Adults with intellectual and developmental disabilities (IDD) frequently become frail earlier than the general population, resulting in higher care needs. This population is at risk for institutionalization, or re-institutionalization, into long-term care (LTC). Using a retrospective cohort design to follow 3,034 individuals (18-99 years) living in Ontario, Canada, and assessed with the Resident Assessment Instrument–Home Care, individuals were characterized with a frailty index (FI) for persons with IDD. Survival analyses determined differences in rates of admission to LTC and survival in the community. Frail individuals had greater rates of admission than non-frail individuals, adjusted HR = 2.19, 95% CI [1.81, 2.64]. The FI predicts institutionalization.
Over the past few decades, the population of persons with intellectual and developmental disabilities (IDD) has been growing larger and older compared to previous generations (Bittles et al., 2002). The prevalence of IDD (e.g., Down syndrome, autism spectrum disorder, fetal alcohol spectrum disorder) is almost 1% (Centers for Disease Control and Prevention, 1996; McConkey, Mulvany, & Barron, 2006; Ouellette-Kuntz & Paquette, 2001; Ouellette-Kuntz et al., 2009; van Schrojenstein Lantman-de Valk et al., 2005), with some living into their seventies and eighties (McConkey, Mulvany, & Barron, 2006). Even though the promotion, management, and delivery of care are advancing (Evenhuis, Henderson, Beange, Lennox, & Chicoine, 2001; Haveman et al., 2011), persons with IDD continue to face increased risks for adverse health outcomes. Genetic predispositions, unfavourable social circumstances, less frequent use of generic health services, and environments that support inactivity and poor lifestyle choices contribute to the risk (Haveman et al., 2011).
Within the Canadian health care system, there is a growing emphasis in care in the community. Ontario, Canada's most populous province, has invested significantly in an Aging at Home Strategy, which aims to help seniors “avoid the unnecessary loss of independence and dignity due to premature admission to higher care long-term care homes or hospitals” (Ontario Ministry of Health and Long Term Care, 2009). Home care is a significant method of achieving this goal, by providing services including visiting health professionals, and help from personal support workers, homemaking services, and community support services.
As governments' management of services for aging adults has shifted towards community-based options, so has the organization and funding of services aimed specifically at supporting persons with IDD. In Ontario, deinstitutionalization was completed on March 31, 2009, transferring the remaining institutionalized individuals into the community (Lemay, 2009; Ontario Ministry of Community and Social Services, 2012). This deinstitutionalization corresponds to Canada's policies for persons with IDD, which are based on principles of community inclusion, choice making, and the improvement of quality of life (Johnson & Traustadottir, 2005). However, policy makers, families, and communities are concerned about the risk of re-institutionalization of younger adults into long-term care (LTC) homes intended for seniors (Burgess, 2014; Priest, 2012).
It is well-supported that frailty plays a large role in determining admission to LTC for the general population (Armstrong, Stolee, Hirdes, & Poss, 2010; Hogan et al., 2012; Jones, Song, Mitnitski, & Rockwood, 2005; Rockwood, Mitnitski, Song, Steen, & Skoog, 2006). Frailty can be an effective indicator of health and vulnerability in aging populations (Clegg, Young, Iliffe, Rikkert, & Rockwood, 2013; Rockwood, Hogan, & MacKnight, 2000; Rockwood, Song, & Mitnitski, 2011), and is often regarded as age-related decline associated with higher risk of negative outcomes. In the population of adults with IDD, there is some evidence that frailty is associated with increased care intensity (Schoufour, Evenhuis, & Echteld, 2014). Individuals with IDD generally have high rates of frailty (Brehmer-Rinderer, Zeilinger, Radaljevic, & Weber, 2013; McKenzie, Ouellette-Kuntz, & Martin, 2015a; Schoufour, Mitnitski, Rockwood, Evenhuis, & Echteld, 2013).
Frailty can be challenging to characterize in this population: often indicators of frailty, such as cognitive impairment, limited mobility, and co-morbidities, are lifelong conditions for adults with IDD and not always indicating age-related decline (Evenhuis, Hermans, Hilgenkamp, Bastiaanse, & Echteld, 2012). Therefore, frailty measures for adults with IDD should be sensitive to changes in health and should reflect factors from multiple domains of health.
The objective of this study was to use a frailty index (FI) created for persons with IDD to predict admission to LTC among persons receiving home care services in Ontario, Canada. A measure that identifies individuals at increased risk for LTC could provide the opportunity, given adequate interventions, for frail persons with IDD to maintain independence in the community and quality of life (Evenhuis et al., 2012).
This study is part of a larger program of research (see www.hcardd.ca). A retrospective cohort design was used to follow 3,034 individuals (aged 18 to 99 years) living in Ontario in 2009/2010, (a) who had their first home care assessment between April 1st, 2010 and March 31st, 2014, (b) who had complete data for all covariates of interest, and (c) who were not living in a residential care facility. A total of 4,628 person-years were analyzed. Individuals with an International Classification of Diseases (ICD) or Diagnostic and Statistical Manual of Mental Disorders (DSM) code indicative of IDD were identified in one or more linked administrative data sets that captured physician claims, hospital visits, and emergency room visits. Additional details on the Health Care Access Research and Developmental Disabilities (H-CARDD) cohort and method for identifying persons with IDD are described elsewhere (E. Lin et al., 2013; Lunsky, Klein-Geltink, & Yates, 2013; Ouellette-Kuntz, Martin, & McKenzie, 2015). The study protocol was reviewed and approved by the Queen's University Health Sciences Research Ethics Board and the institutional review board at Sunnybrook Health Sciences Centre, Toronto, Canada.
Three administrative data sets were used: (1) the Continuing Care Reporting System for Long-Term Care (CCRS-LTC) for dates of admission to LTC (Canadian Institute for Health Information, 2014a), (2) the Home Care Reporting System (HCRS) for RAI-HC data (Canadian Institute for Health Information, 2014b), and (3) the Registered Persons Database (RPDB) for age and date of death (Ontario Population Health Index Database, 2015). Data sets were linked using unique encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (ICES).
The Resident Assessment Instrument- Home Care (RAI-HC; Morris et al., 2009) was developed by interRAI, an international collaboration that serves to collect and interpret data on health and social outcomes (see www.interrai.org). The RAI-HC assessment instrument is used by trained case managers as part of regular clinical practice in various jurisdictions in Canada and elsewhere (Hirdes, Ljunggren, et al., 2008) to assess the preferences and needs of those receiving or about to receive home care. The RAI-HC includes items relating to demographic characteristics, home environment, functioning, health, medications, informal support, and formal health services. Various studies have reported the reliability for interRAI items; these values consistently meet acceptable levels of agreement (Gambassi et al., 1998; Hawes et al., 1995; Hirdes, Ljunggren, et al., 2008; Morris et al., 1997).
The accumulation of deficits approach (Rockwood et al., 2006; Searle, Mitnitski, Gahbauer, Gill, & Rockwood, 2008) to measuring frailty has been applied in home care populations (Armstrong et al., 2010) and in adults with IDD (Schoufour et al., 2013). This approach is based on the principle that the more deficits (e.g., signs, symptoms, diseases, impairments) a person has, the frailer he or she will be (Searle et al., 2008). Using this approach, an (FI) comprised of 42 deficits from the RAI-HC was developed specifically for persons with IDD (McKenzie, Ouellette-Kuntz, & Martin, 2015b). FI scores are based on the proportion of deficits accumulated (Rockwood et al., 2006). To facilitate interpretation, the FI was divided into three categories: (a) non-frail (FI ≤ 0.21; roughly 8 or fewer deficits), (b) pre-frail (0.21 < FI ≤ 0.30; roughly between 8 and 12 deficits), and (c) frail ( > 0.30; 12 or more deficits); methods used to develop these categories are described elsewhere (McKenzie, Ouellette-Kuntz, & Martin, 2015b).
Several demographic characteristics were included as potential covariates: age, sex, rural status, living situation, and residential care history. Postal status listed at the time of assessment informed rural status, defined as living in “locations not included in a [Canadian census metropolitan area or census agglomeration] that have an urban population of less than about 10,000 or are a rural area” (Lunsky et al., 2013, p. 157), which is consistent with the Statistics Canada definition (du Plessis, Beshiri, Bollman, & Clemenson, 2001). Living situation at time of assessment was grouped as lives alone, with spouse and/or child(ren), with other family, or in a group setting with nonrelatives. Residential care history refers to the 5 years prior to assessment. Three caregiver status items were also used: (1) caregiver is unable to continue caring (“caregiver inability”); (2) caregiver is unsatisfied with other informal support (“caregiver unsatisfied”); and (3) caregiver expresses feelings of distress, anger, or depression (“caregiver distress”).
In addition to individual items, two measures embedded in the RAI-HC were included as covariates: the Institutional Risk Clinical Assessment Protocol, and the Cognitive Performance Scale. Clinical Assessment Protocols (CAPs) identify common risks for individuals using home care, such as abuse, functional decline, or LTC placement (Hawes, Fries, James, & Guihan, 2007). When CAPs are triggered, interpretations and potential interventions are offered to case managers to better plan home care services (Hawes et al., 2007; Kwan, Chi, Lam, Lam, & Chou, 2000). The Institutional Risk CAP is triggered for individuals with high risk of institutionalization (Mofina & Guthrie, 2014). The Cognitive Performance Scale (CPS), measures cognition level and characterizes individuals on a scale from 0 (intact cognition) to 3 (moderate or greater impairment; Morris et al., 1994). The CPS is highly correlated with level of IDD (Martin, Hirdes, Fries, & Smith, 2007) and served as a proxy of level of IDD.
Prior to analysis, potential effect modifiers were identified. It was hypothesized that an individual's Institutional Risk CAP score, caregiver status, and/or sex group might modify the effect of frailty on the risk of admission to LTC.
Time Frame and Outcome Measurement
The outcome of interest was the first admission to LTC after an individual's first home care assessment, in the period between April 1st, 2010, and March 31st, 2014. The index point was the date of the individual's first home care assessment and time was measured from this assessment to the date of the event; that is, admission to LTC. Death before an admission to LTC was considered as a competing risk event, as it would prevent the possibility of observing the onset of the outcome of interest (Pintilie, 2011).
Individuals were censored if they were alive in the community at the end of the follow-up period (i.e., had neither event). As this analysis used administrative data to identify admissions and deaths across Ontario, it is assumed that the number of individuals lost to follow-up (e.g., left the province, died out of province) is negligible and not associated with the outcome.
Time-to-event analysis measured the relationship between frailty and admission to LTC. To gain a complete understanding of the effect of frailty on the outcome (with the competing risk of death), two modeling approaches were applied (Latouche, Allignol, Beyersmann, Labopin, & Fine, 2013): (a) one that ignored competing risk events by censoring (cause-specific proportional hazards model; i.e., the Cox model), and (b) one that adjusted for competing risk events (the subdistribution proportional hazards model; i.e., the Fine & Gray model; Fine & Gray, 1999).
Model selection was performed separately for each adjusted model. Starting with saturated models, backwards elimination selected significant covariates. Frailty, age, and sex were forced into the model; additional covariates were retained if parameter estimates differed from zero at a significance level of α = 0.05 using the Wald test.
Potential effect modifiers were decided a priori. Covariates were considered effect modifiers if the interaction between the covariate and frailty was significantly associated with admission to LTC, at a level of α = 0.05, while adjusting for other main effects in the model.
Descriptive analyses examined the frequency of independent variables by outcome status. Bivariate models were derived for all independent variables. The multivariate models provided statistics for covariates that were retained during model selection. The following statistics were generated: cause-specific hazard ratios (CSHRs), 95% confidence intervals, type III p-values from the test of overall differences between levels of categorical variables, and p-values of statistical significance for each level of categorical variables. The level of significance was α = 0.05. Analyses were completed with SAS Enterprise Guide v.6.1.
Cumulative incidence of admission to LTC, and the accompanying 95% confidence intervals (Lin, So, Johnston, & SAS Institute Inc., 2012), was calculated for 6 months and 2 years after first home care assessment. Survival in the community was evaluated using Kaplan-Meier curves (using admission to LTC or death as a combined outcome event) and log-rank tests to determine differences between frailty groups.
Almost a quarter of the sample was admitted to LTC (n = 678; 22%), whereas 283 (9%) died before admission to LTC, and the remaining 2,073 (68%) were censored (Table 1). Individuals who were admitted to LTC were more likely to be pre-frail or frail (68%) compared to individuals who were censored (40%). Similarly, of those who died before admission to LTC, 69% were pre-frail or frail.
Overall, the cumulative incidence of admission was 13.8%, 95% CI [12.5, 15.1] at 6 months and 18.6%, 95% CI [17.2, 20.1] by 2 years. Survival in the community was significantly different between frailty groups (χ2 = 314.2, p < 0.001).
Table 2 presents the bivariate and multivariate associations between independent variables and admission to LTC. The subdistribution hazard ratios for these associations did not significantly differ from the cause-specific hazard ratios (data not shown). Interactions between frailty group and potential effect modifiers (Institutional Risk CAP, caregiver distress, sex) did not change the hazard ratios in any of the models with admission to LTC as the outcome (data not shown).
This study evaluated the relationship between frailty and admission to LTC, in a cohort of persons receiving home care in Ontario, using a frailty index (FI) tailored specifically for adults with IDD. The FI was significantly associated with higher rates of admission to LTC, independent of age, sex, rural status, caregiver inability, living situation, and cognition level. Individuals in the frail group had rates of admission over twice as high as those in the non-frail group. Institutionalization is arguably one of the more severe outcomes of frailty (Jones et al., 2005), and is frequently used as an outcome to validate frailty index measures. Studies of Canadian seniors have reported increased risk of LTC admissions with elevated frailty, recording adjusted hazard ratios from 1.2 to 3.3 (Armstrong et al., 2010; Hogan et al., 2012; Jones et al., 2005). In older adults with IDD, a frailty index was associated with increases in care intensity and health care costs (Schoufour et al., 2014). The current analysis helps to expand our understanding of frailty in this population by identifying significant associations in a cohort of both young and old adults with IDD (aged 18-99 years), suggesting that frailty is a relevant consideration in younger adults.
Caregiver status was a strong predictor of admission to LTC. In the general population, the RAI-HC has identified 2% to 8% of caregivers unable to continue caring (Doran et al., 2009); in this study, 14% report this inability. Adjusting for other factors, individuals with struggling caregivers had twice the risk of admission at any given time. The presence of a system that is responsive to the needs of families may explain higher rates of admission; as caregivers indicate a struggle, the system is able to secure higher levels of care. Alternatively, leaving transition planning until a crisis emerges (e.g., caregiver no longer able to provide adequate care) can lead to emergency, temporary, or inappropriate placements for adults with IDD, who may meet this sudden upheaval with confusion and distress (Prosser, 1997). Future research should explore the relationship between caregiver status and admissions to LTC among persons with IDD, and the best way to support informal caregivers in the community.
One notable deviation from the findings of other studies is the lack of sex differences in the risk of admission to LTC. In the general population, some studies report that women are at greater risk of institutionalization (Martikainen et al., 2009; Puts, Lips, Ribbe, & Deeg, 2005), whereas others report that men have the higher risk (Gaugler, Duval, Anderson, & Kane, 2007; Gaugler, Leach, Clay, & Newcomer, 2004; Gruneir, Forrester, Camacho, Gill, & Bronskill, 2013). We expected to see a significant interaction between frailty and sex, considering that women are understood to tolerate deficits better than men (Rockwood et al., 2006). Our descriptive results suggest that men may be more likely to die before admission to LTC. More research is required to understand the patterns of mortality and service utilization in men and women with IDD.
Chronological age is considered an imprecise predictor of adverse outcomes (Mitnitski, Graham, Mogilner, & Rockwood, 2002), especially in adults with IDD (Carmeli & Imam, 2014). A ten-year increase in age contributed roughly to the same risk of admission to LTC care as being characterized as pre-frail. However, accumulating at least 12 deficits (frail) is a stronger predictor of admission to LTC than age in this sample of adults with IDD. Using measures, such as the FI, can improve care practices that rely heavily on age as an indicator of functioning or need.
After adjusting for other factors, triggering the Institutional Risk CAP did not alter risk of admission to LTC. This finding may indicate that home care services are appropriately responding to individuals identified as being most at risk for institutionalization by modifying care plans and offering more care (Diwan, Shugarman, & Fries, 2004). Alternatively, the Institutional Risk CAP currently in use may be less sensitive among adults with IDD.
The large proportion of individuals with IDD residing in institutions (e.g., long-term care homes, psychiatric hospitals; Chaplin, 2004; Ouellette-Kuntz & Martin, 2014) has raised concerns about re-institutionalization (Priebe, 2004). However, the current study reports an insignificant association between residential care history and admission to LTC. Future research should confirm these findings by better defining “residential care facility” in assessments. The current variable in the RAI-HIC could refer to multiple care settings (e.g. group home).
Strengths of this study include the use of a large, population-based cohort of adults with IDD. Given that individuals with IDD age prematurely (Béange, 2002; Lifshitz & Merrick, 2004; Stax, Luciano, Dunn, & Quevedo, 2010), including young adults in the cohort was a better reflection of aging across the lifespan. Although there is no gold standard for frailty, the FI's strong association with adverse outcomes provides some criterion validity (Rockwood et al., 2000), and can be recommended for inclusion in future studies.
This analysis derived hazard ratios using two time-to-event models: the Cox proportional hazards model and the Fine & Gray subdistribution model, to consider competing risks. The Fine & Gray model is proposed to offer better interpretations for decision makers as it considers multiple outcomes (So, Lin, Johnston, & SAS Institute Inc., 2015). However, risk estimates were not significantly different between the models and either hazard ratio can be interpreted.
A few of the study's limitations require acknowledgement. First, by using a baseline measure of frailty to predict a first admission to LTC, we ignored the changes in frailty that could have occurred during the study follow-up. These changes could be due to positive events, such as receipt of home care services, or negative events, such as a trauma (e.g., a fall; Hseih, Rimmer, & Heller, 2012) or change in informal support (e.g., death of a caregiver). These events may be stronger drivers of the move to LTC (Gaugler et al., 2007). Furthermore, some individuals may have become long-stay patients at hospitals prior to the move to LTC, which could also be considered a form of institutionalization. However, one potential use of the FI is as a screening tool, capable of predicting future outcomes before future events are known (de Vries et al., 2011; Drubbel et al., 2014). On a similar note, another limitation of this study is the unspecified status of individuals after their first admission to LTC; it is not known how soon after admission adults die, or if moving into a LTC care facility is a temporary solution to a crisis. Further research should seek to better understand the relationship between frailty and health care services beyond first interaction with both home care and LTC.
Another limitation of this study is the use of administrative data to identify adults with IDD. Although the cohort was developed using algorithms to balance sensitivity and specificity (Lin et al., 2013), it is still possible that some adults in the cohort do not have an IDD, evidenced by the high proportion of the cohort who had “intact” cognition (11%).
Future research should assess the predictive ability in other populations of individuals with IDD, and consider the feasibility of implementing the FI in a jurisdiction using the RAI-HC. These results should be compared to other RAI-HC embedded indicators of institutional risk, such as the Method for Assigning Priority Levels (MAPLe; Hirdes, Poss, & Curtin-Telegdi, 2008). In addition, further consideration should be given to the complexity of the transition from community to a LTC facility, which frequently entails hospitalizations and time spent as alternate-level of care (ALC) patients. ALC patients remain in hospitals even after acute hospital services are no longer needed, often because of inadequate community support or unavailable beds in LTC facilities (Costa & Hirdes, 2010). These patients have higher risk of high levels of functional deficits, complexity, and disease burden (Costa & Hirdes, 2010). Last, a large-scale analysis of caregivers would expand on the strong association noted in this study between measures of caregiver distress and admission to LTC.
Using a measure of frailty, rather than age, may improve care practices. The FI can be calculated using the RAI-HC assessments, as well as other interRAI assessment tools. Frailty is a dynamic construct (Gobbens, Luijkx, Wijnen-Sponselee, & Schols, 2010), which indicates that an individual can both improve and decline over time. Indeed, frailty is perceived to be potentially modifiable (Rothman, Leo-Summers, & Gill, 2008). Intervention strategies are varied and include optimizing nutritional status (Martone et al., 2013), improving psychological well-being (Ní Mhaoláin et al., 2012), and participating in high-intensity resistance training (Singh et al., 2012), among others. Early identification could lead to community-based interventions, and a meta-analysis of 89 trials reported that such interventions postpone admissions to LTC in the general population (Beswick et al., 2008). Given the emphasis on care in the community of adults with IDD, determining risk factors and adequate predictors of admission to LTC homes could be beneficial for appropriate care planning and service allotment.
This manuscript was presented in an oral presentation at the Global Summit on Innovations in Health and Intellectual and Developmental Disabilities. The research reported in this article was funded by the Ontario Ministry of Health and Long-Term Care (MOHLTC) Health Systems Research Fund (Ministry Grant #06671) and is part of the Health Care Access Research and Developmental Disabilities (H-CARDD) Program. The first author (K. McKenzie) was supported by the Syd Vernon Foundation and Queen's University. Analyses were conducted at the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario MOHLTC. The opinions, results, and conclusions reported in this article are those of the authors and independent of the funding sources. No endorsement by the Ontario MOHLTC, the Ontario Ministry of Community and Social Services, or ICES is intended or should be inferred. Parts of this material are based on data and information compiled and provided by Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions and statements expressed herein are those of the author, and not necessarily those of CIHI. The authors wish to acknowledge the contributions of the staff at the Institute for Clinical Evaluative Sciences at Queen's University, especially Marlo Whitehead for assistance in programming and database management. The study protocol was reviewed and approved by the institutional review board at Sunnybrook Health Sciences Centre, Toronto, Canada, and the Queen's University Health Sciences Research Ethics Board.