Predictors of length of stay in a ward for demented elderly: gender differences

Authors


Dr 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

Background:  In our previous studies, we found both gender differences among care recipients and predictors that influenced outcomes after discharge from a ward for demented elderly. Here, we investigate predictors that influence the length of stay for each sex.

Methods:  We studied the data of 390 patients with dementia who were hospitalized in a ward for demented elderly between 1 April 2000 and 31 March 2008, and treated until 31 March 2009. The patients were divided into groups classified by gender. We analyzed the gender differences of characteristics and evaluated the predictors that influenced the length of stay in the ward for demented elderly using Cox's proportional hazards model. A model using the initial scores of the Revised Hasegawa Dementia Scale (HDS-R), Assessment Scale for Symptoms of Dementia (ASSD) and Nishimura's activity of daily living scale (N-ADL), which were examined on admission, was named Model 1. In Model 1, we checked the effect of each patient's characteristics, except for complications and destinations, on their length of stay. Model 2 used the final scores of HDS-R, ASSD and N-ADL including complications and destinations.

Results:  There was a clear gender difference in the length of stay. The length of stay of women was longer than that of men. It was difficult to predict the length of stay in Model 1. Age was the only predictor in women and no predictor was identified in men. In Model 2, complications and the final HDS-R and N-ADL scores were predictors of the length of stay in men. Age, complications and destinations were predictors of the length of stay in women.

Conclusions:  It was observed that there were gender 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, not by the social variables in men. In women, it was supposed that the caregiver's wish to give care at home reduced the length of stay. Besides, complication was a common predictor of the extension of stay in each sex. We have to decrease the number of complications as much as possible to reduce the length of stay.

INTRODUCTION

In our previous studies,1,2 we investigated the distress of family members caring for demented elderly and the effect of gender differences among care recipients. As a result, the caregiving situation for elderly men was found to differ from that of elderly women. Most men had been cared for by their wives and were hospitalized because of progressive dementia. In contrast, women could be divided into two groups: (i) those who lived alone and could not return to live with their families after discharge from a ward for demented patients; and (ii) those who lived with their families and could return to live with them again. We identified predictors that influenced the outcome after discharge from a ward for demented patients. In the hospital group, the incidence of complications was high for both sexes. The activities of daily life (ADL) score predicted the outcome between the group that returned home for care and the institutionalized group in men. The Revised Hasegawa Dementia Scale (HDS-R) score, the caregiver and the number of cohabitants influenced the outcome in women. In the case of women, it was supposed that the predictors of outcome for those on a ward for the demented elderly were not related to patient condition, but to differences in the system under which they received care. These results indicated that there were some gender differences among the caregiving systems for the demented elderly. Practically, it is important to know the gender differences on the condition where the demented elderly received care. We also need detailed information about the length of stay to establish a better caregiving system. Therefore, we examined the predictor of length of stay in each sex.

METHODS

Subjects and procedures

We analyzed the data of 390 patients with dementia who were hospitalized for psychological and behavioral problems between 1 April 2000 and 31 March 2008, and treated by 31 March 2009. They fulfilled the criteria of the text version of the fourth edition of Diagnostic and Statistical Manual of Mental Disorders (DSM-IVTR) for dementia of the Alzheimer's type (DAT), vascular dementia (VaD), or other types of dementia (OTD). We included data on each patient's age, sex, diagnosis, medical comorbidity (comorbidity), complications during hospitalization (complication), primary caregiver (caregiver), number of cohabitants, care distress, destinations, outcomes and length of stay. Analysis of the data found that 275 patients had comorbidities; 194 had cardiovascular disease, 68 had endocrine-metabolic disease, 43 had orthopedic disease, 37 had urinary disease, 30 had digestive disease, 28 had respiratory disease, 21 had neurological disease and 29 had other comorbidities. It was also found that 176 patients had complications; 74 had respiratory disease, 32 had digestive disease, 24 had urinary disease, 22 had orthopedic disease, 16 had cardiovascular disease, 13 had neurological disease and 38 had other complications. We checked the three main reasons of hospitalization, which were violence (43.8%) wandering (33.6%) and care distress (32.0%) in men, and care distress (34.0%), wandering (24.8%) and hallucination-delusion (19.8%) in women. Then, we used care distress as a marker of care burden. On admission, the destinations of some patients were unstable or variable. Final destinations were decided after consulting social workers. The cognitive status was assessed by HDS-R.3 In addition, the behavioral and psychological symptoms of dementia (BPSD) were evaluated by the Assessment Scale for Symptoms of Dementia (ASSD).4 The ASSD consists of 45 items, and was designed as a measure of dementia severity, with each symptom weighted according to the severity of dementia. Functional status was assessed by Nishimura's activity of daily living scale (N-ADL).5 These assessments were repeated every 3 months. We principally used the initial scores of HDS-R, ASSD and N-ADL as the scores on admission and their final scores as the scores at the time of discharge from our hospital. If the HDS-R, ASSD and N-ADL scores were not available at the time of discharge, we used the scores obtained on admission as a substitute for those at the time of discharge. One-third of final scores was a substitute for initial scores. When severe psychiatric symptoms or behavioral disturbances improved after medical therapy, the patients were advised to leave the ward for demented elderly. Discharge planning was managed by social workers at our hospital.

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

Statistical analysis

The unpaired t-test for continuous variables and χ2-test for categorical data were used to evaluate the gender differences in the characteristics (age, diagnosis, scores of HDS-R, ASSD and N-ADL, comorbidity, complications, caregiver, the number of cohabitants, care distress, destinations and outcomes). The Kaplan–Meier method was used to compare times to discharge between men and women (log–rank test). Then, we evaluated the relationship between the length of stay and each variable. The Kaplan–Meier method was also used to compare the lengths of stay for each variable (diagnosis, comorbidity, complications, caregiver, care distress and destinations) and simple regression analysis was used to quantify the impact of each variable (age, scores of HDS-R, ASSD and N-ADL, and number of cohabitants) on the length of stay. After that, Cox's proportional hazards model with the forward stepwise method was used to analyze the predictors of the length of stay. Cox's proportional hazards model was carried out for all patients and each sex. A model using the initial scores of HDS-R, ASSD and N-ADL, which were examined on admission, was named Model 1. In Model 1, we checked the effect of each patient's characteristics, except for complications and destinations, on their length of stay. Model 2 used the final scores of HDS-R, ASSD and N-ADL including complications and destinations. In this analysis, we used characteristics including age, diagnosis, comorbidity, caregivers, number of cohabitants, care distress, and initial scores of HDS-R, ASSD and N-ADL as the cognitive, behavioral and functional status of patients in Model 1. In Model 2, we used characteristics including age, diagnosis, comorbidity, complications, caregivers, number of cohabitants, care distress, destination and final scores. In the models including men and women, sex was added as an expected predictor. Continuous variables (age, the scores of HDS-R, ASSD, N-ADL and the number of cohabitants) were used in these models. Categorical data were coded for Cox's proportional hazards model. Diagnosis was divided between DAT (0) and others (1), type of caregiver between spouse (0) and others (1), and destination between home (0) and others (1). Comorbidity, complications and care distress were divided between yes (1) and no (0). In each model, the correlation coefficients of each variable were under 0.6. We had a small number of missing data. However, there were no differences in variables between the whole sample and the one used in the Cox's proportional hazards models. Statistical analysis was carried out using spss version 17.0. We regarded P < 0.05 as a statistically significant level.

RESULTS

There was a clear separation between men and women in term of diagnosis, comorbidity, complications, caregiver and outcomes. However, these two groups did not differ in age, scores of HDS-R, ASSD and N-ADL, number of cohabitants, care distress and destination (Table 1). DAT was more common in women (179/262) than in men (65/128). The incidences of comorbidity and complications were significantly lower in women (67.2%, 41.2%) than in men (77.3%, 53.1%). Most men had been cared for by their wives (72/128), but many women had been cared for by daughters-in-law (99/262). Most women were institutionalized after discharge, and the ratio of institutionalization in women (111/262) was higher than that in men (37/128). Kaplan–Meier analysis showed a gender difference in the length of stay (Fig. 1). The median length of stay of 119 days (interquartile range 48–297 days) in women was longer than that of 83 days (interquartile range 35–199 days) in men (P= 0.007).

Table 1.  Clinical characteristics of 390 patients
CharacteristicsTotalMenWomenP
n= 390n= 128n= 262
mean (SD)mean (SD)mean (SD)
  1. ASSD, Assessment Scale for Symptoms of Dementia; DAT, dementia of Alzheimer's type; HDS-R, revised Hasegawa Dementia Scale; Home, Home care-giving; Hospital, transferred to another hospital; Institute, became institutionalized; N-ADL, Nishimura's activity of daily living scale; OTD, other type of dementia; VaD, vascular dementia.

Age (years)81.6 (7.6)80.9 (7.2)81.8 (7.8)0.269
DiagnosisDAT244651790.001
VaD1105258 
OTD361125 
Initial score of HDS-R10.0 (7.4)9.3 (7.3)10.3 (7.4)0.251
Initial score of ASSD36.0 (22.1)39.2 (23.6)34.5 (21.3)0.066
Initial score of N-ADL28.0 (11.5)28.4 (11.6)27.8 (11.4)0.675
Final score of HDS-R9.4 (8.0)8.8 (7.6)9.7 (8.2)0.335
Final score of ASSD38.9 (22.3)40.3 (21.4)38.3 (22.7)0.435
Final score of N-ADL25.5 (12.0)25.2 (12.1)25.7 (12.0)0.732
Comorbidity (%)70.577.367.20.044
Complication (%)45.153.141.20.03
CaregiverSpouse10172290
Son531340 
Daughter611150 
Daughter-in-law1242599 
Relatives31427 
Staff20317 
Number of cohabitants2.3 (1.9)2.4 (1.9)2.2 (1.9)0.268
Care distress (%)34.13235.10.313
DestinationHome12748790.259
Institution22264158 
Others1156 
Unknown301119 
OutcomeHome11544710
Institution14837111 
Hospital793841 
Changed ward21219 
Deceased1679 
Continued11011 
Length of stay (median, days)103831190.007
Figure 1.

Length of stay. Solid line shows the survival rate of men and dotted line shows that of women. The survival rate of men was different from that of women (P= 0.007).

In the univariate examination, patients with complications and younger women were found to have experienced an increased length of stay. High scores of final HDS-R and N-ADL in men and a high score of final HDS-R and a low score of final ASSD in women were shown to be associated with a decreased length of stay. Destinations changed the length of stay (Table 2).

Table 2.  Relationship between the length of stay and each variable
CharacteristicsTotalMenWomen
median (days)/ rPmedian (days)/rPmedian (days)/rP
  1. r, Correlation coefficient. The Kaplan–Meier method was used to compare times to discharge (log–rank test) and simple regression analysis was used to quantify the impact of each variable. ASSD, Assessment Scale for Symptoms of Dementia; DAT, dementia of Alzheimer's type; HDS-R, revised Hasegawa Dementia Scale; N-ADL, Nishimura's activity of daily living scale; OTD, other type of dementia; VaD, vascular dementia. Others included relatives and staff.

Age (years)−0.1280.012−0.0590.508−0.1610.009
DiagnosisDAT1110.2431010.5431260.495
VaD97 65 118 
OTD104 133 96 
Initial score of HDS-R−0.0710.176−0.1060.252−0.0680.285
Initial score of ASSD0.0170.7580.0160.8720.030.649
Initial score of N-ADL0.0450.381−0.0350.7090.0770.218
Final score of HDS-R−0.1610.002−0.2980.001−0.1260.047
Final score of ASSD0.1650.0020.160.1020.1720.008
Final score of N-ADL−0.1590.002−0.3060.001−0.1130.073
ComorbidityYes1020.966880.3441200.959
No107 78 114 
ComplicationYes1540.0011220.0061690.006
No83 58 92 
CaregiverSpouse970.811890.736970.477
Son107 107 100 
Daughter148 96 160 
Daughter-in-law126 70 145 
Others83 57 91 
Number of cohabitants−0.0780.125−0.040.653−0.0860.167
Care distressYes1070.2971010.3731160.565
No100 78 119 
DestinationHome480450580
Institution166 149 169 
Others99 71 344 
Unknown43 31 49 

In the Cox's proportional hazards model including men and women, age (B: 0.020, Wald: 7.203, P= 0.007, Exp (B): 1.020) and sex (B: –0.246, Wald: 4.273, P= 0.039, Exp (B): 0.782) were detected as predictors of length of stay in Model 1. Model 2 indicated that age (B: 0.025, Wald: 9.692, P= 0.002, Exp (B): 1.025), sex (B: −0.332, Wald: 7.226, P= 0.007, Exp (B): 0.717), complications (B: −0.358, Wald: 8.202, P= 0.004, Exp (B): 0.699), final score of ADL (B: 0.011, Wald: 4.509, P= 0.034, Exp (B): 1.011) and destination (B: −0.531, Wald: 17.737, P= 0.000, Exp (B): 0.588) were the predictors of length of stay. These results showed that sex was one of the important predictors of length of stay. Therefore, we evaluated the predictor of length of stay in each sex. According to Model 1, no predictors were detected in men. In contrast, Model 1 showed that age was a predictor of the length of stay (B: 0.019, Wald: 4.886 P= 0.027, Exp (B): 1.019) in women. Model 2 revealed that predictors of the length of stay were complications (B: −0.446, Wald: 4.340, P= 0.037, Exp (B): 0.640), and final scores of HDS-R (B: 0.040, Wald: 8.072, P= 0.004, Exp (B): 1.040) and N-ADL (B: 0.021, Wald: 4.958, P= 0.026, Exp (B): 1.021) in men and age (B: 0.024, Wald: 7.044, P= 0.008; Exp (B): 1.024), complications (B: −0.434, Wald: 9.525, P= 0.002, Exp (B): 0.648) and destination (B: −0.612, Wald: 16.772, P= 0.000, Exp (B): 0.542) in women (Table 3). The estimated median lengths of stay were the same as the observed.

Table 3.  Predictors of the length of stay
 BExp (B)95% Confidence intervalP
  • No predictor was detected. HDS-R, revised Hasegawa Dementia Scale; N-ADL, Nishimura's activity of daily living scale.

Total    
 (Model 1)    
  Age0.021.021.005–1.0350.007
  Sex−0.2460.7820.620–0.9870.039
 (Model 2)    
  Age0.0251.0251.009–1.0410.002
  Sex−0.3320.7170.593–0.9140.007
 Complication−0.3580.6990.547–0.8930.004
 Final score of ADL0.0111.0111.001–1.0210.034
 Destination−0.5310.5880.460–0.7530
Men    
 (Model 1)    
 (Model 2)    
 Complication−0.4460.640.421–0.9740.037
 Final score of HDS-R0.041.041.012–1.0690.004
 Final score of N-ADL0.0211.0211.002–1.0390.026
Women    
 (Model 1)    
  Age0.0191.0191.002–1.0360.027
 (Model 2)    
  Age0.0241.0241.006–1.0430.008
 Complication−0.4340.6480.492–0.8350.002
 Destination−0.6120.5420.405–0.7270

DISCUSSION

In Japan, the traditional family system has changed dramatically and most elderly now live alone or with their spouses. And it can be difficult to provide care for the elderly if they need help with their daily life. A long-term care insurance system started in April 2000. An increase in the number of demented elderly is also a great problem, and is associated with significant stress among their caregivers. Most demented elderly are cared for by their relatives, especially spouses, children and daughters-in-law. Caregivers can be troubled by the demented elderly who often exhibit disquiet, wandering or violence. Some caregivers continue to care beyond their limits. Demented patients who exhibit severe violence and wandering are treated in wards for the demented elderly. After treatment, some patients leave the ward and are again cared for by their relatives. Other patients cannot be cared for by their caregivers and are institutionalized or transferred to another hospital.

The results of our previous studies1,2 showed gender differences in the care situation of the demented elderly. Most women are particularly likely to live alone. There were also gender differences among predictors of outcomes. The differences in outcomes between the hospital group and the others were mainly as a result of complications. In contrast, the difference between the group that returned home for care and the institutionalized group was as a result of N-ADL in men, and the number of cohabitants, caregiver and HDS-R in women. We have now investigated the gender differences and predictors of length of stay in a ward.

There have been many studies6–14 that evaluated the predictors of outcome for patients in acute medical wards. Campbell et al.6 carried out a systematic review to identify factors that had a significant influence on outcome in the older patients admitted to hospital for medical reasons. Functional status score, illness severity, cognitive score, poor nutrition, comorbidity score, diagnosis or presented illness, polypharmacy, age and gender were statistically significant predictors of length of stay. The main objective of the study by Campbell et al. was to evaluate previous studies, which investigated elements of case-mix and outcome assessment in older people admitted from the community to an acute care setting. However, they did not report on social problems.

Bertozzi et al.15 reported that ‘frail’ elderly patients with an increasing burden of comorbidity are at greater risk of prolonged hospitalization and that social support was not related to the length of stay. They described that health-related factors were the main determinants of prolonged stay at a geriatric evaluation and rehabilitation unit; however, they could not rule out the possibility that delayed discharge was a result of social problems possibly caused by relatively poor health. They could not find a gender difference at the geriatric evaluation and rehabilitation unit.

In Japan, Hakoda et al.16 clarified the effects of social resources in the community on the discharge of patients from psychiatric hospital. The length of stay in hospital increased with the rate of inpatients with schizophrenia and mental retardation, inpatients over 65 years old, male inpatients, and the number of inpatients per medical staff. Takeshima et al.17 investigated macroscopic indices to evaluate psychiatric hospital function. They reported that 70.4% of admitted patients were discharged to communities within a year and 50% of admitted patients were discharged within 74.2 days. Fujita and Takeshima18 investigated the discharge curve among psychiatric patients after admission and the risk factors associated with a long-term stay. They reported that patients with dementia tended to stay in hospital for a long period and that the proportion still hospitalized after one year was 27.0%. They reported that dementia was associated with an increased risk of remaining hospitalized after the first year and also that risk factors associated with an increased risk of remaining hospitalized after the first year included a long length of continuous hospitalization, diagnoses of dementia, mental retardation, schizophrenia, male, older age and being in a mental hospital.

The results of the present study showed that there were gender differences in the length of stay and in predictors of the length of stay. The median length of stay was 83 days in men and 119 days in women. The present study also suggested that it was difficult to predict the length of stay on admission, because age was the only predictor of the length of stay in women. In men, no predictor was detected in the present study. Retrospectively, complications and the final HDS-R and N-ADL scores were predictors of the length of stay in men. Age, complications and destinations were predictors of the length of stay in women. The caregiving system of men was simple. Most men were cared for by their spouses and lived with their spouse.1,2 In these caregiving systems, it was supposed that the length of stay was determined by physical and psychological conditions, not by social variables in men. Demented women were cared for by various caregivers. Their caregiving systems were more complex than those of men. It was assumed that these conditions meant that the caregivers' wish to give care at home reduced the length of stay. We could not understand why age was a predictor of the length of stay in women and not in men. It was probably caused by the difference of distribution of age. The standard deviation of age of men was shorter than that of women.

In conclusion, it was observed that there were gender 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, not by the social variables in men. In women, it was supposed that the caregiver's wish reduced the length of stay. Besides, complication was a common predictor of the extension of stay in each sex. We have to decrease the number of complications as much as possible to reduce the length of stay.

The selection of the sample in the present study represents a limitation of the prediction of factors affecting the length of stay, because we did not include economic variables and other factors that could influence the length of stay in the analysis. Further analysis is necessary to investigate the factors that influence the length of stay.

Ancillary