Factors related to readmission to a ward for dementia patients: Sex differences

Authors

Errata

This article is corrected by:

  1. Errata: Erratum Volume 65, Issue 6, 608, Article first published online: 17 October 2011

Toshiyuki Ono, MD, PhD, Department of Psychiatry, Tsuruga Onsen Hospital, 41-1-5 Yoshiko, Tsuruga, Fukui 914-0024, Japan. Email: turugaoh@poem.ocn.ne.jp

Abstract

Aim:  The aim of this study was to investigate the factors related to readmission to a ward for dementia patients with special attention to sex-related issues.

Methods:  We reviewed the data of 326 patients who were hospitalized in a ward for dementia patients between 1 April 2000 and 31 March 2008, and followed up for 24 months after discharge. For univariate and multivariate analyses, patients were divided into: (i) patients who were not readmitted to our ward within 24 months (control); (ii) patients who were readmitted to our ward within 3 months (early readmission); and (iii) patients readmitted within 4 to 24 months (late readmission).

Results:  Factors related to readmission differed between sexes as well as between the early and late stage. A small number of cohabitants and outcome (hospital) were factors related to early readmission in men, while outcome (hospital) and long stay in the ward were related to early readmission in women. High physical function, care distress and short stay in the ward were the factors related to late readmission in women.

Conclusions:  Most patients who were transferred to another hospital were readmitted within 3 months. Some women were readmitted in the late stage. They had relatively high cognitive and physical functions and most of their caregivers had care distress. The causes of readmission were not due to a decline in cognitive function. It is important to prevent complications in dementia patients and to establish a caregiving system for dementia patients that decreases the burden on caregivers in order to reduce the rate of readmission.

IN A PREVIOUS study, we investigated the distress of family members caring for dementia patients and the effect of sex differences among care recipients.1 In a second study, we identified predictors that influenced the outcome after discharge from a ward for dementia patients. In the hospital group, the incidence of complications was high for each sex. In men, the activities of daily living (ADL) score predicted the outcome between the group returning home for care and the institutionalized group. In women, the revised version of Hasegawa's Dementia Scale (HDS-R) score, caregiver, and number of cohabitants influenced the outcome. In women, it was supposed that the predictors of outcome were mainly related to the system under which they received care.2 In a third study, it was observed that there were sex differences among predictors of the length of stay. However, it was difficult to predict the length of stay on admission. Retrospectively, the length of stay was determined by physical and psychological conditions in men. In women, it was supposed that the caregiver's desire to provide care at home reduced the length of stay. However, complication was a common predictor of extension of stay in both sexes.3

There are many studies that have evaluated factors related to readmission to acute medical wards. However, there are few reports investigating readmission to a ward for dementia patients. It is important to understand the psychosocial influences affecting readmission of inpatients with dementia. In this study, we investigated factors related to readmissions.

METHODS

Subjects and procedures

We analyzed data from 326 patients with dementia who were first hospitalized for psychological and behavioral problems between 1 April 2000 and 31 March 2008 and followed for 24 months after discharge. Patients who died during their ward stay were excluded. All patients fulfilled the criteria in DSM-IV-TR for dementia of Alzheimer's type (DAT), vascular dementia (VaD), or other types of dementia (OTD). We checked each patient's age, sex, diagnosis, physical comorbidity (comorbidity), complications during hospitalization (complication), primary caregiver (caregiver), number of cohabitants at home, care distress, outcomes, length of stay and readmission. We checked the three main reasons for hospitalization, which were violence (43.2%), wandering (33.3%) and care distress (32.4%) in men, and care distress (33.0%), wandering (26.5%) and hallucination/delusion (22.3%) in women. Then, we used care distress as a marker of care burden. Cognitive status was assessed with HDS-R,4 while the behavioral and psychological symptoms of dementia (BPSD) were evaluated with the Assessment Scale for Symptoms of Dementia (ASSD).5 Functional status was assessed with Nishimura's ADL scale (N-ADL).6 These assessments were checked every 3 months. We principally used the initial scores of HDS-R, ASSD and N-ADL on admissions and the final scores when the patients were discharged from our ward. If the HDS-R, ASSD, and N-ADL scores were not available at the time of discharge, we substituted the scores on admission.

If a patient's severe psychiatric symptoms or behavioral disturbances improved after medical therapy, it was recommended that they be discharged from the ward for dementia patients. Discharge planning is managed by social workers at our hospital. However, patients with severe physical problems were transferred to another hospital, while patients with psychological problems that needed special treatment were moved to another ward. In our hospital, we have two separate wards; one is for the patients with dementia and another is for patients with psychological problems.

We examined the number of times that patients who were discharged from the dementia ward needed to be readmitted. For analyses, patients were divided into: (i) patients who were not readmitted to our ward within 24 months (control); (ii) patients who were readmitted to our ward within the first 3 months (early readmission); (iii) patients readmitted within 4 to 24 months (late readmission).

We investigated sex differences in their characteristics. Then, we evaluated factors that influenced the early and late readmissions of each sex using univariate and multivariate analyses. For multivariate analysis, the model using the initial scores of HDS-R, ASSD and N-ADL on admission was named Model 1. In Model 1, we checked the effect of each patient characteristic, without complications, outcomes and length of stay, on readmission. Model 2 used the final scores of HDS-R, ASSD, and N-ADL with complications, outcomes and length of stay. Univariate and multivariate analysis indicated that the outcomes were important factors affecting readmission. Therefore, we showed differences in the characteristics among the four groups divided by sex, stage of readmission and outcomes.

Informed consent was obtained from patients and/or their caregivers. The experimental procedure was carried out in line with the Declaration of Helsinki.

Statistical analysis

We used the Mann–Whitney U-test for continuous variables and the χ2-test for categorical data to evaluate sex differences in the characteristics (age, diagnosis, scores on HDS-R, ASSD and N-ADL, comorbidity, complications, caregiver, number of cohabitants, care distress, outcomes and readmissions). The Kaplan–Meier method was used to compare time until discharge (log–rank test). The Kruskal–Wallis rank test, χ2-test and the Kaplan–Meier method were used to test for significance of differences among the three groups (early readmission, late readmission and control). We used the Steel–Dwass test for post-hoc analysis.

Binomial logistic regression with a forward stepwise method was used to analyze the correlation between readmissions (early readmission vs control and late readmission vs control) and characteristics. However, we did not analyze the data on late readmission of men because of the small number of patients in this group. In these analyses, we adopted characteristics including age, diagnosis, comorbidity, caregiver, number of cohabitants at home, care distress and the initial HDS-R, ASSD and N-ADL scores as their cognitive, behavioral and functional status in Model 1. In Model 2, we used characteristics including age, diagnosis, comorbidity, complications, caregiver, number of cohabitants at home, care distress, outcomes, length of stay and final scores. The Hosmer–Lemeshow test was used for goodness-of-fit statistics of these models. There was a small number of missing data. However, there were no differences in variables between the whole sample and the sample used in the binomial logistic regression analysis.

Statistical analysis was carried out using spss version 17.0 (SPSS Japan, Tokyo, Japan) and Statcel3 (OMS Shuppan, Tokorozawa, Japan).7 We regarded P < 0.05 as significant.

RESULTS

We found some differences between men and women in terms of diagnosis, caregiver, and outcomes. However, these two groups did not differ in terms of age, HDS-R, ASSD, and N-ADL scores, comorbidity, complications, number of cohabitants at home, care distress, length of stay and the rate of readmission. DAT was more frequent in women (149/215) than in men (55/111). Most men had been cared for by their spouses (65/111), but many women had been cared for by daughters-in-law (81/215). Most women were institutionalized after discharge, and the ratio of institutionalization was higher for women (97/215) than for men (31/111) (Table 1).

Table 1.  Clinical characteristics of 326 patients
CharacteristicsTotalMenWomenP-value§
n = 326n = 111n = 215
Mean (SD)Mean (SD)Mean (SD)
  • §

    Mann–Whitney U-test for continuous variables and χ2-test for categorical data were used to evaluate sex differences in the characteristics. The Kaplan–Meier method was used to compare time until discharge (log–rank test).

  • ASSD, Assessment Scale for Symptoms of Dementia; DAT, dementia of Alzheimer's type; Home, Home care giving; Hospital, transferred to another hospital; Institution, became institutionalized; N-ADL, Nishimura's activity of daily living scale; OTD, other type of dementia; HDS-R, revised Hasegawa Dementia Scale; VaD, vascular dementia.

Age (years)81.3 (7.2)80.7 (7.3)81.7 (7.2)0.124
Diagnosis    
 DAT204551490.001
 VaD964749
 OTD26917
Initial score of HDS-R9.9 (7.3)9.4 (7.4)10.1 (7.3)0.351
Initial score of ASSD36.7 (21.7)37.4 (23.1)34.9 (21.1)0.455
Initial score of N-ADL28.6 (11.3)28.8 (11.8)28.5 (11.1)0.556
Final score of HDS-R9.5 (7.8)9.1 (7.6)9.7 (8.0)0.566
Final score of ASSD37.9 (21.8)38.1 (20.8)37.9 (22.2)0.808
Final score of N-ADL26.2 (11.8)25.6 (12.2)26.5 (11.6)0.647
Comorbidity (%)68.174.864.70.079
Complication (%)43.949.540.90.158
Caregiver    
 Spouse8965240
 Son41932
 Daughter49940
 Daughter-in-law1032281
 Other relatives23419
 Staff21219
Number of cohabitants2.3 (1.9)2.5 (2.0)2.2 (1.9)0.207
Care distress (%)32.832.4331
Outcome    
 Home10941680
 Institution1283197
 Hospital703733
 Changed ward19217
Length of stay (median, days)92711020.088
Readmission (%)23.325.222.30.582

There were 28 readmissions for men; 22 readmissions were within 3 months (early readmission), and the others were within 24 months (late readmission). For women, there were 48 readmissions; 23 readmissions were within 3 months (early readmission), and the others were within 24 months (late readmission). Late readmissions were more frequent for women than for men (Fig. 1, P = 0.015).

Figure 1.

Number of patients readmitted (men inline imageand women □). There were 28 readmitted; 22 were readmitted within 3 months (early readmission), and the others within 24 months (late readmission). Among the women, there were 48 readmitted; 23 were readmitted within 3 months (early readmission), and the others within 24 months (late readmission). Late readmissions among women were more frequent than those among men (P = 0.015, χ2-test).

On comparison of these groups (control, early and late readmission), the final ASSD score, complication, number of cohabitants and outcomes differed in men, subgroups of women differed in complication, care distress, outcomes and length of stay.

Most men showing early readmission demonstrated complications and were readmitted from another hospital. Men showing late readmission had low cognitive function and a small number of cohabitants at home. Most of the women showing early readmission had comorbidity and complications, and were readmitted from another hospital. These patients were treated for a long time in our hospital. Women showing late readmission had high cognitive and physical function. These patients were treated for a short time in our hospital and most of them were readmitted from home. Their caregivers were burdened.

On comparison between those with early and late readmission, men readmitted later had few complications and most of them were readmitted from home. The women readmitted later demonstrated high cognitive function and had a short stay in the ward. Most of these women were readmitted from home and had a few complications (Table 2).

Table 2.  Comparison of characteristics among the three groups
MenControl (83)ReadmissionP-value§
Early (22)Late (6)
Age (years)80.6 (7.4)82.2 (6.1)75.8 (8.3)0.468
Diagnosis    
 DAT411130.469
 VaD33113
 OTD900
Initial score of HDS-R10.1 (7.3)9.0 (7.8)1.8 (1.7)*0.427
Initial score of ASSD34.7 (23.1)41.2 (20.4)52.3 (27.9)0.127
Initial score of N-ADL28.3 (12.2)29.1 (11.7)34.2 (6.7)0.808
Final score of HDS-R10.0 (7.5)7.8 (8.0)1.7 (1.9)*0.125
Final score of ASSD34.4 (20.0)47.1 (22.2)46.4 (7.7)0.016
Final score of N-ADL25.3 (12.2)24.0 (12.6)33.2 (8.8)0.598
Comorbidity (%)75.977.3500.353
Complication (%)42.286.4*16.7#0
Caregiver    
 Spouse471350.587
 Son810
 Daughter630
 Daughter-in-law1921
 Other relatives220
 Staff110
Number of cohabitants2.8 (2.0)1.8 (1.6)0.8 (0.4)*0.023
Care distress (%)34.927.316.70.553
Outcome    
 Home342*5#0
 Institution2920
 Hospital18181
 Changed ward200
Length of stay (median, days)6570890.726
WomenControl (167)ReadmissionP-value§
Early (23)Late (25)
  • §

    Kruskal–Wallis rank test and χ2-test were used to test for significance regarding the comparison among the three groups (early readmission, late readmission and control). We used Steel–Dwass test for post-hoc analysis. The Kaplan–Meier method was used to compare time until discharge (log–rank test).

  • Significant levels on post-hoc analysis, χ2-test and the Kaplan–Meier method in relation to control, *P < 0.05 and between early and late readmitted patients,#P < 0.05.

  • ASSD, Assessment Scale for Symptoms of Dementia; DAT, dementia of Alzheimer's type; Home, Home care giving; Hospital, transferred to another hospital; Institution, became institutionalized; N-ADL, Nishimura's activity of daily living scale; OTD, other type of dementia; HDS-R, revised Hasegawa Dementia Scale; VaD, vascular dementia.

Age (years)81.9 (7.3)80.6 (8.4)81.3 (5.1)0.405
Diagnosis    
 DAT11120180.238
 VaD4234
 OTD1403
Initial score of HDS-R9.7 (6.9)7.4 (5.5)15.2 (8.5)*,#0.163
Initial score of ASSD35.7 (20.4)31.0 (22.5)33.3 (24.5)0.187
Initial score of N-ADL27.7 (10.9)27.6 (13.8)34.8 (7.4)*0.809
Final score of HDS-R9.4 (7.8)6.8 (6.6)14.0 (8.5)*,#0.156
Final score of ASSD39.1 (21.9)34.6 (22.9)33.0 (23.5)0.309
Final score of N-ADL25.5 (11.2)26.6 (15.5)33.7 (7.5)*0.59
Comorbidity (%)61.782.6*680.134
Complication (%)37.773.9*32#0.003
Caregiver    
 Spouse20220.955
 Son2345
 Daughter3055
 Daughter-in-law6687
 Other relatives1522
 Staff1324
Number of cohabitants2.3 (1.9)1.9 (1.7)1.8 (1.8)0.439
Care distress (%)29.334.856*0.03
Outcome    
 Home503*15*,#0
 Institution9124
 Hospital15162
 Changed ward1124
Length of stay (median, days)107189*44*,#0

According to binomial logistic regression analysis, the factors related to early readmission for men were the number of cohabitants at home (B, −0.305; P = 0.046; odds ratio [OR], 0.737 [Model 1; Cox–Snell R2 = 0.055]), and the number of cohabitants at home (B, −0.411; P = 0.0024; OR, 0.663) and outcome (hospital) (B, 2.950; P = 0.001; OR, 19.115 [Model 2; Cox–Snell R2 = 0.338]). In contrast, Model 1 did not show any factors that distinguished between early readmission and controls for women. Model 2 (Cox–Snell R2 = 0.276) indicated that the factors related to early readmission were length of stay (B, 0.003; P = 0.030; OR, 1.003), and outcome (hospital) (B, 3.215; P = 0.000; OR, 24.901). Finally, Model 1 (Cox–Snell R2 = 0.084) showed that the difference between late readmission and control was linked to care distress (B, 1.069; P = 0.028; OR, 2.913) and the initial score on N-ADL (B, 0.070; P = 0.014; OR, 1.073) for women. Model 2 (Cox–Snell R2 = 0.164) also demonstrated that the differences were influenced by care distress (B, 1.176; P = 0.024; OR, 3.241), length of stay (B, −0.011; P = 0.011; OR, 0.989) and final score on N-ADL (B, 0.059; P = 0.043; OR, 1.061). The P-value on the Hosmer–Lemeshow test of each model was over 0.1 (Table 3).

Table 3.  Factors related to readmission in each model
 BPOR (95% Confidence interval)Cox– Snell R2Classification
  • Binomial logistic regression with a forward stepwise method was used to analyze the relation between readmission (early readmission vs control and late readmission vs control) and characteristics.

  • §

    The Hosmer–Lemeshow test was used for the goodness-of-fit statistics of models.

  • *

    No factor was detected.

  • Home, Home caregiving; Hospital, transferred to another hospital; Institution, became institutionalized; N-ADL, Nishimura's activity of daily living scale; OR, odds ratio.

Men     
Early readmission vs control     
 (Model 1) Goodness-of-fit statistics§: χ2-test = 0.544; P = 0.990   0.05574.7
No. cohabitants−0.3050.0460.737 (0.546–0.995)  
 (Model 2) Goodness-of-fit statistics§: χ2-test = 10.070; P = 0.185   0.33886.4
No. cohabitants−0.4110.0240.663 (0.464–0.948)  
Outcome     
Home 0Reference  
Institution−0.4140.6970.661 (0.083–5.293)  
Hospital2.950.00119.115 (3.455–105.760)  
Changed ward−18.5930.9990  
Men     
Late readmission vs control: not done     
Women     
Early readmission vs control     
 (Model 1)*     
 (Model 2) Goodness-of-fit statistics§: χ2-test = 7.662; P = 0.467   0.27690.5
Length of stay0.0030.031.003 (1.000–1.005)  
Outcome     
Home 0Reference  
Institution−1.4610.1360.232 (0.034–1.584)  
Hospital3.215024.901 (5.702–108.753)  
Changed ward1.2780.2043.588 (0.500–25.744)  
Women     
Late readmission vs control     
 (Model 1) Goodness-of-fit statistics§: χ2-test = 10.808; P = 0.213   0.08487
Care distress1.0690.0282.913 (1.123–7.555)  
Initial score of N-ADL0.070.0141.073 (1.014–1.134)  
 (Model 2) Goodness-of-fit statistics§: χ2-test = 3.589; P = 0.892   0.16487.5
Care distress1.1760.0243.241 (1.166–9.004)  
Length of stay−0.0110.0110.989 (0.980–0.997)  
Final score of N-ADL0.0590.0431.061 (1.002–1.124)  

These findings showed that outcomes were important factors affecting readmission and there were three main groups divided by sex, stage of readmission and outcomes: men and women who returned from another hospital in the early stage; and women who returned from home in the late stage. There was also a small group that included five men readmitted later from home. We compared the characteristics of these four groups divided by sex, stage of readmission and outcomes. All of the men who were readmitted early (n = 18) after being transferred to another hospital demonstrated complications while in the ward. The women who were readmitted early (n = 16) after being transferred to another hospital had a long stay in the ward (181 days). All but one patient had complications. The women who were readmitted late (n = 15) after being cared for at home had good cognitive function (initial score of HDS-R, 15.7 ± 8.4, final score of HDS-R, 15.0 ± 8.3) and a short length of stay (44 days). Many of their caregivers (66.7%) were burdened. The men (n = 5) who were readmitted later from home had low cognitive function (initial score of HDS-R, 2.2 ± 1.6, final score of HDS-R, 2.0 ± 1.9) and a small number of cohabitants at home (Table 4). These findings were similar to those of the groups that were divided by sex and stage of readmission.

Table 4.  Clinical characteristics of four groups that were divided by sex, stage of readmission and outcomes
CharacteristicsMenMenWomenWomen
Early readmitted from hospitalLate readmitted from homeEarly readmitted from hospitalLate readmitted from home
n = 18n = 5n = 16n = 15
Mean (SD)Mean (SD)Mean (SD)Mean (SD)
  • Home caregiving.

  • Transferred from another hospital.

  • ASSD, Assessment Scale for Symptoms of Dementia; DAT, dementia of Alzheimer's type; N-ADL, Nishimura's activity of daily living scale; OTD, other type of dementia. HDS-R, revised Hasegawa Dementia Scale; VaD, vascular dementia.

Age (years)81.3 (6.1)78.4 (6.0)81.0 (9.6)81.1 (5.5)
Diagnosis    
 DAT921311
 VaD9333
 OTD0001
Initial score of HDS-R7.1 (6.3)2.2 (1.6)7.2 (4.8)15.7 (8.4)
Initial score of ASSD44.6 (18.6)52.6 (31.2)32.9 (22.3)32.3 (27.8)
Initial score of N-ADL27.8 (12.0)36.4 (4.3)25.1 (13.9)35.4 (6.5)
Final score of HDS-R5.6 (6.1)2.0 (1.9)5.4 (4.9)15.0 (8.3)
Final score of ASSD51.8 (19.5)46.8 (8.8)35.9 (23.5)31.7 (26.2)
Final score of N-ADL21.6 (12.1)36.5 (5.4)23.3 (15.5)34.9 (6.3)
Comorbidity (%)77.86081.353.3
Complication (%)100093.833.3
Caregiver    
 Spouse12421
 Son1034
 Daughter2034
 Daughter-in-law1155
 Relatives2021
 Staff0010
Number of cohabitants1.6 (1.4)0.8 (0.4)1.7 (1.7)2.4 (1.7)
Care distress (%)27.82043.866.7
Length of stay (median, days)718918144

DISCUSSION

In Japan, the traditional family system has greatly changed and most elderly now live alone or with their spouses.8 In these situations, it is difficult to care for the elderly if they need help to manage routine life. Furthermore, the increasing number of elderly with dementia is a great problem, causing stress for caregivers. Some caregivers try to continue giving care even beyond their limits. Therefore, a long-term care insurance system was started in April 2000 in order to care for the elderly. We have treated patients with behavior and psychological symptoms in the ward for dementia patients, where we have been confronted with the problem that some patients with dementia could not return to their homes and stayed in the ward for a long time. It was also difficult to treat the patients with severe physical problems. Therefore, we evaluated predictors of outcomes and length of stay of the inpatients with dementia in our previous studies.2,3

In the present study, we checked the factors of readmission in the ward for dementia patients because some patients who were discharged were readmitted to our hospital again. Many studies9–13 have evaluated risk factors influencing readmission of patients to acute medical wards. These reports indicated that multiple factors affected readmission to an acute care hospital. However, we found few reports that evaluated factors influencing readmission to the ward for dementia patients. Zanocchi et al.14 conducted an observational cohort study of a group of patients discharged from a geriatric ward and evaluated incidence and predictive factors for hospital readmission. They reported that 107/839 patients (12.8%) had been readmitted and dementia was a factor related to readmission. Andrieu et al.15 reported that 22 patients had one acute hospitalization, seven had two, two had three and one had four hospitalizations during a 12-month follow up of 134 patients with Alzheimer's disease. In our report, 76/326 (23.3%) patients were readmitted within 48 months. These findings suggested that some patients with dementia required readmission.

We checked the factors influencing readmission using binomial logistic regression analysis with Model 1 and Model 2. The percentage of correct classifications of Model 2 was higher than that of Model 1 in each examination. These findings indicated that it was difficult to estimate the factors on admission and that the readmission of each patient was well estimated by the condition at discharge. Therefore, we discussed factors influencing readmissions using the results of Model 2.

The outcome (hospital) was an important factor influencing early readmission to the ward for dementia patients. Univariate analysis indicated that most patients who were readmitted early had complications. In our previous report,2 we determined that the outcome (hospital) was predicted by complication. We consulted another hospital about patients with severe physical condition and sent our patients to them for special treatment. Then we readmitted patients soon after they were treated for their physical problems. It was supposed that these situations probably induced a high rate of readmission from another hospital.

Half of the women readmitted in the late stage showed high physical function and had a short length of stay, but their caregivers were burdened. On univariate analysis, women who were readmitted later had high cognitive and physical functions and outcome (home) was a factor influencing late readmission in this group. A previous study3 indicated that the caregiver's desire to provide care at home reduced the length of stay. The caregiver wanted to live together with the patient as soon as possible, when the patients had high cognitive and physical function. These findings suggested that the caregivers tried to live with dementia patients again but thereafter gave up continuing to provide care to the patients.

For women, the length of stay was shown to be a factor predicting readmission, but the length of stay influences readmissions in a different way. It is assumed that a short length of stay affects the burden on the caregiver and a long stay increases the likelihood of complications. It is supposed that there may be an optimal length of stay that would reduce the likelihood of readmission.

In this study, increased number of cohabitants also reduced the readmission in men. Andrieu et al.15 demonstrated that dependency for ADL-bathing and a low level of education were associated with acute hospitalization. They suggested that the necessity of interventions that support patients and their caregivers to manage the loss of ADL may be a practical approach reducing the need for acute hospital admissions. It was supposed that providing care requires supportive assistance. These results should be taken into account when organizing a caregiving system for patients with dementia.

In Japan, the long-term care insurance system was started in April 2000. Therefore, many elderly use this system. Some use the care services and others are administered by institutions. However, some dementia patients cannot use the caregiving system, and they are repeatedly hospitalized in a ward for dementia patients. The results of our studies suggested that the prognosis of men with dementia in the ward is influenced mainly by their own condition, except that the number of cohabitants at home influenced readmission. However, the prognosis of women mostly depends on the caregiving system available to them.

In conclusion, the factors related to readmission differed between sexes as well as between early and late admission. The causes for readmission were not due to the decline of cognitive function. It is important to prevent complications in patients with dementia and to establish a caregiving system for dementia patients, especially women, that will decrease the care burden in order to reduce the readmission rate.

The selection of our sample is a limitation of this study investigating factors related to readmission because only a small number of readmissions were available for this analysis. There are some differences in the caregiving system of dementia patients among communities and countries. We have discussed the situation affecting the readmission of dementia patients based on data from one facility. Therefore, the results are not generalized, and further analysis is necessary to investigate factors that influence readmission.

ACKNOWLEDGMENT

We thank the staff of Tsuruga Onsen Hospital for their help with our study.

Ancillary