Keywords health promotion - well-being - seniors - geriatric care - eHealth - mHealth
Schlüsselwörter Gesundheitsförderung - Wohlbefinden - Senior:innen - Altenpflege - eHealth - mHealth
Background
Due to the ageing of societies and the increasing need for care, maintaining and
promoting the health of those in need of care is gaining further relevance [1 ]. In 2019, there were 877,200 places in
full inpatient long-term care in Germany [2 ]. Sixteen percent of people with a level of care requirements in
Germany live in nursing homes [1 ]. In
order to comply to this relevance, long-term care insurances in Germany have the
legal mandate to develop proposals for interventions to strengthen the health
resources of residents of nursing homes [3 ]. Comprehensive scientific evaluations of preventive interventions for
residents of nursing homes according to SGB XI § 5 (The Eleventh Book of the German
Social Code contains the law governing social care insurance in Germany.) recommend
focusing on combined interventions from multiple areas of action (physical activity,
violence prevention, cognitive resources, psychosocial health, nutrition) to achieve
health-promoting effects [4 ].
Based on previous findings in health promotion for residents in nursing homes [4 ]
[5 ]
[6 ], the use of mind-body
medicine (MBM), which is based on a connection between mind and body, appears
promising [7 ]. This concept includes,
among other things, areas of action such as exercise and nutrition as well as
training in mindfulness and self-care. Mind-body interventions typically integrate
measures that promote mindfulness indirectly (mindfulness-informed interventions,
MII) or directly (mindfulness-based interventions, MBI) [8 ]. MII aim to improve specific
parameters, such as physical flexibility through breathing exercises or yoga [9 ]. MBI rather “focus on learning or
improving mindfulness” [10 ] (e. g.,
through breathing meditation or body-scan exercises) [9 ].
Positive effects on stress reduction and health promotion through MBM have been shown
in numerous studies e. g. [11 ]
[5 ]
[12 ]. Initial tests of MBI have also been carried out for the target group
of seniors in the setting of nursing homes [4 ]
[6 ]. A study on the
effectiveness of an MII for the target group of residents in nursing homes has shown
that significant increases can be achieved with reference to physical health
(subscale of SF-12) and depressive symptoms (GDS-12R). The authors conducted a
non-randomized study without follow-up survey and emphasized the need for further
adaptions of the course to the needs of seniors [5 ]. Thus, further insights into the
effectiveness of interventions for the target group are still needed [11 ]
[12 ].
The main aims of this pilot study were to develop a MBM-intervention for residents in
nursing homes [13 ] and to explore the
feasibility and effectiveness using qualitative [14 ] and quantitative methods. This paper
describes the quantitative part of the pilot study and contains initial findings on
the completetion rates, the effectiveness of the intervention and the expected
effect sizes.
Methods
Study design
This pilot study is based on a cluster-randomized controlled study design (RCT).
It was registered in the German Register of Clinical Studies (ID: DRKS00030409).
Ethics vote (IRB) was obtained from the ethics committee of the University
Witten/Herdecke (no. 233/2020).
Intervention
The basis for the participatory intervention development process [13 ] was the previously developed and
evaluated MII health promotion course “BERN” (Behavior, Exercise, Relaxation,
Nutrition). Exercises fostering skills in these areas (the BERN intervention)
were adapted so that they were tailored to the resources of residents in nursing
homes. BERN represents a typical MBM program [8 ]
[15 ]
[16 ]. The participatory intervention
development process involved multiple stakeholders (target group, their
relatives, nurses and other experts from nursing homes, health insurance
representatives and app developers) and is described in detail by Michaelsen et
al. [13 ].
The result of the intervention development was an eight week group course
(on-site) and an one-on-one app intervention (video instruction) with identical
content (Online-Appendix 1). The on-site course took place in small groups (3–7
residents). A trainer (qualified therapist for health promotion; experienced
with the target group) came to the facilities for around 50–60 minutes once a
week and led the exercises of the week in person. The app intervention was
completed using the tablets provided. The first exercise of each module was
activated on a weekly basis. The next exercise was unlocked when the previous
one was either completed or skipped. The entire previous module was unlocked at
latest at the start of the new week, i. e., at the beginning of the next module.
Exercises could be skipped or repeated. Progress was saved in the individual,
non-password-protected profiles. It was not possible to interact with other
participants or see other participants´ progress via the app. Project employees
offered technical assistance on an individual basis twice a week for
approximately 20 minutes each session in the residents' rooms.
Study population
Recruitment
In the first step of recruitment, nursing facilities in North
Rhine-Westphalia (Germany) were contacted (N ≈ 250). Facility employees then
identified suitable residents. Potential participants were invited to an
information event led by project employees.
Inclusion and exclusion criteria
Social service employees in the facilities assessed the inclusion and
exclusion criteria (external assessment). The minimum required age of
participants was 70 years. Participants had to be physically and cognitively
(Mini-Mental Status Test/MMST≥20 at T0) capable of taking part in group
activities, as they were required to follow group conversations, articulate
themselves, and follow instructions. Furthermore, as a result of the
participatory intervention development process, people with severe
depression were excluded from participation: this assessment was made via
accessing the relevant information in the resident´s file. There were no
inclusion or exclusion criteria with regard to the length of time the
persons had lived in the nursing home, level of care requirements or
involvement of relatives.
Randomization
Twelve nursing homes took part in the research project. Facilities each were
randomized to the on-site (O), app (A) and control (C) groups using the web
program Jerrydalal by an independent researcher. The result of the
randomization process was reported to the participants after T0.
Data collection
Three quantitative surveys (see [Fig.
1 ]) were conducted in person individually by project employees in the
nursing homes using tablets. The pre-survey (T0) took place immediately before
the intervention period (Nov 2022). The post-survey (T1) was carried out
immediately after the intervention period (Feb 2023), and the follow-up survey
(T2) was carried out three months after the end of the intervention period
(April 2023). The eight-week intervention were performed between T0 and T1. The
survey was performed in German. All outcomes were coded uniformly so that a
higher level could be interpreted as more advantageous. In addition, the trainer
and the technical support documented the participation every week (see As-
treated analysis).
Fig. 1 Three survey timepoints (T0, T1, T2) of quantitative data
collection.
The following socio-demographic information were collected: gender, age, marital
status, highest general school qualification, highest professional degree, level
of care requirements (1=independence slightly impaired to 5=severe impairment of
independence; assessed by nursing care insurance) and move-in dates.
Primary outcome
Subjective well-being [17 ]
[18 ] was measured by five items,
for example: “In the last two weeks I have felt calm and relaxed”. The items
were rated on a six-point Likert scale ranging from (0) at no time to
(5) all the time concerning the past two weeks. The score was: (0)
no well-being to (100) maximum well-being .
Secondary outcomes
The Kentucky Inventory of Mindfulness Skills was used in its short form [19 ] to calculate an average score
for the four fields of mindfulness (observing , describing ,
acting with attention , accepting without judgment ) as well
as an overall score ((1) never or very rarely applies/ low mindfulness
to (5) very often or always applies/ high mindfulness).
Cognitive performance was assessed using the mini-mental status test (MMST)
[20 ], which includes everyday
questions and action-related tasks. One point was awarded for each correct
answer. A maximum of 30 points (sum score) could be achieved [21 ]
[22 ]
Two items of the Salience and Happiness Database (ESH) was included to
measure the current state of happiness (G-1; 21) and life satisfaction (L-1;
22) ((0) not at all happy or satisfied to (10) very happy or
satisfied ; see also [23 ]).
The Geriatric Depression Scale in its short form GDS-8 [24 ] was used to assess mental
health (dichotomous response scale; sum score: (0) lowest mental
health to (8) highest mental health ).
The health-related quality of life was assessed with the first item of the
SF-12 ((1) bad to (5) excellent ) [25 ].
Stress warning signals (SWS) were identified with different items (binary
coded: (1) yes/applicable (0) no/not applicable ) of physical,
emotional, cognitive, social and behavioral symptoms (if a SWS occurred:
frequency (1) very rare to (10) always ) [16 ].
Data analysis
The analysis was performed using IBM software SPSS Statistics .
Pairwise case exclusion was carried out in the analyses. Only a few cases were
missing. If a value was missing (WHO-5, KIMS-D, SWS), the single imputation
method was used by adding the mean of the remaining answers of the (sub-)scale
for the respective collection date.
First, a mean comparison was carried out between the three time points for each
group (one-way repeated measures ANOVAs or Friedman test depending on the
distribution), which indicated a temporal change (Online-Appendix 2a). If these
comparisons were significant, we conducted pairwise comparisons (post hoc tests:
Bonferroni or Dunn-Bonferroni) and estimated the effect size Hedges´ g for each
pair (Online-Appendix 2b).
Second, change scores (ΔT1-T0, ΔT2-T0: positive sign indicates increase) were
calculated for each group to test for differences between groups. Therefore, a
mean comparison (one-factorial ANOVA or Welch-ANOVA depending on the variance
given a normal distribution for all groups or Kruskal-Wallis test without normal
distribution) was carried out between the three groups (study groups or
participation intensity) (Online-Appendix 3a). If these comparisons were
significant, we conducted pairwise comparisons (post hoc tests: Tukey-Kramer,
Games-Howell or Dunn-Bonferroni) and estimated the effect size Hedges´ g for
each pair (Online-Appendix 3b).
The significance level was set at p <.05. For the primary outcome,
results with p <.1 or values close to this threshold were interpreted
as indications. Here |g|=0.2-.49 represents a small effect, |g|=0.5-.79 for a
medium effect and |g|≥.8 for a large effect.
Intention-to-treat analysis (ITT)
The ITT aimed to examine the effects of the intervention by group according
to cluster randomization. No participants were excluded.
As-treated analysis (AT)
The AT aimed to investigate whether people, regardless of the intervention
group, with higher participation intensity (7–8 modules completed) differ
from people with low participation intensity (0–6 modules completed) or
people in the control group. All randomized participants were included.
Sensitivity analysis
During the study, it became apparent that some participant's cognitive
performance decreased to the point below the limit defined in the inclusion
criteria. The sensitivity analyses aimed to check how the results differ if
only people with long-term sufficient cognitive performance were included.
In this sense, the ITT and AT analyses were repeated with the subgroup that
achieved at least 20 points on the cognitive performance test (MMST) at all
three survey times.
Results
The results are presented below. Outcomes not mentioned did not yield any significant
results (MMST, L-1, SF-12) or were evaluated descriptively (SWS).
Description of the sample
Study groups
85 residents were interviewed at T0, of which 77 residents met the inclusion
criteria and 76 were randomized (nO =28; nA =29;
nC =19). 73.7% of participants completed all three surveys
(n=56) (see [Fig. 2 ]).
Fig. 2 Flow chart (participation rate per study group and
survey time).
Completion rate
Overall, 70.2% (n=40) of the participants in the intervention groups
completed the intervention, 59.6% (n=34; nO =13; 46.4%;
nA =21; 72.4%) participated in 7–8 modules.
Sociodemographic data
The mean age was 85.66 years (SD: 6.62; range: 71–100). Most participants
were widowed (n=51; 67.1%) and had children (n=63; 82.9%). Level of care
requirements 2 to 5 were represented in the sample, with care level 3 being
the most common (n=39; 51.3%) (see [Tab. 1 ]). A majority had an equivalent qualification to a
secondary or primary school qualification (n=49; 64.5%). Vocational training
was most frequently cited as the highest level of training (n=33;
43.4%).
Tab. 1 Sample description
Variable
Total n (%)
On-site n (%)
App n (%)
Control n (%)
Study participants
76 (100.0%)
28 (36.8%)
29 (38.2%)
19 (25.0%)
Gender
female
57 (75.0%)
18 (64.3%)
24 (82.8%)
15 (78.9%)
male
19 (25.0%)
10 (35.7%)
5 (17.2%)
4 (21.1%)
Age
70 to 79 years
12 (15.8%)
5 (17.9%)
2 (6.9%)
5 (26.3%)
80 to 89 years
40 (52.6%)
14 (50.0%)
17 (58.6%)
9 (47.4%)
90 to 99 years
23 (30.3%)
8 (28.6%)
10 (34.5%)
5 (26.3%)
Over 100 years
1 (1.3%)
1 (3.6%)
0 (0.0%)
0 (0,0%)
Care Level
1
0 (0.0%)
0 (0.0%)
0 (0.0%)
0 (0.0%)
2
27 (35.5%)
13 (46.4%)
10 (34.5%)
4 (21.1%)
3
39 (51.3%)
13 (46.4%)
11 (37.9%)
15 (78.9%)
4
9 (11.84%)
2 (7.1%)
7 (24.1%)
0 (0.0%)
5
1 (1.3%)
0 (0.0%)
1 (3.4%)
0 (0.0%)
Marital status
widowed
51 (67.1%)
19 (67.9%)
20 (69.0%)
12 (63.2%)
Married or registered civil partnership
10 (13.2%)
4 (14.3%)
3 (10.3%)
3 (15.8%)
Separated/divorced/civil partnership annulled
6
1 (3.6%)
3 (10.3%)
2 (10.5%)
single
9 (11.8%)
4 (14.3%)
3 (10.3%)
2 (10.5%)
Main analyses
Differences at T0
There were no differences at T0 except for the mindfulness subscale
Observing (p =0.006). The post hoc test showed that the
on-site group differed significantly from the app group (p =0.024) and
from the control group (p =0.028). The T0-mean was highest in the
on-site group.
Intention-to-treat analysis
Primary outcome
There was an indication of a different development for subjective
well-being (ΔT0-T1) of medium effect size (p =0.112). In addition,
there was a temporal change in the control group (p =0.052). The
associated post hoc test revealed a significant decrease between T0 and
T1 (p =0.03; |g|=0.82) and indicated an increase between T1 and T2
(p =0.073; |g|=0.60). The intervention groups had no
significant change over time (see [Fig. 3 ]).
Fig. 3 Subjective well-being (WHO-5) at the three survey
times per study group.
Secondary outcomes
The change scores ΔT0-T1 for mindfulness subscale Observing were
different between the three study groups (p =0.014). Participants
with app access showed an increase in the mindfulness subscale
Observing compared to the minimally decreased control group
(p=0. 011; |g|=0.91) and to the almost unchanged on-site group
(p =0.052; |g|=0.72).
As- treated analysis
Primary outcome
When considering the intensity of participation, there was a significant
difference between the three groups (ΔT0-T1, p =0.036),
particularly between participants with high participation intensity and
those in the control group (p =0.039; |g|=0.78), whereas the
values in the latter decreased between T0 and T1 as in the ITT analysis
(see [Fig. 4 ]).
Fig. 4 Subjective well-being (WHO-5) at the three survey
times depending on participation intensity.
Secondary outcomes
The development of mental health between T0 and T2 differed between the
three groups (p =0.027), particularly between participants with
low (decrease) and high (increase) participation intensity
(p =0.010; |g|=0.80).
Sensitivity analysis
Seven people (nO =4; nA =3) achieved less than 20 points on
the MMST at minimum one survey point after T0. The results in the control group
remain consistent because none of the participants met the exclusion criteria
for the sensitivity analysis.
Differences at T0
Groups did not differ at T0 except for the mindfulness overall score
(p =0.029) and the subscale Observing (p =0.028). The
post hoc test showed for both results that the on-site group had
significantly higher values than the app group (overall score:
p =0.024; subscale Observing : p =0.029).
Intention-to-treat analysis
The sensitivity analysis showed some deviations from reported results of the
main analysis. Differences of subjective well-being (ΔT0-T1) increased
between the study groups. Further, there were minor deviations from the main
analysis regarding the mindfulness subscale Observing (ΔT0-T1;
p=0.019). The increased app group differed significantly from the minimally
decreased control group (p=0.018; |g|=0.93) and on-site group (p=0.022;
|g|=0.80).
In addition, the three study groups exhibited a significant difference in
mindfulness (overall score) (ΔT0-T1: p =0.034, ΔT0-T2: p =0.03).
The post hoc tests revealed a significant difference between the on-site
group (minimally decrease) and the app group (increase) (ΔT0-T1: p=0.044;
|g|=0.774). The post hoc tests at ΔT0-T2 were not significant. Additionally,
there was a temporal change regarding the mindfulness score (overall score)
within the app group (p =0.014). The post hoc test demonstrated a
short term increase (ΔT0-T1: p =0.035; |g|=0.572).
As-treated analysis
Almost identical results were found in the sensitivity analysis based on the
as-treated samples with regard to subjective well-being (ΔT0-T1;
p =0.038). The difference regarding mental health, however, was no
longer visible.
In the sensitivity analyses, a difference in happiness was identified
(ΔT0-T2; p =0.03). Residents with a low participation intensity
experienced a decrease and residents with a high participation intensity
showed an increase in happiness (p =0.008; |g|=1.039).
Discussion
We developed and evaluated a MII (on-site and app) for residents in nursing homes.
The aim of the present RCT was to assess the influence of this intervention on
subjective well-being and other health-related outcomes considering the study group
(on-site, app, control), the participation intensity and the cognitive
resources.
It can be summarized that the on-site and app intervention appear to “cushion” a
decrease in subjective well-being that was observed in the control group. Subjective
well-being in the control group worsened during the intervention period (|g|=0.82).
Additionally, the app group showed an increase in mindfulness (subscale
Observing ) during the intervention period in comparison to the control
group (slight decrease; |g|=0.91) and the on-site group (unchanged values;
|g|=0.72).
A high participation was associated with the maintenance or increase of
health-relevant parameters. On the one hand, the difference in subjective well-being
(ΔT0-T1) was identified between the control group and residents with a high
participation intensity (|g|=0.78). On the other hand, residents with low
participation intensity showed a decrease in mental health over the entire study
period (ΔT0-T2), while an increase was observed in the control group (|g|=0.87) and
group of high participation intensity (|g|=0.80).
The discrepancies between the results of the main and sensitivity analyses suggest
that the intervention was particularly more useful for individuals who possess a
sufficient level of cognitive ability, as assessed by the MMST. The described
beneficial effect of the app intervention regarding the mindfulness subscale
Observing (ΔT0-T1), for example, seems to be particularly applicable for
people without decreased cognitive abilities (sensitivity analysis: app and control
group: |g|=0.64; app and on-site group: |g|=0.77). In this sample, a high
participation intensity was also associated with an increase in happiness, while
participants with a low participation intensity experienced a decrease
(|g|=1.04).
Lastly, none of the analyses of the mindfulness subscales Describing, Acting with
attention and Accepting without evaluation , as well as cognitive
performance, life satisfaction, and health-related quality of life indicated a
significant effect of the intervention for the target group.
The high standardization (on-site courses by one qualified trainer; uniform
video instructions in the app) of the intervention should be positively
highlighted.
The results does not prove, but already indicate that the intervention improves the
subjective well-being and mindfulness of the target group. Individual support could
explain the higher effectiveness in the app group. The effectiveness of the
intervention is consistent with results in other target groups. For example, Compen
et al. (2020) also found that group-based and digital MBI are effective in reducing
psychological stress compared to passive control groups [26 ].
The participant completion rate of over 70% can be seen as an indication of the high
level of acceptance and feasibility of the intervention. A comparative study shows
that in a mindfulness intervention the completion rate is higher for people with
stress symptoms than for pain patients [27 ]. Ernst et al. (2008) noticed that the study population of older
people is more characterized by chronic complaints, which is how the authors explain
the higher dropout rate in their study. In relation to the vulnerable target group
at hand, the authors already consider a completion rate of 60% to be acceptable
[5 ].
The “cushioning” effect on residents regarding a psychosocial outcome (here:
subjective well-being) was also observed in preventive exercise interventions –
here, although depressive symptoms were not reduced by the intervention, these
remained constant for six months in comparison to a decline in the control group
[28 ]. However, in our RCT, the
subjective well-being of the control group almost settled back to the original level
at T2. This development could potentially be explained by a seasonal effect
(winter). Long-term studies remain necessary.
The results of the as-treated analysis confirm one of the mindfulness principles of
the necessity of regular practice to achieve desired outcomes [29 ]. Parsons et al. (2017) reported in
their meta-analysis small to moderate effects of exercise units in mindfulness
interventions [30 ]. Contrary, Keng et al.
(2022) found no significant relationship between training duration and changes in
the parameters examined [31 ]. Overall,
the dose-response relationship in meditation warrants further research.
A plausible explanation for the slightly better results of the app intervention
regarding mindfulness (subscale Observing ), in addition to the lower initial
level (see differences from T0), could be the individual technical support available
to these participants. A meta-analysis of digital-based mindfulness interventions
showed, based on 97 included RCTs, that the stress-reducing effect was higher when
the intervention was accompanied by specialist staff [32 ].
Limitations
The generalizability of the results to all residents is limited because most of
the participants had care levels 2–3 (86.8%), whereas in Germany only 54% of
those in need of care in inpatient facilities have these care levels [1 ]. Most residents in Germany have
higher care levels. It would be useful to develop an adapted intervention or
alternative health promotion intervention for residents with care level 4 and 5.
The sample size also limits generalizability. However, due to the nature of a
pilot study, the analyses are exploratory in nature and must be interpreted
accordingly.
The study was conducted in the context of the Covid-19 pandemic, which had a
potential impact on the results. Political regulations, such as quarantine
measures, were also present in the context of the study. For example, in one
nursing facility (control group), a quarantine was prescribed immediately before
T0, as in one of the on-site groups immediately after module 7, which meant that
the last module could not be offered and the T1 survey was postponed. A
systematic review shows that Covid-related visitor restrictions, in general, led
to increased loneliness among residents [33 ].
It should be taken into account that the originally planned double-blinding
(participants and interviewer) could not be implemented due to structural
constraints. However, our alternative cluster-randomization enabled better
feasibility: this also ensured that there was no mutual influence within an
institution.
The randomization was largely successful, as shown by the differences from T0.
However, for taking assumed differences into account, we used the reported
change scores.
Limitations with regard to the data collection process apply because the surveys
were provided in person, which could have led to socially desirable answers.
This is, however, generally difficult to rule out in psychosocial studies [34 ]. The additional use of objective
measurements (e. g., for stress processing: salivary cortisol, etc.) could
improve the reliability of the results in future studies.
Another limitation is that there can be daily fluctuations in the parameters
collected (e. g., MMST) [35 ]. It
should also not be neglected that the request for a break during the survey was
fulfilled (T0: n=3; T1: n=1). In such a case, the survey was continued on
another day.
With respect to data preparation and statistical analysis, the single imputation
method can be criticized, among other things, because replacing missing values
can reduce the actual variance of the data. [36 ]
Another methodological limitation in this study concerns the multiple testing
problem, we only corrected for post hoc testing. This was accepted, because the
RCT is a pilot study (exploratory).
The results of the AT, unlike those of the ITT, cannot be interpreted as a
cause-and-effect relationship because the allocation for the AT was not
randomized. It can be assumed that confounders had an influence on the
residents' participation intensity and effectiveness: For example,
individuals with a low participation intensity showed a greater reduction in
mental health (ΔT0-T2). It is possible that individuals practiced less because
of this health development. It should also be taken into account that peers and
companions (e. g., project employees, social service employees) could have
influenced the motivation to participate in both directions.Further research
with larger samples and a wider variety of participants is needed to improve the
generalizability and robustness of the findings. In future research, it would be
helpful to introduce a waiting control group and an active control group as
comparison with another behavioral prevention intervention to see whether the
results can be explained by the MII or mainly by, e. g., the increased personal
attention.
Moreover, it can also be assumed that the structural conditions of the nursing
homes have an influence on the intervention. Opportunities (e. g. technical
support) and obstacles (e. g. missing group room) to the implementation of the
intervention and the influence on effectiveness were discussed as part of the
qualitative sub-study [14 ]. In a
further evaluation, those results of individual interviews should also be taken
into account when constructing questionnaires.
Conclusion
Our RCT aimed to explore the feasibility and effectiveness of an eight-week
mindfulness-informed mind-body intervention, which focuses on physical activity,
strengthening of mental/cognitive resources, psychosocial health, and nutrition in
residents of nursing homes. The results indicate, among others, that the
intervention (on-site and app) is effective, e. g., by buffering against negative
influences on subjective well-being, particularly when participation intensity was
high. In addition, there was an increase regarding mindfulness (subscale
Observing ) in the app group immediately after the intervention. In
addition to its effectiveness, there was also a high level of feasibility and
acceptance (over 70% completion rate) among residents. Further evidence-based
development and scientific evaluation of MBM interventions in the setting of nursing
homes is recommended. Nevertheless, the present pilot study provides suitable
initial indications for the positive health promoting effect of mind-body medicine
interventions on residents in nursing homes and give a solid basis for a subsequent
effectiveness study including sample size calculation.
Fundref Information
gefördert durch den Verband der Ersatzkassen e. V. (vdek) im Namen und Auftrag der
Techniker Krankenkasse (TK), BARMER, DAK-Gesundheit, KKH Kaufmännische Krankenkasse,
hkk – Handelskrankenkasse und der HEK – Hanseatische Krankenkasse
