We examined the frequency and stability of family contact with long-term institutional residents during a major deinstitutionalization project. Movers relocated to community accommodation between Assessments 1 (baseline) and 2. Stayers remained institutionalized. We investigated family contact longitudinally over four annual assessments. There was no significant between-group difference in frequency of family contact at baseline, but, subsequently, movers had more frequent family contact than did stayers. There was a significant decline over time in the proportion of stayers with family contact and a significant increase in movers with family contact, with a marked resumption of contact by movers during the first year of community living. Carefully supporting family (re)involvement during the relocation process may be associated with stable, long-term family contact.
Maintaining regular contact with family members throughout life, long after moving out of the family home, is a normative expectation in our society. Family contact is even more important for people with intellectual disability who live outside the family home. First, because people with intellectual disability typically have very small social networks (Robertson et al., 2001; Stancliffe & Lakin, 1999), absence of family contact is likely to have a more marked impact. Second, families provide social, emotional, and practical support, as well as information and advocacy (Robertson et al., 2001). Third, a substantial proportion of the social network of people with an intellectual disability is usually made up of (residential) staff, but staff turnover tends to be high, so the long-term consistency of family contact can act as a buffer against the stress associated with staff leaving (Seltzer, Krauss, Hong, & Orsmond, 2001).
Most people with intellectual disability live at home with family (Fujiura, 1998, Larson, Doljanac, & Lakin, 2005), where access to family contact is usually not a problem. For those who reside in the formal long-term residential system, family contact has been the subject of some research attention, which has shown that the majority of people living in out-of-family-home accommodation have some degree of family contact. The frequency of such contact is very high for people who have recently made a planned (i.e., noncrisis) move out of the family home and whose parents are well (Seltzer et al., 2001).
Other environmental factors associated with increased family contact include living geographically closer to the family (Baker & Blacher, 1993; Baker, Blacher, & Pfeiffer, 1996), living in community (i.e., noninstitutional) settings (Spreat & Conroy, 2002; Spreat, Conroy, & Rice, 1998; Stancliffe & Lakin, 1998), living in smaller community residences (Robertson et al., 2001; Spreat et al., 1998), having moved to one's current residence from the family home (Robertson et al., 2001), family involvement in planning the move from family home to community residence (Schwartz, 2005), and lower staffing ratios (Robertson et al., 2001). However, Emerson et al. (2001) found no significant difference between supported living, small group homes (1 to 3 residents), and large group homes (4 to 6 residents) in the number of contacts with family in the last 3 months.
Younger individuals have more family contact (Robertson et al., 2001; Stancliffe, Dew, Gonzalez, & Atkinson, 2001). This aside, contact seems to be unrelated to resident characteristics. For example, neither adaptive nor challenging behavior are significantly correlated with family contact (Robertson et al., 2001; Stancliffe et al., 2001).
Much research has been focused on children or younger adults who had moved out of the family home relatively recently (Baker & Blacher, 1993; Baker et al., 1996; Blacher, Baker, & Feinfield, 1999; Schwartz, 2005). These researchers reported that almost all participants (93% to 100%) had family contact after leaving home. Blacher et al. (1999) followed families for 3.5 years after placement and found continuing, stable family involvement. In a study with middle-aged individuals involved in a planned (noncrisis) move out of the family home, Seltzer et al. (2001) reported similar results, with high and stable levels of family contact up to 3.8 years after relocation.
Other investigators have examined family contact with younger adults or middle-aged individuals, many with a history of institutionalization (Robertson et al., 2001; Spreat & Conroy, 2002; Spreat et al., 1998; Stancliffe & Lakin, 1998). Because of the way data were reported in these studies, the percentage of participants with family contact is not specified, so it is difficult to draw many conclusions about the relative level of contact in noninstitutional settings. However, Robertson et al. (2001) found that across their institutional and community samples, 72% of residents had interacted with a family member in the previous month. Spreat and Conroy (2002), Spreat et al. (1998), and Stancliffe and Lakin (1998) all reported higher levels of family contact for former institution residents now living in the community. Spreat and Conroy, in a study of younger adults, found that the increased level of family contact was maintained for up to 4 years of community living.
Family contact during and after institutionalization remains an important topic given that there were about 67,000 people with intellectual disability still institutionalized in the United States in 2005 (Prouty, Smith, & Lakin, 2006). Many of these individuals are aging and have been institutionalized for a very long time. We need to clearly document and understand the level and stability of family contact for those in institutions and evaluate the impact of deinstitutionalization on this contact. Other researchers have examined the longitudinal stability of family contact (Blacher et al., 1999; Seltzer et al., 2001; Spreat & Conroy, 2002), but we know of none that have focused on middle-aged adults with several decades of institutional living.
Our purpose was to document the percentage of individuals who had contact with immediate family members, as well as the frequency and longitudinal stability of that contact, for a sample of middle-aged adults with a history of long-term institutionalization, some of whom moved to community placements and others of whom remained institutionalized. We also sought to examine the effect on family contact of moving to the community and identify the factors associated with having or not having family contact
Minnesota Longitudinal Study
The data examined here were drawn from the Minnesota Longitudinal Study which was designed to examine deinstitutionalization of adults with intellectual disability by utilizing annual follow-along assessments (Hayden, DePaepe, Soulen, & Polister, 1995). For movers, the timing of individual participant's move to the community was beyond the control of the researchers. Consequently, some movers had several annual assessments in the institution before they relocated to the community.
Participants moved to the community at various times throughout the entire 6-year duration of the Minnesota Longitudinal Study. We conducted a maximum of four annual community assessments, but many participants had fewer than four because they moved later in the study period (see Stancliffe, Hayden, Larson, & Lakin, 2002, for details). Movers with fewer than three community assessments were excluded from the current study. For the present paper we selected participants with four consecutive annual assessments: stayers with four annual institutional assessments and movers with three annual community assessments following their last annual institutional assessment(s). Included among movers were individuals who had received several annual institutional assessments prior to moving, but for the present study their last institutional assessment before relocation was designated as Assessment 1 (even though it may not have been their first assessment), with their subsequent annual community assessments being identified as Assessments 2, 3, and 4, with the result that all movers relocated to the community between what we label Assessments 1 and 2.
Data collection took place on a rolling schedule, with data for different individuals collected at different times throughout each year. To maintain consistency in the time elapsed between assessments, we conducted annual follow-up assessments for a given individual one year from the date of that person's previous assessments. Across the entire sample, this approach meant that possible seasonal effects associated, for example, with Minnesota's harsh winter climate should have been cancelled out.
There were 155 service users with data from four consecutive annual assessments. This group was made up of 84 movers and 71 stayers. All 155 participants were institutionalized at the time of Assessment 1. Movers relocated to community accommodation between Assessments 1 and 2 and remained living in the community during each subsequent annual assessment. For movers, Assessment 1 was their final institutional assessment. Stayers remained institutionalized for all four annual assessments. Descriptive information about participants is shown in Tables 1 and 2.
Stayers and movers did not differ significantly on age, length of institutionalization, or adaptive behavior. However, stayers had significantly more problematic challenging behavior than did movers, t(153) = −4.93, p < .01. Gender, ethnic group, and level of intellectual disability were not significantly related to group membership.
There were two participating state institutions, both of which were certified as ICFs/MR, located 1 to 1.5 hours drive outside the Minneapolis/St. Paul metro area. At Assessment 1 (baseline), institutional “households” averaged 12.02 residents (SD = 2.67, range = 5 to 26). Most institutional residential buildings contained four households. There were multiple residential buildings on each institutional campus. At Assessment 2, for the 84 movers, there were 54 different participating community residences located throughout Minnesota that averaged 6.36 residents (SD = 3.26, range = 2 to 15). Study participants made up only a part of the total population of the institutions and the community settings. All institutional and community residences were staffed at all times when residents were present.
The main instrument used in the Minnesota Longitudinal Study in relation to residential services was the Minnesota Longitudinal Study Residential Service Provider Interview. This detailed staff interview survey is described in more detail in previous publications (e.g., Stancliffe & Lakin, 1998). Its psychometric properties are unknown. The section on social network includes an item on family contact: “How many times has s/he (the service user) spent time with someone in the immediate family in the past 6 months? (e.g., face-to-face visits and telephone conversations).” Immediate family members were defined as mother, father, brother, or sister. Data provided by staff in response to this question served as our dependent variable.
Missing Family Contact Data
Having repeated assessments for the same individuals, it seemed reasonable to impute missing family contact data from the nearest available assessment for that individual (e.g., if Assessment 4 was missing but Assessment 3 was available, Assessment 3 was imputed as the value for Assessment 4). If two options were available (e.g., Assessment 3 missing, with Assessments 2 and 4 present), we averaged the adjoining scores. The median of the imputed values calculated in this manner was zero.
This method was not appropriate when family contact data for all 4 assessments were missing (14.8% of cases: 16.9% of stayers, 13.1% of movers). We could omit these cases or we could impute missing scores. If staff were unable to provide family contact data for four consecutive assessments, then the family likely had little or no contact. Omitting these cases would overestimate the typical level of family contact because we would systematically exclude cases with zero or near zero contact. Alternatively, we could impute zero contact to these cases. The latter approach seemed appropriate given that most values imputed using available data were zero (see previous paragraph).
We used both approaches and analyzed a sample of 132 (excluding the 23 participants with all assessments missing) as well as the full sample of 155 (missing data for these 23 imputed to zero). The results from both analyses were almost identical; therefore, we have reported the data for the full sample.
Frequency of Family Contact
Challenging behavior was not significantly correlated with frequency of family contact on any of the four assessments, so there was no need to include this type of behavior as a covariate when analyzing family contact frequency.
Frequency of family contact for movers and stayers did not differ significantly at baseline in the institution, t(153) = −1.35, p = .18. Mean frequency of contact at each assessment in shown in Table 3.
Repeated measures multivariate general linear model (GLM) ANOVAS across all four assessments showed that there was a significant overall change in frequency of family contact over time, F(1.54, 235.59) = 3.73, p = .036 (Greenhouse-Geisser corrected significance test). The change over time was best described by a quadratic trend, F(1, 153) = 4.49, p = .036, indicating an increase followed by a decrease. Post-hoc contrasts showed that contact frequency at Assessment 2 was significantly higher than at Assessment 1, but Assessment 2 did not differ significantly from Assessments 3 or 4 (the same pattern was found when movers' data were analyzed alone, whereas there were no significant differences between assessments when stayers were analyzed alone). The Time × Group interaction was not significant. There was a significant difference by group, with movers having significantly more frequent family contact than stayers, F(1, 153) = 4.92, p = .028.
Taken together, these results indicate that movers and stayers had similar levels of family contact before deinstitutionalization; however, following the move to the community, movers had significantly increased family contact overall. The time course of the movers' increase was an initial marked growth at the first community assessment (Assessment 2), followed by a nonsignificant decline to levels above movers' baseline level and above stayers' current level.
Stability of Family Contact
To what extent were the changes in overall frequency of contact actually associated with changes in the number of movers and stayers with family contact? There was no significant difference in the proportion of movers (58%) and stayers (45%) with immediate family contact at baseline, χ2 (1, N = 155) = 2.72, p = .10. For the total sample, over half (52%) had family contact (see Table 4). This overall incidence of family contact remained fairly stable over time, with 56%, 59%, and 52% of the total sample having contact at Assessments 2, 3, and 4, respectively (see Table 5).
Family contact was highly consistent longitudinally. Those in contact at baseline tended to remain so, whereas those not in contact also retained this status over time. At annual Assessment 2, one year after baseline, 136 participants (88%) retained the same family contact status as at baseline, 133 (86%) at Assessment 3, and 130 (84%) at Assessment 4 (Table 4).
The percentage of participants retaining the same contact status each year relative to their status at baseline is shown in Figure 1, separately for movers and stayers and for those with and without contact. As can be seen, over time a consistently high 94% and over of movers with family contact at baseline retained that contact, whereas retention among stayers with family contact dropped to as low as 69% by Assessment 4. These data suggest that, overall, there was a high level of consistency over time in whether participants had contact with their immediate family, although there was a decline among stayers with family contact by Assessment 4.
Figure 2 shows the percentage of movers and stayers with family contact at each annual assessment. There was a significant decline over time in the proportion of stayers with family contact, Q(3) = 11.16, p = .011, and there was a significant increase in the proportion of movers with family contact, Q(3) = 8.16, p = .043.
Figure 2 illustrates the increasing discrepancy over time between movers and stayers in the percentage with family contact. At baseline the percentage of movers with family contact (58.3%) was 1.29 times the percentage of stayers (45.1%), whereas by Assessment 4, this ratio was 1.97 (movers, 66.7%; stayers, 33.8%). Did this change occur because community services were associated with greater resumption of contact for service users who had previously lost contact and/or because stayers experienced more loss of contact?
Resumption of contact
To evaluate resumption of contact, we selected participants with no family contact at a given assessment and examined the proportion that resumed contact at the next annual assessment. Figure 3 shows these percentages for movers and stayers who resumed family contact at each annual assessment. For example, at Assessment 2, one year after baseline, 7.7% of stayers and 25.7% of movers who had no contact at baseline had resumed contact. A significantly greater proportion of movers had resumed contact on moving from institutional to community services (i.e., at Assessment 2), χ2(1, N = 74) = 4.41, p = .036. However, there was no significant difference between movers and stayers on resumption of family contact at Assessment 3, χ2(1, N = 69) = 0.17, p = .68, or Assessment 4, χ2(1, N = 64) = 2.83, p = .093. It appears that there was significantly greater resumption in family contact for movers soon after the time they relocated to the community, but in later years resumption of contact did not differ significantly for movers and stayers.
Loss of contact
To assess loss of contact, we looked at participants with family contact at a given assessment and examined the proportion that no longer had contact at the next annual assessment. These percentages are shown in Figure 4. For example, the pair of bars on the left show that among participants with family contact at baseline, 12.5% of stayers did not have contact at Assessment 2 compared to 6.1% of movers, a nonsignificant difference. There was also no significant difference at Assessment 3, but at Assessment 4 there was a significantly higher proportion of stayers who lost family contact, χ2(1, N = 91) = 10.14, p = .001.
Overall, these findings on resumption and loss of contact show that stayers experienced significantly greater loss of family contact over time, whereas movers encountered significantly higher resumption of such contact. These two factors appear to account for the increasingly divergent proportions of movers and stayers with family contact, as shown in Figure 2.
Only univariate comparisons between movers and stayers have been reported so far. We used backward logistic regression to undertake a multivariate examination of the relation between a number of predictor variables (including status as a mover or stayer) and family contact status (a) at baseline (when all participants were institutionalized) and (b) at Assessment 2 (when movers had relocated to the community). The predictor variables evaluated were consumer characteristics such as adaptive behavior, challenging behavior, gender, and age, as well as environmental variables such as years in the institution and group (mover or stayer).
Younger age, OR = 0.971, and better developed adaptive behavior, OR = 1.011, were predictors of family contact at baseline, with all other predictor variables being eliminated during backward regression. Importantly, group was not a predictor, echoing the univariate finding reported previously of no significant difference in contact status between movers and stayers at baseline. However, at Assessment 2, when movers now lived in the community, group was a significant predictor, OR = 0.451, with stayers being 55% less likely to have family contact, whereas adaptive behavior, OR = 1.010, and age, OR = 0.975, remained predictors of family contact as in the baseline logistic regression. The other predictor variables were eliminated during backward regression for Assessment 2. The absence of challenging behavior as a predictor in either analysis shows that this variable was not significantly related to family contact status. This means that between-group differences in challenging behavior at baseline (Table 1) did not significantly affect the analyses of family contact status.
We examined the frequency and stability of family contact with adults who have intellectual disability and several decades of institutionalization over four annual assessments during a major deinstitutionalization project. Movers transferred to community living between annual Assessment 1 (when all participants were still institutionalized) and Assessment 2. Except for stayers having more problematic challenging behavior, movers and stayers' personal characteristics did not differ significantly at baseline. Challenging behavior was not significantly related to frequency of family contact or to the participant's family contact status. Therefore, it was not necessary to take challenging behavior into account statistically.
Frequency of Family Contact
Frequency of family contact did not differ between movers and stayers at baseline, when all participants were living in an institution. Once living in the community, movers experienced significantly more frequent family contact, with a marked increase at Assessment 2 (the first assessment after the move to the community), followed by a nonsignificant decline (but not to the level of stayers). In other cross-sectional analyses of family contact from the Minnesota Longitudinal Study (Stancliffe & Lakin, 1998) and other longitudinal deinstitutionalization studies (Spreat & Conroy, 2002; Spreat et al., 1998), researchers reported similar findings, so our longitudinal findings reconfirm that community living is more conducive to family contact than is institutional living. These findings suggest that aspects of the relocation process and/or the nature of the new community living arrangements may have supported more families to have more contact or, conversely, that some aspects of the institutional environment reduced contact.
It is difficult to make valid comparisons of the overall frequency of family contact with findings from other studies because of differences in data measurement (what was counted as a contact) and reporting (composite scores, means, percentage of respondents by frequency range), the time period during which contact was measured, as well as the characteristics and circumstances of the study participants. Annual frequency of contact for the present study may be calculated by doubling the medians and means shown in Table 2, which contains data on contacts in the preceding 6 months. Thus, mean annual contact for movers in the present study was between bi-weekly contact (20.2 annual contacts at Assessment 2) and monthly contact (11.6 annual contacts, Assessment 4), with medians of 4 annual contacts in both cases. With appropriate cautions because of the imprecise nature of the comparisons, the amount of contact in the current study appears to be lower than that found by most researchers where annual contact can be estimated (Baker & Blacher, 1993; Baker et al., 1996; Blacher et al., 1999; Schwartz, 2005; Seltzer et al., 2001). All of these studies involved younger participants who had moved out of the family home recently (mostly between 1 and 4 years, previously). This comparative outcome is consistent with research results showing that contact is more frequent with (a) younger participants (as was also found in the present study) and (b) having moved into one's current placement from the family home (Robertson et al., 2001; Stancliffe et al., 2001). However, family contact frequency in the present study appears broadly similar to Spreat and Conroy's (2002) deinstitutionalization data (allowing for the fact that these authors excluded participants with no family).
One important factor in the lower average frequency of contact in the present study is that we found only a little over half (52% to 59%) of the participants had family contact, whereas other researchers reported 94% to 100% of participants with family contact (Baker & Blacher, 1993; Baker et al., 1996; Blacher et al., 1999; Schwartz, 2005; Seltzer et al., 2001). As noted, these studies involved younger participants who had moved out of the family home in recent years. Interestingly, Robertson et al. (2001), whose sample was very similar in age and years in out-of-home placement to the present study, reported 72% of their participants had received contact from families, a figure much closer to our findings.
Seltzer et al. (2001) followed middle-aged adults who moved out of the parental home mostly into community settings. All mothers in this sample were in relatively good health, so the move was not precipitated by a crisis, such as declining maternal health. Seltzer et al.'s reported high levels of contact (99% contact at the Wave 1 assessment) were partly because all mothers were well enough to make contact, whereas family health problems or death during the decades since our participants had left home may have caused some participants in the present study to have no contact.
Stability of Family Contact
Family contact status was highly stable over time, albeit with differing temporal patterns for movers and stayers. Overall, people in contact with their immediate family tended to retain contact over time, and those without such contact continued to be isolated from family. In addition to this underlying stability, a significantly higher proportion of movers had contact with their family once living in the community than they did in the institution at baseline, whereas this proportion declined significantly over time for stayers.
In light of the evident stability of contact, it seems likely that successful efforts to re-establish contact with family could well have long-term benefits. Together with the apparent surge of family contact around the time of moving to the community, our findings suggest that carefully supporting family involvement (and re-involvement) during the relocation process may be beneficial, not only in the short term, but also may be associated with stable, long-term family contact. Research about parent contact with adult offspring living in service settings outside the family home has shown that parental involvement in the relocation process is associated with high, stable levels of parental contact (Schwartz, 2005; Seltzer et al., 2001). Such outcomes are consistent with our finding of generally stable family contact, although we did not measure family involvement in the relocation process per se.
At baseline, participants had lived in their current institution for an average of 27 years, and their mean age was about 45, suggesting that their parents were in their 70s. It is notable that after all that time, more than half still had contact with their immediate family. Although we have no data about family contact during that long period of institutionalization, given our findings on the longitudinal stability of such contact, it seems highly likely that many families had maintained regular contact throughout these years. This somewhat speculative conclusion is a testimony to the remarkable perseverance and determination of these families keep their family connected despite living apart. Other researchers have examined the longitudinal stability of family contact (Blacher et al., 1999; Seltzer et al., 2001; Spreat & Conroy, 2002), but we know of none who have focused on adults with several decades of institutional living.
One notable limitation of our study concerns the amount of missing data. Missing data and participant attrition are an inherent part of longitudinal research, and many researchers deal with this issue by omitting participants with missing data. For example, Spreat and Conroy (2002) dropped 15% of the participants from their longitudinal study for this reason. We chose to impute missing data and retain our full sample, although we also ran the same analyses with a reduced sample omitting those with high levels of missing data and found almost the same results (details available from the authors). The median imputed score was zero (i.e., no family contact), so omitting cases with missing data would have biased our results by systematically eliminating participants with low or no family contact.
In our question on family contact, we asked how many times there had been contact in the preceding 6 months. This means that contacts 7 to 12 months previously were not counted. This should have only a modest impact on the accuracy of our frequency data (doubling our figures should yield a reasonable estimate of annual contact). However, participants' status as having or not having family contact in a given year doubtless was affected. Those with contacts 7 to 12 months previously would be identified as having no contact, so the number of people we identified as having contact is an underestimate. This factor affects the absolute number of those with/without contact to an unknown degree, but it does not invalidate comparisons between movers and stayers or longitudinal comparisons because these were all conducted using equivalent data.
There were several issues for which additional information would have enabled greater depth of understanding. We did not ask who among immediate family members had contact, so we cannot assert confidently that it was necessarily parents who had maintained contact over an average of several decades. We had no data on parents/immediate family who were dead, too ill to visit, or had moved away, so we cannot quantify the extent to which these factors account for participants having no family contact. Future researchers using in-depth interviews about the reasons for family contact could identify family and contextual factors associated with changes in contact.
It seems highly likely that most movers' community residence was located closer to family than the institution they had lived in, but we did not have adequate data to determine this empirically. Community housing was typically located in the participant's county of origin (i.e., where the family had originally lived), whereas the two institutions were 30 to 60 miles from most families, so it is probable that most movers lived closer to their family once they had relocated to the community. Greater geographical proximity is, therefore, likely to be a contributing factor to the increased family contact experienced by movers (see Baker & Blacher, 1993; Baker et al., 1996).
Future researchers should attempt to identify specific reasons for low or no family contact. For those issues amenable to intervention, investigators should evaluate the effectiveness of interventions in reinstating, increasing, and enriching family contact. Further, studies of different forms of community living and support should be designed to assess family contact (and social network more generally) to identify features of various community services that are related to greater or lesser family contact.
This report was supported by Cooperative Agreements H133B80048 and H133B031116 from the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (NIDRR). The contents and opinions expressed herein do not necessarily reflect the position or policy of the U.S. Department of Education, and no official endorsement should be inferred. The authors thank the Department of Human Services (Division of Developmental Disabilities and Guardianship Services for Mentally Retarded), county case managers, and residential staff from state-operated and community-based facilities for their assistance. We also thank Mary Hayden, Barb Polister, Paris DePaepe, Tim Soulen, Jordan Orzoff, and other project staff for their dedicated efforts in conducting the Minnesota Longitudinal Study.
Authors: Roger J. Stancliffe, PhD (firstname.lastname@example.org), Consultant Research Fellow, 176 Holden St., Ashfield, NSW 2131, Australia. K. Charlie Lakin, PhD, Director, Research and Training Center on Community Living, University of Minnesota, Institute on Community Integration, 150 Pillsbury Dr., SE, 102 Pattee Hall, Minneapolis, MN 55455