Introduction
In response to the 1987 Omnibus Reconciliation Act, nursing homes have increasingly implemented restorative care programs, aiming to enhance resident functionality. Since 1998, Medicare has offered additional reimbursement for skilled nursing facilities providing at least two restorative care activities – such as walking, range of motion exercises, and mobility training – for a minimum of 15 minutes daily, six days a week. This reimbursement model was informed by the Nursing Home Case Mix and Quality Demonstration project (Reilly, Mueller, & Zimmerman, 2007). Despite the financial incentives and the widespread adoption of these programs, a comprehensive evaluation of their effectiveness has been lacking, creating a critical gap in guiding best practices and future research in restorative care.
Restorative care, broadly speaking, is a philosophy focused on evaluating and optimizing residents’ functional capabilities (Resnick, Galik, & Boltz, 2013). Two primary approaches exist: dedicated and integrated. Dedicated programs, often aligned with Medicare guidelines, involve trained staff delivering specific activities in 15-minute sessions to residents experiencing functional decline or needing to maintain gains post-therapy. Nursing staff typically identify candidates based on observed decline and perceived potential benefit, yet detailed understanding of participant selection and optimal beneficiary profiles remains limited (Vahakangas, Noro, & Bjorkgren, 2006). The integrated approach advocates for training all staff to incorporate function-promoting activities into all resident interactions. However, robust evidence supporting either approach is scarce. A significant randomized clinical trial (Resnick et al., 2009) exploring an integrated approach showed improvements in specific areas like mobility and balance but not in overall Activities of Daily Living (ADL) function. Quasi-experimental studies have yielded mixed results, ranging from ADL improvement (Chang, Wung, & Crogan, 2008; Morris et al., 1999) to maintenance (Galik et al., 2008) and even deterioration (Resnick et al., 2006). This study aims to contribute to this body of knowledge by evaluating the impact of Medicare-supported restorative care programs, providing valuable insights for program development and implementation.
Long-stay nursing home residents, constituting over 70% of the population on any given day, are particularly relevant to restorative care (Center for Disease Control and National Center for Health Statistics, n.d.). This demographic often experiences progressive ADL dependency, influenced by both resident and nursing home characteristics (Arling, Kane, Mueller, Bershadsky, & Degenholtz, 2007; McConnell et al., 2003; Wang, Kane, Eberly, Virnig, & Chang, 2009). Resident-specific factors associated with increased dependency include advanced age, female gender, and extended stays (Ang, Au, Yap, & Ee, 2006; Peres, Verret, Alioum, & Barberger-Gateau, 2005). Comorbidities like arthritis, diabetes, heart disease, COPD, depression, multiple chronic conditions, and stroke also contribute significantly (Ang et al., 2006; Arling et al., 2007; Fried & Guralnik, 1997; Frytak, Kane, Finch, Kane, & Maude-Griffin, 2001). Physical impairments affecting balance, gait, and range of motion are additional risk factors (Ang et al., 2006; Arling et al., 2007; Fried & Guralnik, 1997; McConnell et al., 2003; Sakari-Rantala, Era, Rantanen, & Heikkinen, 1998; Wang et al., 2009). Nursing home characteristics, such as size, staffing levels, clinical leadership certification, and ownership type, can also play a role (Arling et al., 2007; Wang et al., 2009). Demonstrating the effectiveness of restorative care for long-stay residents can significantly aid nursing homes in participant selection and program design.
The Minimum Data Set (MDS), a standardized assessment tool mandated for all Medicare and Medicaid-reimbursed nursing homes, includes data on restorative care provision. The availability of MDS data linked to the 2004 National Nursing Home Survey (NNHS) offers a unique opportunity to study restorative care participants and evaluate Medicare-supported programs on a national scale. This study leverages this data to: (a) determine the prevalence of restorative care programs in nursing homes and the participation rate of long-stay residents; (b) compare the characteristics of residents participating and not participating in these programs; and (c) evaluate the impact of restorative care on changes in ADL dependency. The central hypothesis is that participants in restorative care will exhibit a slower decline in ADL dependency over 18 months compared to nonparticipants, after accounting for various resident (age, length of stay, cognitive impairment, frailty, comorbidities, mood, social engagement, pain, physical impairments, baseline ADL function, nurse assessment of improvement potential) and nursing home characteristics (Medicare/Medicaid percentages, nursing contact hours, medical director/nursing director certification, facility accreditation). The findings are expected to inform future directions in restorative care research and practice.
Design and Methodology
This research employed a longitudinal analysis of MDS data integrated with the 2004 NNHS, a nationally representative survey. Data access was granted by the National Center for Health Statistics (NCHS), with analyses conducted at the Minnesota Census Research Data Center due to the use of restricted NCHS variables. Resident data originated from the MDS, while nursing home characteristics were sourced from the NNHS.
Sample Selection
The study focused on long-stay nursing home residents aged 65 and older deemed likely to benefit from restorative care, requiring a minimum residency of six months. Exclusion criteria included residents who were bedfast, in a vegetative state, with a life expectancy of fewer than six months, or with end-stage diseases, as functional improvement or maintenance was considered unlikely. Residents receiving occupational, physical, or speech therapy were also excluded to isolate the effects of restorative care programs.
Longitudinal Data Collection
An 18-month longitudinal dataset was constructed using admission, quarterly, significant change, and annual MDS assessments from 2003-2006 for eligible residents. Data points were structured quarterly (baseline, 3, 6, 9, 12, 15, and 18 months), utilizing the most recent assessment within each 3-month period. Significant change assessments constituted only 7% of the data. The 18-month timeframe was chosen to capture long-term effects, aligning with recommendations for evaluating rehabilitation sustainability (Forster et al., 2009). Baseline variables were derived from the initial full or admission MDS assessment, and NNHS data, treated as baseline characteristics, provided nursing home variables.
Figure 1.
Figure 1: Illustrates the sample selection process and attrition rates throughout the 18-month study period.
Attrition analysis revealed that 60% of residents remained in the dataset at 18 months.
Table 1. Study Variables
| Restorative care (MDS items) | Activities of daily living (ADL) dependency (MDS items) | Resident characteristics (MDS items) | Nursing home characteristics (NNHS items) |
|---|---|---|---|
| Passive range of motion | ADL self-performance scale | Age | % Residents with Medicare reimbursement |
| Active range of motion | Length of stay | % Residents with Medicaid reimbursement | |
| Splint/brace assistance | Cognitive Performance Scale | Hours of patient contact with nursing staff | |
| Bed mobility | Frailty | Medical Director certification | |
| Transferring | Number of disabling diseases | Director of Nursing certification | |
| Walking | Mood | Facility accreditation | |
| Dressing/grooming | Social engagement | ||
| Eating/swallowing | Pain | ||
| Amputation/prosthesis care | Number of physical impairments | ||
| Communication | Staff assessment of resident’s ability to improve ADL dependency | ||
| Other |
Table 1: Details the variables utilized in the study, categorized by their source and focus.
MDS Variable Measures
MDS Psychometric Validation
The MDS 2.0, employed in this study, has demonstrated robust psychometric properties in numerous studies. Over 85% of MDS items exhibit adequate inter-rater reliability (κ > .6) (Mor, 2004). Scales measuring ADLs, cognitive function, and medical diagnoses are particularly reliable and valid. Measures of pain, mood, and social engagement show less ideal psychometric properties (Casten, Lawton, Parmelee, & Kleban, 1998; Frederiksen, Tariot, & De Jonghe, 1996; Gambassi et al., 1998; Hartmaier et al., 1995; Lawton et al., 1998; Mor, Intrator, Unruh, & Cai, 2011, Morris et al., 1990; Williams, Li, Fries, & Warren, 1997). Data collection by nursing home staff, through interviews and record reviews, introduces potential inter-facility variability in measurement quality (Lum, Lin, & Kane, 2005; Shin & Scherer, 2009). Facilities tending to over- or under-report data often do so consistently across items (Wu, Mor, & Roy, 2009). Accounting for individual nursing homes in analyses is therefore crucial. Despite these limitations, the MDS offers comprehensive, systematically collected data valuable for real-world evaluations of restorative care.
ADL dependency was quantified using the MDS ADL-7 scale, an additive scale comprising seven MDS items assessing self-performance in bed mobility, transferring, dressing, eating, toilet use, hygiene, and bathing. Each item is scored on a 5-point Likert scale (0=independent to 4=total dependence), with total scores ranging from 0-28 (higher scores indicating greater dependency). The ADL-7 exhibits strong internal consistency (Cronbach’s alpha > .85) (Mor, Intrator, Unruh, & Cai, 2011), moderate to strong correlations with other ADL measures (r = .58–.79) (Frederiksen et al., 1996; Lawton et al., 1998; Snowden et al., 1999), and predictive validity for nursing assistant time utilization (Morris et al., 1999). It is also sensitive to change in observational and interventional studies (Carpenter, Hastie, Morris, Fries, & Ankri, 2006; Grando et al., 2009; Morris et al., 1999).
Restorative care activities are documented in the MDS as the number of days within the past week a resident received at least 15 minutes of specific activities: passive and active range of motion, splint/brace assistance, and skill practice in bed mobility, transferring, walking, dressing, grooming, eating, swallowing, amputation-prosthesis care, communication, or other skills. Preliminary analyses explored three operationalizations of restorative care: (a) dichotomous (any activity received), (b) count of activities, and (c) continuous variable summing days of activity provision. Only the dichotomous variable predicted ADL dependency in univariable models (p = .02) and was thus used in the multivariate model as a time-varying predictor of restorative care receipt. NNHS data provided information on the prevalence of nursing homes employing specially trained personnel for restorative care programs.
Resident Characteristics Measures
Resident characteristics were measured using MDS items including age, gender, race/ethnicity, and length of stay. A comorbidity variable, counting ten disease categories associated with ADL decline (dementia, stroke/paralysis, arthritis, cancer, COPD, heart disease, diabetes, neurological disease, depression, and eye disease), was created from MDS items.
Cognitive impairment was assessed with the MDS Cognitive Performance Scale (CPS), a six-level scale (scores 0-6, higher scores indicating greater impairment) with high sensitivity and specificity (.94) (Hartmaier, Sloane, Guess, & Koch, 1994; Hartmaier et al., 1995; Morris et al., 1994). Physical impairment was scored on a 0-4 scale, awarding one point for impairment in each of balance, mobility, range of motion, and voluntary movement (Cronbach’s alpha = .70). Frailty was measured using the Edmonton Frailty Scale, a multidimensional scale incorporating MDS items related to cognition, general health, function, social support, medication use, nutrition, mood, continence, and performance (scores 0-17, higher scores indicating greater frailty) (Armstrong, Stolee, Hirdes, & Poss, 2010).
Social engagement was measured by the MDS social engagement scale (six items on interactions and activity engagement, scores 0-6, higher scores indicating greater engagement, Cronbach’s alpha = .79) (Dubeau, Simon, & Morris, 2006; Mor et al., 1995). Depressive symptoms were assessed with the Burrow’s mood scale (seven MDS items, scores 0-14, Cronbach’s alpha > .70, validated against Hamilton and Cornell scales) (Burrows, Morris, Simon, Hirdes, & Phillips, 2000). Pain was categorized (none, mild, moderate, severe) using the MDS Pain Scale (two items on frequency and intensity, 93% agreement, κ = .71 concurrent validity with nurse-administered visual analog scale in post-acute settings) (Fries, Simon, Morris, Flodstrom, & Bookstein, 2001). A dichotomous MDS item indicating staff perception of a resident’s potential for ADL independence improvement was also included (Vahakangas et al., 2006).
NNHS Variable Measures
Nursing home characteristics from the NNHS included: percentage of residents with Medicare/Medicaid reimbursement, nursing staff patient contact hours, medical director and director of nursing certification status, and facility accreditation. Nursing staff contact hours were measured as average full-time equivalent hours spent on patient care by registered nurses, licensed practical nurses, and nursing assistants. Dichotomous variables indicated medical director certification in relevant specialties (family medicine, internal medicine, geriatrics, palliative care) and director of nursing certification from recognized bodies. Facility accreditation by organizations like the Joint Commission was also represented as a dichotomous variable. State of nursing home residence was incorporated as a random effect to account for state-level policy variations and regional differences in MDS data collection.
Statistical Analysis
Descriptive statistics were used to determine the prevalence of restorative care provision and participation, and to characterize the study sample. Univariable chi-square and t-tests compared baseline characteristics between restorative care participants and nonparticipants. Linear mixed models evaluated the effect of restorative care on ADL dependency over 18 months. An autoregressive covariance structure was used to model the within-subject correlation over time. Resident variables were treated as time-varying predictors, and nursing home variables as static predictors. A linear model was found to best represent ADL dependency change over time. Variables with p < .05 in univariable analyses were included in the multivariable model.
The multivariable model is specified as:
| ADLkij=β0+b0k+b0i+β1Xkij1+…+βpXkijp +εkij, k=1,…,K, i=1,…,nk, j=1,…,nki |
|---|
Where k = state, i = subject ID, j = time point, p = number of covariates, nk = subjects in state k, nkj = time points for subject i in state k.
Ethical Considerations
The 2004 NNHS data collection received approval from the NCHS Research Ethics Review Board. The University of Minnesota deemed this analysis of de-identified survey data exempt from federal human research participant protection regulations. Analysis of restricted data at the NCHS Research Data Center was also approved by the NCHS Ethical Review Board.
Results
Restorative Care Prevalence
Approximately 67% of nursing homes reported offering restorative care programs. Resident participation in restorative care increased after baseline, stabilizing thereafter (24% at baseline, rising to 35-37% in subsequent quarters). Walking, range of motion exercises (passive and active), and dressing/grooming were the most frequently provided restorative care activities.
Table 2. Restorative Care Activity Participation Rates Over Quarters
| Restorative care activity | Baseline, N = 7,735 | Quarter 1, N = 6,719 | Quarter 2, N = 6,462 | Quarter 3, N = 6,052 | Quarter 4, N = 5,664 | Quarter 5, N = 5,279 | Quarter 6, N = 4,676 |
|---|---|---|---|---|---|---|---|
| Passive range of motion | 8.9% | 11.0% | 10.9% | 11.6% | 11.6% | 13.4% | 12.9% |
| Active range of motion | 10.7% | 16.8% | 17.7% | 17.2% | 17.3% | 17.4% | 16.8% |
| Splint/brace assistance | 2.2% | 2.6% | 2.7% | 2.8% | 2.9% | 3.6% | 3.3% |
| Bed mobility | 2.4% | 2.2% | 2.0% | 2.3% | 2.2% | 2.5% | 2.4% |
| Transferring | 4.4% | 6.0% | 5.7% | 5.5% | 5.3% | 5.9% | 5.9% |
| Walking | 9.3% | 15.7% | 16.1% | 16.0% | 15.1% | 14.4% | 14.7% |
| Dressing/grooming | 4.7% | 6.3% | 6.3% | 6.3% | 6.3% | 6.9% | 7.0% |
| Eating/swallowing | 2.5% | 3.3% | 3.3% | 3.5% | 3.3% | 4.0% | 4.1% |
| Amputation/prosthesis | 0.1% | 0.2% | 0.2% | 0.3% | 0.3% | 0.3% | 0.2% |
| Communication | 0.3% | 0.5% | 0.5% | 0.5% | 0.5% | 0.6% | 0.5% |
| Other | 1.7% | 2.8% | 3.0% | 3.1% | 3.0% | 3.1% | 2.7% |
| Received any activity | 24.1% | 35.0% | 35.9% | 35.9% | 35.3% | 36.9% | 36.9% |
Table 2: Shows the percentage of residents participating in various restorative care activities across the study quarters.
Participant vs. Non-participant Characteristics
The majority of long-stay residents were white females, averaging 85 years old, residing in urban, for-profit nursing homes for approximately 3.2 years. Restorative care participants differed significantly from nonparticipants in several baseline characteristics. Participants exhibited greater cognitive impairment (p < .001), a higher number of disabling diseases (p < .001), more physical impairments (p < .001), and higher baseline ADL dependency (p < .001). Conversely, they were less likely to be assessed by nurses as having the ability to improve ADLs (p < .001), less likely to experience pain (p < .001), and more likely to reside in for-profit nursing homes (p < .001). Nursing homes providing restorative care to participants were less likely to be non-profit (p = .008) and less likely to have directors of nursing with specialty certifications (p = .02) or facility accreditation (p = .006).
Table 3. Baseline Characteristics Comparison: Restorative Care Participants vs. Non-participants
| Characteristic | Total sample, N = 7,735 | Restorative care participants, N = 1,864 | Non-participants, N = 5,871 | Chi-square | t-test | p-value |
|---|---|---|---|---|---|---|
| % / Mean ± SD | % / Mean ± SD | % / Mean ± SD | ||||
| Resident traits | ||||||
| Age (years) | 84.8 ± 8.0 | 85.2 ± 8.1 | 84.6 ± 8.0 | -2.8 | .005 | |
| Length of stay (years) | 3.2 ± 3.4 | 3.3 ± 3.5 | 3.1 ± 3.3 | -2.1 | .04 | |
| Female (%) | 75.4% | 74.5% | 75.7% | 1.2 | .27 | |
| White race (%) | 88.9% | 89.4% | 88.7% | 0.8 | .38 | |
| Urban nursing home (%) | 52.2% | 47.5% | 61.1% | 25.0 | <.001 | |
| For-profit nursing home (%) | 59.2% | 52.8% | 53.8% | 57.5 | <.001 | |
| Cognitive Performance Scale (0-6) | 2.3 ± 1.4 | 2.5 ± 1.3 | 2.2 ± 1.4 | -9.0 | <.001 | |
| Frailty (0-15) | 6.3 ± 2.2 | 6.2 ± 1.9 | 6.3 ± 2.3 | 1.5 | .15 | |
| Disabling diseases (0-10) | 2.5 ± 1.4 | 2.8 ± 1.4 | 2.4 ± 1.3 | -10.3 | <.001 | |
| Mood scale (0-14) | 0.9 ± 1.6 | 1.0 ± 1.7 | 0.9 ± 1.5 | -4.4 | <.001 | |
| Social engagement (0-6) | 2.5 ± 1.7 | 2.5 ± 1.7 | 2.5 ± 1.6 | 0.3 | .76 | |
| Pain: None (%) | 60.3% | 63.6% | 59.2% | 26.5 | <.001 | |
| Pain: Mild (%) | 24.0% | 24.0% | 24.0% | |||
| Pain: Moderate (%) | 13.7% | 11.4% | 14.4% | |||
| Pain: Severe (%) | 2.0% | 1.0% | 2.4% | |||
| Physical impairments (0-4) | 2.7 ± 1.2 | 3.0 ± 1.1 | 2.5 ± 1.2 | -14.7 | <.001 | |
| Baseline ADL dependency (0-28) | 15.3 ± 7.6 | 17.5 ± 7.2 | 14.6 ± 7.7 | -14.2 | <.001 | |
| Nurse assessed ADL improvement ability (%) | 26.9% | 17.5% | 29.9% | 111.3 | <.001 | |
| Nursing home traits | ||||||
| % Medicare reimbursement | 3.0% | 1% | 3.0% | 5.2 | .02 | |
| % Medicaid reimbursement | 20.0% | 26% | 18% | -7.0 | <.001 | |
| Nursing staff contact hours | 73.3 ± 110.2 | 70.5 ± 107.1 | 74.2 ± 111.2 | 1.2 | .22 | |
| Medical Director certification (%) | 84.4% | 82.6% | 84.9% | 5.7 | .02 | |
| Director of Nursing certification (%) | 37.3% | 34.6% | 38.2% | 7.7 | .006 | |
| Facility accreditation (%) | 11.1% | 9.4% | 11.7% | 7.0 | .008 |
Table 3: Presents a detailed comparison of baseline characteristics between residents who received restorative care and those who did not.
Impact of Restorative Care on ADL Dependency
The multivariable model results (Table 4) indicated that restorative care programs did not significantly affect the change in ADL dependency over 18 months (p = .12) after adjusting for resident and nursing home characteristics. Figure 2 illustrates the predicted change in ADL dependency for both groups. The adjusted mean baseline ADL dependency score was 17.9 for participants and 14.0 for nonparticipants. Both groups showed similar rates of ADL dependency increase over 18 months, with participants increasing by 0.5 points and nonparticipants by 1.0 point.
Table 4. Multivariable Analysis: Restorative Care Effect on ADL Dependency Change
| Characteristic | Coefficient | SE | p-value |
|---|---|---|---|
| Intercept | -1.6 | .42 | N/A |
| Resident Traits | |||
| Age (years) | .01 | .003 | .15 |
| Length of stay (years) | .03 | .01 | .001 |
| Cognitive Performance Scale (0-6) | .38 | .03 | <.001 |
| Frailty (0-15) | .63 | .01 | <.001 |
| Disabling diseases (0-10) | .10 | .02 | <.001 |
| Mood scale (0-14) | -.16 | .02 | <.001 |
| Social engagement (0-6) | -.39 | .02 | <.001 |
| Pain: Mild (ref: Severe) | -.57 | .20 | .004 |
| Pain: Moderate (ref: Severe) | -.37 | .20 | .07 |
| Pain: None (ref: Severe) | -.55 | .20 | .006 |
| Physical impairments (0-4) | .97 | .03 | <.001 |
| Baseline ADL score (0-28) | .68 | .01 | <.001 |
| Nurse indicated no ADL improvement ability | -.13 | .07 | .06 |
| Restorative care receipt over time | .09 | .06 | .12 |
| Nursing home traits | |||
| % Medicare reimbursement | -.07 | .22 | .76 |
| % Medicaid reimbursement | -.02 | .08 | .75 |
| Nursing staff contact hours | -.0001 | .0003 | .68 |
| Medical Director no certification | .19 | .09 | .04 |
| Director of Nursing no certification | .07 | .07 | .31 |
| Facility no accreditation | -.11 | .10 | .29 |
Table 4: Presents the coefficients, standard errors, and p-values from the multivariable model evaluating the effect of restorative care on ADL dependency change.
Figure 2.
Predicted change in activities of daily living (ADL) dependency by receipt of restorative care after adjusting for confounding resident and nursing home traits.Figure 2: Illustrates the predicted changes in ADL dependency over 18 months for residents who received restorative care compared to those who did not, after adjusting for confounding factors.
Discussion
Despite the majority of nursing homes offering Medicare-supported restorative care programs, only about one-third of long-stay residents participated. Participants exhibited greater baseline ADL dependency, cognitive and physical impairments, and more comorbidities. However, both participants and nonparticipants showed similar rates of ADL dependency decline, suggesting a broader need for functional maintenance interventions among long-stay residents.
Baseline restorative care participation was 24%, increasing to 37% over 18 months. Reasons for low participation rates may include resident choice, perceived lack of value, or organizational barriers (Benjamin, Edwards, Ploeg, & Legault, 2014). Nursing staff-identified barriers include time constraints, difficulty motivating residents (especially those with cognitive impairment), learned dependency, and fear of falls (Resnick et al., 2008). State Medicaid reimbursement policies, which vary and may not always cover restorative care, also likely play a role, particularly as 20% of the sample were Medicaid recipients. The inclusion of state of residence as a random effect in the model aimed to mitigate this confounding factor.
The most clinically significant baseline difference was higher ADL dependency in participants (mean ADL score 17.5 vs. 14.6 for nonparticipants), consistent with prior findings (Berg et al., 1997). This likely reflects targeted program enrollment for residents with more apparent functional impairments. However, the similar ADL decline in both groups highlights the functional maintenance needs of most long-stay residents. Reimbursement policies favoring higher-dependency residents might also contribute to the exclusion of less dependent individuals.
The lack of a differential ADL dependency decline in restorative care participants suggests that Medicare-reimbursed programs, as currently structured and implemented, may not achieve their intended functional outcomes. The intensity and structure of these programs may be insufficient. The 15-minute daily/six days a week threshold, while reimbursable, may be inadequate, particularly for activities like walking, where 30 minutes is recommended for older adults (Nelson et al., 2007). The study also lacked information on program structure (dedicated vs. integrated approach). Dedicated approaches, where specialized staff provide care at specific times, are likely more common. While limited evidence exists comparing approaches, a systematic review suggests integrated restorative care in controlled settings can improve ADL dependency, physical function, and activity (Resnick et al., 2013). Alternatively, participants’ higher baseline dependency may have limited their potential for improvement. However, a recent trial demonstrated that integrated restorative care improved physical function and reduced falls in long-stay residents even with moderate-severe cognitive impairment (Galik et al., 2014).
These findings should not justify program elimination or reimbursement cessation. A significant portion of moderately dependent long-stay residents did not participate but could likely benefit. These results, alongside recent evidence, support shifting towards restorative care as an overarching philosophy rather than isolated activities (Resnick, Galik, & Boltz, 2013). Future research should explore the efficacy of expanding restorative care to all long-stay residents and compare integrated and dedicated approaches (Resnick, Galik, Remsburg, & Pretzer-Aboff, 2009).
Limitations
Study limitations include the focus on ADL dependency as the primary outcome. Activity-specific outcomes (e.g., gait speed for ambulation programs) could offer more granular insights (Forster, Lambley, & Young, 2010), but were unavailable in the dataset. MDS restorative care measures also lacked the sensitivity to detect dose-response relationships. The use of the latest MDS assessment per quarter might introduce bias towards residents experiencing significant changes, though this is minimized by the low proportion of significant change assessments (7%). An admission cohort design would have been ideal but was infeasible due to the low number of newly admitted long-stay residents (<1.5%). Sample attrition (40% at 18 months due to death or discharge) introduces potential bias, as differences between censored and surviving residents could affect ADL dependency progression. Despite these limitations, the study provides valuable information on restorative care participation and effects, informing practice and future research.
Conclusion
This national study of nursing home MDS data revealed that while most facilities offer restorative care programs, less than a third of long-stay residents participate. Participants had higher ADL dependency but similar progression rates compared to nonparticipants, suggesting potential benefits for nonparticipants. Implementing restorative care as an integrated philosophy, rather than a program of discrete activities, warrants consideration. Future research should compare integrated and dedicated approaches and assess the impact of offering restorative care to all long-stay residents.
Funding
This project was supported by grants from the National Institutes of Child Health and Human Development (Grant Number K12HD055887) and the National Institute on Aging (grant number 1R03AG037127-01A1). The findings and conclusions are those of the authors and do not necessarily reflect the views of the Census Bureau, Research Data Center, National Center for Health Statistics, Centers for Disease Control and Prevention, or National Institutes of Health. K. M. C. Talley was supported by the John A. Hartford Foundation and the Claire M. Fagin post-doctoral fellowship.
Acknowledgements
Research was conducted while K. M. C. Talley, K. Savik, & H. Zhao were Special Sworn Status researchers at the Minnesota Census Research Data Center, U.S. Census Bureau. Data was provided by the National Center for Health Statistics, Centers for Disease Control and Prevention. This article has been screened to ensure no confidential data is revealed.
