Does the length of institutionalization matter? Longitudinal follow‐up of persons with severe mental illness 65 years and older: shorter‐stay versus longer‐stay

As part of the process of de‐institutionalization in the Swedish mental healthcare system, a reform was implemented in 1995, moving the responsibility for services and social support for people with severe mental illness (SMI) from the regional level to the municipalities. In many ways, older people with SMI were neglected in this changing landscape of psychiatric care. The aim of this study is to investigate functional levels, living conditions, need of support in daily life, and how these aspects changed over time for older people with SMI.

� After correction for the effects of ageing, older adults with severe mental illness and more accumulated time in a mental hospital demonstrated decreased functioning and increased need for support over time, compared to older adults with severe mental illness but less time in a mental hospital � The longer-stay group was much more likely to have been hospitalized prior to the 1980s, when significant changes in care began to be implemented � It is important for clinical practice to be aware of the lingering effects of long-term stay in mental hospitals under potentially out-dated treatment methods and expectations

| INTRODUCTION
The process of deinstutionalization initiated in the second half of 20th century impacted not only the organization and provision of mental health services, but it also affected everyday lives of people living with mental illness. Generally, studies about deintitutionalization indicate that former mental hospital patients benefit from the transition from mental hospitals to community care. 1 The studies following the process of deinstutionalization differ, however, in the assessment instruments used, duration of follow-up time, and rehabilitation efforts. 2,3 Differences are also noted in relation to the type of populations followed. For instance, the commonly applied terms of short-and long-stay patients are problematic. Typically, a 6-month 3,4 or 12-month cut-off is used to differentiate between long-and shortterm stay, 5,6 but at least one study used a 3-year cut-off. 7 In general, long-stay patients refer to a group of people with extensive needs of care and support. They have experienced long periods of inpatient care, typically, due to severe and persistent symptoms, 4 but also due to the lack of appropriate and sufficient alternatives to inpatient care. 8 Further, despite the ambition to reduce inpatient care in favour of community-based care, inpatient care still takes a majority of the budget for psychiatric care 9 and extended periods of hospitalization are not uncommon.
International studies following the effects of deinstitutionalization on long-stay patients, in particular, provide mixed results ranging from reports of improved social and psychiatric functioning (e.g., 10,11,12 ) to no changes (e.g., 13 ) or deterioration. 14,15 These mixed results can be explained by not only different diagnoses but also different ages at baseline and different follow-up periods. 3,11 Across studies, mean baseline age ranged from 41 to 67, but in most studies mean baseline age was less than 65 (mean = 55), meaning that most individuals studied experienced de-insitutionalization prior to late adulthood. Length of follow-up in these studies ranged from 6 months to 10 years for prospective studies, with an average of about 4 years of follow-up across studies.
To our knowledge, none of the follow-up studies take into account that the patients age over time and that functional levels, psychiatric symptoms and social contacts change with increasing age.
Taking age into account is especially important in studies that cover longer follow-up periods and where the studied population is older at the beginning of the studied period. Based on our review of the literature, there are no studies that have older adults with severe mental illness (SMI) as the target group for the investigation and follow-up. More knowledge is needed about this group because both research and social policy recognize that the number of older people with SMI will increase sharply during the first decades of the 21st century. [16][17][18][19] The aim of this study is to investigate functional levels, living conditions, need of support in daily life, and how these aspects changed over time for older people with SMI. The focus is on the impact of length of time of inpatient care in mental hospitals. We address two important concerns. First, we focus particularly on the de-institutionalization experiences of older adults with SMI. Second, these adults were assessed over a period of 15 years regarding changes in life situation, functional level and needs and support in everyday life.

| METHODS
In this study we used data from repeated surveys and national registers.

| Data
At the outset of the Mental Health Care Reform in 1995, the Na-  In addition, based on the survey participants' personal identification numbers, data were drawn from National registers; Statistics Sweden, Cause of death register and National Patient Register. The current study was approved by the Regional Ethical Review Board in Linköping. men were more likely to be diagnosed with psychosis and women were more likely to be diagnosed with depression or neurosis.

| The National Board of Health's Inventory form
Nine items from the National Board of Health's Inventory (NBHI) assessing functioning were combined to create a daily functioning measure: hygiene, household, food preparation, finances, telephone, daily activity, travel to another county, travel to another city, and contact with authorities. 22 Each item was rated by the interviewer on a scale from 1 (manage without help) to 3 (cannot manage), so higher scores indicate more functional difficulties. This measure was not included in the 2011 interview.

| General assessment of functioning
The Global Assessment of Functioning (GAF) is a measure of symptom and social disability. GAF is a numeric scale to rate subjectively the social, occupational, and psychological functioning of an individual. 23 Scores range from 100 (extremely high functioning) to 1 (severely impaired), so higher scores indicate better functioning.    were diagnosed as psychotic, compared with 43.9% of the shorterstay group; the second group was more likely to be diagnosed as neurotic (19.6% vs. 6.1%). Living situations for the two groups over time are presented in Figure 2. Differences between groups were significant at each wave: the shorter-stay group was more likely to living in their own dwelling, with or without support, and the longer-stay group was more likely to be living in an institution or special accommodation.

| Shorter-stay versus longer-stay
There were no significant differences between groups in work situation or primary source of income. NBHI was significantly different between the two groups at all three waves, and differences remained significant after regression-correction for age. Moreover, the differences between the groups increased somewhat over time, so that difficulties as measured by the NBHI of the longer-stay group increased more over time (see Figure 3a). The shorter-stay group actually improved on the NBHI from 1996 to 2001. Both groups declined in functioning as measured by the GAF, but the longer-stay group declined faster over time (see Figure 3b). Again, differences between the groups remained significant after regression-correction for age. The difference between the two groups was most evident on the CAN. Group means in client ratings after regression-correction for age are presented in -1227 which resulted in the longer-stay group being more likely to have never married. That particularly applies to people with psychosis.
Psychosis was the dominant diagnosis in the longer-stay group and was previously considered an incurable chronic disease.
The fact that there were no differences between the longer-stay group and the shorter-stay groups in being under-employed and thus

| Strengths and limitations
One of the main strengths of this study is not only a follow-up period of 15 years, but also a negligible drop-out level: over 90% of targeted population was interviewed at each wave. Still, the drop-out due to mortality was marked and resulted in reduced power at later waves.
The composite nature of the data set (both interview and register data) could be a limitation. However, the national register data was used only to identify the length of time individuals had spent in a mental hospital, data that could not reliably be collected via selfreport. Thus, use of national register data added to the validity of the method for grouping SMI-O:P individuals into shorter-stay and longer-stay groups. Another limitation was that length of stay was confounded with diagnosis (longer-stay and psychosis); however, the relatively small sample size prevented further investigation of that confound. Finally, in the current study longer-stay was defined as more than 3 years of accumulated hospitalization, as opposed to the more typical 6-month cut-off seen in the literature. To test the impact of this decision, analyses were repeated using a 6-month limit and results were similar, although reduced sample size in the shorterstay group resulted in reduced power.