Introduction

Hip fracture is a major injury among older adults that can lead to substantial loss of healthy life-years [1]. Older adults with dementia are also more likely to incur hip fractures than those without dementia [2–4]. Due to the rapidly increasing prevalence of dementia in the general population [5], its occurrence among patients with hip fractures has also increased in acute care settings. Systematic reviews suggested that about 20–50% of the individuals admitted to acute care hospitals due to hip fractures have dementia or a cognitive impairment [6, 7]. Among people who underwent hip surgery, retrospective cohort studies investigated the association between dementia and patient outcomes, in Australia, Canada, the Netherlands, and Japan [8–12]. Compared to those without dementia, people with dementia were more likely to have a higher delirium rate [8], more post-surgical complications [12], be admitted to long-term care [13], and be readmitted within 30 days after discharge [9].

However, reports on the association between dementia and the length of hospital stay after hip surgery are inconsistent. A shorter hospital stay among people with dementia than in those without dementia has been indicated [8, 10]; however, this association varied depending on patients’ social factors such as type of residence before the hip fracture—a longer hospital stay was reported for community-dwelling people than for long-term care facility users [13]. A better understanding of the association between dementia and patient outcomes after hip surgery, considering residence type before and after hospitalization, will help promote integrated care for people with dementia in aging societies.

Moreover, providing adequate care for individuals with dementia in hospital settings has been challenging [14, 15]. Nursing care environment factors such as nurse staffing level, skill mix, and advanced skill for dementia and gerontological nursing care are essential for providing adequate care to people with dementia [16]. However, few studies have considered the effect of the nursing care environment when investigating the association between dementia and patient outcomes [14, 17]. We, therefore, aimed to investigate the associations between dementia, adverse events, and length of hospital stay among older adults who underwent hip surgery considering the participants’ social and nursing care environment.

Methods

Results

We obtained data from 48,049 eligible patients from 404 hospitals (S1 Fig). Of them, 20,638 (43.0%) had dementia (Table 1). The dementia group had more women, long-term care facility users before admission, and patients with higher CCI scores (Table 1) compared to those in the without dementia group.

Table 1

Characteristics of the study population with and without dementia.

	Total		With dementia		Without dementia		P
	N = 48,797		n = 20,638		n = 28,159
	n/mean	%/SD	n/mean	%/SD	n/mean	%/SD
Female, n, %	38,830	79.6	16,648	80.7	22,182	78.8	<0.001
Age, years, mean, SD	82.2	8.3	86.1	6.7	79.3	8.1	<0.001
Body mass index, mean, SD	21.2	3.8	20.3	3.5	21.9	3.9	<0.001
Charlson comorbidity index, n, %							<0.001
0	22,493	46.1	6,801	33.0	15,692	55.7
1	15,289	31.3	8,006	38.8	7,283	25.9
2	7,122	14.6	3,939	19.1	3,183	11.3
≥ 3	3,893	8.0	1,892	9.2	2,001	7.1
Place of residence before admission, n, %							<0.001
Home	37,619	77.1	12,415	60.2	25,204	89.5
Long-term care facility	7,654	15.7	6,323	30.6	1,331	4.7
Other (hospital, clinic, etc.)	3,524	7.2	1,900	9.2	1,624	5.8
Place of residence after discharge, n, %							<0.001
Home	15,839	32.5	3,700	17.9	12,139	43.1
Long-term care facility	6,741	13.8	5,572	27.0	1,169	4.2
Other (hospital, clinic, etc.)	26,217	53.7	11,366	55.1	14,851	52.7
Psychotropic drug prescription, n, %	32,207	66.0	14,776	71.6	17,431	61.9	<0.001
Type of surgery, n, %
Osteosynthesis	26,327	54.0	13,763	66.7	12,564	44.6	<0.001
Bipolar hip arthroplasty	13,865	28.4	6,397	31.0	7,468	26.5	<0.001
Total hip arthroplasty	8,743	17.9	540	2.6	8,203	29.1	<0.001

SD, standard deviation. Chi-square test or t-test.

Table 2 shows the characteristics of all the 404 hospitals. Most were public and had no special care for dementia. Moreover, most of the nursing staff comprised nurses and, in average, 6.3 patients were allocated per nurse and associate nurse.

Table 2

Hospital characteristics (N = 404).

	Mean/%	SD
Number of beds	439.0	167.9
Number of patients per nurse and associate nurse	6.3	0.8
Percentage of nurses among all nursing staff	83.1	6.0
	n	%
Types of established organization of hospitals
National	41	10.2
Public	168	41.6
Social	20	5.0
Private	73	18.1
Othersa	102	25.3
Addition to special care for dementia in the fee schedule
None	205	50.7
Type 2 (trained nurses)	99	24.5
Type 1 (multidisciplinary dementia care team)	100	24.8

SD, standard deviation.

^a public interest corporations, private university corporations, social welfare corporations, medical co-op, companies, and other corporations.

The overall number (proportions) of in-hospital deaths, in-hospital pneumonia, and in-hospital fractures were 748 (1.53%), 51 (0.10%), and 1,634 (3.35%), respectively. The prevalence of in-hospital death, in-hospital pneumonia, and in-hospital fracture for people with and without dementia were 2.11% vs. 1.11% (p < .001), 0.15% vs. 0.07% (p = .007) and 3.76% vs. 3.05% (p < .001), respectively (Table 3). The medians (inter quartile range) length of hospital stay in the total sample and in people with and without dementia were 26 (19–38), 26 (19–39) and 25 (19–37) days, respectively (p = .758).

Table 3

Length of hospital stay and prevalence of outcomes in people with and without dementia.

	With dementia	Without dementia	P
	n = 20,638*	n = 28,159*
In-hospital death, n (%)	436 (2.11)	312 (1.11)	< .001a)
In-hospital pneumonia, n (%)	31 (0.15)	20 (0.07)	.007a)
In-hospital fracture, n (%)	776 (3.76)	858 (3.05)	< .001a)
Length of hospital stay (days), median (IQR)	26 (19–39)	25 (19–37)	.758b)

*For the length of hospital stay, 20,202 patients with dementia and 27,847 patients without dementia were included after excluding in-hospital death; IQR, inter quartile range

a), chi-square test

b), Mann-Whitney U test.

The adjusted ORs (95% CI) of dementia for in-hospital death, in-hospital pneumonia, and in-hospital fracture were 1.12 (0.95–1.33), 0.95 (0.51–1.80), and 1.08 (0.92–1.25), respectively (Table 4 and S1 Table).

Table 4

Results of multilevel logistic regression analysis for in-hospital death, in-hospital pneumonia, and in-hospital fracture among people with and without dementia (n = 48,797).

	Univariate regression				Multivariate regression
	OR	95% CI		P	Adjusted OR	95% CI		P
In-hospital death	1.88	1.62	2.19	<0.001	1.12	0.95	1.33	0.181
In-hospital pneumonia	1.97	1.11	3.49	0.021	0.95	0.51	1.80	0.885
In-hospital fracture	1.37	1.20	1.57	<0.001	1.08	0.92	1.25	0.344

Adjusted for sex, body mass index, Charlson comorbidity index, type of surgery, psychotropic drug use, residence of before admission, number of in-patients per nurse and associate nurse, percentage of nurses among all nursing staff, addition to special care for dementia in the fee schedule, types of established organization of hospitals and number of hospital beds.

CI, confidence interval; OR, odds ratio.

Table 5 and S2 Table show that dementia was not significantly associated with the length of hospital stay after adjusting for individual and hospital characteristics (-0.7%, 95% confidence interval (CI) = -1.6–0.3%). Shorter hospital stays were associated with individuals’ place of residence before admission and after discharge compared with from/to home, admission from long-term care facilities (-20.4%, 95% CI = -21.5–-19.3%) and other places including hospitals (-10.6%, 95% CI = -12.0–-9.2%), and discharge to long-term care facilities (-9.8%, 95% CI = -11.3–-8.4%) and to other institutions including hospitals (-10.7%, 95% CI = -11.6–-9.8%). Regarding hospital nursing care environment, a higher number of in-patients per nurse and associate nurse was significantly associated with 7.8% (95% CI = 4.2–11.5%) longer length of hospital stay.

Table 5

Results of multilevel analyses for length of hospital stay (n = 44,883).

	Coefficient	95% CI		P
With dementia (ref. without dementia)	-0.007	-0.016	0.003	0.184
Individual characteristics
The type of residence before admission (ref. home)
Long-term care facilities	-0.228	-0.242	-0.214	<0.001
Other (hospital, clinic, etc.)	-0.112	-0.128	-0.096	<0.001
The type of residence after discharge (ref. home)
Long-term care facilities	-0.103	-0.119	-0.087	<0.001
Other (hospital, clinic, etc.)	-0.113	-0.124	-0.103	<0.001
Hospital characteristics
Number of in-patients per nurse and associate nurse	0.075	0.042	0.109	<0.001
Percentage of nurses among all nursing staff	0.001	-0.005	0.005	0.937
Addition to special care for dementia in the fee schedule (ref. none)
Type 2 (trained nurses)	0.045	-0.020	0.111	0.175
Type 1 (multidisciplinary dementia care team)	-0.020	-0.085	0.045	0.542

Adjusted for sex, body mass index, Charlson comorbidity index, type of surgery, psychotropic drug use, types of established organization of hospitals and number of hospital beds. The exact percentage change, %Δy = 100 * [exp (bjΔXj)−1] gives a more accurate prediction of the change in length of hospital stay. CI, confidence interval.

Discussion

We conducted a retrospective observational study with a large sample using nationwide administrative data in Japan to explore the association between dementia and patient outcomes after hip surgery among older adults. Although in-hospital death, in-hospital pneumonia, and in-hospital fracture were more likely to occur in people with dementia than in those without dementia in the present study, they were not significantly associated with dementia after adjusting for potential confounders such as individual social factors, nurse staffing level, and dementia care expertise. The improvement of dementia care in acute care settings could explain our findings. A previous Japanese study using 2007–2010 data from the same database utilized in our research revealed that people with dementia are more likely to have post-surgical complications during hospitalization [12]. This discrepancy might be explained by the promotion of policies for dementia care in hospitals. In 2015, the Japanese government renewed its comprehensive strategy for the promotion of dementia care—the New Orange Plan—which aimed to create a friendly community for older adults with dementia [32]. This national strategy established the critical goal of providing appropriate medical care for patients with dementia in acute care facilities. Following this, financial incentives for appropriate medical care for patients with dementia were introduced in April 2016, which are included in a revision concerning the reimbursement of medical fees [29]. Such policy changes in dementia care may have reduced the disadvantages faced by people with dementia in acute care facilities; however, further studies are necessary to investigate the effect of policy changes on health outcomes of the patients.

In addition, there is another reason why dementia was not associated with longer hospital stays in the present study. Participants’ social factors may be key determinants of length of hospital stay in Japan rather than the dementia status. In the present study, people admitted from home or discharged to home were more likely to have longer hospitalizations than those admitted from and discharged to long-term care facilities and other hospitals, regardless of dementia status. Past research in other countries also reported that patients often quickly return to their previous nursing home and therefore have a shorter length of stay compared to those admitted from their own homes [8, 10].

In addition, Japan is well known for recording prolonged hospital stays than other countries in the Organization for Economic Co-operation and Development (OECD) [33]. The median length of hospital stay for people with and without dementia in the present study was 26 days; almost the same as the average length of hospital stay in hip fracture cases overall in acute care hospitals in Japan [34] but longer than those in Canada (no dementia = 17.5 days, dementia = 24.0 days [13]) or Australia (dementia < 15 days [10]). Currently, the government has promoted the transition of the elderly from hospitalization to welfare homes, which is expected to contribute to reduced length of hospital stay. Fee-schedule introduced financial incentive to encourage hospital discharges and use post-discharge protocols, fostering co-operation among hospitals, clinics, and long-term-care facilities, especially for patients with hip fractures [35]. In this context, the places of residence before admission and after discharge are key determinants on the decision of time of discharge for elderly people after hip surgery. People with dementia who were admitted from home need more time to prepare the adequate post-hospitalization residence compared to those admitted from long-term care facilities [13]. Regarding hospital discharge, individuals going to another hospital and/or long-term care facility transition with no delays and seamlessly compared to those being discharged to their own homes, given that existing post-discharge protocols between acute and post-acute hospitals or long-term-care facilities work well [35].

Our study also contributed new findings concerning the factors related to older adult in-patients’ health outcomes after hip surgery. Remarkably, our study suggested that not only individual characteristics but also the nursing staffing level is related to these outcomes. The increase of one in-patient per nursing staff was associated with longer length of hospital stay, in other words, increased nursing workload might disturb early discharge. This finding supported the results of Kane’s meta-analysis, which concluded that an increase by one registered nurse per patient day was associated with a 31% shorter length of hospital stay in surgical patients (OR, 0.69; 95% CI, 0.55–0.86 [27]).

This study had several limitations. First, the data were obtained from relatively large hospitals in Japan. We selected hospitals that had an average of > 200 in-patients per day in order to adjust for the effect of surgeon procedure volumes, which are related to patient outcomes [36]. Our findings might thus not be generalizable. Second, there may be some unmeasured confounders that might affect patient outcomes (e.g., whether it was a scheduled or emergency surgery and the severity of dementia). Patients with severe dementia are more likely to choose non-operative treatment for hip fractures [37]. In this study, we focused on cases with hip surgery, which might have involved those with relatively mild dementia. The ADL of patients were also related to the length of hospital stay [8, 38]. Although our dataset included ADL, some ADL data were missing and thus had to be excluded from the analysis. Third, for the database used in this study, the validation study reported that the sensitivity of diagnoses was low and varied across conditions [22]. The prevalence of non-critical adverse events, such as mild pneumonia, might have been underestimated. This, however, would not affect our findings since underestimations could occur with or without dementia.

Conclusion

Adverse events were more likely to occur in people aged 65 or older with dementia than in those without dementia after hip surgery. These events were not significantly associated with dementia after adjusting for individual social and nursing care environment. Further studies are necessary to identify factors that mitigate the effect of dementia on poor outcomes.