Marital and Labor Market Status in the Long Run in Schizophrenia

SINGLENESS, UNEMPLOYMENT, and labor market marginalization may independentlyincrease the risk of schizophrenia,^{1 - 3} orinsidious or untreated psychotic symptoms^{4 - 5} maylead to singleness or unemployment or may hinder development of personal relationshipsor gaining a foothold on the labor market.^{2 ,6 - 8} Individualswho have been admitted to a psychiatric hospital with schizophrenia are subsequentlymore likely to become single, unemployed, or recipients of social benefits.^{9 - 10} These findings could be explainedby the influence of cognitive deficits or permanent behavioral traits thathave been connected with schizophrenia^{11 - 12} orby the effects of labeling and stigma leading to singleness and labor marketmarginalization.^{13 - 14}

The long-term effect of singleness and labor market affiliation on therisk of developing schizophrenia, as well as the long-term effects of schizophreniaon marital and labor market status, have rarely been examined^{15 - 19} andnever, to our knowledge, simultaneously in an unselected population.

This study aims to investigate long- and short-term associations betweenmarital status and labor market affiliation before and after the first admissionwith schizophrenia compared with population-based controls.

Data were obtained by linking 3 Danish population-based registers bymeans of the unique personal identification number, which is assigned to allpersons living in Denmark, thus ensuring accurate linkage of information betweenregisters.²⁰ The Danish Psychiatric CentralRegister²¹ includes all admission and dischargedates and diagnoses according to the World Health Organization International Classification of Diseases, Eighth Revision²² and International Classification of Diseases, 10th Revision²³ classification of all psychiatric inpatient facilitiesin Denmark since 1969. There are no private psychiatric hospitals in Denmark,and all treatment is free of charge. Registration of outpatient activitieswas established in 1995. The Danish Medical Register on Vital Statistics containsdates and causes of all deaths in Denmark recorded from the Cause-of-DeathCertificates since 1976 and for suicide since 1970.²⁴ TheIntegrated Database for Longitudinal Labour Market Research covers the totalDanish population and contains detailed year-by-year information for the period1980 and onward with information from administrative registers.²⁵ Informationwas recorded only as long as the person was living in the country on December31, thus excluding people who emigrated or died within the year in question.

In total, 5341 cases were identified, which is the total sample of individualswho were diagnosed as having schizophrenia at their first admission in theperiod 1970 through 1999 at an age greater than 14 years, and who were aliveand living in Denmark for at least 1 year during the period 1980 through 1997,and who had not previously been recorded as outpatients. A first admissionwas defined as the date on which the individual appeared in the Danish PsychiatricCentral Register for the first time, where the records showed that the personhad not been admitted before the beginning of the register in 1969. Schizophreniawas defined as International Classification of Diseases,Eighth Revision code 295 and International Classificationof Diseases, 10th Revision code F20.

Each individual admitted with schizophrenia was matched to a representativerandom subsample of exactly 10 persons of the same sex, who were born thesame year, who were alive and never admitted or who had never been recordedas an outpatient at the particular admission date and age (in days), and whowere living in the country for at least 1 year during the period. This processyielded a total of 53410 control subjects.

A person admitted in 1999 (and his or her matched controls) could befollowed up in the Integrated Database for Longitudinal Labour Market Researchfrom 1980 to 1997, which was from 19 until 2 years before admission. Similarly,a person admitted in 1970 could be followed up from 10 (1980) until 27 (1997)years after admission. However, individuals were only followed up until 25years after the first admission date (and matching date), as the informationwas too sparse during the last 2 years.

When odds ratios associated with a particular year were calculated,only individuals who were alive and who were residing in Denmark were included.To minimize survival bias, controls matched to a particular case were removedif the case individual died or emigrated. Controls admitted after the matchingdate were kept in the analyses.

Because of the method of sampling controls, odds ratios associated witha year before the admission year can be interpreted as incidence rate ratios,and, furthermore, because of the rarity of schizophrenia, they can also beinterpreted as risk ratios of being first admitted with schizophrenia betweenlevels of marital and labor market status. The odds ratios associated witha year before or after admission can be interpreted as the odds ratio of thespecific marital or labor market status between those who have been admittedwith schizophrenia and those who have not been admitted.²⁶

Annual information on marital status as of December 31 was categorizedas married and living with spouse, cohabiting and living with cohabitee, livingalone, or being a child (which was excluded from the analysis because of agematching). Labor market affiliation during a given year was categorized into6 disjoint categories: (1) fully employed or self-employed, (2) unemployedmore than 1% of the year, (3) student, (4) recipient of social benefits, (5)outside the labor market (eg, housewives, children, and adolescents), and(6) receiving a pension for age or disability. This information was obtainedfrom the Integrated Database for Longitudinal Labour Market Research.

Data were analyzed by conditional logistic regression using the PHREGprocedures in SAS software (version 8.1; SAS Institute Inc, Cary, NC), whereeach case (with matched controls) formed a separate stratum. For each year,odds ratios and 95% Wald confidence bands were computed.

Table 1 shows the numberof cases and controls who formed the background for the present analysis.In any given row of Table 1, thereis a range of ages at admission and cohorts of births. At the first admission,the average age of schizophrenic patients was 29.2 years (SD, 11.0 years)and 36.4 years (16.1 years) for males and females, respectively.

Figure 1 outlines the oddsratios of being single compared with being married for cases and controlsassociated with selected years before and after admission with schizophrenia.Odds ratios for cohabiting individuals were not shown. Up to 19 years before,individuals who were single or cohabiting were more likely to be admittedwith schizophrenia, and single men had a particularly high risk. There wereno sex differences among those who were cohabiting. Up to 25 years after admission,the odds of being single or cohabiting were higher for people with schizophreniathan for controls and were particularly high for single men. The odds ratiosseemed to increase toward the admission year and peak 2 to 3 years after thatyear, after which they fell to the level before admission for both men andwomen. The odds ratio of being single was less elevated for those who hadbeen discharged and not readmitted compared with controls, but the sex differencewas maintained. Note that the reference category consisted of married individualswho were living with their spouse, and furthermore, that it is mandatory toreport a change of address within 5 weekdays.

Odds ratios of (1) being a student, (2) receiving a pension for ageor disability, (3) receiving social security, (4) being outside the labormarket, (5) being unemployed, and (6) being fully employed or self-employedfor cases vs controls were calculated. As long as 15 years before admission,individuals who were not fully employed or self-employed had a significantlyhigher risk of being admitted with schizophrenia. In general, the odds ratiosincreased from those who were students through the unemployed, and from thoseoutside the labor market and those who were pension recipients to those whowere recipients of social security benefits; furthermore, the rates increasedin the years toward the admission year. In the years succeeding the admissionyears, the odds of not being fully employed or self-employed increased, whereaslater they declined to some extent. This pattern was particularly pronouncedfor students, pension recipients, social security recipients, and those outsidethe labor market, but was also evident for the unemployed.

Generally, the odds of being a pension recipient, being a social securityrecipient, and being outside the labor market were less marked for those whohad been discharged and not yet readmitted. Figure 2 displays the odds ratios associated with being unemployedmore than 1% of the year for cases vs controls. The odds ratios associatedwith unemployment increased steadily toward the admission year, after whichthey decreased both for those who were still admitted, or had been readmitted,and for those who had been discharged and not readmitted, when compared withcontrols. Figure 3 shows that individualswho would be or who had been admitted with schizophrenia had a higher oddsratio of receiving a pension than healthy controls. The odds ratios increasedslowly until the matching year, where there was a steep increase, especiallyfor those who were still admitted or had been readmitted, after which theydecreased gradually as individuals in the general population retired.

This population-based study shows that individuals with schizophreniadiffer from the general population with respect to marital and labor marketbehavior 15 to 20 years before, as well as up to 20 to 25 years after, theirfirst admission to a psychiatric hospital. The main finding is the stronglong-term association between schizophrenia, singleness, disadvantaged socioeconomicposition, and labor market marginalization. Furthermore, these relationshipswere relatively unaffected by admission to a psychiatric hospital—exceptfor the transition into disability. These findings have strong implicationsfor understanding the onset and course of schizophrenia.

Although untreated psychosis and acute and insidious onset of illnessare indistinguishable in our study, evidence is added to the conjecture thatschizophrenia does not appear suddenly,^{12 ,27} sinceour study shows that the social disadvantage is present up to 15 to 20 yearsbefore the actual first admission. In the ABC (Age, Beginning, and Course)Schizophrenia Study, which includes information on a sample of 232 first-admittedpatients with schizophrenia, Hafner and colleagues^{16 ,28 - 30} suggestedthat negative symptoms, and presumably associated social disadvantage, appearup to 5 years before admission, and psychotic symptoms up to 2 years beforeadmission. As opposed to the ABC studies, our study findings suggest thatsocial disadvantage, and presumably associated negative symptoms, might emergeearlier, which has been suggested in some studies of premorbid factors.^{31 - 34} Apparently,the etiologically relevant period is very long and the effects of single status,or low social status, accumulate very slowly to an etiologic threshold, whicheventually precipitates an episode of hospitalization.

The long-term association between social disadvantage and schizophreniais not in conflict with the neurodevelopmental hypothesis of schizophrenia,which assumes a disruption in the normal development of the brain, secondaryto genetic and environmental factors.^{35 - 37} Centralto the neurodevelopmental hypothesis of schizophrenia is the idea that neurologicor behavioral abnormalities or deficits preceding overt clinical symptomsof adult schizophrenia characterize those at risk during childhood and adolescence.^{31 ,36 - 40} Althoughour socioeconomic measures apply only to the adult population, our findingsare not in keeping with models of schizophrenia that hypothesize that abnormalitiesdevelop relatively close to the illness onset.⁴¹

Individuals who are young at the admission or matching date were childrenduring the years before, which implies that their marital status and labormarket status are recorded as "child" and "outside the labor market," respectively,and the matching by age accounts for this. Therefore, it is primarily informationon those who are older at the admission or matching date that contributesto the odds ratios measured several years before, which means that the onsetof schizophrenia must be relatively late for these cases. Social isolationand withdrawal are recognized as premorbid and prodromal syndromes,^{15 ,42 - 44} andour finding suggests that individuals with a late disease onset could havehad a prolonged premorbid or initial prodromal phase, where they were livingalone or where they were marginalized from the labor market. Our study couldnot examine whether the length of these early phases of schizophrenia hada predictive value for the illness course,^{45 - 48} andit should be noted that the first hospitalization is only an indicator ofthe first illness episode.¹⁵ However, individualswhose socioeconomic and marital status could be observed several years beforehad an admission or matching date later in the calendar period, which furtherensures that the admission actually is the first admission.

Although patients with early- and late-onset illness might have differentpathways to admission, the finding in the present study cannot be used toresolve the ongoing controversies of whether early- and late-onset schizophreniaare different or similar disorders.^{28 ,49} Onthe basis of our findings, it could be argued that patients with late-onsetdisease might have needed treatment years earlier, which adds weight to thepoint of view that early- and late-onset schizophrenia are more similar disorders.On the other hand, the fact that individuals with late onset manage to stayout of the hospital could mean that they suffer less severe symptoms and,therefore, that there might be differences regarding the symptoms of schizophrenia.However, on the basis of our findings, it could be claimed that early- andlate-onset schizophrenia become indistinguishable over time, which is in accordancewith earlier reports.^{50 - 53} Onthe basis of our findings, it could be argued that patients with late-onsetschizophrenia might have suffered a decline from an already achieved socialstatus (ie, social drift),⁵⁴ or our findingscould suggest a less-than-expected achievement years before the first admission(ie, social selection),⁵⁵ which is in keepingwith other studies.^{56 - 57}

The first hospital episode with schizophrenia is preceded by a periodof years, where future patients increasingly often live alone or are marginalizedfrom the labor market. Three explanations seem immediately evident: The incidenceof untreated psychosis or insidious symptoms is higher close to the admission,or the duration of illness before the first hospital admission is skewed towardshort durations, or future patients might be more likely to remain unemployed,marginalized, or single after entering the labor and marital markets. Thedata from our analysis suggest that the first admission episode with schizophreniais followed by a period of leveling, during which the social status is notfurther deteriorated. However, the odds ratio of becoming a disability pensionrecipient is overwhelming, which could reflect that psychiatrists at mentalhospitals tend to endorse a disability pension once the diagnosis has beenestablished, or that the social welfare system recognizes the disabling impactof schizophrenia. The decreasing rates associated herewith primarily reflectthe transition into age pension in the general population, but possibly alsothat the disability attributed to schizophrenia generally ameliorates.^{58 - 59} One plausible explanation for theleveling could be that hospital treatment is actually beneficial, which couldalso explain the differences between individuals who have been dischargedand not readmitted and those who are readmitted or still admitted.

The odds ratios associated with being single or recipients of disabilitypensions or social security benefits are smaller among individuals who havebeen discharged and not readmitted. This leads us to conclude that the moreseverely ill schizophrenic patients are also more likely to be readmittedor to stay in the hospital. In addition, the data from the present study showdecreasing odds ratios associated with singleness, which could reflect thatpatients find a spouse or a cohabitee around the time of hospitalization,or it could reflect high rates of divorce in the general population duringthe period, an argument that generally applies in our study. We found a sexdifference insofar as the odds ratios associated with singleness were greaterin men than women, which is a well-established finding.^{3 ,60} However,the mean first-admission age was higher in women, which has previously beenreported,^{61 - 62} whereas a decreasingfirst admission rate for schizophrenia in Denmark also was reported, whichprobably could be attributed to a shift to outpatient cases. In our study,the first-admission age was rather high, as only patients who were diagnosedas having schizophrenia at the first admission were enrolled, and the first-admissionrates of schizophrenia in Denmark have been significantly increasing sincethe late 1980s.⁶³

Studies on the cost of illness have shown that schizophrenia imposesan enormous economic burden on both society and the individual person.^{59 ,64 - 65} Our study suggeststhat these costs, and in particular the indirect costs in terms of lost incomeand productivity, could be biased and conservatively estimated, as the costsassociated with the period before the first hospitalization might be underestimated.Furthermore, our study shows that the long-term indirect cost in patientswith relapses is higher than in those who are not readmitted, and presumablythat the quality of life and the social functioning are also higher in thesepatients. Hence, effective treatments used early in the course of schizophreniamay help reduce the costs associated with schizophrenia beyond the immediatereduction in direct costs and in alleviating the personal burden of the illness.^{66 - 68}

In this study, social patterning such as early social drift and putativebiological risk factors are indistinguishable. However, sustained low socioeconomicstatus, rather than acute social problems, is associated with hospital admission,or, alternatively, schizophrenia deteriorates or hinders social achievementlong before the actual admission to a psychiatric hospital.

Corresponding author and reprints: Esben Agerbo, MSc, National Centrefor Register-Based Research, University of Aarhus, Taasingegade 1, Aarhus8000 C, Denmark.

Submitted for publication September 3, 2002; final revision receivedJune 2, 2003; accepted June 12, 2003.

This study was supported by the Stanley Medical Research Institute,Bethesda, Md, and by grant 9600264 from the Danish National Research Foundation,Copenhagen, and the Danish Research Council, Copenhagen. Dr Eaton was supportedby grant MH53188 from the National Institute of Mental Health, Bethesda.

We thank an anonymous reviewer for helpful discussions and commentson the manuscript.