Table 3

Factors associated with (1) hospitalisation of nursing home residents in last month of life and (2) in-hospital death

	Hospitalisation*		In-hospital death*
	Exp(Coefficient)†; (95% CI)	P value for factor	Exp(Coefficient)†; (95% CI)	P value for factor
Country, resident and nursing home characteristics (independent variables)
Intercept‡ Country §	8.02 (1.90 to 33.8)	0.005	1.37 (0.19 to 9.82)	0.75
Belgium (reference)	Ref		Ref	Ref
Finland	0.51 (0.24 to 1.11)	0.09	0.35 (0.11 to 1.23)	0.08
Italy	0.36 (0.16 to 0.85)	0.02	1.55 (0.50 to 4.81)	0.45
The Netherlands	0.27 (0.12 to 0.62)	0.002	0.33 (0.10 to 1.17)	0.09
Poland	0.39 (0.16 to 0.97)	0.04	1.33 (0.40 to 4.43)	0.64
Clinical characteristics
Dementia at time of death¶	0.86 (0.55 to 1.37)	0.53	0.78 (0.43 to 1.41)	0.41
Functional/cognitive status 1 month before death (BANS-S) ¶	0.90 (0.86 to 0.95)	<0.001	0.89 (0.84 to 0.95)	<0.001
Nursing home stay characteristics
Type of nursing home§
Type 1 (reference)	Ref	Ref	Ref	Ref
Type 2	0.91 (0.47 to 1.75)	0.77	1.90 (0.73 to 4.94)	0.19
Nursing home size§	1.00 (0.99 to 1.00)	0.23	0.99 (0.99 to 1.00)	0.10
Opioids available 24/7 for residents with PC needs§
Yes, for all residents	Ref	Ref	Ref	Ref
Yes, for most residents	2.73 (1.25 to 5.95)	0.01	1.85 (0.49 to 6.92)	0.36
Yes, for some residents	1.84 (0.84 to 3.99)	0.13	1.30 (0.44 to 3.86)	0.64
No, never	1.01 (0.43 to 2.38)	0.98	0.40 (0.10 to 1.55)	0.18
Care process characteristics
Conversation between nurse and relative about preferred course of care was held¶	0.47 (0.31 to 0.72)	<0.001	0.54 (0.31 to 0.95)	0.03
Resident had a written advance directive with regard to ‘do not transfer to a hospital’¶	0.17 (0.06 to 0.50)	0.001	0.40 (0.12 to 1.34)	0.14
Contact person for residents/relatives to ensure coordinated health and social care §
Yes, for all residents	Ref	Ref	Ref	Ref
Yes, for most residents	0.87 (0.29 to 2.62)	0.80	0.73 (0.12 to 4.64)	0.74
Yes, for some residents	0.43 (0.13 to 1.39)	0.16	0.94 (0.22 to 4.00)	0.93
No, never	1.24 (0.73 to 2.11)	0.42	0.90 (0.40 to 2.00)	0.74
Physician’s level of expectation about death**
Death was expected	Ref	Ref	Ref	Ref
Death was expected, yet sooner than anticipated	1.18 (0.73 to 1.93)	0.50	2.82 (1.48 to 5.39)	0.002
Death was neither expected nor unexpected	1.04 (0.53 to 2.03)	0.91	3.90 (1.78 to 8.55)	0.001
Death was unexpected	0.45 (0.19 to 1.04)	0.06	5.21 (2.22 to 12.23)	<0.001

We were able to include 790 residents for analysis of hospitalisation and 814 residents for analysis of in-hospital death in the complete-case multivariate analysis.
Due to low numbers of complete cases in England, we excluded data from England from the multivariate analyses. Therefore, also type 3 nursing homes were not included.
Data in bold indicate statistically significant p-values.
Multivariate logistic mixed-effect regression models reporting p value with hospitalisation in the last month of life and place of death as independent variables and nursing home as random factor, α<0.05.
*Hospitalisation ‘no’ and in-hospital death ‘no’ are reference categories.
†Exponentiation of the coefficient, which is an OR.
‡Expected mean value of the dependent variables (hospitalisation and in-hospital death) when all factors equal zero.
§Reported by administrator/manager of nursing home. For 44 out of 1384 residents, no questionnaire was returned by the nursing home administrator.
¶Reported by the nurse/care assistant most involved in care.
**Reported by the physician. For 397/1384 residents, no questionnaire was returned by physician.
BANS-S, Bedford Alzheimer Nursing Severity Scale; PC, palliative care.