Influence of Cognitive Impairment on Rehabilitation Received and Its Mediating Effect on Functional Recovery

Abstract

Background:

Cognitive impairment (CI) has been reported to negatively impact rehabilitation outcomes. Knowledge about differences in rehabilitation received in dependence of CI as a potential mediating factor is limited.

Objective:

To analyze whether CI affects amount and frequency of rehabilitation received and if associations between CI and rehabilitation outcome are mediated by the provided amount of therapy.

Methods:

Observational cohort study in ward-based geriatric rehabilitation consecutively including 373 patients (mean age 82.0±6.69 years, mean MMSE 23.66±5.31). Outcome measures were amount, frequency, and type of multi-professional therapy sessions and rehabilitation outcome assessed with the Barthel Index (BI). Cognitive status was measured with the Mini-Mental-State Examination (MMSE) classifying three patient subgroups according to cognitive status.

Results:

Patients with more severe CI received least total therapy hours (TTH) (MMSE < 17, 13.67±6.58 versus MMSE 17–26, 16.12±7.19 and MMSE > 26, 17.79±8.88 h, p = 0.014) and were less often included in occupational therapy (MMSE < 17, 48.9%versus MMSE 17–26, 65.5%and MMSE > 26, 71.4%, p = 0.019) and group-based physiotherapy (MMSE < 17, 73.3%versus MMSE 17–26, 88.5%and MMSE > 26, 81.2%, p = 0.027). Regression models showed that CI negatively impacted TTH (β= 0.24, p = 0.003) and rehabilitation outcome (β= 0.41, p = 0.008). In the mediation model, TTH accounted for 23.18%(p < 0.001) of the relationship between CI and rehabilitation outcome.

Conclusion:

Cognitive impairment negatively impacted rehabilitation received. The lower TTH partly mediated the negative association between CI and rehabilitation outcome. Future research should identify specific barriers to therapy provision and optimal length, intensity, and dosage of rehabilitation programs to optimize rehabilitation outcomes in CI.

Keywords

Cognitive impairment functional outcome rehabilitation therapy provision

INTRODUCTION

Demographic change has led to a growing demand for Geriatric Rehabilitation services to support the increasing number of older people who experience acute or sub-acute loss in function after illness and hospitalization [1 –3]. A consensus has been achieved that multimodal and multiprofessional rehabilitation programs are crucial for older people to protect or improve health and quality of life, to optimize functional independence, and to avoid admission to nursing home [2 , 5]. Among multi-morbid conditions, cognitive impairment (CI) is highly prevalent in geriatric rehabilitation and is associated with negative consequences on the rehabilitation process. Depending on sample specifics and cut-off criteria up to 70%of geriatric patients undergoing rehabilitation have been classified as cognitively impaired [6 –8].

Although many studies in the rehabilitation setting could show that cognitively impaired patients have the potential to achieve functional improvements during rehabilitation, CI has also been reported to negatively impact rehabilitation. Patients with CI, especially those with severe impairment, were often excluded from rehabilitation and related research [9]. However, despite a proven positive effect of rehabilitation, it is this most affected group of patients which presents with increased rates of institutionalization and mortality [10] as well as increased length of stay, low functional outcomes, and high rates of adverse incidents [10 –13].

As the elderly population including persons with cognitive impairment increases rapidly, a deeper understanding of whether and how rehabilitation in patients with CI can be improved and specifically tailored to the target group, is dearly needed.

Up to now, the mechanisms by which CI patients have poorer rehabilitation outcomes is not well understood. Mechanisms driving this association may be multifactorial. Functional outcomes may be impacted due to reduced capacity to adapt to the hospital environment [14, 15], to lower mental and physical ability to fully participate in rehabilitation interventions [16], to recover from stressors, or to compensate the risks associated with acute illness [17]. But other factors leading to worse outcomes may also be amenable to modifiable organizational factors and processes of care including clinicians’ decision-making on therapy interventions or time provision [18 –20]. If so, this would have important clinical implications for the management of patients with CI in the rehabilitation setting and the provision of appropriate rehabilitation regimen.

However, despite its clinical relevance, comprehensive and differentiated data on therapies received in rehabilitation patients with different cognitive impairment levels is scarce. Moreover, knowledge about the potential mediating effect of amount of therapy received, as a mechanism explaining lower functional recovery in patients with CI, is lacking so far. The majority of studies examining the association between cognitive impairment and rehabilitation success does not specify nor take into account the actual extent of therapies received during rehabilitation, implying identical rehabilitation regimes for patients with and without cognitive impairment. Geriatric rehabilitation most often took the form of interdisciplinary or multidisciplinary care, but the amount and the types of actual therapies provided were rarely defined [21, 22]. Only a few studies, mostly from stroke or hip fracture rehabilitation, reported actual amount of therapy received during rehabilitation with heterogeneous results. While some studies found no evidence that patients with CI received less therapy [23], others suggest that patients’ cognitive status affected the extent of therapy received during rehabilitation [20 , 25].

The interpretation and generalization of this literature is limited because studies have focused on selected groups of patients within specific rehabilitation settings, such as hip fracture- [23 –26] or stroke rehabilitation [20], used multiple methods to determine CI, or did not differentiate between CI levels [20, 25] and amount or type of therapies received [20 , 26]. Furthermore, these studies left unanswered the key question, if differing therapy provision in dependence of cognitive status impacted rehabilitation outcomes and can therefore be identified as a mechanism underlying the association between CI and lower rehabilitation outcome.

Based on the identified knowledge gap the aim of the study was 1) to analyze whether CI affects amount, frequency, and type of therapeutic interventions received during rehabilitation and 2) to determine, whether the relationship between CI and rehabilitation outcome is mediated by the amount of therapy received.

MATERIALS AND METHODS

Study design and setting

We present an observational cohort study conducted at two rehabilitation wards of a geriatric hospital. The wards were identical in regard to size, number of beds, staffing, and usual care. Random allocation to one of the two wards was based solely on the availability of rehabilitation beds in each ward. All individuals admitted for rehabilitation had been transferred from an acute setting after being assessed by a multidisciplinary team determining if the patient shows the need, rehabilitation potential, and motivation to further increase in functional level and to participate in an intensive rehab program. Inpatient geriatric rehabilitation was recommended for medically stable, older patients affected by multimorbidity and geriatric syndromes. Chronological age, place of residence and the presence of cognitive impairment were not used as exclusion criteria. Keeping with the principle “rehabilitation before care” and to allow a most unrestricted access to rehabilitation in the highly vulnerable geriatric population, patients with uncertain rehab prognosis in the face of clinical complexity and frailty could also be transferred to rehabilitation “on a trial basis” [1]. Regular duration of rehabilitation was limited to a maximum of 21 days by the statutory health insurance with a limited opportunity for restricted extension if further inpatient rehabilitative therapy is necessary in case of a delayed rehabilitation progress. Patients were recruited consecutively within 48 h after admission for geriatric rehabilitation between 1/2011 and 12/2012. Exclusion criteria were medical and/or psychological conditions not allowing the application of cognitive and functional assessments such as acute confusion (delirium), aphasia, severe visual or auditory impairment, severe psychiatric disorders, severe functional-motor deficits, or inadequate language level. Patients with a length of stay < 14 days were excluded from analyses representing participants with premature discharge by different reasons, such as unexpected medical events and transition to acute care not representing the target group of patients receiving regular geriatric rehabilitation. The study was approved by the Ethics Committee of the Medical Faculty at the Heidelberg University in accordance with the Helsinki Declaration. Written informed consent was obtained from all participants or their legal representatives

Rehabilitation program

Each participant received geriatric rehabilitation, depending on personal goals and rehabilitation needs following a structured multimodal and multi-professional rehabilitation approach. A patient’s individualized rehabilitation plan was based on Comprehensive Geriatric Assessment initiated at admission to rehabilitation with regular evaluation of progress during weekly team case conferences [1]. The main focus of the rehabilitative program was functional recovery to gain the highest possible level of independence in activities of daily living. Each patient received formal input from physiotherapy according to the patients’ needs, with the therapy plan complemented by occupational- and art therapy, speech therapy, and psychological interventions, including individual as well as group-based treatments. Formal therapies were offered only on weekdays. All patients received informal nurse-led therapy by activating care with early participation in self-care and practicing activities throughout the day. Therapists were familiar with and trained in treatment methods and the rehabilitation of multi-morbid geriatric patients and patients with CI. Type, frequency, and duration of therapies were objectively recorded for each patient and therapy session in the hospital’s electronic documentation system (ORBIS, Agfa HealthCare, Bonn Germany).

Patient characteristics

Characteristics of the study participants included admission age, sex, living arrangement before admission (community dwelling versus institutionalized), indication for geriatric rehabilitation by diagnostic groups, and length of stay (LOS). The Mini-Mental State Examination (MMSE) was used as a screening instrument to assess participants’ cognitive function on admission (range 0 to 30, higher scores indicating better cognitive functioning). The MMSE is a well-established, reliable, valid, and comprehensive cognitive screening instrument to grade patients’ cognitive status, which has high reliability and is easy to administer. The instrument examines attention, orientation, immediate and short-term recall, language, drawing, and the ability to follow simple verbal and written commands [27]. Functional independence in the domains of personal care and mobility were evaluated by the Barthel Index for Activities of Daily Living (BI) administered by observation of the patients’ status by a trained nurse (maximal score 100, a higher score denotes greater independence). The BI is well-established as a valid assessment instrument in geriatric care and for evaluation of geriatric rehabilitation as well as in other public health services. It records 10 domains of basic daily functions, vital from the point of view of discharge planning and destination [28, 29].

Motor capacity was assessed by Tinetti’s Performance Oriented Motor Assessment (POMA) including sub-tests on gait and balance with a maximal score of 28 [30]. The POMA is a commonly used, valid clinical test [31] for assessing mobility deficits in older persons by external expert ratings with higher scores indicating less functional deficits. Descriptive clinical variables were extracted from patient‘s charts and assessed by trained research practitioners.

Definition of outcomes

Primary outcome measures were total therapy hours (TTH) calculated by combining amount of all multidisciplinary therapy sessions in hours received during rehabilitation and rehabilitation outcome operationalized as improvement of functional status (BI discharge –BI admission). Secondary outcome measures were total frequency of therapy visits and discipline-specific type, frequency, and number of visits for physiotherapy (PT) and occupational therapy (OT). The following outcomes were documented: 1) percentage of patients who received PT and OT; 2) individual PT including physical exercise, strengthening, balance, mobilizing exercises, and individual instructions to patients in gait and transfer training; 3) group-based strength, flexibility, and balance exercises; 4) passive physical therapy such as massage, heat application, or lymphatic drainage; 5) total number and length of PT (overall PT) comprising individual and group based PT, machine-based strength and resistance training, passive PT; and 6) total number and length of OT. Additionally, physiotherapists documented number of planned but not successfully executed sessions (so called therapeutic attempts) due to rejection, lack of motivation or acute adverse events.

Data analysis

Characteristics of patients were summarized by frequencies (%) for categorical variables or means with standard deviations (SD) for continuous variables for the whole sample and stratified by cognitive status groups. Three cognitive groups were distinguished: no cognitive impairment (NCI) MMSE 30-27, mild to moderate cognitive impairment (MCI) MMSE 26-17, or moderate to severe cognitive impairment (SCI) MMSE ≤16 [32]. Distribution of continuous variables was checked by comprehensive descriptive analysis and visual inspection to justify normality assumption. We considered the BI as quasi-interval scaled and used parametric analyses for admission and change scores [33].

Differences in patient baseline characteristics and in number and amount of therapy visits received during rehabilitation between the three groups according to cognitive status were analyzed by one-way analyses of variance (ANOVA) or Chi-squared test with respective post-hoc tests for pairwise comparisons (LSD, or Chi-squared) in case of overall significance (p-value < 0.05). As sample sizes were unequal between groups, we used Welch’s ANOVA. As these analyses were exploratory, p-values have to be interpreted descriptively.

To evaluate the association between CI and TTH, linear regression models (LM) were run with the independent variable MMSE score (continuous). Model 1 considers the unadjusted effect (no covariates included) and model 2 includes the relevant clinical confounder functional status at admission (measured by the BI) and age to get adjusted effects on TTH. All variables were forced into the models in a single step. Regression coefficient β with standard error (SE) and p-values are given for the effect of interest for the MMSE score and for the covariates in model 2. Additionally, the same regression models were run with MMSE as categorical variable (cognitive groups NCI, MCI, SCI) to get adjusted means per cognitive group (least square means) with SE. Impact of CI on improvement of functional status was assessed the same way as described above for TTH.

To test whether any relationship between MMSE and improvement of functional status (BI discharge-BI admission) is mediated by TTH, a mediation model was run including the following steps to yield direct, indirect (a^*b), and total effects [34]: 1) LM of TTH (mediator) as a function of MMSE (yielding effect a = first factor of the indirect effect); 2) LM of improvement of functional status, as a function of MMSE (yielding the total effect); 3) LM of improvement of functional status as a function of MMSE and TTH (yielding the direct effect) and effect b = second factor of the indirect effect. In each model, the BI admission score was included as a covariate as this was found to be a strong predictor for functioning at discharge in other studies [13]. To get comparable estimates, standardized regression coefficients were reported alongside the respective p-values. Additionally, 95%confidence intervals (CI) for the indirect effect were reported by applying bootstrapping (2000 bootstrap samples were used) as proposed by Preacher and Hayes (2008) [35]. The null hypothesis of no mediation effect was rejected when the 95%did not contain 0. A two-sided p-value ≤0.05 was considered statistically significant. Statistical analysis was performed using SPSS statistics 27.0 (IBM, Armonk, NY, USA) and SAS version 9.4 for the regression models and mediation analysis.

RESULTS

Participants

Among 476 persons consecutively and inclusively screened for eligibility, 50 patients (10.3%) were not included into the study due to predefined exclusion criteria and other 53 patients (11.1%) dropped out due to early discharge from rehabilitation. Excluded patients did not differ in age or sex but had a significantly shorter length of stay (13.98±9.31 versus 22.5±5.32, p < 0.001), were more cognitively impaired (MMSE score 21.79±6.11 versus 23.66±5.31, p < 0.01), and had worse BI admission scores (46.45±24.25 versus 56.50±19.99, p < 0.001).

Patient characteristics

The analyzed sample population comprised of 373 older (mean age 82±6.7 years), mainly female patients (58.2%) with impaired functional and motor status indicated by mean BI admission score of 56.5±20.0 and mean POMA score of 11.82±7.33. The most frequent main diagnoses were musculoskeletal disorders including fractures (44.2%) and cardiovascular diseases (24.1%), 9.8%of patients were institutionalized before admission. Mean length of stay (LOS) was 22.51±5.32 days.

Cognitive subgroups differed significantly with respect to MMSE scores (NCI 28.09±1.04 (n = 154) versus MCI 22.65±2.50 (n = 174) versus SCI 12.38±2.70 group (n = 45), p < 0.001).

Comparisons of subgroups showed significant differences in age, functional status and motor status, and prevalence of institutionalization. Cognitive intact patients (NCI) were significantly younger than patients with MCI or SCI (NCI 80.38±6.93 versus MCI 83.25±6.22 and SCI 83.13±6.47 years, p < 0.001). The highest prevalence of institutionalization before admission was found in the group with SCI (5.3%in NCI versus 11.2%in MCI and 20%in SCI, p < 0.001). BI score (NCI 60.84±18.77 versus MCI 56.67±19.47 versus SCI 41.00, p < 0.001) and POMA score (NCI 13.18±7.18 versus MCI 11.53±7.35 versus SCI 8.29±6.58, p < 0.001) differed significantly between the three cognitive groups with patients having more pronounced CI showing lower BI and POMA scores (see Table 1).

Table 1

Characteristic for the total sample and subgroups stratified by cognitive status

Characteristics	Total (n = 373)	NCI (n = 154)	MCI (n = 174)	SCI (n = 45)	p
MMSE, score (0–30), mean (SD)	23.66 (5.31)	28.42 (1.04)	22.65 (2.50)	12.38 (2.70)	< 0.001^a +++
Age, y, mean (SD)	82.05 (6.69)	80.38 (6.93)	83.25 (6.22)	83.13 (6.47)	< 0.001^a ++
Sex, female, (%)	58.2	57.1	60.3	53.3	0.658^b
Admission main diagnosis, (%)					0.073^b
Musculoskeletal disorders	44.2	40.3	43.7	60.0
Cerebrovascular disease	7.5	9.7	2.4	8.9
Cardiovascular disease	24.1	20.1	29.9	15.6
Other internal disease	17.7	22.1	15.5	11.1
Miscellaneous	6.4	7.8	5.7	4.4
Institutionalized before admission (%)	9.8	5.3	11.2	20.0	0.017^b +
BI admission, score (0–100), mean (SD)	56.50 19.99)	60.84 (18.77)	56.67 (19.47)	41.00 (18.73)	< 0.001^a +++
POMA, score (0–28), mean (SD)	11.82 (7.33)	13.18 (7.18)	11.53 (7.35)	8.29 (6.58)	< 0.001^a +++
LOS, days, mean (SD)	22.51 (5.32)	22.79 (5.87)	22.34 (5.87)	22.18 (4.70)	0.066^a

Presented are comparisons of subgroups according to cognitive status. p-values for the effect of group were analyzed by a) univariate ANOVA or b) Chi-squared test with respective post-hoc tests; NCI, no cognitive impairment (MMSE ≥27); MCI, mild cognitive impairment (MMSE 26-17); SCI, moderate to severe cognitive impairment (MMSE≤16); POMA, Tinetti’s Performance Oriented Motor Assessment, MMSE, Mini-Mental State Examination; LOS, length of stay; BI, Barthel Index; SD, standard deviation;⁺significant difference between SCI and both other groups (MCI = NCI);⁺⁺significant difference between NCI and both other groups (MCI = SCI); ⁺⁺⁺NCI > MCI > SCI.

Differences between subgroups according to cognitive status in rehabilitation therapy received

Frequency of total therapy visits differed significantly between cognitive groups with patients with SCI receiving least visits during rehabilitation (SCI 33.71±14.78 visits versus NCI 41.49±17.21 and MCI 39.06±15.24, p = 0.014). For PT, a significant difference for amount and frequency in overall PT received was found, again with patients in the SCI-group receiving least PT therapy hours (SCI 10.84±5.15 versus NCI 13.86±6.07 and MCI 13.15±4.98 PT hours, p = 0.005) and least therapy visits (SCI 28.86±11.93 versus NCI 32.62±12.89 and MCI 31.53±11.76 visits, p = 0.023). When examining specific types of PT, CI showed a significant negative impact on amount and frequency of therapies received. Patients with CI received significantly less individual PT hours (4.03±2.11 hours in SCI and 4.20±2.44 in MCI versus 4.87±2.48 in NCI, p = 0.020) while differences in individual PT frequency did not reach significance (p = 0.124).

Table 2

Comparison of frequency and amount of therapy received stratified by subgroups according to cognitive status

Type of therapy		Total	NCI	MCI	SCI	p
Total therapy	hours, mean (SD)	16.51 (6.57)	17.79 (8.88)	16.12 (7.19)	13.67 (6.58)	0.004^a +
	visits, mean (SD)	39.42 (16.17)	41.49 (17.21)	39.06 (15.24)	33.71 (14.78)	0.014^a +
PT overall	hours, mean (SD)	13.16 (5.54)	13.86 (6.07)	13.15 (4.98)	10.84 (5.15)	0.005^a +
	visits, mean (SD)	31.42 (12.39)	32.62 (12.98)	31.53 (11.76)	26.89 (11.93)	0.023^a +
PT individual	Included, n (%)	373 (100)	154 (100)	148 (100)	71 (100)	1.0^b
	hours, mean (SD)	4.46 (2.44)	4.87 (2.48)	4.20 (2.44)	4.03 (2.11)	0.020^a ++
	visits, mean (SD)	10.29 (4.63)	10.86 (4.44)	9.96 (4.77)	9.64 (4.60)	0.124^a
PT group-based	Included n (%)	312 (83.64)	125 (81.2)	154 (88.5)	33 (73.3)	0.027^b +
	hours, mean (SD)	3.98 (2.39)	4.28 (2.86)	3.87 (1.93)	3.39 (2.29)	0.116^a
	visits, mean (SD)	8.02 (4.76)	8.59 (5.63)	7.82 (3.90)	6.79 (4.59)	0.116^a
PT passive	Included, n (%)	159 (42.6)	68 (44.2)	78 (44.8)	13 (28.9)	0.138^b
	hours, mean (SD)	2.29 (1.80)	2.25 (1.67)	2.47 (2.00)	1.38 (0.82)	0.129^a
	visits, mean (SD)	6.77 (4.36)	6.71 (4.23)	7.23 (4.64)	4.31 (2.21)	0.081^a
PT therapeutic attempt	visits, mean (SD)	3.03 (2.95)	2.64 (3.90)	3.12 (2.69)	2.95 (3.15)	0.731^a
OT overall	Included, n (%)	246 (66.0)	110 (71.4)	114 (65.5)	22 (48.9)	0.019^b +
	hours, mean (SD)	4.79 (4.38)	5.09 (4.85)	4.77 (4.16)	3.39 (2.59)	0.253^a
	visits, mean (SD)	9.01 (7.10)	9.62 (7.80)	8.84 (6.63)	6.82 (5.27)	0.227^a

Presented are comparisons of the cognitive status groups analyzed by a) univariate ANOVA or b) Chi-squared test with post-hoc LSD or Chi-squared tests; ⁺significant difference between SCI and both other groups (MCI = NCI); ⁺⁺significant difference between NCI and both other groups (MCI = SCI); NCI, no cognitive impairment (MMSE ≥27); MCI, mild cognitive impairment (MMSE 26-17); SCI, moderate to severe cognitive impairment (MMSE ≤16); PT, physiotherapy; OT occupational therapy; therapeutic attempt refers to number of planned but not successfully executed sessions.

Additionally, patients in the SCI-group were significantly less often included in group-based PT with 73.3%of patients in the SCI-group versus 81.2%in the NCI-group and 88.5%in the MCI-group (p = 0.027). Prevalence of patients receiving passive PT and OT and differences in amount and frequency of group-based and passive PT and OT were lowest in the SCI-group but differences between cognitive groups did not reach significance. Therapeutic attempts, indicative for planned, but not successfully executed therapy sessions, did not differ between cognitive groups.

Association between CI and TTH and improvement of functional status

Linear regression analysis identified CI (MMSE scores) as a significant explanatory variable for the primary outcome TTH (β= 0.24, p = 0.003) (Table 3) and improvement of functional status (β=0.41, p < 0.008) (Table 4). Patients with lower cognitive status received less TTH and received lower functional improvements. The adjusted models showed that higher age was associated with lower TTH (β= –0.21, p = 0.001) (Table 3, model 2). A decrease of mean TTH was observed along the three cognitive status groups (NCI > MCI > SCI) which was stable over both models.

Table 3

Association between CI and TTH

	β (SE) β standardized	p	Mean TTH (SE) for subgroups according to cognitive status
	MMSE score continuous		NCI	MCI	SCI
Unadjusted model 1			Unadjusted mean (SE)
MMSE score	0.24 (0.08)	0.002	17.79 (0.63)	16.12 (0.60)	13.67 (1.17)
	0.16
Adjusted model 2			Adjusted mean (SE)
MMSE score	0.24 (0.08)	0.003	17.61 (0.64)	16.37 (0.60)	13.30 (1.20)
	0.16
BI admission score	–0.04 (0.02)	0.063
	–0.10
Age	–0.21 (0.06)	0.001
	–0.17

Presented is association between CI (MMSE continuous scores) and TTH (total therapy hours) with corresponding p-values analyzed by an unadjusted linear model without covariates (model 1) and an adjusted one (model 2) with BI score at admission and age as clinical covariates; mean TTH per cognitive status group was determined by running the same models with MMSE as categorical variable (NCI, MCI, SCI); β, regression coefficient; SE, standard error; BI, Barthel Index; NCI, no cognitive impairment (MMSE ≥27); MCI, mild cognitive impairment (MMSE 26-17); SCI, moderate to severe cognitive impaired (MMSE ≤16); MMSE, Mini-Mental-State Examination.

Table 4

Association between CI and improvement of functional status

	β (SE) β standardized	p	Mean BI gain (SE) for cognitive subgroups
	MMSE core continuous		NCI	MCI	SCI
Unadjusted model 1			Unadjusted mean (SE)
MMSE score	0.14 (0.15)	0.338	12.47 (1.22)	11.88 (1.15)	11.16 (2.30)
	0.05
Adjusted model 2			Adjusted mean (SE)
MMSE score	0.41 (0.15)	0.008	13.27 (1.20)	11.99 (1.11)	7.89 (2.28)
	0.14
BI admission score	–0.25 (0.07)	<0.001
	–0.32
Age	–0.09 (0.11)	0.443
	–0.04

Presented is the association between CI (MMSE continuous scores) and improvement in functional status with corresponding p-values analyzed by an unadjusted linear model without covariates (model 1) and an adjusted one (model 2) with BI score at admission and age as clinical covariates; mean improvement of functional status (BI discharge –BI admission scores) per cognitive status group was determined by running the same models with MMSE as categorical variable (NCI,MCI,SCI); β, regression coefficient; SE, standard error; BI, Barthel Index; NCI, no cognitive impairment (MMSE ≥27); MCI, mild cognitive impairment (MMSE 26-17); SCI, moderate to severe cognitive impaired (MMSE ≤16); MMSE, Mini-Mental-State Examination.

Functional improvement during rehabilitation decreased according to cognitive status (NCI > MCI > SCI). Functional status at admission was a significant confounder (β= –0.25, p < 0.001) with higher improvements in patients with lower functional admission scores (Table 4, model 2). The effect of cognitive status was more pronounced in the adjusted model, taking into account relevant covariates.

Mediating effect of TTH on the relationship between CI and improvement of functional status

A mediation analysis was performed to analyze whether the relationship between CI and improvement of functional status was mediated by TTH. Figure 1 presents the results of the mediation model: the total effect of cognitive performance on rehabilitation outcome was β_c = 0.148, p < 0.001. The significant association of CI on TTH was β_a = 0.193, p < 0.001, which in turn was associated with lower improvements of functional status (β_b = 0.178, p < 0.001). The indirect effect (product of β_a and β_b) was β_axb = 0.035 with a 95%CI of 0.013 to 0.061 based on 2000 bootstrap samples. The indirect effect accounts for 23.28%of the total effect, reducing the direct effect of CI to β_c = 0.114, p = 0.030. Results indicated that TTH partially mediated the relationship between CI and improvement of functional status.

Fig. 1

Mediation model on the relationship between CI and improvement of functional status through TTH. Presented are results of analyses of the mediation of total therapy hours (TTH) on the relationship of cognitive impairment (MMSE scores) gains in Barthel Index (BI) scores. Given are standardized coefficient β for total (c), direct unmediated (c‘) and indirect effects (product of coefficients (a), and (b)) of CI on improvement of functional status (BI discharge –BI admission); (a), association between CI and TTH; (b), association between mediator TTH and improvement of functional status; effects are given adjusted for BI admission scores; ^*p≤0.05, ^**p≤0.001.

DISCUSSION

This study explored differences in rehabilitation therapy received between patient groups with varying cognitive impairment levels in post-acute inpatient geriatric rehabilitation and its mediating effect on the relationship between CI and rehabilitation outcome. Results demonstrate that CI negatively influenced amount, number and type of multidisciplinary interventions received during rehabilitation. The total amount of therapies received was associated with rehabilitation outcome and partially mediated the relationship between CI and improvement in functional status.

Differences between subgroups according to cognitive status in therapy received and impact of CI on TTH

Data on actual therapy provision and possible disparities between elder patients with differing cognitive states in rehabilitation are scarce and provide conflicting results. While some studies found no evidence that patients with CI received different therapy regimes [23], others suggest that patients’ cognitive status negatively affected the extent of therapy received during rehabilitation [20 , 25]. The interpretation and generalization of this literature is limited. Previous studies focused on selected groups of patients within specific rehabilitation settings, such as hip fracture- [23 –26] or stroke rehabilitation [20], had methodological shortcomings such as multiple or invalid assessment of cognitive impairment [20], exclusion of patients with SCI or lack of staging of CI [20, 25], or did not differentiate interventions received [20 , 26].

The present study avoided the above-mentioned shortcomings. Regression analysis supported the notion that CI negatively influenced overall amount of therapy received and generalized it to an unselected sample of geriatric inpatients with varying cognitive impairment levels. This finding was specified by a comprehensive report and analysis of therapeutic interventions, differentiated for discipline-specific visits. Comparisons between patient groups stratified according to cognitive status showed that CI impacted therapy provision in different ways, resulting in lowest amount and frequency of multidisciplinary and PT interventions in patients with SCI. First, differences between groups were obvious in highest exclusion rate in group-based PT and OT in patients with SCI. Second, inclusion rates for passive PT as well as number and amount of passive and group-based PT and OT sessions for those who were included in these interventions showed a non-significant trend in the same direction, increasing the effect mentioned above. Third, even though all patients received individual PT, cognitive intact patients received the highest amount of individual PT. The number of visits did not significantly differ between groups, implying shorter single therapy sessions in patients with CI.

The reported disparities might be explained by patient- as well as therapist-related factors. Previous studies suggested that besides cognitive deficits, challenging behavioral and psychological symptoms, most pronounced in persons with advanced CI, might lead to problems in communication and complying with therapist’s instruction hampering motivation for and successful participation in therapies [16 , 37]. These factors also make it challenging for therapists to tailor and individualize interventions to effectively engage these patients in therapies [14 , 38–42] and, therefore, may have triggered higher exclusion rates for patients with SCI group-based PT and OT. Interestingly, in line with this, number of therapeutic attempts as indicator for planned but not successfully executed therapy sessions did not differ between cognitive groups, suggesting lower number of planned sessions in patients with CI. On the other hand, patients in the SCI group had also lowest functional and motor status at admission, probably causing early physical exhaustion which was identified as a risk factor for poor adherence to therapies in other studies [43]. If so, the seeming inequality in number and amount of TTH could possibly be interpreted as an appropriate and adaptive component of a personalized rehabilitation approach to tailor intensity and quantity of interventions to mental and functional restrictions of the patients, for example, through shortened therapy sessions. Therefore, reasons for non-participation or lower inclusion rates should be a research objective in further studies. In the present data analyses, we made a further step to clarify whether therapy provision as affected by cognitive status would also represent a missing link to connect CI and rehabilitation outcome.

Association between CI and improvement in functional status and the mediating role of TTH

In the present study, patients across all groups according to cognitive status improved their level of functioning during geriatric rehabilitation. This finding adds evidence to the rehabilitation potential of elder patients with moderate to severe CI, a negatively stigmatized patient group, often excluded from rehabilitation and related research [9 , 44]. However, improvement in functional status was not independent of the degree of cognitive impairment with patients with more pronounced CI achieving lower improvements. This is in line with other studies assessing the impact of CI and dementia on the effects of rehabilitation and efficacy of physiotherapy in elderly patients with mild CI [10 , 45]. In addition, multivariate analyses, adjusting for potential confounders, validated previous results [13] that besides CI, functional improvement was strongly associated with lower functional status at admission and generalized it to older inpatients including those with SCI.

Up to now, it is unclear how CI might lead to poorer functional outcomes. The finding that patients in the present study received lower actual number of therapies dependent on their cognitive status might indicate such a mechanism. Although former studies suggested a relationship between CI, rehabilitation outcome, and the amount of therapy received [20 , 24–26], we are not aware of studies examining a possible mediating effect of TTH on the relationship between CI and rehabilitation outcome.

The applied mediation contributed to a better understanding of the underlying mechanism by showing that TTH was significantly and positively associated with functional improvements and that lower TTH in patients with lower cognitive status, to a considerable part, contributed to lower outcomes. The indirect effect of CI through TTH accounted for almost one quarter of the total effect of CI on improvement in functional status. The results convincingly demonstrated the negative clinical effects of poorer therapy provision in patients with CI. These patients improved less in their functional status, partly because they received less therapy. Therefore, in terms of rehabilitation effectiveness, any amount of therapy may be valid. Lower therapy provision in patients with CI did not indicate an appropriate means to adapt programs but accounted for lower functional improvements in patients with CI.

The findings demonstrated that actual amount of therapy received has to be taken into account, when evaluating the rehabilitation potential and effectiveness of rehabilitation of older rehabilitants with CI and have direct implications for further rehabilitation research. Even though some studies suggest that intensifying programs [43, 46] and enhanced interdisciplinary inpatient rehabilitation may have some benefits over conventional rehabilitation for people with CI [47, 48], there still remains uncertainty and paucity of best practice rehabilitation programs [49]. Guidelines for therapists are lacking so far. Regular duration of rehabilitation in the present study was limited to 3 weeks by health insurances and therefore did not differ significantly between cognitive groups. Future studies will have to determine, whether, contrary to the general economically driven trend toward shorter rehabilitation periods, increasing total length of rehabilitation, would allow to counteract the negative impact of CI on amount of therapy received by more individualized and adaptive approaches. Such a view was supported by a study by Poynter et al. [14] which could demonstrate that individuals with more severe cognitive impairment showed low improvements in BI in the first two weeks of rehabilitation, but continued showing improvement for up to 6 weeks, whereas people who are more cognitively intact progressed only in the initial stages of rehabilitation. Consequently, optimal frequency, amount, and content of rehabilitation programs adapted to the different physical and mental capacity of patients have to be determined and should be a research priority.

The present study demonstrated that patients with CI and among those, in particular patients with SCI, represent a highly vulnerable patient group with lower functional improvements partly due to a higher exclusion rate and a lower amount of therapy received. Yet typically, patients with SCI were often excluded from rehabilitation research [9, 11]. The influence of patient-related barriers, such as age or comorbid diseases including frailty, and therapist-related factors, such as experience, attitude toward dementia patients, on actual therapy provision and outcomes requires further research. Knowledge about innovative approaches to engage these individuals are rare but integration in rehabilitation approaches is mandatory to optimize rehabilitation outcomes in these patients [18 , 50].

Limitations

The present observational study documented the course of geriatric rehabilitation in representative subgroups according to cognitive status. As patients with severe cognitive impairment are underrepresented in rehabilitation, sample sizes of subgroups differed with potential consequences for statistical analysis. It also resulted in natural differences in age and BI admission scores between groups, which we controlled for in our analyses. However, in our view, the classification to subgroups reflects the huge diversity with respect to cognitive status representative for geriatric inpatients and may increase generalizability and relevance of the results. Data on therapies received in the present study were based on clinical routine documentation. Number of planned but not performed therapy sessions did not differ between cognitive groups. However, reasons for non-participation or lower number of therapies received were not routinely documented, limiting interpretability of results. Therefore, barriers for therapy provision in patients with CI should be identified in future prospective studies and reflected in rehabilitation programs.

Conclusion and recommendations

The study is the first to analyze comprehensive data on actual therapy provision, detailed for overall as well as discipline-specific visits for a large representative cohort of multi-morbid patients with varying levels of cognitive impairment in inpatient geriatric rehabilitation.

Results demonstrated that CI negatively affected amount, frequency, and type of therapies received. Cognitive impairment predicted lower rehabilitation outcome, which was partly accounted for by lower TTH. The study findings imply that TTH, as a mediator for the relationship between CI and rehabilitation outcome, has to be taken into account, when evaluating the rehabilitation potential of patients with CI. Future research is required to explore optimal length of rehabilitation, dosage, and content of programs as well as barriers for therapy provision and participation in patients with varying cognitive impairment levels to optimize rehabilitation outcomes.

Footnotes

ACKNOWLEDGMENTS

This study was supported by the Baden Württemberg Foundation and the Dietmar-Hopp Foundation. The funding source had no role in the design and conduct of the study; collection, management, analyses, and interpretation of the data; preparation, review or approval of the manuscript.

Authors’ disclosures available online ().

References

Grund

, van Wijngaarden

, Gordon

, Schols

, Bauer

(2020) EuGMS survey on structures of geriatric rehabilitation across Europe. Eur Geriatr Med 11, 217–232.

Achterberg

, Cameron

, Bauer

, Schols

(2019) Geriatric rehabilitation-state of the art and future priorities. J Am Med Dir Assoc 20, 396–398.

Bean

, Orkaby

, Driver

(2019) Geriatric rehabilitation should not be an oxymoron: A path forward. Arch Phys Med Rehabil 100, 995–1000.

Bachmann

, Finger

, Huss

, Egger

, Stuck

, Clough-Gorr

(2010) Inpatient rehabilitation specifically designed for geriatric patients: Systematic review and meta-analysis of randomised controlled trials. BMJ 340, c1718.

Wang

, Chou

, Liang

, Peng

, Chen

, Loh

(2019) Post-acute care as a key component in a healthcare system for older adults. Ann Geriatr Med Res 23, 54–62.

Welz-Barth

, Stella

, Füsgen

(2007) Incidence of cognitive dysfunctions in geriatric rehabilitation. Phys Med Rehab Kuror 17, 94–97.

Shah

, Evans

, King

(2000) Prevalence of mental illness in a rehabilitation unit for older adults. Postgrad Med J 76, 153–156.

Meziere

, Blachier

, Thomas

, Verny

, Herbaud

, Bouillanne

, Henry

, David

, Le Thuaut

, Canoui-Poitrine

, Paillaud

(2013) Neuropsychiatric symptoms in elderly inpatients: A multicenter cross-sectional study. Dement Geriatr Cogn Dis Extra 3, 123–130.

Taylor

, DeMers

, Vig

, Borson

(2012) The disappearing subject: Exclusion of people with cognitive impairment and dementia from geriatrics research. J Am Geriatr Soc 60, 413–419.

10.

Poynter

, Kwan

, Sayer

, Vassallo

(2011) Does cognitive impairment affect rehabilitation outcome? J Am Geriatr Soc 59, 2108–2111.

11.

Vassallo

, Poynter

, Kwan

, Sharma

, Allen

(2016) A prospective observational study of outcomes from rehabilitation of elderly patients with moderate to severe cognitive impairment. Clin Rehabil 30, 901–908.

12.

Poynter

, Kwan

, Vassallo

(2013) How does cognitive impairment impact on functional improvement following the rehabilitation of elderly patients? Int J Clin Pract 67, 811–815.

13.

Cameron

, Schaafsma

, Wilson

, Baker

, Buckley

(2012) Outcomes of rehabilitation in older people–functioning and cognition are the most important predictors: An inception cohort study. J Rehabil Med 44, 24–30.

14.

McGilton

, Wells

, Davis

, Rochon

, Calabrese

, Teare

, Naglie

, Biscardi

(2007) Rehabilitating patients with dementia who have had a hip fracture: Part II: Cognitive symptoms that influence care. Topics Geriatr Rehabil 23, 174–182.

15.

McGilton

, Wells

, Teare

, Davis

, Rochon

, Calabrese

, Naglie

, Boscart

(2007) Rehabilitating patients with dementia who have had a hip fracture: Part I: Behavioral symptoms that influence care. Topics Geriatr Rehab 23, 161–173.

16.

Lenze

, Munin

, Dew

, Rogers

, Seligman

, Mulsant

, Reynolds III

(2004) Adverse effects of depression and cognitive impairment on rehabilitation participation and recovery from hip fracture. Int J Geriatr Psychiatry 19, 472–478.

17.

Hartley

, Gibbins

, Saunders

, Alexander

, Conroy

, Dixon

, Lang

, Luckett

, Luddington

, Romero-Ortuno

(2017) The association between cognitive impairment and functional outcome in hospitalised older patients: A systematic review and meta-analysis. Age Ageing 46, 559–567.

18.

Hall

, Watkins

, Lang

, Endacott

, Goodwin

(2017) The experiences of physiotherapists treating people with dementia who fracture their hip. BMC Geriatr 17, 91.

19.

Isbel

, Jamieson

(2017) Views from health professionals on accessing rehabilitation for people with dementia following a hip fracture. Dementia (London) 16, 1020–1031.

20.

Longley

, Peters

, Swarbrick

, Rhodes

, Bowen

(2019) Does pre-existing cognitive impairment impact on amount of stroke rehabilitation received? An observational cohort study. Clin Rehabil 33, 1492–1502.

21.

Resnick

, Beaupre

, McGilton

, Galik

, Liu

, Neuman

, Gruber-Baldini

, Orwig

, Magaziner

(2016) Rehabilitation interventions for older individuals with cognitive impairment post-hip fracture: A systematic review. J Am Med Direct Assoc 17, 200–205.

22.

Allen

, Koziak

, Buddingh

, Liang

, Buckingham

, Beaupre

(2012) Rehabilitation in patients with dementia following hip fracture: A systematic review. Physiother Can 64, 190–201.

23.

Horgan

, Cunningham

(2003) Impact of cognitive impairment on hip fracture outcome in older people. Br J Ther Rehabil 10, 228–232.

24.

Bellelli

, Frisoni

, Pagani

, Magnifico

, Trabucchi

(2007) Does cognitive performance affect physical therapy regimen after hip fracture surgery? Aging Clin Exp Res 19, 119–124.

25.

Rösler

, Krause

, Niehuus

, von Renteln-Kruse

(2009) Dementia as a cofactor for geriatric rehabilitation-outcome in patients with osteosynthesis of the proximal femur: A retrospective, matched-pair analysis of 250 patients. Arch Gerontol Geriatr 49, e36–e39.

26.

Barnes

, Conner

, Legault

, Reznickova

, Harrison-Felix

(2004) Rehabilitation outcomes in cognitively impaired patients admitted to skilled nursing facilities from the community. Arch Phys Med Rehabil 85, 1602–1607.

27.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

28.

Mahoney

, Barthel

(1965) Functional evaluation: The Barthel Index. Md State Med J 14, 61–65.

29.

Lubke

, Meinck

, Von Renteln-Kruse

(2004) [The Barthel Index in geriatrics. A context analysis for the Hamburg Classification Manual]. Z Gerontol Geriatr 37, 316–326.

30.

Tinetti

(1986) Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 34, 119–126.

31.

Faber

, Bosscher

, van Wieringen

(2006) Clinimetric properties of the performance-oriented mobility assessment. Phys Ther 86, 944–954.

32.

(2019) ICD-10-GM: Systematisches Verzeichnis; Internationale statistische Klassifikation der Krankheiten und verwandter Gesundheitsprobleme, pictura Werbung, Lich.

33.

Song

, Jerosch-Herold

, Holland

, Drachler Mde

, Mares

, Harvey

(2006) Statistical methods for analysing Barthel scores in trials of poststroke interventions: A review and computer simulations. Clin Rehabil 20, 347–356.

34.

Preacher

, Hayes

(2004) SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav Res Methods Instrum Comput 36, 717–731.

35.

Preacher

, Hayes

(2008) Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 40, 879–891.

36.

Kazui

, Yoshiyama

, Kanemoto

, Suzuki

, Sato

, Hashimoto

, Ikeda

, Tanaka

, Hatada

, Matsushita

, Nishio

, Mori

, Tanimukai

, Komori

, Yoshida

, Shimizu

, Matsumoto

, Mori

, Kashibayashi

, Yokoyama

, Shimomura

, Kabeshita

, Adachi

, Tanaka

(2016) Differences of behavioral and psychological symptoms of dementia in disease severity in four major dementias. PLoS One 11, e0161092.

37.

Srikanth

, Nagaraja

, Ratnavalli

(2005) Neuropsychiatric symptoms in dementia-frequency, relationship to dementia severity and comparison in Alzheimer’s disease, vascular dementia and frontotemporal dementia. J Neurol Sci 236, 43–48.

38.

Longley

, Peters

, Swarbrick

, Bowen

(2018) What influences decisions about ongoing stroke rehabilitation for patients with pre-existing dementia or cognitive impairment: A qualitative study? Clin Rehabil 32, 1133–1144.

39.

Shibasaki

, Asahi

, Mizobuchi

, Akishita

, Ogawa

(2018) Rehabilitation strategy for hip fracture, focused on behavioral psychological symptoms of dementia for older people with cognitive impairment: A nationwide Japan rehabilitation database. PloS One 13, e0200143.

40.

Hall

, Burrows

, Lang

, Endacott

, Goodwin

(2018) Are physiotherapists employing person-centred care for people with dementia? An exploratory qualitative study examining the experiences of people with dementia and their carers. BMC Geriatr 18, 63.

41.

Goodwin

, Allan

(2019) ‘Mrs Smith has no rehab potential’: Does rehabilitation have a role in the management of people with dementia? Age Ageing 48, 5–7.

42.

Fjellman-Wiklund

, Nordin

, Skelton

, Lundin-Olsson

(2016) Reach the person behind the dementia - physical therapists’ reflections and strategies when composing physical training. PLoS One 11, e0166686.

43.

Schwenk

, Dutzi

, Englert

, Micol

, Najafi

, Mohler

, Hauer

(2014) An intensive exercise program improves motor performances in patients with dementia: Translational model of geriatric rehabilitation. J Alzheimers Dis 39, 487–498.

44.

Seematter-Bagnoud

, Lécureux

, Rochat

, Monod

, Lenoble-Hoskovec

, Büla

(2013) Predictors of functional recovery in patients admitted to geriatric postacute rehabilitation. Arch Phys Med Rehabil 94, 2373–2380.

45.

Kowalska

, Mazurek

, Kubasik

, Rymaszewska

(2019) Effectiveness of physiotherapy in elderly patients with dementia: A prospective, comparative analysis. Disabil Rehabil 41, 815–819.

46.

Uda

, Matsui

, Fushimi

, Yasunaga

(2019) Intensive in-hospital rehabilitation after hip fracture surgery and activities of daily living in patients with dementia: Retrospective analysis of a nationwide inpatient database. Arch Phys Med Rehabil 100, 2301–2307.

47.

Stenvall

, Berggren

, Lundström

, Gustafson

, Olofsson

(2012) A multidisciplinary intervention program improved the outcome after hip fracture for people with dementia—Subgroup analyses of a randomized controlled trial. Arch Gerontol Geriatr 54, e284–e289.

48.

Lenze

, Skidmore

, Dew

, Butters

, Rogers

, Begley

, Reynolds

, 3rd, Munin

(2007) Does depression, apathy or cognitive impairment reduce the benefit of inpatient rehabilitation facilities for elderly hip fracture patients? Gen Hosp Psychiatry 29, 141–146.

49.

Smith

, Gilbert

, Sreekanta

, Sahota

, Griffin

, Cross

, Fox

, Lamb

(2020) Enhanced rehabilitation and care models for adults with dementia following hip fracture surgery. Cochrane Database Syst Rev 2, CD010569.

50.

McGilton

, Vellani

, Zheng

, Wang

, Yeung

, Escrig-Pinol

(2020) Healthcare professionals’ perspectives on rehabilitating persons with cognitive impairment. Dementia (London) 20, 1772–1790.