Key words
psychosocial intervention - dementia - nursing homes - Treatment adherence and compliance
- Covid-19
PWSDs People with severe dementia
Background and research question
Background and research question
People with severe dementia in nursing homes
Dementia is a chronic progressive disease of the brain involving disturbances in many
higher cortical functions (ICD-10). In Germany, the prevalence rate of dementia in
the over-65 population is 8.5% [1]. In nursing homes, however, 68.6% of the residents suffer from dementia, and one
third have severe dementia [2]. Severe dementia means that the cognitive impairments of these individuals have
already progressed to such an extent that language is limited to a few words and even
basic everyday activities can no longer be performed (independently). Often, these
individuals are no longer able to move or eat independently. In addition, altered
behaviours, such as aberrant motor behaviour (e. g. nesting, unpacking and packing,
wandering), aggression, or apathy can be observed in a large proportion of these individuals
[3], thus posing great challenges to caregivers in everyday care.
Quality dimensions in healthcare
According to Donabedian [4]
[5], three dimensions of quality in medical care can be distinguished: structural, process,
and outcome quality. Structural quality includes material and personnel resources,
organisational structures, and financial framework conditions. Process quality represents
the execution of an intervention, while outcome quality describes the effect. In the
case of a psychosocial intervention in nursing homes, structural quality includes
the spatial conditions, the material equipment, and the qualifications of the staff.
Process quality involves the proper implementation of the intervention (i. e. adherence
to the time and content specifications). Outcome quality, or the quality of the results,
is reflected in participants' satisfaction or gains in resources.
Psychosocial interventions for people with severe dementia
Psychosocial interventions include procedures that promote everyday practical, cognitive,
social, or behavioural skills and thus enable those who are affected to live their
lives as independently as possible [6]. A variety of psychosocial interventions are available for people with dementia.
Reviews and meta-analyses have shown that the majority of these therapies have beneficial
effects on cognition, daily living skills, or behavioural symptoms in people with
dementia with different degrees of severity [7]
[8]. For people with severe dementia (PWSDs), non-verbal treatments such as basal stimulation,
aromatherapy, light therapy, or music therapy seem to be more suitable [9]. However, there are hardly any analyses on psychosocial interventions specifically
for PWSDs. Thus far, the only meta-analysis that included mainly PWSDs showed an improvement
in daily activities and a reduction in depressive symptoms in PWSDs through interventions
such as music therapy, physical exercises, or massage [10]. Specially adapted to the needs and abilities of PWSDs, the multi-component MAKS-s
(Motor, Everyday Practical, Cognitive, Social for people with severe dementia) intervention
was developed. MAKS-s is executed in small groups of 3 to 6 people [11], and trained therapists carry out the four components in the order S-M-K-A during
a one-hour intervention. In the randomised controlled trial conducted in German nursing
homes during the COVID-19 pandemic, MAKS-s was examined in a standardised manner in
nursing homes. The outcome measures quality of life and psychological and behavioural
symptoms were assessed with observer rating scales by trained nursing staff who were
not involved in the intervention. However, within the scope of the study, no significant
effect on quality of life, behavioural symptoms, or daily living skills of the PWSDs
could be determined in comparison with the control group [12].
Therapy fidelity
In the DeTaMAKS study, which investigated people with mild or moderate dementia in
day care, the open study phase showed a lower effect size than was found during the
controlled phase [13]
[14]. The open study phase means that the intervention was no longer carried out under
controlled conditions (all facilities were trained, the application of the intervention
was optional). Other studies have also shown that the effectiveness of an intervention
is substantially influenced by the “proper” implementation of the intervention [15]. Therapy fidelity (i. e. when an intervention is carried out in exact accordance
with the manual) thus seems to have a significant influence on the effect of an intervention.
Research question
Since the MAKS-s intervention could not be shown to be effective in terms of quality
of life and psychological and behavioural symptoms during the RCT phase, the question
that arises is whether the lack of effectiveness might be related to a lack of therapy
fidelity during the COVID-19 pandemic or whether other factors are responsible for
the lack of effectiveness. Therefore, the present study was designed to investigate
the following questions: 1) Did therapy fidelity as a predictor influence the benefits
(the social, everyday practical, and emotional gain) that people with severe dementia
received?; 2) Which predictors influenced the process quality?
Methods
Design and sample
The MAKS-s baseline study [11]
[12] was a two-arm, cluster-randomised, controlled intervention study with a waitlist
control group design. The study was conducted in 26 nursing homes (13 intervention,
13 control groups) in different federal states of Germany. The intervention phase
lasted 6 months (June to December 2020). The intervention was a psychosocial group
intervention conducted three times a week for one hour by previously trained MAKS-s
therapists. Each session consisted of four components: motor stimulation, ADL training,
cognitive stimulation, and social functioning. The baseline study included 144 people
with severe dementia. Severe dementia was defined as a Mini-Mental-State-Examination
score<10. The full description of the intervention and methods can be found in the
internationally published and freely available study protocol [11]. After the end of the controlled phase, the nursing and care staff who were caring
for the control groups were also trained to implement the MAKS-s intervention in accordance
with the study protocol (waitlist control group design) so that all 26 nursing homes
were able to implement MAKS-s afterwards. From this point—in the so-called open phase
of the study—all participating nursing homes were able to decide independently whether
and how often they wanted to carry out MAKS-s. In order to obtain additional information
about the possible effects of MAKS-s, a questionnaire was sent to all trained MAKS-s
therapists by post 6 months after the end of the RCT phase. The study coordinators
in the nursing homes distributed the questionnaires to the trained MAKS-s therapists
and collected them again. Four individuals were trained as MAKS-s therapists in each
nursing home. These individuals belonged to one of the following professional groups:
occupational staff, head of social care, nurse, or therapist.
Assessment
Unfortunately, validated instruments could not be used to assess the quality dimensions
and the impact of the COVID-19 pandemic. At the time of the survey, there were no
suitable and validated scales for assessing the stress caused by the COVID-19 pandemic.
According to Donabedian, the quality criteria must always be individually adapted
to the setting, and thus, there were no ready-made scales for the quality criteria
either. The development of the questionnaire was concept-based. The 38 items captured
the three quality dimensions of health care according to Donabedian [4]
[5]: structural quality (3 items), process quality (2 items), and outcome quality (10
items). In addition, we assessed the impact of the COVID-19 pandemic (19 items) and
the evaluation of the intervention (4 items). Each item was rated on a 5-point Likert
scale (see the Additional Material for the questionnaire).
Structural quality was assessed with items such as “How do you rate your spatial and
material conditions?” Process quality consisted of the two items “How often did the
MAKS-s intervention take place per week?” and “Were you able to implement the intervention
in accordance with the instructions?” Process quality was dichotomously defined from
these two items as therapy fidelity versus deviations from the manual. Therapy fidelity
included therapists who delivered the MAKS-s intervention at least twice a week without
changing the order or duration of the modules. Outcome quality was subdivided into
two areas reflecting whether the outcome benefitted the PWSDs (5 items) or the therapists
(5 items). According to Donabedian, outcome quality covers participants’ (i. e. therapists
or patients’) satisfaction and gains. PWSDs’ satisfaction was surveyed in accordance
with Clarke with the domains relevant to well-being, such as positive emotions, social
participation, and social relationships [16]. An example satisfaction item is “PWSDs showed positive emotions.” Benefits for
the therapists included items such as, “Since I have been doing the MAKS-s intervention,
I am more satisfied with my job.” The effects of the COVID-19 pandemic were measured
with items such as “Due to the pandemic, distance rules had to be followed during
group sessions” or “Since the pandemic began, I have felt more psychologically stressed.”
The intervention was evaluated with items such as “I will recommend MAKS-s to others.”
Statistical analyses
First, the approval rates for each item were determined and presented descriptively.
The programme IBM SPSS version 28 was used for all statistical analyses.
-
Principal component analysis to prepare the data
To form summed values in an empirically supported way, the domains structural quality,
benefits for the PWSDs, benefit for the therapists, influence of the COVID-19 pandemic,
and evaluation of the intervention were first subjected to a principal component analysis
(PCA) with an orthogonal rotation (VARIMAX) for each domain. The Kaiser-Meyer-Olkin
criterion was applied to test the prerequisites for a PCA. Items that did not load
clearly on a factor or had a factor loading<0.50 were removed. To determine the internal
consistency of each domain, Cronbach's alpha was then computed for each domain. At
the item level, the discriminatory power and Cronbach's alpha “if item deleted” were
calculated for each item. Items with a discriminatory power<0.5 and items for which
Cronbach's alpha improved when it was deleted were removed from the scale.
-
Linear regression analysis to determine the factors influencing outcome quality
To determine the predictive power of the potential predictors evaluation of the intervention,
benefits for the therapists, structural quality, normative constraints and psychosocial
burdens from the COVID-19 pandemic, and process quality (therapy fidelity) on outcome
quality (benefits for the PWSDs), a hierarchical linear regression model was calculated,
and all variables were tested for multicollinearity (r≥0.70). In the first step, the
two potential bias variables benefits for the therapists and evaluation of the intervention
were included in the regression model. In the next step, the effects of the COVID-19
pandemic and structural quality were added, and in the last step, the therapy fidelity
variable was inserted.
-
Binary logistic regression to identify the factors that influence process quality
To identify the factors that could have a potential influence on process quality,
group differences between therapy fidelity and deviations from the manual were first
calculated for all potential predictors using a t-test for unconnected samples. The
independent variables for which there were significant group differences (p<0.05)
with regard to process quality were tested for multicollinearity. If multicollinearity
(r≥0.70) was identified between two variables, the variable with the lower correlation
with the target variable was not included in the regression analysis.
The remaining variables were included as predictors in a binary logistic regression
with process quality (therapy fidelity) as the dependent variable.
Results
Of the 26 nursing homes that originally participated in the study, 18 were willing
to participate in the follow-up survey 12 months after the study began and 6 months
after the open study phase began (t12). In 14 of these 18 nursing homes, the MAKS-s
intervention was implemented in the open phase between t6 and t12. They were equally
divided between the former intervention and control groups. Of the 104 trained MAKS-s
therapists, 58 responded to the questionnaire, thus corresponding to a response rate
of 56%. On average, 3 therapists from each nursing home responded (M=3.22, SD=0.73). With regard to the quality of outcomes (benefits for the PWSDs), there was
an agreement rate (“fully agree”; “tend to agree”) of 71% across all 5 items. The
item with the highest level of agreement (84%) was “The participating PWSD showed
positive emotions during the MAKS-s intervention” ([Fig. 1]).
Fig. 1 Benefits for PWSDs participating in MAKS-s from the perspective of the MAKS-s therapists
(n=51).
-
Principal component analyses: preparation of the data
The principal component analyses (see the Additional Material) identified one factor
each for the dimensions structural quality, benefits for the PWSDs, and evaluation
of the intervention; all items had a loading of>0.7 and could thus be retained. The
dimension benefits for the therapists also showed only one factor, but it had to be
reduced by one item (“Since I have been carrying out the MAKS-s intervention, I feel
more burdened”) due to a lack of loading. The scree plot for the principal component
analysis on “effects of the COVID-19 pandemic”, which had a total of 19 items, revealed
two different factors: “normative constraints due to COVID-19” (12 items) and “psychosocial
effects of the COVID-19 pandemic” (7 items).
With the exception of structural quality, the internal consistency of the individual
scales was above 0.80, which can be considered high. The value of the structural quality
was in the acceptable range with Cronbach's alpha=0.784. At the item level, discriminatory
power>0.50 was achieved for all items with one exception: “The use of MAKS-s makes
my workflow easier” (Cronbach's alpha=0.329). Therefore, this item was removed from
the benefits for the therapists scale. In addition, one item had to be removed from
the evaluation of the MAKS-s intervention scale, as it had both low discriminatory
power and an unfavourable Cronbach's alpha value “if item deleted”.
After all the dimensions had been checked with PCAs, a sum value was formed for each
dimension by adding up the respective item values. The structural quality dimension
ranged from 0–12 points; the outcome quality dimension and benefits for the PWSDs
both ranged from 0–20 points; benefits for the therapists ranged from 0–12; evaluation
of the intervention ranged from 0–12; psychosocial burdens from COVID-19 ranged from
0–28; and the normative constraints due to COVID-19 ranged from 0–48. Means and standard
deviations for the total sample can be found in [Table 1].
-
Linear regression analysis: Factors influencing outcome quality
The regression analysis showed that therapy fidelity had a significant influence on
the benefits (i. e. the social, everyday practical, and emotional gains) for the PWSDs
(see [Table 2]). Of the potential bias variables that were added in the first step of the hierarchical
regression, benefits for the therapist was a significant predictor (β=0.46, p<0.001).
The two variables together explained 42.6% of the variance. The variables added in
the next step were unable to bring about a significant change in the amount of explained
variance. The therapy fidelity added in the last step showed a significant effect
(β=0.45, p=0.019) and explained an additional 9.3% of the variance. The final model
explained 59.5% of the variance and was statistically significant, F(6,27)=6.61, p<0.001
([Table 2]).
-
binary-logistic regression: factors influencing process quality
Table 1 Mean comparisons of the implementation of the intervention with or without deviations
from the manual.
|
Variable
|
therapy fidelity
|
deviations from the manual
|
total
|
p
|
|
structural quality (range 0–12) M (SD); (n=51)
|
10,2
|
(1,4)
|
8,8
|
(1,8)
|
9,4
|
(1,8)
|
,004
|
|
Benefit PWSDs (range 0–20), M (SD); (n=51)
|
16,9
|
(3,3)
|
13,5
|
(4,0)
|
15,0
|
(4,0)
|
,002
|
|
Benefit therapist (range 0–12), M (SD); (n=50)
|
9,3
|
(2,4)
|
7,52
|
(2,4)
|
8,3
|
(2,5)
|
,013
|
|
evaluation of the intervention (range 0–12), M (SD); (n=48)
|
11,1
|
(1,2)
|
9,6
|
(2,6)
|
10,27
|
(2,2)
|
,011
|
|
Normative constraints due to COVID-19 (range 0–48), M (SD); (n=42)
|
17,9
|
(10,7)
|
29,8
|
(11,0)
|
25,3
|
(12,2)
|
,002
|
|
Psychosocial effects of the COVID-19 pandemic (range 0–28), M (SD); (n=55)
|
11,2
|
(5,9)
|
13,9
|
(6,3)
|
15,2
|
(7,0)
|
,164
|
Table 2 Hierarchical linear regression for the variables predicting the benefits for PWSDs;
model: ENTER.
|
|
B
|
SE(B)
|
β
|
p
|
ΔR
2
|
|
Step 1
|
|
|
|
|
|
.426, p<.001
|
|
Benefit therapist
|
0,628
|
0,232
|
0,455
|
.007
|
|
|
Evaluation of the intervention
|
0,322
|
0,242
|
0,223
|
.099
|
|
|
Step 2
|
|
|
|
|
|
.076, p=.257
|
|
Benefit therapist
|
0,444
|
0,259
|
0,322
|
.038
|
|
|
Evaluation of the intervention
|
0,442
|
0,242
|
0,307
|
.078
|
|
|
Structural quality
|
-0,060
|
0,402
|
-0,024
|
.846
|
|
|
Normative constraints due to COVID-19
|
-0,038
|
0,063
|
-0,096
|
.570
|
|
|
Psychosocial effects of the COVID-19 pandemic
|
0,259
|
0,121
|
0,366
|
.064
|
|
|
Step 3
|
|
|
|
|
|
.093, p=.019
|
|
Benefit therapist
|
0,461
|
0,238
|
0,334
|
.027
|
|
|
Evaluation of the intervention
|
0,306
|
0,228
|
0,213
|
.176
|
|
|
Structural quality
|
-0,407
|
0,394
|
-0,164
|
.286
|
|
|
Normative constraints due to COVID-19
|
0,061
|
0,070
|
0,155
|
.390
|
|
|
Psychosocial effects of the COVID-19 pandemic
|
0,105
|
0,127
|
0,149
|
.479
|
|
|
Therapy fidelity
|
3,831
|
1,528
|
0,447
|
.019
|
|
|
Total R2
|
|
|
|
|
|
.595
|
a Therapy fidelity: 1 (yes); 0 (no).
The t-test showed significant differences in almost all dimensions between the therapy
fidelity group and the group that deviated from the manual, in the sense that the
therapy fidelity group scored significantly better ([Table 1]). The psychosocial impact of the pandemic was not significantly different and was
therefore not included in the regression model. There was a high degree of multicollinearity
(r=0.72) between the variables benefits for the PWSDs and evaluation of MAKS-s, which
is why the variable evaluation of MAKS-s was excluded from the regression. The binary
logistic regression model was statistically significant, χ²(4)=34.25, p<0.001. It
had a high variance resolution of Nagelkerke's R²=.834, meaning that the predictors
we examined explained 83.4% of the variance in therapy fidelity ([Table 3]). Significant predictors were normative constraints due to COVID-19 and benefits
for the PWSDs. For each point increase in the normative constraints due to the COVID-19
scale, therapy fidelity (i. e. the likelihood that the MAKS-s intervention was delivered
as specified) decreased by 42%. For each point increase in the benefits for the PWSDs
scale, the likelihood that the intervention was delivered in accordance with the manual
increased by a factor of four.
Table 3 Binary logistic regression with therapy fidelity (process quality) as dependent variable
(1=therapy fidelity, 0=deviations from the manual).
|
95% CI for Odds Ratio
|
|
B
|
SE
|
Wald
|
p
|
Odds Ratio
|
lower value
|
upper value
|
|
Structural quality
|
0,72
|
0,75
|
0,91
|
0,341
|
2,04
|
0,47
|
8,91
|
|
Benefit therapist
|
0,23
|
0,42
|
0,30
|
0,585
|
1,26
|
0,55
|
2,90
|
|
Benefit PWSDs
|
1,14
|
0,64
|
5,36
|
0,021
|
4,42
|
1,26
|
15,71
|
|
Normative constraints due to COVID-19
|
-0,54
|
0,26
|
4,28
|
0,039
|
0,58
|
0,35
|
0,97
|
Discussion
The aim of the present study was to determine whether implementing the intervention
in accordance with the manual had an influence on the fact that PWSDs could benefit
from the psychosocial multi-component MAKS-s intervention in the open phase after
the end of the RCT. In addition, the aim was to identify the predictors that influenced
whether or not a nursing home had implemented the MAKS-s intervention in the open
phase in accordance with the manual.
This current analyses provide additional results that complement the results of the
RCT. The RCT did not find efficacy in terms of quality of life or psychological and
behavioural symptoms. By contrast, the follow-up survey answered by the therapists
6 months after the end of the RCT showed positive effects of the intervention on the
PWSDs’ outcomes.
Outcome quality
During the open phase of the MAKS-s study, almost three quarters of the therapists
observed positive effects of the MAKS-s intervention on the PWSDs. In particular,
84% of the interviewed therapists agreed that positive emotions emerged during the
implementation of the MAKS-s intervention. The analysis showed that therapy fidelity
(at least twice a week, all four modules in the given order, without time cuts) had
a significant influence on whether PWSDs benefited from MAKS-s or not. This result
is in line with findings from other studies, which also only achieved a positive effect
when the intervention was carried out at the required intensity [17]
[18]. During the randomised controlled phase of the MAKS-s trial, elements of the MAKS-s
intervention might not have been delivered with therapy fidelity due to the restrictions
from the COVID-19 pandemic. Unfortunately, therapy fidelity could not be adequately
verified, also due to the restrictions from the pandemic (e. g. monitoring visits
were not possible). Low therapy fidelity could therefore be one potential reason for
why no significant effects on the quality of life and behavioural symptoms of the
PWSDs were observed in A Kratzer et al. [19] during the controlled phase.
The fact that positive effects on PWSDs were observed during the intervention is in
contrast with the observations of the nurses who did not perceive any positive changes
during the controlled phase of the MAKS-s study [19]. How can these differences in the results be explained? The survey technique (observer
rating scale) was the same in both cases. It is possible that the different observation
times by different observers were responsible for the observed discrepancy. During
the RCT phase, the primary nurses who were not involved in the MAKS-s intervention
retrospectively assessed the PWSDs’ daily behaviour across a period of several days.
During the open phase, on the other hand, the MAKS-s therapists who carried out the
intervention themselves reported their immediate perceptions during the intervention.
Thus, the short-term effects of the intervention were assessed. These short-term and
directly observable effects were predominantly assessed positively, which speaks for
a positive effect of the MAKS-s intervention on the current well-being of the PWSDs.
It seems that long-term changes in quality of life are no longer possible in PWSDs,
evidence for which have also been found in other intervention studies involving PWSDs
[10]. This tendency could also explain the lack of changes in longer-term quality of
life during the RCT phase of the MAKS-s study.
Process quality
The second step was to examine which factors contributed to whether the intervention
was carried out in accordance with the manual or whether substantial deviations from
the manual occurred. The more positive the effects of the MAKS-s intervention that
the therapists perceived in the PWSDs, the greater the probability that the therapists
had carried out the intervention in accordance with the manual. In addition, there
was also a negative correlation between the degree of normative constraints from COVID-19
regulations and implementation in accordance with the manual. That is, the more regulations
there were in a nursing home to observe hygiene measures and distance rules or to
refrain from certain social activities, the greater the likelihood that the MAKS-s
intervention was not carried out in accordance with the manual. This finding is consistent
with a finding from the 2021 Nursing Report: Measures such as distance rules, contact
restrictions, and bans on social group activities, which were intended to protect
against COVID-19 infection, have conversely led to cuts in the health care of those
in need of care [20].
In summary, it can be concluded that the normative constraints from COVID-19 significantly
influenced the way in which the MAKS-s psychosocial intervention was implemented.
These changes led to the fact that PWSDs who were subject to very strict normative
constraints were temporarily not allowed to participate in the intervention in some
nursing homes and therefore could not benefit from the possible positive effects.
Strengths and limitations
The questions included on the assessment instrument were newly developed for the survey;
the main reasons were to be able to react to the current situation (COVID-19 pandemic)
with targeted questions and to keep participants’ efforts in filling out the survey
(number of questions) as low as possible. In order to at least meet the requirements
of content validity, a concept-based approach that used Donabedian’s dimensions of
quality and Clarke´s domains of well-being was used. In addition, all constructs were
examined for internal consistency by computing a PCA so that the formation of sum
values was empirically supported. Although all nursing homes and all trained therapists
were invited to participate in the survey 12 months after the study began, the present
study did not survey all potential participants. The response rate for the nursing
homes was just under 70%. It can be assumed that selection factors, such as current
staffing levels and personal attitudes towards the MAKS-s intervention, influenced
participation. To reduce this bias, the evaluation of the MAKS-s intervention and
benefits for the therapists were included as control variables in the first step of
the hierarchical regression model. In general, all statements were subjective perceptions
of individual groups of people (MAKS-s therapists) who made these statements retrospectively.
In addition, the Hawthorne effect, recall, and social desirability biases, which are
typical for a survey study, cannot be completely excluded. However, it can be assumed
that a possible Hawthorne effect was reduced by controlling for the variables evaluation
of the intervention and benefits for the therapists. A possible recall bias can work
in both positive and negative directions, which is why it can be assumed that any
effects should have averaged out. In addition, the social desirability bias could
have been minimised by the completely anonymous survey, although not completely.
The strength of the present study lies in the fact that the results reflect the reality
of care after the “rigid” requirements of an RCT have ended.
Conclusions for practice
-
Therapy fidelity seems to be a decisive factor in whether people with severe dementia
benefit from the MAKS-s psychosocial intervention. Therefore, it is recommended that
the MAKS-s manual be followed without shortcuts or changes.
-
Future randomised controlled trials on the effects of psychosocial interventions should
include therapy fidelity as a mediating variable.
-
In order to examine the effect of a psychosocial intervention on PWSDs, future studies
should not look primarily for lasting effects but should rather focus on the short-term
effects that can be directly observed. Contact restrictions, distance bans, and bans
on certain social activities meant that the MAKS-s psychosocial group intervention
often could not be carried out in accordance with manual. Since social contacts —
especially for people with severe dementia — are a central element for establishing
contact with the environment, socially restrictive measures reduce the potential benefits
of a psychosocial intervention.
Compliance with ethical guidelines
All of the human studies described were conducted with the approval of the relevant
ethics committee, in accordance with national law, and in accordance with the Declaration
of Helsinki of 1975 (in its current, revised version) (Ref.295_19B). Informed consent
was obtained from all individuals involved.
Contributor’s Statement
None
