predictors of long-term work disability in major depressive disorder: a prospective study

Predictors of long-term work disability inMajor Depressive Disorder: a prospectivestudy

H. J. Rytsala, T. K. Melartin, U. S. Leskela, T. P. Sokero, P. S. Lestela-Mielonen, E. T. Isometsa. Predictors of long-term work disability inMajor Depressive Disorder: a prospective study.

Objective: Major Depressive Disorder (MDD) is a major cause oflong-term work disability. However, factors predicting this are not wellknown.Method: In the Vantaa Depression Study, predictors for being granteda disability pension during an 18-month follow-up were examinedamong the 186 psychiatric MDD in- and out-patients belonging to thelabour force at baseline.Results: The 21 patients (11.3%) granted a disability pension weresignificantly older, more hopeless, had worse social andoccupational functioning, and spent more time depressed duringfollow-up. After adjusting for these predictors, being on sickleave at baseline still strongly predicted disability pension duringfollow-up.Conclusion: Disability pension is predicted by multiplesociodemographic and clinical factors. Baseline level of functioningand duration of depressive episodes are key clinical predictors. Thepositive and negative consequences of sick leave warrant closerattention.

H. J. Ryts�l�1,2, T. K. Melartin1,3,U. S. Leskel�1,2, T. P. Sokero1,P. S. Lestel�-Mielonen1,2,E. T. Isomets�1,3

1 Department of Mental Health and Alcohol Research,National Public Health Institute, Helsinki, 2 Departmentof Psychiatry, Helsinki University Central Hospital, PeijasHospital, Vantaa and 3 Department of Psychiatry,Helsinki University Central Hospital, Helsinki, Finland

Key words: depressive disorder; major; work; sickleave; insurance; disability

Erkki T. Isomets�, Department of Mental Health andAlcohol Research, National Public Health Institute,Mannerheimintie 166, FIN-00300 Helsinki, Finland.E-mail: [email protected]

Accepted for publication June 7, 2006

Significant outcomes

• Disability pension for depression is predicted by multiple factors, not exclusively by clinical ones.• Baseline level of functioning and duration of depression are key clinical predictors of future disability

and disability pension.• Sick leaves are often necessary, but they may also have negative consequences.

Limitations

• The number of patients eligible for analysis (21 pensioned compared to 165 not pensioned) was onlymoderate.

• Some patients on long sick leaves at 18 months may not return to work, thus potentially increasingthe number of disability pensions later.

• Motivational factors might predict both being on sick leave while depressed and work outcome later,but could not be measured.

Introduction

Major Depressive Disorder (MDD) is worldwidethe fourth leading illness causing functionalimpairment (1). By the year 2020, it has been

predicted to be the second major cause of func-tional disability after ischaemic heart diseases (2).Several epidemiological and other studies in theUS demonstrate that depression is one of the mostimportant factors associated with sickness absence

Acta Psychiatr Scand 2007: 115: 206–213All rights reservedDOI: 10.1111/j.1600-0447.2006.00878.x

Copyright � 2006 The AuthorsJournal Compilation � 2006 Blackwell Munksgaard

ACTA PSYCHIATRICASCANDINAVICA

206

from work (3–5). This problem has also beenrecognized in many European countries. In a largeepidemiological study conducted in six Europeancountries (6), MDD was among the 10 mostdisabling mental disorders. In another Europeanstudy, a significant proportion of patients receiv-ing treatment for depression and anxiety disorderswere found not to belong to the labour force (7).The overall costs of work disability are high (4, 8)both to employees and employers, and to societyas a whole (4, 9). However, factors predictinglong-term work disability among psychiatricMDD patients are not well known. In ourprevious cross-sectional study (10), we foundthat older age, comorbidity, severity and durationof depression and number of previous episodes ofdepression were the predominant factors predict-ing overall disability. Other cross-sectional studieshave also suggested that patients with comorbidMDD are more disabled than those with pureMDD (11, 12).Several studies have reported shortcomings in

treatment of depression, and poor adherence totreatment (13, 14), which both may hinderrecovery from depression, and thus contributeto functional and work disability (15). In Fin-land, particular attention has been paid to therapid increase of the long-term work disabilitypensions due to depression. Salminen et al. (16)noted the rapid increase in number of long-termwork disability pensions between 1987 and 1994,raising the issue, how and why the functionalcapacity of depressive patients could have mark-edly deteriorated despite new drugs and othertreatments that had become available. In twoindependent, representative record-based studieswe investigated the quality and intensity oftreatment provided to patients with clinicallydiagnosed depression, who were granted a dis-ability pension in Finland in 1993 (11), or inpsychiatric care in the city of Vantaa in 1996(17). Both studies documented the low intensityof treatment provided. Nevertheless, patients whowere granted a disability pension still had morevisits to professionals than the others (17).Overall, given the high human and economicalcosts of long-term disability, continuous evalua-tion of quality of treatment provided to thosedisabled due to depression is warranted.

Aims of the study

The aim of this study was to prospectivelyinvestigate factors predicting long-term work dis-ability in a large sample of psychiatric patientswith MDD.

Material and methods

The Vantaa Depression Study (VDS) is a colla-borative depression research project between theDepartment of Mental Health and AlcoholResearch of the National Public Health Institute,Helsinki, Finland, and the Department of Psychi-atry of the Peijas Medical Care District (PMCD),Vantaa, Finland. Its background and methodologyhave been described in detail elsewhere (18–20).

Screening, baseline and follow-up methods

In brief, in the first phase of the study, 806psychiatric patients were screened for the presenceof depressive symptoms during a 18-month periodstarting on 1 February 1997. Screened and eligiblepatients who had given their written informedconsent were interviewed face-to-face by one ofthe researchers (UL, PL-M, TM, HR or PS) usingthe WHO Schedules for Clinical Assessment inNeuropsychiatry (SCAN), version 2.0 (21). Allresearchers have received relevant training by aWHO-certified training centre. Patients fulfillingthe current mood episode for (unipolar) DSM-IVMajor Depressive Disorder comprised the VDSMDD cohort of 269 patients. Diagnostic reliabil-ity was found to be excellent [kappa for MDD0.86 (0.58–1.0)] (18). All patients in the studycohort were interviewed with the entire SCAN togive a full picture of Axis I current comorbiddisorders, and with the Structured Clinical Inter-view for DSM-III-R personality disorders (SCID-II) (22) to assess current diagnoses on Axis II. Inaddition to these, baseline and follow-up inter-views included the 21-item Beck DepressionInventory (BDI) (23), the 17-item HamiltonRating Scale for Depression (Ham-D) (24), theSocial and Occupational Functioning AssessmentScale of DSM-IV (SOFAS) (25), the SocialAdjustment Scale-Self Report (SAS-SR) (26), theBeck Anxiety Inventory (BAI) (27), the PerceivedSocial Support Scale-Revised (PSSS-R) (28), theBeck Hopelessness Scale (HS) (29), the Scale forSuicidal Ideation (SSI) (30) and social, occupa-tional, and treatment-related factors.After baseline, patients were investigated at 6

and 18 months from baseline with the life chartmethodology described in detail in Melartin et al.(19) and scales mentioned above. We employed aLikert scale to investigate patients� self reportedtreatment adherence with the following responseitems: been on antidepressants i) regularly, treat-ment compliance adequate with respect to treat-ment goals, ii) somewhat irregularly, it is unclearwhether this would affect treatment goals, iii) very

Predictors of work disability in MDD

207

irregularly, the treatment did not proceed accord-ing to plan and iv) not at all, the providedtreatment could not be implemented. Hospitaltreatment periods during follow-up wereascertained in the medical records (20). At the 18-month follow-up 13 of the 269 (5%) switched tobipolar disorder during the follow-up and 58patients (20%) dropped out. Thus, at the 18-month interview, 198 unipolar patients remained(19).

Follow-up of subjects in this study

Of the 198 patients who were prospectively fol-lowed up, we excluded those already pensioned atbaseline. The remaining 186 patients were classifiedinto two groups as follows; i) those who at18 months were still in the labour force (n ¼ 99)or were unemployed (n ¼ 29), on sick leave (n ¼14), students (n ¼ 13) or others (n ¼ 10) and ii)those who were granted a long-term work disabil-ity pension due to MDD during follow-up (n ¼21). The majority of subjects in the cohort werefemales (134, 72%), and the mean age of all was42.1 years (SD 10.8). In Finland, MDD sick leavesof over 2 months are issued by psychiatrists,usually in several consecutive periods duringtreatment if the patient is judged incapable ofwork. Medical certificates issued by a psychiatristfor work disability allowances are referred to andgranted by the Social Insurance Institution ofFinland. After having received this daily allowancefrom sickness insurance for 300 days (counted at6 days/week), employees under 63–65 years areeligible for disability pension, if necessary. Medicalcertificates are referred to and pensions granted bythe pertinent insurance company.

Statistical analyses

We performed the chi-squared analyses with Yates�continuity correction in categorical comparisonsbetween pensioned vs. not pensioned groups. Incross-tabulation analyses where some cells con-tained less than five cases, we used Fisher’s exacttest. Analysis of variance (anova) was used incomparisons of continuous variables. In all analy-ses, we omitted item 7 in Ham-D and item 15 inBDI, capacity for work, in order to avoid circu-larity. We employed three different logistic regres-sion models to investigate i) the baseline factorspredicting pension, ii) the effect of duration ofdepression and iii) whether being on sick leave atbaseline independently predicts being granted adisability pension. We first adjusted for sex, age,marital status, severity and number of previous

episodes of depression, alcoholism, personalitydisorders, BAI, PSSS-R, HS, SSI, SAS-SR andSOFAS scores at baseline and duration of depres-sion, and then omitted non-significant factors. Weused spss version 13.0 (31).

Results

Sociodemographic and clinical differences

Of the 186 patients, 21 (11.3%) were granted adisability pension during the 18-month follow-up.The patients who had received a pension weresignificantly older, more seldom had vocationaleducation and were more often on sick leave thanthose not pensioned, but did not differ with regardto any other sociodemographic or clinical factors(Table 1). However, at the 6-month follow-up,these groups had differed in nearly all variables,with these differences diminishing over the follow-ing year (Table 2). Two-thirds (14 of 21, 67%) ofpensioned patients vs. one-fourth (42 of 161, 25%)of those not pensioned (df ¼ 1, v2 ¼ 13.6,P < 0.001) were on sick leave at baseline. Themean time spent in major depressive episodes(MDEs) was 10.6 months in pensioned patientsand 4.1 months in non-pensioned patients (anovaF ¼ 34.4, P < 0.001, medians 13.7 and2.5 months respectively). One-third (7 of 21) ofpensioned patients vs. only 3% of the others (5 of165, df ¼ 1, v2 ¼ 23.5, P < 0.001) did not reacheven partial remission during follow-up. Socialphobia was the only Axis I or II comorbid disorderto be more common among pensioned patients (8of 21, 38% vs. 25 of 165, 15%, df ¼ 1, v2 ¼ 5.2,P ¼ 0.022).In a logistic regression analysis adjusted for sex,

age and severity of depression, greater age, hope-lessness, disability and no vocational educationpredicted disability pension (Table 3). The timespent in MDEs during the 18 months emerged asvery significant when included in the model, withhopelessness remaining only a trend (Table 4).Finally, we analysed the same model, including thevariable being on sick leave at baseline, andomitting SOFAS to avoid circularity. In thismodel, male gender becomes significant and hope-lessness disappears. After all adjustments, the riskof being granted a disability pension during thenext 18 months is markedly elevated (OR 6.1)among patients on sick leave at baseline (Table 5).

Treatment

Patients who were granted a work disabilitypension were treated somewhat more intensively

Rytsala et al.

208

Table 1. Sociodemographic and clinical data atbaseline of patients pensioned or not pensioned atthe 18-month follow-up

Not pensioned(n ¼ 165)

Pensioned(n ¼ 21)

Total(n ¼ 186)

Female, n (%) 120 (72.7) 14 (66.7) 134 (72.0)Age at 18-month, mean (SD)* 40.8 (10.5) 52.4 (7.6) 42.1 (10.8)Married or cohabiting, n (%) 87 (52.7) 13 (61.9) 100 (53.8)Vocational education, n (%)� 104 (63.0) 8 (38.1) 112 (60.2)In-patients, n (%) 22 (13.3) 4 (19.0) 160 (86.0)On sick leave, n (%)� 40 (24.2) 14 (66.7) 54 (29.0)Severity of depression

Mild, n (%) 9 (5.5) 2 (9.5) 11 (5.9)Moderate, n (%) 92 (55.8) 7 (33.3) 99 (53.2)Severe without psychotic sympt. n (%) 61 (37.0) 12 (57.1) 73 (39.2)Severe with psychotic sympt. n (%) 3 (1.8) 0 (0.0) 3 (1.6)

Recurrence of depressionFirst episode, n (%) 61 (37.0) 4 (19.0) 65 (34.9)Recurrent depression, n (%) 104 (63.0) 17 (81.0) 121 (65.1)MDD without comorbidity, n (%) 39 (23.6) 3 (14.3) 42 (22.6)

Axis I comorbidityDysthymia, n (%) 15 (9.1) 4 (19.0) 19 (10.2)Any anxiety disorder, n (%) 88 (53.3) 13 (61.9) 101 54.3)Any alcohol use disorder, n (%) 35 (21.2) 3 (14.3) 38 (20.4)

Axis II comorbidityAny personality disorder, n (%) 68 (41.2) 12 (57.1) 80 (43.0)Cluster A, n (%) 31 (18.8) 5 (23.8) 36 (19.4)Cluster B, n (%) 23 (13.9) 4 (19.0) 27 (14.5)Cluster C, n (%) 49 (29.7) 9 (42.9) 58 (31.2)

*anova F ¼ 24.1, P < 0.001.�v2 ¼ 3.8, P ¼ 0.050.�v2 ¼ 14.3, P < 0.001.

Table 2. Mean and standard deviation (SD) valuesof univariate analyses of patients pensioned or notpensioned at the 18-month follow-up

Not pensioned Pensioned F P

BDI at baseline, mean (SD) 25.5 (7.7) 27.8 (8.1) 1.7 0.188BDI at 6 months, mean (SD)* 10.7 (8.7) 19.4 (8.4) 18.9 <0.001BDI at 18 months, mean (SD)� 9.9 (9.4) 15.2 (9.8) 5.9 0.016Ham-D at baseline, mean (SD) 16.9 (5.3) 19.6 (4.3) 4.9 0.028Ham-D at 6 months, mean (SD)* 7.4 (6.5) 13.3 (6.5) 15.3 <0.001Ham-D at 18 months, mean (SD) 6.7 (6.2) 10.4 (6.9) 6.4 0.012SOFAS at baseline, mean (SD)� 53.6 (10.2) 47.2 (6.7) 7.7 0.006SOFAS at 6 months, mean (SD)* 69.4 (13.8) 53.9 (8.0) 25.1 <0.001SOFAS at 18 months, mean (SD)� 73.9 (13.8) 65.6 (14.8) 6.5 0.012SAS-SR overall at baseline, mean (SD) 2.4 (0.4) 2.6 (0.3) 3.5 0.062SAS-SR overall at 6 months, mean (SD)� 2.0 (0.5) 2.4 (0.3) 13.7 <0.001SAS-SR overall at 18 months, mean (SD)§ 2.0 (0.5) 2.1 (0.4) 1.2 0.271BAI at baseline, mean (SD) 21.1 (10.8) 27.6 (8.2) 7.0 0.009BAI at 6 months, mean (SD)* 11.9 (9.3) 22.0 (7.4) 22.8 <0.001BAI at 18 months, mean (SD) 10.9 (10.0) 15.4 (10.5) 3.7 0.057PSSS-R at baseline, mean (SD) 39.6 (12.7) 34.3 (12.6) 3.2 0.075PSSS-R at 6 months, mean (SD)* 42.9 (12.8) 35.5 (12.4) 6.3 0.013PSSS-R at 18 months, mean (SD) 43.5 (13.0) 37.6 (16.6) 3.6 0.058HS at baseline, mean (SD) 9.9 (4.6) 12.8 (4.1) 7.7 0.006HS at 6 months, mean (SD)* 6.8 (5.0) 11.7 (4.2) 18.2 <0.001HS at 18 months, mean (SD)§ 6.1 (4.8) 8.6 (5.4) 5.1 0.025SSI at baseline, mean (SD)� 5.5 (7.4) 10.6 (9.6) 8.2 0.005SSI at 6 months, mean (SD)–,** 1.6 (4.1) 3.6 (7.5) 2.6 0.107SSI at 18 months, mean (SD)*,�� 1.2 (3.8) 3.0 (6.7) 3.0 0.084

BDI, Beck Depression Inventory; Ham-D, Hamilton Rating Scale for Depression; SOFAS, Social and OccupationalFunctioning Assessment Scale; SAS-SR, Social Adjustment Scale-Self Report; BAI, Beck Anxiety Inventory; PSSS-R,Perceived Social Support Scale-Revised; HS, Beck Hopelessness Scale; SSI, Scale for Suicidal Ideation.*Data of five patients are missing in not pensioned group.�Data of one patient is missing in not pensioned group.�Data of four patients are missing in not pensioned group.§Data of two patients are missing in not pensioned group.–Data of 56 patients are missing in not pensioned group.**Data of four patients are missing in pensioned group.��Data of one patient is missing in pensioned group.


209

than the others. All pensioned patients had anti-depressants at baseline, whereas 23 of 165 non-pensioned patients (14%) had none (Fisher’s exacttest, P ¼ 0.081). At 6 months, 17 of 21 pensionedpatients (81%) had used antidepressants uninter-ruptedly from baseline vs. 87 of 165 their counter-parts (53%, df ¼ 1, v2 ¼ 4.930, P ¼ 0.026); at18 months the corresponding figures were 14 of 21(67%) and 70 of 165 (42%, df ¼ 1, v2 ¼ 3.496,P ¼ 0.062). The median number of visits to doc-tors during the 18 months was also higher (6.0)among the pensioned patients than among thosenot pensioned (2.0) (Mann–Whitney test, Z ¼)4.051, P < 0.001). The median number of visitsto any personnel was 19.0 vs. 14.0 visits respec-tively (Mann–Whitney Test, Z ¼ )1.829, P ¼0.067).In other aspects of treatment, attitudes towards

treatment, or adherence to it, significant differenceswere not found. All but one (95.2%) of thepensioned and 116 of 142 (81.7%) of the non-pensioned subjects had adequate antidepressanttreatment in the acute phase of MDE (Fisher’sexact test, P ¼ 0.205). Antidepressant combina-tions were received during follow-up by 5 of 21(23.8%) vs. 25 of 165 (15.2%, df ¼ 1, v2 ¼ 0.491,P ¼ 0.483). Two of 21 pensioned patients (10%)vs. 26 of 165 non-pensioners (16%) received weekly

psychotherapy (Fisher’s exact test P ¼ 0.745).Patient attitudes towards medication did notdiffer significantly either; one of 21 (5%) ofpensioned patients vs. 27 of 165 (16%) of othersreported a negative attitude towards antidepres-sants (Fisher’s exact test, P ¼ 0.209). Of thepensioned subjects, 19 of 20 (95.0%) reportedhaving taken antidepressants regularly or some-what irregularly, of the non-pensioned 96 of 106(90.6%) (Fisher’s exact test, P ¼ 1.000). Of thepensioned patients five of 21 (23.8%) had beenhospitalized (one to four times) either duringbaseline or follow-up vs. 33 of 165 (20%, 1 to 14visits) of those not pensioned (df ¼ 1, v2 ¼ 0.015P ¼ 0.904). The mean length of hospital stay inpensioned patients was 33.8 days (SD 14.9) vs.44.9 days respectively (SD 57.3, anova F ¼ 0.183,P ¼ 0.671).

Discussion

Main findings

Many sociodemographic and clinical factorsclearly predict long-term work disability amongpsychiatric patients with MDD. Even after adjust-ing for clinical variables, the sociodemographicfactors of older age and lack of vocational educa-

Table 3. Logistic regression model of significantclinical baseline variables. The dependent variableis either being pensioned or not pensioned at the18-month follow-up

Variable B SE Wald P OR 95% CI for B

Age 0.1438 0.0410 12.3272 0.0004 1.1546 1.0656; 1.2511Hopelessness scale at baseline 0.2035 0.0734 7.6833 0.0056 1.2257 1.0614; 1.4154SOFAS at baseline )0.0748 0.0300 6.1996 0.0128 0.9279 0.8748; 0.9842Vocational education )1.2414 0.5803 5.5764 0.0324 0.2890 0.0927; 0.9012

SOFAS, Social and Occupational Functioning Assessment Scale.

Table 4. Logistic regression model of significantclinical baseline variables and time in define MDEduring the 18-month follow-up. The dependentvariable is either being pensioned or not pensionedat follow-up


Age 0.1442 0.0451 10.2267 0.0014 1.1551 1.0574; 1.2618Hopelessness scale at baseline 0.1500 0.0789 3.6160 0.0572 1.1618 0.9954; 1.3560SOFAS at baseline )0.0692 0.0314 4.8545 0.0245 0.9331 0.8774; 0.9924Time in MDE 0.1520 0.0482 9.9481 0.0016 1.1641 1.0592; 1.2794Vocational education )1.4487 0.6442 5.0564 0.0245 0.2349 0.0664; 0.8303

SOFAS, Social and Occupational Functioning Assessment Scale; MDE, major depressive episode.

Table 5. Logistic regression model of significantclinical baseline variables, time in MDE during the18-month follow-up and being on sick leave atbaseline. The dependent variable is either beingpensioned or not pensioned at follow-up


Sex )1.7723 0.8125 4.7580 0.0292 0.1699 0.0346; 0.8354Age 0.1407 0.0466 9.1370 0.0025 1.1511 1.0507; 1.2611Time in MDE 0.1838 0.0490 14.0790 0.0002 1.2018 1.0918; 1.3229Vocational education )1.2415 0.6481 3.6696 0.0554 0.2889 0.0811; 1.0292Being on sick leave at baseline 1.8040 0.6884 6.8672 0.0088 6.0738 1.5758; 23.4115

MDE, major depressive episode.

Rytsala et al.

210

tion independently predicted disability pension. Ofthe baseline clinical predictors, hopelessness wasthe strongest, but level of disability and lack ofvocational education also had an effect. Thus thosewho are most hopeless about their future appear tomore often eventually be granted a work disabilitypension. During follow-up, slow recovery fromdepression (time spent in MDEs) was one of thestrongest factors affecting patients� work ability. Amajor public health policy issue is the role of sickleaves. Being on sick leave at baseline stronglypredicted (OR 6.1) disability pension, even afteradjusting for all other significant predictors.

Strengths, limitations and generalizability

This study included the majority of the VDSpatient cohort comprising psychiatric in- and out-patients with MDD, and representing secondary-level psychiatric care in a Finnish city. We studiedthe effect of the whole spectrum of Axis I and IIcomorbid disorders and other factors associatedwith depression, disability and its duration onpatients� employment status. The life chart (19)integrated information on recovery from depres-sion during the follow-up.Our study sample comprised only 69% (186 of

269) of the unipolar MDD patients of our initialcohort as only those patients who were potentiallabour force contributors at baseline were inclu-ded. Moreover, part of the initial cohort droppedout, and some switched to bipolar disorder. Therelatively small number of pensioned patientslimits the statistical power of results and mightalso increase the risk for spurious findings. How-ever, our findings were found to be statisticallyhighly significant. The main measure of disabilitywas SOFAS, which may have some contaminationbetween symptoms of depression and those ofcomorbid disorders. As this measure containsability to work as one of its major domains, itwas clearly circular with being on sick leave, andcould not be used as a predictor in all analyses.To our knowledge, no other prospective, com-

prehensive long-term work disability studies havebeen conducted in psychiatric settings. The fewwork disability studies are mainly general popula-tion (4, 32, 33) or primary care (9) studies focusingon income losses, functional disability or unem-ployment. The general population studies havedocumented depressed patients to have significantsalary losses (4) and more new unemployment thanothers (32). The primary care study by Simon et al.(9) found lack of recovery from depression to bestrongly associated with absence from paidemployment. Of the scant studies in psychiatric

settings, Paykel et al. (34) focused in their follow-up study on residual symptoms in depressedpatients, finding that the more symptoms patientshave, the more they are disabled overall. Ofresearch focusing on long-term disability andgranted disability pensions, all has been eitherrecord-based (17, 35) or retrospective (11). Thecomparability of the clinical characteristics ofpatients in the VDS cohort has been reviewedearlier (18, 19) and found not to be different fromother depression literature. However, the extent towhich our findings related to disability pensionsand sick leaves are generalizable to other settingsremains unknown at present.

Sociodemographic variables

Older age was a major factor predicting workdisability and disability pension, even after adjust-ing for other predictors. In our cross-sectionalstudy (10), females were more often acutely workdisabled, but in previous studies of long-term workdisability (11), no sex differences were present.After adjusting for other possibly significant fac-tors, lack of vocational education clearly predictedlong-term work disability. This finding was con-vergent with a US study by Elinson et al. (36),although we found no differences related to maritalstatus. Being granted a disability pension fordepression is obviously dependent on other factorsin addition to clinical ones.

Clinical variables

Several clinical factors were found to significantlypredict long-term work disability. Of the baselinevariables, hopelessness was more pronounced andfunctional disability worse among the pensionedthan among the other patients. In our view,subjects who see their future in negative termsmay be more inclined to cope with their depres-sion by seeking long-term disability pension.Furthermore, poor level of functioning in theacute phase is, hardly unexpectedly, a strongpredictor of disability also during follow-up.Contrary to the findings of Spijker et al. (33) inthe Dutch general population, duration of MDEwas in pensioned patients significantly longer, andone-third did not reach even partial remission. At6 months, the differences between groups weregreatest in nearly all measures, which also reflectsthe slow recovery from depression. In contrast toprevious cross-sectional findings (10), severity andnumber of previous episodes of depression andcomorbidity were not effective predictors of long-term work disability; their role was overshadowed


211

by the impact of time spent in MDEs duringfollow-up.

Role of sick leave

The strong role of being on sick leave at baseline asa predictor of disability pension later, was surpris-ing. This effect persisted even after adjusting forother predictors. Thus, of patients with apparentlysimilar clinical characteristics, those who are cur-rently on sick leave seem to have markedly higherprobability of pension in the future. This finding isunlikely to be tautological, i.e. merely due topreceding long sick leaves being a precondition fora pension to be granted. The study design wasprospective, and the majority of patients on sickleave at baseline returned to work or unemploy-ment during follow-up. However, a major limita-tion in our study was that we were not able tomeasure motivational factors related to work, thatmay also influence seeking a pension. To ourknowledge, this research area has not been inves-tigated in other studies concerning depression anddisability, but it is obviously a major health policyissue. Studies of somatic diseases (37) have foundthat returning to work is difficult after time out ofwork-life. The extent to which sick leaves haveadverse, disability-reinforcing or negative motiva-tional consequences among patients with depres-sion warrants further investigation.

Role of treatment

In our previous studies focusing on the early andmid-1990s (11, 17), we found pharmacotherapyamong subjects granted a disability pension fordepression to have been largely suboptimal. Hereall of the pensioned patients had antidepressants atbaseline, and in nearly all cases, the treatment wasclassified as adequate. Problems in the intensityand monitoring of treatment have thus diminishedcompared with the preceding era (17). In thepresent study disability was not related to theinferior treatment. However, given the high costsrelated to disability, the intensity of treatmentprovided was far from optimal.In conclusion, disability pension for depression

is predicted by multiple sociodemographic andclinical factors, not exclusively by clinical factors.Baseline level of functioning and duration ofdepressive episodes are key clinical predictors offuture disability. Reducing time spent depressed inacute phase by optimal treatment is one of themost important aims for reducing long-term dis-ability. The positive and negative consequences ofsick leaves also require elucidation.

References

1. Murray CJL, Lopez AD. Global mortality, disability, andthe contribution of risk factors: Global Burden of DiseaseStudy. Lancet 1997;349:1436–1442.

2. Murray CJL, Lopez AD. Alternative projections of mor-tality and disability by cause 1990–2020: Global Burden ofDisease Study. Lancet 1997;349:1498–1504.

3. Broadhead WE, Blazer DG, George LK, Tse CK. Depres-sion, disability days, and days lost from work in aprospective epidemiologic survey. JAMA 1990;264:2524–2528.

4. Kessler RC, Barber C, Birnbaum HG et al. Depression inthe workplace: effects on short-term disability. Health Aff(Millwood) 1999;18:163–171.

5. Wells KB, Stewart A, Hays RD et al. The functioning andwell-being of depressed patients. Results from the MedicalOutcomes Study. JAMA 1989;262:914–919.

6. The ESEMeD/MHEDEA2000 investigators. Disabilityand quality of life impact of mental disorders in Europe:results from the European Study of the Epidemiology ofMental Disorders (ESEMeD) project. Acta PsychiatrScand 2004;109(Suppl. 420):38–46.

7. Waghorn G, Chant D. Labour force activity by people withdepression and anxiety disorders: a population-level sec-ond-order analysis. Acta Psychiatr Scand 2005;112:415–424.

8. Stewart WF, Ricci JA, Chee E, Hahn SR, Morganstein D.

Cost of lost productive work time among US workers withdepression. JAMA 2003;289:3135–3144.

9. Simon GE, Revicki D, Heiligenstein J et al. Recovery fromdepression, work productivity, and health care costsamong primary care patients. General Hospital Psychiatry2000;22:153–162.

10. Rytsala HJ, Melartin TK, Leskela US, Sokero TP, Lestela-

Mielonen PS, Isometsa ET. Functional and work disabilityin major depressive disorder. J Nerv Ment Dis2005;193:189–195.

11. Isometsa ET, Katila H, Aro T. Disability pension for majordepression in Finland. Am J Psychiatry 2000;157:1869–1872.

12. McDermut W, Mattia J, Zimmerman M. Comorbidity burdenand its impact on psychosocial morbidity in depressedoutpatients. J Affect Disord 2001;65:289–295.

13. Katon W, von Korff M, Lin E et al. Collaborative man-agement to achieve treatment guidelines. Impact ondepression in primary care. JAMA 1995;273:1026–1031.

14. Lingam R, Scott J. Treatment non-adherence in affectivedisorders. Acta Psychiatr Scand 2002;105:164–172.

15. von Korff M, Katon W, Rutter C et al. Effect on disabilityoutcomes of a depression relapse prevention program.Psychosom Med 2003;65:938–943.

16. Salminen JK, Saarijarvi S, Raitasalo R. Depression anddisability pension in Finland. Acta Psychiatr Scand1997;95:242–243.

17. Rytsala HJ, Melartin TK, Leskela US, Lestela-Mielonen

PS, Sokero TP, Isometsa ET. A record-based analysis of 803patients treated for depression in psychiatric care. J ClinPsychiatry 2001;62:701–706.

18. Melartin TK, Rytsala HJ, Leskela US, Lestela-Mielonen

PS, Sokero TP, Isometsa ET. Current comorbidity ofDSM IV major depressive disorder in psychiatric care –Vantaa Depression Study. J Clin Psychiatry 2002;63:126–134.


PS, Sokero TP, Isometsa ET. Severity and comorbiditypredict episode duration and recurrence of DSM-IV

Rytsala et al.

212

major depressive disorder. J Clin Psychiatry 2004;65:810–819.


PS, Sokero TP, Isometsa ET. Continuity is the main chal-lenge in treating major depressive disorder in psychiatriccare. J Clin Psychiatry 2005;66:220–227.

21. Wing JK, Babor T, Brugha T et al. Schedules for clinicalassessment in neuropsychiatry. Arch Gen Psychiatry1990;47:589–593.

22. Spitzer RL, Williams JBW, Gibbon M, First MB. Structuredclinical interview for DSM-III-R personality disorders(SCID-II). New York: New York State Psychiatric Insti-tute, Biometrics Research, 1987.

23. Beck AT, Ward CH, Mendelson M, Moch JE, Erbaugh JK.

An inventory for measuring depression. Arch Gen Psy-chiatry 1961;4:561–571.

24. Hamilton M. A rating scale for depression. J Neurol Neu-rosurg Psychiatry 1960;23:56–62.

25. Goldman HH, Skodol AE, Lave TR. Revising axis V forDSM-IV: a review of measures of social functioning. Am JPsychiatry 1992;149:1148–1156.

26. Weissman MM, Bothwell S. Assessment of social adjust-ment by patient self-report. Arch Gen Psychiatry1976;33:1111–1115.

27. Beck AT, Brown G, Epstein N, Steer RA. An inventory formeasuring clinical anxiety: psychometric properties. JConsult Clin Psychol 1988;56:893–897.

28. Blumenthal JA, Burg MM, Barefoot J, Williams RB, Haney

T, Zimet G. Social support, type A behavior, and coronaryartery disease. Psychosom Med 1987;49:331–340.

29. Beck AT, Weissman A, Lester D, Trexler L. The measure ofpessimism: the hopelessness scale. J Consult Clin Psychol1974;42:861–865.

30. Beck AT, Kovacs M, Weissman A. Assessment of suicidalintention: the scale for suicide ideation. J Consult ClinPsychol 1979;47:343–352.

31. Statistical Package for the Social Sciences for Windows.Statistical package for the social sciences for windows, Re-lease 13.0. Chicago, IL, Copyright SPSS Inc. (1989–2004).

32. Lerner D, Adler DA, Chang H et al. Unemployment, jobretention, and productivity loss among employees withdepression. Psychiatr Serv 2004;55:1371–1378.

33. Spijker J, Graaf R, Bijl RV, Beekman AT, Ormel J, Nolen

WA. Functional disability and depression in the generalpopulation. Results from the Netherlands Mental HealthSurvey and Incidence Study (NEMESIS). Acta PsychiatrScand 2004;110:208–214.

34. Paykel ES, Ramana R, Cooper Z, Hayhurst H, Kerr J, Ba-

rocka A. Residual symptoms after partial remission: animportant outcome in depression. Psychol Med 1995;25:1171–1180.

35. Sorvaniemi M, Helenius H, Salokangas R K. Factors asso-ciated with being granted a pension among psychiatricoutpatients with major depression. J Affect Disord2003;75:43–48.

36. Elinson L, Houck P, Marcus SC, Pincus HA. Depression andthe ability to work. Psychiatr Serv 2004;55:29–34.

37. Rabkin JG, McElhiney M, Ferrando SJ, Van Gorp W, Lin SH.

Predictors of employment of men with HIV/AIDS: a lon-gitudinal study. Psychosom Med 2004;66:72–78.


213

predictors of long-term work disability in major depressive disorder: a prospective study

Documents