This study showed that residents' characteristics differ in small-scale living facilities and regular psychogeriatric wards, although all residents required a similar nursing home level of care. Residents in small-scale living facilities had a higher cognitive and functional status than residents in regular wards. Demographic characteristics such as living condition prior to admission and length of stay could explain these results to some extent. Length of stay in small-scale living facilities was inevitably shorter, since these are relatively new facilities (newest facility was open for one year), whereas regular nursing home wards are located in long established facilities. This explains the large difference (i.e. 17 months) in mean length of stay between the two care settings. However, while controlling for this and other demographic variables, the association remained between dementia care setting and cognition and functional status. Although some studies have found similar results regarding functional status [12, 14, 28] and cognition , other studies did not find significant differences [4, 29].
An explanation for our findings may be that selection has occurred in allocating residents to small-scale living facilities, despite similar admission criteria for both dementia care settings as determined by a standardized assessment procedure performed by a governmental agency. Most of these residents were transferred from a regular psychogeriatric ward. As residents in small-scale living had better cognitive and ADL performance, it seems that residents with the best cognitive and functional abilities were selected for the small-scale living facilities. A recent study by te Boekhorst et al. (2009) confirms this explanation . They found that residents admitted in small-scale living were in a slightly earlier stage of dementia than residents admitted in traditional nursing homes, as reflected in significantly higher cognitive performance and functional abilities.
A selection process is probably related to the innovative concept of small-scale living facilities. Although small-scale living is currently expanding in the Netherlands, these facilities are still relatively new compared with traditional nursing homes. Over time, residents' characteristics may change resulting in an increased care dependency and decreased cognitive and functional status. Research conducted in Sweden supports this assumption. In Sweden, group living is a long-established dementia care setting, in which residents have become more ADL dependent over the years [11, 16]. These results support a clinical experience in Sweden that over time, residents were admitted in a later stage in their dementia . However, our study identified that already 42.5% of residents in small-scale living had a low level of cognition and functional status. These results highlight the importance of research into suitability of small-scale living for residents with more cognitive and functional impairments.
In our study, the level of care dependency, as measured with DCSP scores, did not differ between the two settings. This is in line with the standardized assessment procedure to determine the level of care: all residents in our study require a similar intensive nursing home level of care. However, we found that residents in small-scale living facilities were more independent in ADL and had a better cognitive performance. Since we derived DCSP scores from the medical record, this might not correspond completely in time with the assessment of ADL and cognition during the screening. Therefore residents might have deteriorated due to the progressive nature of their disease which could explain the differences. Moreover, care dependency constitutes more than just cognition and ADL dependency, including behavioral problems for example. In the DCSP scores, behavioral problems are incorporated among others, where a higher score indicates more (behavioral) problems. However, previous research suggested that DCSP items relating to behavior were possibly more difficult to interpret and had a lower reliability than other DCSP items . The overall DCSP scores' validity or reliability was not studied. More research is needed to confirm that DCSP scores are a valid and reliable measure of care dependency and how this measure is related to other validated measures of care dependency.
Additionally, health care policy and economic issues might have had an influence, since financing of care settings is based on these DCSP scores. An adequate score on the DSCP measure might have been a selection criterion for intake in a small-scale living facility, without residents really being as care-dependant as in a regular nursing home ward. Most residents in our sample, approximately two third in both care settings, had a relatively low level of care dependency (DCSP scores 1-5). It might be that for small-scale living facilities, this is an underestimation and that actually residents now classified as having a relatively high care dependency (DSCP scores 6-8) are actually in a lower need of care.
Some limitations regarding this study must be considered. This study focused on cognition and functional status and therefore assessed only a limited number of variables. Other relevant characteristics such as behavioral problems and social functioning need to be investigated as well. Additionally, residents in other care settings could be included, for example residential care, to cover the whole continuum of dementia care in the Netherlands. Furthermore, a cross-sectional design was used, since this study's objective was to compare residents in two dementia care settings. This design limits causal interpretation of our results. For example, it might be possible that at admission ADL and cognition were the same for residents in both care settings, which may imply a positive effect of small-scale living facilities. However, in our sample no standardized information regarding these patient characteristics at admission was present, which is a drawback. Therefore no inferences can be drawn regarding effects of small-scale living facilities regarding the variables ADL and cognition. Longitudinal research is needed to investigate effects of dementia care setting on residents, addressing several important outcome measures such as quality of life, functional status, behavioral problems and social functioning. This is important, since dementia care settings are increasingly directed towards small-scale and homelike facilities. A few studies have been reported regarding these measures showing promising results [4, 7, 12, 13, 28]. However, methodological limitations such as small sample sizes, differences at baseline between groups or a relatively short follow-up period, hinder interpretation of results.
Our results suggest that functional status and cognition of residents living in small-scale, homelike facilities is better than in regular psychogeriatric wards of nursing homes. These differences in baseline characteristics have implications for research and practice. Effectiveness of new dementia care settings is hard to predict. Research focusing on effects of care settings on residents, family members and nursing staff should take baseline differences in residents' characteristics into account, since these could influence outcome measures. Matching of residents based on a profile of functional status and cognition could form a solution for this challenge. This procedure will increase a study's internal validity and therefore enhance the prognostic comparability of the study groups. In addition, statistical analyses can be used to correct for remaining baseline differences between groups.
Furthermore, development of small-scale living facilities may influence daily practice in more traditional nursing homes. Our results suggest that residents with better cognitive and functional abilities were transferred from traditional nursing homes. As a result, care dependency in traditional nursing homes may increase. Our results highlight the importance of research into optimal environments in the continuum of dementia care.