Living condition, weight loss and cognitive decline among people with dementia

Abstract Aims The aim of this study was to investigate cognitive performance and BMI of patients with dementia living in their own homes with family members, nursing homes and alone. Design A prospective observational cohort study with a quantitative design. Method Mini–mental state examination (MMSE) scores and BMI were examined with a sample of Slovak patients (N = 428). Patients were followed up 12 months later after the first examination. Results Cognitive decline was significantly faster for patients living in nursing homes and for solitary patients. BMI consistently decreased in the follow‐up examination and this drop was stronger in patients living alone and in nursing homes. Patients with VaD manifested a stronger BMI decline as compared with AD patients. This study suggests that impoverished conditions such as nursing homes or social isolation of solitary people contribute to stronger progress in dementia. Healthcare professionals need to implement meaningful activities for institutionalized people and for people who are living alone to eliminate the negative impact of an impowerished environment on patient's cognitive functioning.

that health workers lack adequate knowledge and skills. There is a frequent absence of physicians in nursing homes, a lack of attention to symptom control and spiritual needs, poor hygiene and lack of care planning. These variables are in all probability responsible for lower family satisfaction with nursing homes compared with home or hospital settings (Teno et al., 2004). Furthermore, NH can be a source of psychological stress to their residents; more than 80% of NH staff in the USA observed at least one psychologically abusive incident (Pillemer & Moore, 1989) and 79% of NH staff surveyed in Germany confessed that they had abused or neglected a resident at least once over the previous 2 months (Goergen, 2001).
Moving to a nursing home where the older people no longer permanently live with their relatives is stressful for them (An & Jo, 2009;Givens et al., 2012). Furthermore, institutionalization aggravates cognitive decline, probably due to the impoverished environment of nursing homes, defined by Volkers and Scherder (2011) as an environment with limited possibilities for physical and social activity (González-Colaço Harmand et al., 2014;Scocco, Rapattoni, & Fantoni, 2006;Wilson, McCann, et al., 2007;Winocur & Moscovitch, 1990). Indeed, a sample of research revealed a negative association between social isolation and cognitive functioning (Conroy, Golden, Jeffares, O'Neill, & McGee, 2010;O'Luanaigh et al., 2012;Tilvis et al., 2004). A cognitive decline was also documented in non-human animals, particularly in older individuals (for a review see Volkers & Scherder, 2011). Old rats, for example, living in an impoverished environment for 92 days showed a decline in learning and memory compared with rats in a standard environment (Winocur, 1998). The same can be applied to sedentary and lonely people who have, due to an impoverished environment, worse cognitive functions and faster cognitive decline than physically and socially active people (Ayalon, Shiovitz-Ezra, & Roziner, 2016).
Several authors have demonstrated that loneliness is also related to dietary inadequacies (Walker & Beauchene, 1991) and undernutrition (Pilrich & Lochs, 2001;Nogay & Akıncı, 2012;Eskelinen, Hartikainen, & Nykänen, 2016). We suggest that body mass index (BMI) will drop amongst solitary people and those living in NH more rapidly as compared with people living with families.

| Aim of the study
The aim of this study is twofold. First, we examined the cognitive performance of patients with vascular dementia (VaD) and Alzheimer's disease (AD) with respect to three different living conditions (homes with family members, nursing homes and alone). We hypothesize that residents living in nursing homes and alone have a worse cognitive performance compared with residents living in their homes. Secondly, we examined possible differences in BMI in these patients with respect to the same conditions to discover whether weight loss is faster in NH residents or in solitary patients compared with those living in homes with their families.

| Design
This prospective observational cohort study adopted a quantitative research design.

| Sample
The findings were derived from data collected through two waves

| Data collection
We examined patients two times each at approximately 12-month intervals. All the examinations were performed by the same person (MH).

| Weighing
Each patient first undressed down to their underwear and was weighed on a digital certificated scale (OMRON HN 288) to the nearest ±0.1 kg during each examination. All patients were weighed on the same scale. The height of the patients was measured to the nearest 1 cm during the first examination.

| Cognitive performance
All the patients were examined with a mini-mental state examination (MMSE) 30-point questionnaire which is used extensively in clinical and research to screen for dementia (e.g., Meaney, Croke, & Kirby, 2005;Mortimer, Ebbitt, Jun, & Finch, 1992;Pangman, Sloan, & Guse, 2000). Repeated use of the MMSE in patients with diagnosed dementia was also reported by other authors (e.g., González-Colaço Harmand et al., 2014;Samuel et al., 2000;Scocco et al., 2006). The test administration obviously did not exceed 10-30 min. Higher MMSE scores indicate a better cognitive performance. The mean MMSE scores from the first examination were almost always higher than the scores from the second examination. This suggests that patient's cognitive performance declined. A high difference meant that the progress in dementia was rapid while a low difference suggested that the progress in dementia was slower.

| Data analysis
Differences in the MMSE scores between the patients (dependent variable) living in NH, with families and solitary (categorical predictor), were compared with the general linear model (GLM) with patients' age defined as covariate. Mean difference scores from the MMSE questionnaire (first minus second examination) were not normally distributed (Kolmogorov-Smirnov test, d = 0.15, p < .001) and advanced examination of the data revealed that they corresponded to the Gamma distribution. This is a two-parameter family of continuous probability distributions. More details about parameters can be found in Boland (2007, p. 43). In this case of distribution, however, all data should have higher values than zero. In our case, two patients (0.5%) showed no differences in the first and second examination scores (i.e., their final score was 0); thus, we used X + 0.1 transformation of the data to obtain positive values. The data were finally analysed with a generalized linear model with gamma distribution. The categorical predictors in the model were gender, type of care and type of dementia with age defined as covariate. Differences between the means were subsequently carried out with analysis of contrasts. Nonsignificant interaction terms were removed, and the model was run again (according to Zuur et al., 2009). Regarding potentially confusing variables (relationship status, education, having children or not), their inclusion into the statistical models did not change the results presented below.
Differences between the first and second examination of BMI always resulted in negative values; thus, we could not employ the same approach as with the data from the questionnaire. Data on BMI were Box-Cox transformed and normality was achieved. The general linear model where BMI was measured during the first and second examinations was consequently defined as the within-subject variable.
Categorical predictors in the model were gender, type of care and type of dementia while age was defined as covariate. Differences between means were examined with the Tukey post hoc test. Statistical tests were performed with the software Statistica (Version 8, StatSoft 2007, Tulsa, Oklahoma, USA, http://www.statsoft.com).

| Ethical considerations
Ethics approval was obtained from the Faculty of Health Care and Social Work, Trnava University. while the remaining 80 patients (19%) were alone. Further details regarding differences between groups can be found in Table 1. Table 2 indicates mean MMSE scores for three groups of patients.

| Basic differences in MMSE test scores
GLM with the MMSE score from the first examination as the dependent variable revealed that the scores significantly differed between the three groups of patients (F(2,42) = 5.94, p = .003). Solitary patients had significantly lower mean scores than patients living in NH (Tukey post hoc test, p = .02), but other differences were not significant (all p > .24). Differences in the follow-up examination were also statistically significant (GLM, F(2,42) = 30.4, p < .001). Patients living with their families scored significantly better than patients from NH and patients living alone (Tukey post hoc tests, all p < .01).
The latter group of patients scored lower than patients from NH (Tukey post hoc test, p < .001).

| Differences in MMSE test scores after 12 months
The Cronbach's alpha of the MMSE questionnaire for the firstand second-examination phase was high (0.71 and 0.90 respectively). The test-retest reliability was also acceptable (Guttman split-half reliability = 0.67, Spearman-Brown coefficient = 0.79), indicating that measurements during both the first and second examinations were reliable. The patient's age was significantly associated with the final MMSE score meaning that progress in dementia was more rapid amongst younger patients compared with older patients (Table 3, Figure 1). Gender did not influence the final MMSE score. The type of dementia was also associated with the final MMSE dementia scores (Table 3) When considering the interaction terms between the variables (Table 3), the type of care × type of dementia interaction suggests that under family care, patients with Alzheimer's diagnosis scored lower (i.e., the progress in dementia was slower) than patients with vascular diagnoses (Figure 3).

Family
Nursing homes Alone  (Cypriani et al., 2015). In line with our hypothesis, patient's cognitive performance decreased significantly more when they lived alone or in nursing homes compared with patients living in homes with their families. This finding is particularly important because a majority of dementia patients in western countries live in nursing homes (Givens et al., 2012;Mitchell et al., 2005;Teno et al., 2004). Our sample revealed a similar trend since, compared with solitary patients and those living at home, a majority of patients (45%) lived in nursing homes.
To date, few studies have shown that living in nursing homes is associated with cognitive decline (González-Colaço Harmand et al., 2014;Scocco et al., 2006;Wilson, McCann, et al., 2007;Winocur & Moscovitch, 1990). The results of these studies suggest that the transition from the community to a nursing home is cognitively deleterious. However, these findings can potentially be explained by pre-existing differences between patients, since dementia is one Loneliness is associated with limited social networks and social engagements and is consequently associated with a faster cognitive decline in older people with or without cognitive impairment (Andrew & Rockwood, 2010;Bennett, Schneider, Tang, Arnold, & Wilson, 2006;Fratiglioni et al., 2004;Karp et al., 2006;Krueger et al. 2009). In line with our findings, Meaney et al. (2005) found that lower MMSE scores were associated with loneliness. Furthermore, Newall, Chipperfield, Bailis, and Stewart (2013) revealed that loneliness is an independent risk factor for mortality and reduced physical activity among older adults. Low physical activity (Hawkley, Thisted, & Cacioppo, 2009), unmet social and emotional demands (Meaney et al., 2005), an impoverished environment (Volkers & Scherder, 2011) and probably an absence of care from family members associated with lower emotional support (Ellwardt, Aartsen, Deeg, & Steverink, 2013) may negatively influence both the emotional and cognitive performance of solitary people.
The second aim of this research was to compare BMI between patients living in NH, alone and in their own homes with family members. Patients with VaD showed a stronger decline in BMI over 12 months compared with patients with AD. This result can be explained by more frequent dietary restrictions recommended by physicians to reduce the incidence of ischaemic heart disease and stroke in patients with VaD (Zimetbaum, Frishman, & Aronson, 1991). AD patients frequently suffer from appetite loss (Finkel, 2000;Finkel, Costa E Silva, Cohen, Miller, & Sartorius, 1996) which can explain the BMI drop in this study.
Both weight loss and cognitive decline can be in mutual association; weight loss is one of factors included in the operational definitions of frailty in older age, a syndrome particularly common in nursing homes (Kaiser et al., 2010). Given that cognitive impairment and dementia are causally linked with frailty (Panza et al., 2015); it seems that weight loss and/or other forms of physical frailty are associated with the development of dementia and AD (Panza et al., 2015).
Patients who lived alone showed the highest BMI compared with patients from the remaining two groups. These results are surprising at first glance because loneliness is obviously associated with malnutrition (Donini et al., 2013;Ramic et al., 2011;Shahar, Shai, Vardi, & Fraser, 2003). We suggest that their diet is not under a F I G U R E 4 Descriptive statistics for distribution of BMI among patients across two examinations similar control as with patients in nursing homes or with families. It can therefore consist of high calories although unhealthy foods may promote high BMI (Neumark-Sztainer, Wall, Story, & Standish, 2012).
Perceived social isolation and a negative affect are associated with binge eating (Mason, Heron, Braitman, & Lewis, 2016) which can also contribute to higher BMI in solitary patients. It should be noted that food intake amongst solitary people depends on a great extent on their attitudes towards cooking (Hughes, Bennett, & Hetherington, 2004), distance from food markets and the patient's socio-economic status (Donini et al., 2013). These variables were not examined in this study but could at least partly influence the presented results.
Family care was associated with the smallest BMI decline compared with patients living in nursing homes or alone. The diet preservation recommended by physicians may be significantly lower amongst these patients compared with the strict conditions in nursing homes.
There is no data, however, to support this claim although our personal experiences strongly support this possibility (M. Harsányiová, personal experiences). In this case, the lower drop in BMI needs not have positive outcomes on the health of patients. An increased decline in BMI in patients from nursing homes could be, conversely, influenced by inadequate care in nursing homes (Donini et al., 2013;Stange et al., 2013).
This latter claim can be supported by a comparable major BMI drop amongst solitary people, who are unable to obtain and cook foods of the required amount and quality (Donini et al., 2013;Hughes et al., 2004;Ramic et al., 2011). Further in-depth research involving family members is clearly required to support our hypothesis.

| Limitations
The main limitation of this study is lack of patients living alone.
Unfortunately, most of family members or neighbours are not interested in this cohort of people, which finally resulted in smaller sample size. Secondly, we have not enough childless patients who deserve special attention. According to our own experiences, childless patients who have lived for a long time alone enjoy NH more than those who have children. Childless patients obviously enjoy relatively rich social ties, thanks to the interest of other patients or staff in NH. We suggest that the investigation of NH from the perspective of psychosocial and health benefits for childless patients would be an avenue for further research.

| Relevance to clinical practice
Patients in NH should be provided with a daily programme with diverse activities to stimulate cognitive functions. Physical activities (e.g., rehabilitation, working therapy and physical training) should be planned for various parts of the day, to prevent physical decline.
Patients should live with other patients who are at a similar level (not more progressive) of dementia and, if possible, rooms should be equipped with furniture from their homes.
In case of patients living alone, family members and/or social authorities should be contacted and trained to provide adequate home care for them. Social authorities should regularly monitor patients living alone and secure adequate health care and/or recommend any adequate NH.
If the patient is living with family members, they should also be educated to provide adequate health care and rehabilitation. Any

ACK N OWLED G EM ENTS
We thank David Livingstone for improving the English and anonymous referees for constructive comments.

CO N FLI C T O F I NTE R E S T
Authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N S
MH and PP were involved in the design of the study; MH and PP were collected and analysed the data; and MH and PP were prepared the manuscript.