prevalence andcorrelates of major depression after SCI, far less is knownabout the longitudinal course of major depression in peoplewith SCI. Most longitudinal studies of depression reported totalgroup mean values or other descriptive data that did not showthe clinical course in the subgroup that was depressed.10-14 Of the longitudinal studies that provided data for the clinicalcourse of people who have become depressed, several empha-sized a high rate of recovery from depression after SCI, butfocused primarily on adjustment to injury within the first yearafter injury. Judd et al15 reported that 38% of people with SCIscored higher than 14 on the BDI at least once during inpatientrehabilitation, whereas 18% (47% of the depressed group)recovered before discharge from inpatient rehabilitation. Kishiet al16 reported that 21.7% of 60 inpatients with SCI met

DSM-III diagnostic criteria for major depression while hospi-talized, but half of these acute-onset depressive episodes hadremitted by 3 months. In a study based on International Clas-sification of Diseases–9th Revision–Clinical Modificationscodes from a population-based administrative data set in Al-berta, Canada, Dryden et al17 reported that depression wasdiagnosed by a physician in 34 of the 201 patients studied(17%) during the initial hospitalization. Of these, 50% discon-tinued depression treatment with the physician by discharge,implying that they had recovered.

Others found more lasting rates of depression. For example,Craig18 and Hancock 19 and colleagues reported that 30% of patients with SCI were significantly distressed (depressed andFrom the Department of Rehabilitation Medicine, University of Washington,

Seattle, WA (Hoffman, Bombardier); Department of Physical Medicine and Reha-bilitation Medicine, Baylor College of Medicine, Houston, TX (Graves); Department

of Physical Medicine and Rehabilitation Medicine, University of Michigan, AnnArbor, MI (Kalpakjian); and College of Health Professions, Medical University of South Carolina, Charleston, SC (Krause).

Supported by the Department of Education, National Institute on Disability andRehabilitation Research, SCI Model Systems: University of Washington (grant no.H133N060033), Baylor College of Medicine (grant no. H133N060003), University of Michigan (grant no. H133N060032), and Shepherd Model System (grant no.H133N060009).

No commercial party having a direct financial interest in the results of the researchsupporting this article has or will confer a benefit on the authors or on any organi-zation with which the authors are associated.

Correspondence to Jeanne Hoffman, PhD, Dept of Rehab ilitation, University of Washington, Box 356490, Seattle, WA 98195-6490, e-mail: [email protected]. Re-prints are not available from the author.

0003-9993/11/9203-00001$36.00/0doi:10.1016/j.apmr.2010.10.036

List of Abbreviations

AIS American Spinal Injury Association Impairment

Scale

BDI Beck Depression Inventory

CHART Craig Handicap Assessment and Reporting

Technique

DSM Diagnostic and Statistical Manual for Mental

Disorders

MDD major depressive disorder

PHQ-9 Physician Health Questionnaire-9

SCI spinal cord injury

411

Arch Phys Med Rehabil Vol 92, March 2011

mailto:[email protected]






anxious) during their initial hospital stay and remained dis-tressed through 1 to 2 years after SCI. Two treatment studiesalso emphasized that depression after SCI tends not to improve.Craig et al20 reported that 10 people who scored higher than 14on the BDI during acute rehabilitation and did not receive anearly cognitive-behavioral intervention continued to reportmoderate to severe depression through 12 months postinjury.Kemp et al21 described 15 persons with major depression whorefused a treatment trial. These people were 5 to 37 yearspost-SCI and an unspecified number had chronic depression. Inthis untreated group, depression severity scores were un-changed during a 24-week follow-up period.

Research on the course of depression after SCI is conflictingand incomplete. Moreover, the literature is limited in terms of the overall number of studies, small sample sizes, and relianceon administrative data, and there are even more sparse data forthe course of depression beyond 1 year after injury. It isimportant to improve our understanding of the course of de-pression after SCI because if depression after SCI is confinedto people who become depressed soon after injury and arechronically depressed, mental health care should focus on earlyidentification of at-risk persons and lifelong treatment focusedon that subgroup. If depression after SCI occurs in a larger

segment of the population who move in and out of depressiveepisodes, universal depression screening should occur as partof both acute and postacute care.

Because there is a paucity of data about the course of depression beyond the first year after SCI, we used the SCIModel System longitudinal data set to examine the course of depression from 1 to 5 years after SCI. Our research questionswere as follows. (1) What percentage of people with probablemajor depression at 1 year after SCI are recovered by 5 yearspostinjury? (2) What percentage of those with probable majordepression at 1 year have evidence of chronic or recurrentdepression at 5 years? (3) What percentage of people who arenondepressed at 1 year after SCI have probable major depres-sion at year 5? (4) What percentage of people with SCI do nothave probable major depression at either time? (5) What de-

mographic, injury-related, or other clinical variables predictclinically significant depression improvement or developmentof depression between 1 and 5 years postinjury?

Based on prior research suggesting that depression in thepostacute phase may be largely chronic20,21 and recent evi-dence that depression is undertreated,22 we hypothesized that ahigher proportion of people with SCI would be depressed atboth 1 and 5 years compared with the general population, forwhich the prevalence of depression is 2.7%.23 We also exam-ined potential risk and protective factors. On an a priori basis,we predicted that improvement in pain,24-26 substanceabuse,17,27,28 community integration,29,30 and subjective healthstatus1,31 would predict clinically significant improvement indepressive symptoms between 1 and 5 years after SCI. For thesame reasons, we predicted that worsening in each of these

areas would predict the development of depression from 1 to 5years after SCI.

METHODS

Participants

Participants in the present analysis are from 16 SCI ModelSystem centers located throughout the United States. The SCIModel System program, funded by the U.S. Department of Education, National Institute on Disability and RehabilitationResearch, has been in existence for more than 30 years. Allsites contribute to a uniform national data set with the goal of examining the course of recovery and outcomes of persons

after SCI. The study sample consisted of 1035 persons withtraumatic SCI who completed both 1- and 5-year follow-upinterviews between 2000 and 2009. This longitudinal studyfollows up people over time, and data were collected during the1- and 5-year follow-up assessments, which included thePHQ-9. Participants were 17 years or older, sustained a trau-matic SCI, were admitted to each model system within 1 yearof injury, completed inpatient rehabilitation, and discharge wasin the geographic catchment area of the system. All participantsprovided informed consent for data collection, and researchprotocols were approved by each model system’s local insti-tutional review board.

Because we were interested in examining change over time,only people who completed interviews at years 1 and 5 wereincluded in the present analysis. An additional 612 peoplecompleted year 1 but not year 5, and 160 people completedyear 5 but not year 1. Those who completed year 1 but not year5 were significantly younger (mean SD age, 34.114.7y;P.001), were more likely to be men (81%; P.002), andwere less likely to be white (67%; P.001), but had depressionscores similar to those in the present sample. Those whocompleted year 5 but not year 1 were similar in age and sex, butless likely to be white (66%; P.001) and had higher depres-

sion scores at year 5 than those in the present sample (mean

SD score, 5.675.77; P.031).

Measures

Demographic and injury-related characteristics were ob-tained from interviews, physical examinations, and hospitalrecords near the time of the person’s initial rehabilitationhospitalization.

Injury characteristics. Cause of injury was divided into 5categories: motor vehicle, violence, sports, fall, or pedestrian.Subjects were categorized into 2 broad levels of injury, tetra-plegia versus paraplegia. The AIS32 was used to classify injuryseverity into complete (AIS grade A) versus incomplete (AISgrades B–E).

Probable MDD. We used the PHQ-9 to identif y cases with

probable MDD and measure depression severity.2

The PHQ-9is based on the 9 symptom criteria of DSM-IV MDD.33 Each of the 9 depression items reflects the persistence of the symptomduring the past 2 weeks: 0 (not at all), 1 (several days), 2 (morethan half the days), or 3 (nearly every day). Item scores can besummed, resulting in a total score of 0 to 27. Prior researc h hasshown that the PHQ-9 conforms to a unidimensional scale34 oris characterized by 2 factors that are highly correlated (r .69–.73).14 Participants who score 10 or more on the PHQ-9 areclassified as having probable MDD, whereas those with scoresless than 10 are considered nondepressed. We chose this cutoff because for medical patients, a cutoff of 10 or higher yieldedthe best combination of sensitivity (88%) and specificity (88%)compared with an independent DSM-IV diagnosis of majordepression made by a mental health professional.35 Addition-

ally, in a large SCI survey study, this cutoff score resulted in adepression prevalence rate consistent with results derived fromSCI studies using structured diagnostic interviews.1 We baseddepression severity categories on the ranges proposed in theoriginal PHQ-9 validity study: 0 to 4 indicates minimal; 5 to 9,mild; 10 to 14, moderate; 15 to 19, moderately severe; and 20or higher, severe.35 Accordingly, those with probable MDD atyear 1 were considered remitted if they scored in the minimalrange (0–4) at year 5. Based on widely used clinical guidelines,we defined clinically significant improvement as a decreaseof5ormore points on PHQ-9 score from year 1 to year 5.36

Subjective health status. A single item from the 36-ItemShort Form Health Survey was used to assess subjective health

412 SPINAL CORD INJURY AND DEPRESSION, Hoffman




status.37 Participants were asked to indicate their current healthstatus from excellent (1) to poor (5) on a 5-point scale. Changein health status was calculated by subtracting year 1 ratingsfrom year 5 ratings, such that a higher score would indicateworsening health status.

Pain. We used a numerical rating scale to measure currentpain level. Participants rated their level of pain from 0 (no pain)to 10 (pain so severe you could not stand it). Change in pain

was calculated by subtracting year 1 ratings from year 5 rat-ings, with higher scores indicating increasing pain levels.

Community integration. The 19-item CHART Short Formwas used to measure this domain. Higher scores correspond togreater participation in each domain (mobility, occupation,social integration, physical integration). The original CHARThas high test-retest correlations (.93 for the total CHART,.80–.95 for the subscales) and validity established throughcorrelations with independent observers (.83).38,39 Change ineach domain was calculated by subtracting year 1 scoresfrom year 5 scores, with higher scores indicating greaterparticipation.

Unsafe alcohol use. Our measure of substance abuse wasunsafe alcohol use. Participants were asked about the number

of occasions of drinking per week during the past month,average number of drinks consumed per occasion, and fre-quency of binge drinking (5 drinks/occasion) in the pastmonth. Based on the National Institute on Alcohol Abuse andAlcoholism standards for safe drinking, participants were clas-sified as having unsafe alcohol use if they were women andconsumed more than 7 drinks a week, were men and consumedmore than 14 drinks a week, or reported any episodes of bingedrinking during the past month.40 Change in unsafe alcohol usewas coded as follows: 0 indicates no unsafe alcohol use ateither time; 1, unsafe alcohol use at year 1 and no unsafealcohol use at year 5; 2, no unsafe alcohol use at year 1 andunsafe alcohol use at year 5; and 3, unsafe alcohol use at bothtimes.

Statistical AnalysesDescriptive statistics provide information about the popula-

tion. Two logistic regression models were used to examine therelationship between the change in our a priori predictor vari-ables (pain, health status, community integration, unsafe alco-hol use) and outcomes after controlling for demographic vari-ables (age, sex, race, cause and level of impairment). In the firstregression equation, we predicted clinically significant im-provement in depression severity (decrease 5 points) at year5 in the group that had probable MDD at year 1. With thesecond regression equation, we predicted the development of MDD by year 5 in the group that was nondepressed at year 1.Additional tests were conducted to determine whether site(nesting) had an influence or variables predicted to affectdepression correlated highly (tests of intraclass correlations forboth year 1 and year 5 PHQ scores were .00, suggesting norelationship by site). In addition, correlations among variablesof interest used in the logistic regression were less than .15,except for the correlation between change in mobility andchange in occupation, which was .29, reducing concerns formulticollinearity.

RESULTS

Demographic data for the 1035 participants are listed in table1. Participants on average were aged 37 years, 74.5% weremen, and 77.2% were white. Most injuries occurred in motorvehicle collisions (54.8%) or falls (23.5%).

Prevalence and Natural History of Probable Major

Depression

Percentages of the sample with probable MDD were 20.6%at year 1 and 18.1% at year 5. Most (N724) participants(70.0%) scored in the nondepressed range at both times,whereas 90 (8.7%) reported probable MDD on both occasions(chronically depressed). Of those not depressed at 1 year, 97(11.8%) met criteria for probable MDD at 5 years. Of those

with probable MDD at year 1, at 5 years, 124 (57.9%) scoredin the nondepressed range, 135 (63.1%) experienced a clini-cally significant improvement in depression severity, and 71(6.9%) were in remission from probable MDD. Table 2 listsadditional detail about how depression severity changed overtime by using the 5 ranges of PHQ score severity.

Rates of probable MDD did not vary significantly based onsex at year 1. However, at year 5, the rate of probable MDDwas significantly greater in women than men (23% of womenvs 16% of men; P.05).

Predictors of Clinically Significant Improvement in

Depression or Development of Depression

The first logistic regression predicting clinically significantimprovement in depression from year 1 to year 5 showed no

significant predictors from those chosen a priori (improvementin pain, community integration, alcohol use, health status).Results are listed in table 3. The second logistic regressionmodel to determine possible protective or risk factors fordeveloping depression between year 1 and year 5 is listed intable 4. Results from the second model suggest that increasingpain, declining health status, and decrease in unsafe alcohol usewere risk factors for the development of depression. For ex-ample, a participant who has decreased unsafe alcohol use isalmost 3 times more likely to become depressed than someonewho has never engaged in unsafe alcohol use. In addition,being African American was a protective factor compared withbeing white for the development of depression. African Amer-

Table 1: Demographics

Variable At Enrollment

Age (y) 37.114.8

Sex: men 74.5

Race

White 77.2

African American 18.1

Other 4.7

Ethnicity: non-Hispanic 91.8 (N1033)Cause

Motor vehicle 54.8 (N1004)

Fall 23.5

Violence 11.0

Sports 9.7

Pedestrian 1.1

Injury severity (N1012)

Paraplegia incomplete 17.8

Paraplegia complete 28.2

Tetraplegia incomplete 35.9

Tetraplegia complete 18.0

Minimal deficit 0.2

Education: high school 81.2 (N1005)

Married 40.4 (N1034)

NOTE. Values expressed as mean SD or % of total. Total N whenless than 1035 is noted.

413SPINAL CORD INJURY AND DEPRESSION, Hoffman




icans were 2.7 times less likely than whites to become de-pressed at either 1 or 5 years postinjury.

DISCUSSION

This study provides a detailed examination of depressivesymptoms at 1 and 5 years after SCI and potential risk andprotective factors for the development of depression. Begin-ning with the most positive findings, these data show that70% of the sample scored in the nondepressed range at both

1 and 5 years after SCI. Other longitudinal studies also haveemphasized that most people with SCI are characterized byresilience, not depression.15-17

Next, although prevalences of depression at years 1 and 5were similar, there was considerable change in depressionstatus over time. More than half of those depressed at year 1were not depressed at year 5, and almost 12% of those notdepressed at year 1 were depressed by year 5. Somewhat moreparticipants became nondepressed (n124) than became de-pressed (n97) from 1 to 5 years. This was reflected in thesmall overall decrease in depression prevalence (21% to 18%)over time. The finding that 12% developed probable MDDbetween 1 and 5 years suggests that ongoing screening fordepression is needed in postacute and long-term care settings.

These longitudinal data show for the first time that a signif-icant minority of participants with SCI (8.7%) may have eitherchronic or recurrent depression from 1 to 5 years postinjury.Epidemiologic data indicate that only 2.7% of the general adultpopulation report chronic depressive symptoms.23 High rates of chronic or recurrent depression in people with SCI may beattributable to several factors. Major depression appears to beboth underrecognized and undertreated in people with SCI.22

Potential side effects (eg, sexual dysfunction or increased spas-

ticity41) and poor tolerability may make physicians or patientswith SCI reluctant to use antidepressants.42 Another possibilityis that people with SCI may be less responsive to sta ndardtreatment. Characteristics such as medical comorbidities,43 co-morbid traumatic brain injury,44 bereavement-like reactions,45

low socioeconomic status,46 low activities of daily living sta-tus,47 and poor social support47 are associated with poor re-sponse to depression treatment and are common in people withSCI. Finally, pain after SCI is common and may influencedepression onset and treatment.48 The lack of well-controlledefficacy studies of treatment of major depression in people withSCI also may contribute to inadequate treatment and morechronic depression. Studies of antidepressant efficacy and tol-erability are needed in persons with SCI, as are studies of

Table 2: Depression Severity at 1 and 5 Years After SCI

Year 1 PHQ-9 Score Ranges

Year 5 PHQ-9 Score Ranges

0–4 5–9 10–14 15–19 20–27 Totals

0–4 462 (44.6) 87 (8.4) 39 (3.8) 17 (1.6) 5 (0.5) 610 (58.9)

5–9 115 (11.1) 60 (5.8) 25 (2.4) 8 (0.8) 3 (0.3) 211 (20.4)

10–14 45 (4.3) 30 (2.9) 28 (2.7) 12 (1.2) 9 (0.9) 124 (12.0)

15–19 19 (1.8) 12 (1.2) 16 (1.5) 6 (0.6) 4 (0.4) 57 (5.5)

20–27 7 (0.7) 11 (1.1) 9 (0.9) 4 (0.4) 2 (0.2) 33 (3.2)

Totals 648 (62.6) 200 (19.3) 117 (11.3) 47 (4.5) 23 (2.2) 1035 (100)

NOTE. Values expressed as N (% of total).

Table 3: Logistic Regression: Prediction of Clinically Significant Improvement in Depression From Year 1 to 5

Variable SE Significance Odds Ratio 95% Confidence Interval

Age at injury 0.001 0.017 .974 1.001 0.97–1.03

Sex .213 0.386 .581 0.808 0.38–1.72

Race 0.069

African American (vs white) 1.160 0.582 0.046 3.189 1.02–9.98

All other races (vs white) 1.268 1.077 0.239 0.281 0.03–2.32

Hispanic (yes vs no) 0.190 0.592 0.748 1.209 0.38–3.86

Cause of injury 0.982

Level and severity of SCI 0.062

Paraplegia complete (vs paraplegia incomplete) 0.111 0.520 0.830 0.895 0.32–2.48

Tetraplegia incomplete (vs paraplegia incomplete) 0.530 0.523 0.311 1.699 0.61–4.73

Tetraplegia complete (vs paraplegia incomplete) 1.338 0.641 0.037 3.810 1.08–13.38Change in health status (higher scores indicate declining health) 0.186 0.157 0.238 0.831 0.61–1.13

Change in pain (higher scores indicate higher pain) 0.082 0.065 0.211 0.922 0.81–1.05

Change in physical independence* 0.002 0.006 0.720 1.002 0.99–1.01

Change in mobility* 0.003 0.007 0.646 1.003 0.99–1.02

Change in occupation* 0.002 0.005 0.660 0.998 0.99–1.01

Change in social integration* 0.009 0.008 0.273 1.009 0.99–1.02

Change in unsafe alcohol use† 0.590

Constant 0.209 0.823 0.800 1.232

NOTE. All change scores calculated as year 5 level year 1 level.*Measured by using the CHART Short Form (higher scores indicate better integration).†No unsafe alcohol use at either time (0) was compared with 1, unsafe alcohol use at year 1 and no unsafe alcohol use at year 5; 2, no unsafealcohol use at year 1 and unsafe alcohol use at year 5; and 3, unsafe alcohol use at both times.





psychotherapeutic interventions and combined interventions.49

Multimodal treatment may be more effective in cases charac-terized by chronic depression.50

Contrary to our hypotheses, we were unable to predict whowould have clinically significant improvement in depressionover time. Although we have tended to think that treatingassociated conditions, such as pain, poor health, and poorcommunity integration, will have a positive impact on depres-sive symptoms in the postacute setting, these data suggest thatprobable MDD is a more independent condition. This may

mean that probable MDD requires specific treatments thattarget depression directly. This interpretation is consistent withthe findings of Kemp et al,21 who showed that without specificevidence-based depression treatment, people with SCI andMDD may remain depressed over an extended period. We donot have data for treatment use and therefore are unable todetermine the extent to which treatment received in the interimperiod contributed to depression improvement. In addition,because depression and grief symptoms also may overlap, onecould speculate that a decrease in depression symptoms overtime may reflect resolution of grief. Research is needed todefine grief processes in people with SCI and disentangle grief from depression.

In contrast, we found that increasing pain, decreasing healthstatus, and decreasing unsafe alcohol use were risk factors for

the development of depression at year 5 in those not depressedat year 1. These results are consistent with other research thatdocumented the presence of later developing depression afterSCI17 and the association between pain and depression afterSCI.48 One clinical implication is that health care providersshould be attentive to the possibility that worsening pain orhealth may be associated with the development of probableMDD in the postacute setting. Surprisingly, a decrease inunsafe alcohol use also may represent a risk for worseningdepression. Although this may seem paradoxical, alcohol abusemay represent a means of coping with the stresses of SCI.Giving up alcohol abuse as a habitual coping strategy does notmean that the person has learned alternative strategies to cope

with the stresses of loss or disability. Cessation of unhealthyalcohol use also may unmask the presence of a depressivedisorder. Alternatively, cessation of alcohol use may be asso-ciated with switching to more harmful drug abuse or loss of social support (“drinking buddies” or social outlets being al-cohol establishments). However, we have no data to verifythese potential explanations. These results highlight the needfor depression treatment and coping skills training as elementsof any plan to treat alcohol abuse in the aftermath of SCI.

We found that African American race was a protective factor

for the development of depression at 5 years post-SCI. Incontrast, a large cross-sectional study of people almost 10 yearspost-SCI showed that African American men were at signifi-cantly greater risk (odds ratio, 1.43) of probable major depres-sion compared with white men, although the relationship wasincompletely mediated by lower education and income in theminority sample.51 More research is needed about the potentialrelationship of race or ethnicity to depression incidence andremission in longitudinal studies, especially given that thestudy sample had lower rates of minority representation thanthe group that did not complete 1 of the 2 follow-up interviews.

Study Limitations

Important study limitations should be mentioned as caveats

to the interpretation of these data and guides for future re-search. Probable MDD was assessed by using a depressionscreening measure, not a structured diagnostic interview. ThePHQ-9 has good sensitivity and specificity compared withdiagnostic interviews in primary care patients.35 However, thePHQ-9 lacks rigorous validity testing in people with SCI and avariety of alternative scoring methods have been suggested. Acutoff score of 10 or higher was used to indicate probableMDD because this cutoff has the most validity data to supportit,35 and depression prevalence estimates derived from thiscutoff more closely approximate the estimated prevalence of MDD from diagnostic studies.1 More research is needed tovalidate depression screening measures in people with SCI.52 If

Table 4: Logistic Regression: Prediction of Becoming Depressed at Year 5 if Not Depressed at Year 1

Variable SE Significance Odds Ratio 95% Confidence Interval

Age at injury 0.007 0.010 .457 1.007 0.99–1.03

Sex 0.369 0.288 .200 1.446 0.82–2.54

Race* .032*

African American (vs white)* 1.001 0.488 .040* 0.369* 0.14–0.96*

All other races (vs white)* 0.769 0.534 .150 2.157 0.76–6.14

Hispanic (yes vs no) 0.485 0.456 .288 1.624 0.66–3.97

Cause of injury .427Level and severity of injury .675

Change in health status (higher score indicates worsening health)* 0.327 0.125 .009* 1.386* 1.09–1.77*

Change in pain (higher scores indicate higher pain)* 0.094 0.044 .031* 1.099* 1.01–1.20*

Change in physical independence† 0.000 0.004 .947 1.000 0.99–1.01

Change in mobility†0.004 0.006 .499 0.996 0.98–1.01

Change in occupation†0.005 0.004 .259 0.995 0.99–1.00

Change in social integration†0.006 0.006 .292 0.994 0.98–1.00

Change in unsafe alcohol use* .033*

Reducing unsafe use (vs no unsafe use)* 1.080 0.426 .011* 2.946* 1.28–6.79*

Beginning unsafe use (vs no unsafe use)* 0.395 0.433 .385 1.469 0.62–3.50

Continued unsafe use (vs no unsafe use)* 1.260 1.042 .226 0.284 0.04–2.18

Constant 2.597 0.523 .000 0.076

NOTE. All change scores calculated as year 5 level – year 1 level.

*Significant relationships.†Measured by using the CHART Short Form (higher scores indicate better integration).





feasible, efforts to replicate this study should use valid diag-nostic interviews for MDD.

This study does not address the validity of DSM-IV symp-toms of MDD in people with SCI. Somatic symptoms, such asfatigue, insomnia, and poor appetite, may be attributable tophysical conditions associated with SCI or to MDD, althoughmeasurement of somatic items may be the most problematicduring inpatient rehabilitation.53 Including these symptoms

may spuriously inflate rates of depression, whereas excludingthese symptoms may lead to underdiagnosis and undertreat-ment. We chose to include somatic symptoms because thisinclusive approach is considered a reliable and valid means of diagnosing depression in people with a wide variety of comor-bid physical illnesses.54

Our approach and ability to predict depression were limited.For example, information for preinjury history of depressionmay have considerable prognostic value.17 However, such in-formation is not included in the SCI Model System database.Certain symptoms of depression might predict future depres-sion. Krause et al53,55,56 have examined differences betweensomatic and nonsomatic symptoms on the PHQ-9 completedduring inpatient rehabilitation and at approximately 17 and 29

months after injury. They found a distinctive somatic factorduring inpatient rehabilitation53 that was not predictive of either somatic or nonsomatic symptoms at 17 or 29 monthspostinjury.55,56 However, analysis of the predictive validity of specific symptoms was beyond the scope of this report.

The SCI Model Systems are not population based and there-fore cannot be assumed to be representative of the full SCIpopulation. All possible participants were not included becausesome people did not complete follow-up questionnaires andthere were differences between those who completed the 2follow-up interviews and those who completed only 1. There-fore, all absolute estimates of the prevalence of depressivedisorders must be interpreted with caution, particularly as weattempt to generalize beyond the SCI Model System.

There are limitations associated with the fact that assess-

ments of probable MDD occurred approximately 4 years apart.Major depressive episodes could have developed and fullyremitted during the interim period. Therefore, these data mayunderrepresent the total burden of probable MDD during the4-year period. The design also prevents us from distinguishingrecurrent from chronic depression in those depressed at bothtimes. Future researchers should consider reassessing depres-sion more frequently and using valid retrospective reportingmethods.57

More research is needed on subgroups that develop depres-sion soon after SCI versus in later years because depression inthese different subgroups may have distinct causes and mayvary in response to treatment.16 The prevalence of late devel-oping depression makes it incumbent on health care providersto continue screening for major depression as part of routine

follow-up for at least the first 5 years after SCI and probablybeyond.

This study collected no data about the use of antidepressantsor psychotherapy. Health services research on the prevalenceand efficacy of mental health treatment in persons with SCI islacking. More information about depression treatment underusual-care conditions can aid in the planning of more effectivemental health services for people with SCI. Future researchshould include information about mental health treatment use,depression treatment preferences, treatment accessibility, bar-riers to mental health treatment, and ways to overcome thosebarriers. Future research also should examine the possibledepressive effects of commonly used medications (antispasm,

antiseizure, opioids) and drugs of abuse (cocaine, amphet-amines) in people with SCI.

In addition, future research should examine additionalrisk factors for chronic, recurrent, or treatment-resistantdepression after SCI and evaluate methods for treating thismost concerning subgroup. Although those with a priorhistory of major depression are at risk for later depression,there are people who develop major depression by 5 years

despite being nondepressed at 1 year. Characteristics of latedeveloping depression also need to be elucidated. Finally,interventions designed to maintain subjective health andprevent increases in pain may have the effect of preventingthe onset of depressive symptoms. The impact of healthmaintenance and pain management on depression onset maybe a fruitful area of research.

CONCLUSIONS

Prior cross-sectional depression prevalence studies have notbeen able to elucidate the course of depression in people withSCI. This study shows that the 18% of the sample with prob-able MDD at 5 years postinjury consists of half with chronic orrecurrent depression and half with depression not present at 1

year post-SCI. The presence of both new-onset and chronic orrecurrent depression suggests that ongoing systematic screen-ing for depression is needed beyond the acute treatment phase.Declining health, worsening pain, and cessation of unhealthyalcohol use should alert clinicians to an increased risk forprobable MDD. A variety of treatment-related research isneeded to address the high rate of chronic or recurrent depres-sion, as well as new-onset depression affecting the lives of people with SCI.

References1. Bombardier CH, Richards JS, Krause JS, Tulsky D, Tate DG.

Symptoms of major depression in people with spinal cord injury:

implications for screening. Arch Phys Med Rehabil 2004;85:

1749-56.2. Spitzer RL, Kroenke K, Williams JB. Validation and utility of a

self-report version of PRIME-MD: the PHQ primary care study.

Primary Care Evaluation of Mental Disorders. Patient Health

Questionnaire. JAMA 1999;282:1737-44.

3. Kessler R, Berglund P, Demler O, et al. The epidemiology of

major depressive disorder: results from the National Comorbidity

Survey Replication (NCS-R). JAMA 2003;289:3095-105.

4. Herrick S, Elliott TR, Crow F. Social support and the prediction of

health complications among persons with spinal cord injuries.

Rehabil Psychol 1994;39:231-50.

5. Schulz R, Decker S. Long-term adjustment to physical disability:

the role of social support, perceived control, and self-blame. J Pers

Soc Psychol 1985;48:1162-72.

6. Elliott TR, Shewchuck R. Social support and leisure activitiesfollowing severe physical disability: testing the mediating effects

of depression. Basic Appl Soc Psychol 1995;16:471-587.7. Fuhrer MJ, Rintala DH, Hart KA, Clearman R, Young ME.

Depressive symptomatology in persons with spinal cord injurywho reside in the community. Arch Phys Med Rehabil 1993;74:255-60.

8. MacDonald MR, Nielson WR, Cameron MG. Depression andactivity patterns of spinal cord injured persons living in the com-munity. Arch Phys Med Rehabil 1987;68:339-43.

9. Krause J, Carter R, Pickelsimer E, Wilson D. A prospective studyof health and risk of mortality after spinal cord injury. Arch PhysMed Rehabil 2008;89:1482-91.





10. Richards J. Psychologic adjustment to spinal cord injury during

first postdischarge year. Arch Phys Med Rehabil 1986;67:362-5.11. Kennedy P, Rogers B. Anxiety and depression after spinal cord

injury: a longitudinal analysis. Arch Phys Med Rehabil 2000;81:932-7.

12. Pollard C, Kennedy P. A longitudinal analysis of emotional im-pact, coping strategies and post-traumatic psychological growthfollowing spinal cord injury: a 10-year review. Br J Health Psy-chol 2007;12(Pt 3):347-62.

13. Charlifue S, Gerhart K. Changing psychosocial morbidity inpeople aging with spinal cord injury. Neurorehabilitation 2004;19:15-23.

14. Richardson E, Richards J. Factor structure of the PHQ-9 screen fordepression across time since injury among persons with spinalcord injury. Rehabil Psychol 2008;53:243-9.

15. Judd F, Stone J, Webber J, Brown D, Burrows G. Depressionfollowing spinal cord injury. A prospective in-patient study. Br JPsychiatry 1989;154:668-71.

16. Kishi Y, Robinson R, Forrester A. Prospective longitudinal studyof depression following spinal cord injury. J Neuropsychiatry ClinNeurosci 1994;6:237-44.

17. Dryden D, Saunders L, Rowe B, et al. Depression followingtraumatic spinal cord injury. Neuroepidemiology 2005;25:55-61.

18. Craig A, Hancock K, Dickson H. A longitudinal investigation intoanxiety and depression in the first 2 years following a spinal cordinjury. Paraplegia 1994;32:675-9.

19. Hancock K, Craig A, Dickson H, Chang E, Martin J. Anxiety anddepression over the first year of spinal cord injury: a longitudinalstudy. Paraplegia 1993;31:349-57.

20. Craig A, Hancock K, Dickson H, Chang E. Long-term psycho-logical outcomes in spinal cord injured persons: results of acontrolled trial using cognitive behavior therapy. Arch Phys MedRehabil 1997;78:33-8.

21. Kemp B, Kahan J, Krause J, Adkins R, Nava G. Treatment of major depression in individuals with spinal cord injury. J SpinalCord Med 2004;27:22-8.

22. Fann J, Bombardier C, Richards J, Tate D, Wilson C, Temkin N.Depression after spinal cord injury: comorbidities, mental health

service use and adequacy of treatment. Arch Phys Med Rehabil2011;92:352-60.

23. Satyanarayana S, Enns M, Cox B, Sareen J. Prevalence andcorrelates of chronic depression in the Canadian CommunityHealth Survey: mental health and well-being. Can J Psychiatry2009;54:389-98.

24. Nicholson Perry K, Nicholas M, Middleton J, Siddall P. Psycho-logical characteristics of people with spinal cord injury-relatedpersisting pain referred to a tertiary pain management center. JRehabil Res Dev 2009;46:57-67.

25. Craig A, Hancock K, Dickson H. Spinal cord injury: a search fordeterminants of depression two years after the event. Br J ClinPsychol 1994;33(Pt 2):221-30.

26. Craig A, Tran Y, Middleton J. Psychological morbidity and spinalcord injury: a systematic review. Spinal Cord 2009;47:108-14.

27. Young M, Rintala D, Rossi C, Hart K, Fuhrer M. Alcohol andmarijuana use in a community-based sample of persons withspinal cord injury. Arch Phys Med Rehabil 1995;76:525-32.

28. Orenczuk S, Slivinski J, Mehta S, Teasell R. Depression followingspinal cord injury. In: Eng J, Teasell R, Miller W, et al, editors. Spinalcord injury rehabilitation evidence. Available at: www.scireproject.com/rehabilitation-evidence. Accessed October 14, 2010.

29. Tate D, Forchheimer M, Maynard F, Dijkers M. Predicting de-pression and psychological distress in persons with spinal cordinjury based on indicators of handicap. Am J Phys Med Rehabil1994;73:175-83.

30. Tawashy A, Eng J, Lin K, Tang P, Hung C. Physical activity isrelated to lower levels of pain, fatigue and depression in individ-

uals with spinal-cord injury: a correlational study. Spinal Cord

2009;90:301-6.

31. Banerjea R, Findley P, Smith B, Findley T, Sambamoorthi U.

Co-occurring medical and mental illness and substance use disor-

ders among veteran clinic users with spinal cord injury patients

with complexities. Spinal Cord 2009;47:789-95.

32. Marino RJ. International Standard for Neurological Classification

of Spinal Cord Injury. Atlanta: ASIA; 2000.

33. American Psychiatric Association. Major depressive episode. In:Diagnostic and Statistical Manual of Mental Disorders. 4th ed.

Washington (DC): American Psychiatric Association; 1994. p

320-7.

34. Williams R, Heinemann A, Bode R, Wilson C, Fann J, Tate D.

Improving measurement properties of the Patient Health Ques-

tionnaire-9 with rating scale analysis. Rehabil Psychol 2009;54:

198-203.

35. Kroenke K, Spitzer R, Williams J. The PHQ-9: validity of a brief

depression severity measure. J Gen Intern Med 2001;16:606-13.

36. MacArthur initiative on depression in primary care 2010. Avail-

able at: http://www.depression-primarycare.org/clinicians/

toolkits/materials/forms/phq9/treatment_response/. Accessed July

7, 2010.

37. McHorney C, Ware JJ, Raczek A. The MOS 36-Item Short-FormHealth Survey (SF-36): II. Psychometric and clinical tests of

validity in measuring physical and mental health constructs. Med

Care 1993;31:247-63.

38. Hall K, Dijkers M, Whiteneck G, et al. The Craig Handicap

Assessment and Reporting Technique (CHART): metric proper-

ties and scoring. Top Spinal Cord Injury Rehabil 1998;4:16-30.

39. Mellick D. The Craig Handicap Assessment and Reporting Tech-

nique-Short Form 2000. Available at: http://www.tbims.org/

combi/chartsf. Accessed October 30, 2008.

40. National Institute on Alcohol Abuse and Alcoholism. FAQs 2009.

Available at: http://www.niaaa.nih.gov/FAQs/General-English/

default.htm. Accessed October 23, 2009.

41. Stolp-Smith K, Wainberg M. Antidepressant exacerbation of spas-

ticity. Arch Phys Med Rehabil 1999;80:339-42.42. Smith B, Weaver F, Ullrich P. Prevalence of depression diagnoses

and use of antidepressant medications by veterans with spinal cord

injury. Am J Phys Med Rehabil 2007;86:662-71.

43. Iosifescu D. Treating depression in the medically ill. Psychiatr

Clin North Am 2007;30:77-90.

44. Dinan T, Mobayed M. Treatment resistance of depression after

head injury: a preliminary study of amitriptyline response. Acta

Psychiatr Scand 1992;85:292-4.

45. Reynolds CR, Miller M, Pasternak R, et al. Treatment of bereave-

ment-related major depressive episodes in later life: a controlled

study of acute and continuation treatment with nortriptyline and

interpersonal psychotherapy. Am J Psychiatry 1999;156:202-8.

46. Cohen A, Houck P, Szanto K, Dew M, Gilman S, Reynolds CR.

Social inequalities in response to antidepressant treatment in older

adults. Arch Gen Psychiatry 2006;63:50-6.

47. Bosworth H, Hays J, George L, Steffens D. Psychosocial and

clinical predictors of unipolar depression outcome in older adults.

Int J Geriatr Psychiatry 2002;17:238-46.

48. Cairns D, Adkins R, Scott M. Pain and depression in acute

traumatic spinal cord injury: origins of chronic problematic pain?

Arch Phys Med Rehabil 1996;77:329-35.

49. Elliott T, Kennedy P. Treatment of depression following spinal

cord injury: an evidence-based review. Rehabil Psychol 2004;49:

134-9.

50. Manber R, Kraemer H, Arnow B, et al. Faster remission of chronic depression with combined psychotherapy and medica-



http://www.scireproject.com/rehabilitation-evidence


http://www.depression-primarycare.org/clinicians/toolkits/materials/forms/phq9/treatment_response/


http://www.tbims.org/combi/chartsf


http://www.niaaa.nih.gov/FAQs/General-English/default.htm












tion than with each therapy alone. J Consult Clin Psychol2008;76:459-67.

51. Krause J, Kemp B, Coker J. Depression after spinal cord injury:relation to gender, ethnicity, aging, and socioeconomic indicators.Arch Phys Med Rehabil 2000;81:1099-109.

52. Kalpakjian C, Bombardier C, Schomer K, Brown P, Johnson K.Measuring depression in persons with spinal cord injury: a sys-tematic review. J Spinal Cord Med 2009;32:6-24.

53. Krause J, Bombardier C, Carter R. Assessment of depressivesymptoms during inpatient rehabilitation for spinal cord injury: isthere an underlying somatic factor when using the PHQ? RehabilPsychol 2008:53:513-20.

54. Williams JJ, Noël P, Cordes J, Ramirez G, Pignone M. Is thispatient clinically depressed? JAMA 2002;287:1160-70.

55. Krause J, Reed K, McArdle J. Prediction of somatic and non-somatic depressive symptoms between inpatient rehabilitation andfollow-up. Spinal Cord 2010;76:239-44.

56. Krause J, Reed K, McArdle J. Factor structure and predictivevalidity of somatic and non-somatic symptoms from the PHQ-9: alongitudinal study after spinal cord injury. Arch Phys Med Reha-bil 2010;91:1218-24.

57. Eaton W, Shao H, Nestadt G, et al. Population-based study of firstonset and chronicity in major depressive disorder. Arch GenPsychiatry 2008;65:513-20.



Documents

A Longitudinal Study of Depression From 1 to 5 Years After Spinal Cord Injury