Typology of residential long-term care units in 1 Germany: An explorative hierarchical clustering on 2 principal components analysis

Background Organizational health care research focuses on describing structures and processes in organizations and investigating their impact on the quality of health care. In the setting of 21 residential long-term care, this effort includes the examination and description of structural 22 differences among the organizations (e.g., nursing homes). The objective of the analysis is to 23 develop an empirical typology of living units in nursing homes that differ in their structural 24 characteristics. Data from the DemenzMonitor Study were used. The DemenzMonitor is an observational study 27 carried out in a convenience sample of 103 living units in 51 nursing homes spread over 11 28 German federal states. Characteristics of living units were measured by 19 variables related to 29 staffing, work organization, building characteristics and meal preparation. Multiple 30 correspondence analysis (MCA) and agglomerative hierarchical cluster analysis (AHC) are 31 suitable to create a typology of living units. Both methods are multivariate and explorative. We 32 present a comparison with a previous typology (created by a nonexplorative and nonmultivariate 33 process) of the living units derived from the same data set.

Data from the DemenzMonitor Study were used. The DemenzMonitor is an observational study 27 carried out in a convenience sample of 103 living units in 51 nursing homes spread over 11 28 German federal states. Characteristics of living units were measured by 19 variables related to 29 staffing, work organization, building characteristics and meal preparation. Multiple 30 correspondence analysis (MCA) and agglomerative hierarchical cluster analysis (AHC) are 31 suitable to create a typology of living units. Both methods are multivariate and explorative. We 32 present a comparison with a previous typology (created by a nonexplorative and nonmultivariate 33 process) of the living units derived from the same data set. 34

35
The MCA revealed differences among the living units, which are defined in particular by the size 36 of the living unit (number of beds), the additional qualifications of the head nurse, the living 37 concept and the presence of additional financing through a separate benefit agreement. Three 38 clusters could be identified; these clusters occur significantly with a certain combination of 39 characteristics. In terms of content, the three clusters can be defined as "house community", 40 "dementia special care units" and "usual care". 41

Conclusion 42
The typology of living units allows to identify more suitable outcomes and to develop more tailor-43 Introduction 53 In Germany, nursing homes are an important part of health care organizations. At present, more 54 than 11,000 institutions are providing service for more than 800,000 people [1]. Nursing homes 55 vary enormously with regard to their structural characteristics. 56 Nursing homes may be affiliated with owners who have different business objectives (for-profit 57 vs. nonprofit); they can be organized as chains with superordinate policies and regulations; they 58 can provide more than 300 beds or fewer than 10, and they may be organized in separable 59 units with different teams and philosophies of care. Additionally, their service mission is 60 multilayered: they deliver professional nursing care, provide opportunities for social interaction 61 and participation for their residents, and ensure that medical care by general physicians and 62 specialists is being delivered and prescribed therapy is received. Likewise, nursing homes are 63 expected to provide an environment that maintains their residents' preserved skills and that 64 supports people with dementia in acting and making decisions autonomously for as long as 65 possible. Nursing homes are also expected to provide an environment in which residents feel at 66 homenot institutionalizedand can thus maintain their quality of life on the highest possible 67 level [2]. 68 German nursing home residents all share the attribute of being approved as care-dependent by 69 the Long Term Care Insurance entity, which enables them to receive benefits. Because people 70 wish to stay at home as long as possible, they do not move to a nursing home until their need 71 for care exceeds what can be provided at home. As a result, nursing home residents are 72 predominantly severely care-dependent; more than 70% are affected by the consequences of a 73 dementia [3,4]. 74 In recent years, as problems with outcome quality have become public, the quality of care in 75 nursing homes has attracted more political and scientific attention. In particular, it was reported 76 that the needs of people with dementia were not being sufficiently addressed [5]. As a result, 77 nursing homes implemented various approaches to dementia care that necessitated some 78 changes in organizational structure. One major change was to use designated care units to 79 separate residents with dementia from residents without cognitive impairments, under the 80 assumption that care for residents with dementia could be better provided in a special 81 environment. The implementation of "Dementia Special Care Units" (DSCUs) was a worldwide 82 development that had its origin in the United States. In Germany, it is estimated that 30%-50% 83 of nursing homes have implemented at least one DSCU [5,6]. 84 The question of whether DSCUs provide better outcomes has been the subject of large 85 research projects throughout the world. Leading researchers from the United States concluded 86 after 20 years of research that "(D) SCUs  One reason why evaluation studies of DSCUs failed to produce explicit results may be that the 95 interpretation of existing quasi-experimental studies is complicated because they are prone to 96 many sources of nonrandom error [9]. One described challenge for the interpretation of the 97 study results is the lack of definitional clarity for DSCUs [10]. Whereas in the U.S., DSCU 98 typologies have been developed in response to this lack of clarity [11,12], such typologies are 99 still missing in Germany. 100 We conducted a study in German nursing homes that aimed to identify resident-and facility-101 related factors that are associated with the nursing home residents' health care outcomes 102 (DemenzMonitor study) [13]. One goal was to answer the question of whether we can find 103 defined according to the German statutory long-term care insurance law, under which people in 153 need of care are reimbursed by the statutory long-term care insurance. 154 Beyond that, there were no inclusion or exclusion criteria for the participation of nursing homes; 155 diversity was intended. The nursing homes that declared their interest in participating were 156 included. All nursing homes participated voluntarily. 157 158 Data collection 159 The data were collected by the nursing home staff using a standardized questionnaire. 160 Therefore, specific questionnaires were developed and tested. The details of the questionnaire 161 development are described in depth elsewhere [13,14]. 162 In separate questionnaires, the data were collected at the level of the nursing home, the living 163 units and the residents. The living unit questionnaire was completed by the head nurse; the 164 nursing home questionnaire was completed by the nursing home manager; and the resident 165 questionnaire was completed by a registered nurse familiar with the resident. More details on 166 data collection can also be obtained from previous reports [6,14]. 167 168 Definition and measures 169 For the present study, we evaluated the same items on the structural characteristics of the living 170 units that were used in the previously published results [14]. These are variables for structural 171 characteristics, such as the organization of meal services, size of the living unit, interior design, 172 architectural characteristics, staffing, etc. The data level of the variables is exclusively 173 categorical. An overview of the variables and their distributions is provided in the results section 174 (see Table 2). In addition to the structural characteristics, resident variables were included to 175 determine age, sex, presence of dementia diagnosis and severity of dementia [16]. This variable 176 was used exclusively to further describe the identified clusters and did not contribute to their 177 calculation. 178 179 Statistical analysis 180 Multiple correspondence analysis (MCA) and agglomerative hierarchical cluster analysis (AHC) 181 were used to develop the typology of living units. First, an MCA was used whose principal 182 components represent synthetic quantitative variables that summarize all categorical variables 183 [17]. This is a dimension-reducing procedure that selects a few characteristic combinations from 184 the many possible characteristics so that as much information as possible is retained from the 185 data. Second, an AHC is performed with the dimensionally reduced data; this method is suitable 186 for identifying groups of living units that are mapped in the geometric structures of the MCA [18]. 187 The statistics that are applied in this study are not restricted to a certain sample size, so it can 188 also be used to describe structures in data with small sample sizes, see [19] for an example 189 with n=12. The statistical software R was used to conduct the statistical analyses [20]. MCA and 190 AHC analyses were performed with the R package "FactoMineR" using the MCA and HCPC 191 functions [21]. The plots of the results were generated using the R Package "factoextra" [22]. 192 The R-code and the raw data are available in the supplemental information. 193 To make the procedure transparent and the graphical results comprehensible, the following 194 sections contain a brief description of the methods used. This includes methodical analysis 195 steps that provide a basis for decision-making regarding the presentation of results. We decided 196 how much information is retained by the MCA and how many clusters are formed by the AHC. The researcher has to choose how many of the axes and eigenvalues he or she wants to omit 225 to reduce the dimensions of the data cloud. Here, the inertia provides guidance. This means 226 that the best low-dimensional solution is calculated that is capable of distinguishing geometric 227 patterns in the data by mapping each structural characteristic and living unit as a point in a 228 nuclear space [26]. To determine the number of axes (dimensions) to be analyzed, various 229 information about the percentage of explained inertia and the interpretability of each axis is 230 taken into account. Table 1 illustrates the proportion of explained inertia for each axis in 231 decreasing order and thus provides the information needed to make decisions about the 232 number of axes to be analyzed. For high-dimensional data sets, the modified inertia rates 233 should also be considered because the inertia rates of the first dimension are usually low.  The proportion of "between-clusters inertia" that can be measured using formula 1 is 25.41% for 270 a three-cluster solution. 271 To understand how this value is derived, the following information is helpful: It is always true 272 that the "between clusters inertia" of partitioning into two clusters is less than the first eigenvalue 273 for the subsequent performance of the AHC that only those axes are included in the 283 analyses that explain a high proportion of total inertia (approx. 80% to 90% in total). 284 The calculation of the total inertia of the data amounts to / − 1 = 1.053 and is distributed 307 over a total of − = 20 eigenvalues. The average eigenvalue is ̅ = 1/ = 0.052 and 308 explains 4.93% of the total inertia. 309 The first axis 1 explains 17.44% of the total inertia, and the second axis 2 explains 13.76% of 310 the total inertia. Thus, the MCA map ( Fig. 2) represents 31.21% of the total inertia. For the 311 interpretation of the principal axes, the categories that contribute significantly to the explanation 312 of the principal axis are informative. These include all categories whose contribution exceeds 313 the average contribution of 2.56%. The first principal axis applies to the following categories: "living unit has a size ≤ 15 beds" (Size 319 0), "living unit is additionally financed" (Finance 1), "living unit has only single rooms" (SRoom 320 1), "nurses do not work exclusively in one unit" (AssignN 0), "lunch is cooked in the kitchen of 321 the unit" (Selfcook 1), "a registered nurse is not always present" (PresenceRN 0), "all meals are 322 served homestyle on the table" (Mealserv 1), "segregated living concept" (Segregative 1), "do 323 not exclusively have single rooms" (SRoom 0), "living unit has a size > 15" (Size 1), "residents-324 per-service staff member ratio is less than or equal to the median" (SSMRatio 1), "residents-per-325 service staff member ratio is greater than the median" (SSMRatio 0), and "integrative living 326 concept" (Segregative 0). The categories are sorted according to their contributions, so that the 327 first category Size 0 explains the main contribution to the first axis. A substantial contribution to 328 the second principal axis is made by the following categories: "no special qualification in 329 psychogeriatric care" (Jobqual 0), "segregated living concept" (Segregative 1), "built specially for 330 people with dementia" (Build 1), "is additionally financed" (Finance 1), and "living unit has a size 331 ≤ 15" (Size 0). These categories each explain between seven and ten percent of the second 332 principal axis. 333 The cosine angle, which can be measured between two categories at the centroid, represents 334 the tetrachoric point correlation. This property is useful for describing the explored data structure 335 in Figure 2. The categories that are close to each other, such as "living unit is additionally 336 financed" (Finance 1), "special qualification in psychogeriatric care" (Jobqual 1), "living unit is 337 protected by exit controls" characterized. The test results show that each of the three clusters in Figure 3 occurs with a 358 specific combination of categories. Table 3 illustrates these combinations, which leads us to the 359 content-related definition of our three cluster types. We designate the three clusters as "home 360 community", "dementia special care units" and "usual care". 361  The percentages in brackets of Table 3  Furthermore, three cases of attributions in the categories can be differentiated for the 380 interpretation of the clusters in Table 3. The first case of attribution concerns categories that are 381 only informative for a particular cluster. We describe this case as a "unique characteristic". This 382 applies, for example, to the "living unit is protected by exit controls" (Guarded 1) category in the 383 "dementia special care units" cluster. 384 The second case concerns dichotomous categories relating to different clusters. We define this 385 case as "strong difference". This is valid for the categories "living unit was not specially built for 386 people with dementia" (Build 0) and "living unit was built specially for people with dementia" 387 (Build 1) because Build 1 relates to the cluster "dementia special care unit" and Build 0 to the 388 cluster "usual care". 389 The third case will be applicable when a category is related to two or more clusters. We define 390 this case as "intersection". This applies to the category "do not exclusively have single rooms" 391 (SRoom 0), which is indicative of both the cluster "dementia special care unit" and the cluster 392 "usual care". However, it should be noted that the second case also applies to the category 393 "living units do not exclusively have single rooms" (SRoom 0) because "living units have only 394 single rooms" (SRoom 1) is informative for the cluster "house community". 395 Categories describing the second case are particularly suitable for describing differences 396 between two clusters. 397 Table 3 shows that these category combinations allow clear distinctions to be made from the 398 cluster "usual care". The five top categories of the cluster "dementia special care unit" and 399 cluster "house community" can be distinguished by the dichotomous categories of the cluster 400 "usual care". 401 In contrast, the differences between the clusters "dementia special care units" and "house 402 community" are distinguished more by their unique characteristics. This distinction is 403 exemplified by the fact that categories such as "special qualification in psychogeriatric care" 404 (Jobqual 1), "segregated living concept" (Segregative 1) and "living unit is protected by exit 405 controls" (Guarded 1) are informative for the cluster "dementia special care units", but, including 406 their dichotomous category, have no significance for the cluster "house community". 407 408

409
The examination of the resident data in Table 4 shows that no differences with regard to the 410 variables "gender", "age" and "diagnosis of dementia" were found, despite the large number of 411

421
The aim of this study was to empirically develop a typology of living units based on their 422 structural characteristics. Using an explorative clustering technique on data from 103 living units 423 in 51 nursing homes, we identified three different clusters (types). We designated the types as 424 "house community", "dementia special care units" and "usual care". The three categories that 425 have the greatest influence on the formation of these types are named below. 426 The categories that showed the strongest influence on the first type, "dementia special care 427 units," were "additionally financed" (Finance 1), "special qualification in psychogeriatric care" 428 (Jobqual 1) and "segregated living concept" (Segregative 1). 429 The categories that contributed most to the second type, "usual care," were "large size" (Size 1), 430 "no special qualification in psychogeriatric care" (Jobqual 0) and "not additionally financed" 431 (Finance 0). The categories that showed the strongest influence on the third type, "house 432 community," were "small size" (Size 0), "living unit with only single rooms" (Sroom 1) and 433 "cooked lunch in the kitchen of the living unit" (Selfcook 1). Prior to this study, we used a 434 deductive approach to define living unit types and used the variables size, living concept, and 435 finance (Palm et al. 2014). 436 If we compare the types identified with the multivariate inductive method to these a priori 437 defined types, we can see that some categories that were used for definition also have a strong 438 impact on the types developed in the MCA model, whereas others have not. 439 Two types were defined using the categories "large size" (Size 1), "segregative living concept" 440 (Segregative 1) and the variable "additional financing regulated by a special agreement" 441 (Finance 0 and Finance 1). Hence, they differed with respect to the additional financing variable, 442 which was present in one type but not in the other. In the MCA model, the categories "no 443 additional financing regulated by a special agreement" (Finance 0) and "large size" (Size 1) 444 correlate with each other, but there is no correlation between the categories "segregative living 445 concept" (Segregative 1) and "large size" (Size 1). However, the category "segregative living 446 concept" (Segregative 1) correlates strongly with the category "additional financing regulated by 447 a special agreement" (Finance 1) but not with "large size" (Size 1). 448 The categories "small size" (Size 0) and "segregated living concept" (Segregative 1) that were 449 also used to define the type "small segregated living units without extra funding" (SSLU) a priori 450 showed no correlation in the MCA model. 451 If we look at the variables that were significant in determining the empirically developed types, it 452 becomes apparent that other variables play roles that were not considered in the a priori 453 definition. This observation applies to "building specific for residents with dementia", "special 454 qualification of the head nurse in psychogeriatric care", "availability of single rooms", "resident-455 per-service staff member ratio (is less or equal than the mean)", "possibilities to cook lunch in 456 the living unit", etc. 457 The different types of development techniques had an impact on the affiliation of the 103 living 458 units to the types. To illustrate this,  One can see that all of the living units that were formerly affiliated with the type "large 465 segregated living units with additional financing regulated by an agreement" (LSLU II) are now 466 affiliated with the type "dementia special care units". 467 However, three living units that were formerly affiliated with the type "large segregated living 468 units without extra funding" (LSLU I) are also affiliated with the type "dementia special care 469 units". It is surprising that one living unit that was formerly affiliated with the type "large 470 integrated living units without extra funding" (LILU) is now also affiliated with "dementia special 471 care units". This may be explained by the fact that this living unit does not have the 472 characteristic "segregative living concept" (Segregative 1) but is defined by the type-specific 473 characteristics "built specially for people with dementia" (Build 1), "special qualification in 474 psychogeriatric care" (Jobqual 1), residents-per-registered nurse ratio is less than or equal to 475 the median (RNRatio 1), "do not exclusively have single rooms" (SRoom 0), "residents-per-476 service staff member ratio is greater than the median" (SSMRatio 0), "lunch is not cooked in the 477 kitchen of the living unit" (Selfcook 0) and "large size" (Size 1). 478 When looking at the type "usual care", it is clear that the majority (47 of 59) were formerly 479 affiliated with the type "large integrated living units without extra funding" (LILU). However, 11 480 living units from the type "large segregated living units without extra funding" (LSLU I) are now 481 affiliated with the "usual care" type. The type "house community" is more or less consistently 482 compounded by living units that were formerly affiliated with the small living units (integrated 483 and segregated without extra funding). 484 Again, what is surprising is that one living unit that was formerly affiliated with the type "large 485 segregated living units without extra funding" (LSLU I) is now affiliated with the type "house 486 community". This can be explained by the categories "lunch is cooked in the kitchen of the living 487 unit" (Selfcook 1), "all meals are served home style on the table" (Mealserv 1), "residents-per-488 service staff member ratio is less than or equal to the median" (SSMRatio 1) and "living unit is 489 not additionally financed" (Finance 0), which were evident in this living unit. 490

491
In the present study, we also showed which variables and categories do not contribute to the 492 cluster model "constant assignment of service staff" (AssignSSM 0 and AssignSSM 1), "certified 493 nursing assistant ratio" (CNARatio 0 and CNARatio 1), "residents-per-nursing assistant ratio" 494 (NARatio 0 and NARatio 1), "furnishing of public rooms" (Furniture 0 and Furniture 1), "living unit 495 is located in a separate building" (Separate 1), "living unit is not protected by an exit control" 496 (Guarded 0). 497 Some of these variables ("furnishing of public rooms" and "constant assignment of service staff") 498 also did not show significant differences between the formerly defined five living unit types. 499 500 In contrast to the previous results, "intersections", "unique characteristics" and "strong" 501 differences" between the clusters can be identified for the current cluster solution. 502 This is evident in the classification of the categories that are described for the results of Table 3. 503 These attribution possibilities result from the multivariate static model, enabling the relationships 504 between the clusters to be described in detail. 505 Furthermore, the cluster association in the current results is not determined by the fact that the 506 living units have all the cluster-specific characteristics in Table 3. Rather, the probability that a 507 living unit belongs to a particular cluster increases with the presence of each additional cluster-508 specific characteristic. Thus, in terms of the data, it is probable (92% chance) that a living unit 509 with the characteristic "small size" (Size 0) belongs to the cluster "house community". 510 The probability increases to 95% if the characteristic "lunch is cooked in the kitchen of the living 511 unit" (Selfcook 1) is specified in addition to the characteristic "small size" (Size 0). When a living 512 unit has the first three characteristics of the cluster "house community", the affiliation is 100%. 513 With this application, the typology can also be applied to living unit characteristic combinations 514 that are not contained in our data. 515 516 Finally, a comparison of the "dementia special care units" cluster in Table 3  The analysis of the living units shows that systematic differences based on the interrelationships 529 of numerous characteristics can be identified. These results lead to a complex type formation, 530 as seen from the fact that the types are described by the interaction of nine or more 531 characteristics. This supports the assumption that definitions that are solely based on size or 532 living concept ignore the diversity within these groups [14]. 533 A main result of the comparison is that the five a priori types would not be formed in the 534 multivariate model because there are major groups of characteristics that correspond more to 535 each other than to other characteristics and thus lead to a more stable cluster solution. If the 536 intersections of the five cluster solutions and the three cluster solutions are considered, it 537 becomes apparent that the variable "additional financing regulated by a special agreement" and 538 "size of the living unit" are particularly suitable for distinguishing between them. The variable 539 "living concept" has a significantly lower impact on the differentiation of clusters. This can be 540 seen, on the one hand, in the ranking of the categories and, on the other hand, in the result that 541 the variable is insignificant for the cluster "house community". 542 Regarding a classification of living units based on the present study, the following practical 543 recommendation can be made: It can be assumed that a living unit belongs to a cluster if it has 544 three or more of the characteristics shown in Table 3. If we look at the first three characteristics 545 of the clusters, we see the following allocation probability: 546 547 1. If a living unit is assigned the characteristics "additional financing regulated by a special 548 agreement" (Finance 1), "special qualification in psychogeriatric care" (Jobqual 1) and 549 "segregative living concept" (Segregative 1), then it is 100% in cluster 1. 550 1. If a living unit has the characteristics "living unit has a size > 15" (Size 1), "no special 551 qualification in psychogeriatric care" (Jobqual 0) and "not additionally financed" (Finance 552 0), then it is 96.49% in cluster 2. 553 2. If a living unit with the characteristics "living unit has a size ≤ 15 beds" (Size 0), "living 554 unit has only single rooms" (SRoom 1) and "cooked lunch in the kitchen of the living unit" 555 (Selfcook 1), then it is 100% in cluster 3. 556 If the characteristics in the ranking of the table are higher, the classification of the corresponding 557 living unit is more reliable. 558 559 From a methodological perspective, it should be noted that the formation of a typology of living 560 units based on complex characteristic correlations can be more appropriately described using a 561 multivariate statistical method. A methodological approach such as the one applied in the 562 present study is suitable to map the multiple interrelationships in the care landscape [26,37]. An 563 advantage of this explorative analysis is that it delivers a cluster solution that fits the data. In the 564 previously published results, eight possible types were defined a priori, of which only five types 565 could be realized in the data [14]. 566 567 There are methodological limitations of the DemenzMonitor study and the present study that 568 limit the external validity of the results. The participating institutions are spread over 11 federal 569 states. It should be noted here that the distribution of institutions among the federal states in the 570 data set does not correspond to the actual distribution of inpatient geriatric care institutions in 571 Germany. Therefore, the results cannot be considered representative of German care 572 institutions in general. A further methodological limitation relates to the dichotomization of 573 variables. For the staffing variables, the ratio was split using the median. Such a definition is 574 difficult to justify and is normative. This causes information to be lost. An alternative would be to 575 use methods that can map both categorical and metric variables in a model. Building specific for residents with dementia The living unit was not specially built for people with dementia.
Build 0 54% (56) The living unit was built specially for people with dementia.
Build 1 46% (47) Architectural segregation from other units The living unit is not located in a separate building or floor and is not separated by a closed door.
Separate 0 31% (32) The living unit is located in a separate building, floor or is separated by a closed door.
Separate 1 69% (71) Exit control The living unit is not protected by an exit control.
Guarded 0 83% (86) The living unit is protected by exit controls. Guarded 1 17% (17) Furnishing of public rooms Furnishings are solely functional (Functional furniture is provided by the institution and designed for a special use.) Furniture 0 13% (13) Furnishings are functional and individual (Individual furniture is purchased from private individuals.) Furniture 1 87% (90) Opportunities to cook lunch in the living unit Lunch is not cooked in the kitchen of the living unit.
Selfcook 0 73% (75) Lunch is cooked in the kitchen of the living unit.
Selfcook 1 27% (28) Meal serving system All meals (breakfast, lunch and dinner) are not served homestyle on the table (tray system, dish system, buffet system or mixed system).

Constant assignment of nurses
Nurses do not work exclusively in one designated living unit.
AssignN 1 93% (96) Constant assignment of service staff Service workers do not work exclusively in one designated living unit.
AssignSSM 1 75% (77) Continuous presence of a registered nurse A registered nurse is not always present during the day shift in the living unit.
PresenceRN 0 9.7% (10) A registered nurse is always present during day shift in the living unit.
PresenceRN 1 90% (93) Special qualification of head nurse in psychogeriatric care The head nurse of the living unit has no special qualification in psychogeriatric care.
Jobqual 0 75% (77) The head nurse of the living unit has a special qualification in psychogeriatric care.
Jobqual 1 25% (26) Additional financing regulated by a special agreement Living unit is not additionally financed. Finance 0 84% (87) Living unit is additionally financed. Finance 1 16% (16) Living concept Integration (residents with and without dementia live together in one living unit).
The RNRatio is greater than the median (cutoff: median = 18).
RNRatio 1 49% (50) Certified nursing assistant ratio (defined as nurses with a minimum education of one year) There are no Certified nursing assistants working on the living unit.
CNARatio 0 30% (31) There are Certified nursing assistants working on the living unit.

Accessible outdoor area
There is no accessible outdoor area. Outdoor 0 6.8% (7) The residents can go out alone. Outdoor 1 80% (82) The residents can only go out in the presence of a caregiver.