33. Caregiver pre‐bereavement psychological well‐being: home palliative care versus usual care.

Study	Measure	Analysis	Follow‐up	Significance and direction	Details
Jordhøy 2000 Norway (high quality)   and   Ahlner‐Elmqvist 2008 Sweden	SF‐36 2 subscales: role limitation due to emotional problems (3 items) and mental health (5 items) transformed scores: from 0 to 100; higher scores equal better health; caregiver report	Backwards from death	44 to 13 months	Increased difference over time in scores for the 2 subscales favours intervention   Role limitation due to emotional problems group effect: n.s. b ‐6.67 SE 5.36 P value ≥ 0.10   Group by time interaction: favours intervention P value < 0.05   Mental health Group effect: n.s. b ‐5.09 SE 3.02 P value ≥ 0.10   Group by time interaction: favours intervention P value < 0.05	Pooled analysis of data from the 2 studies (n = 517); longitudinal analysis of trajectories included pre‐ and post‐bereavement measurements   Findings showed a strong decline in scores before the death of the patient, and rising scores thereafter, especially for the intervention group   b values and SEs for each time interval (in group by time analysis) and graphs with descriptive data on trajectories in scores over time provided in Ringdal 2004 (Jordhøy 2000 and Ahlner‐Elmqvist 2008)
			12 to 6 months
			5 to 0 months
		Forwards from death	1 to 2 months
			3 to 5 months
			6 to 12 months
			13 to 17 months
McWhinney 1994 US	Center for Epidemiological Studies Depression (CES‐D) scale score: range and interpretation not stated; caregiver report	Forwards from enrolment	1 month	n.s.	"There were no clinically or statistically significant differences between the experimental and control groups on any of the measures at one month" (McWhinney 1994); no data provided   High attrition (53/146) mainly due to death; 2 month data not analysed due to further attrition
Hughes 1992 US	Philadelphia Geriatric Center Morale Scale (short‐version) score: range not stated; higher scores equal lower morale; caregiver report	Forwards from enrolment	1 month	n.s. Beta ‐0.02 t 0.57 P value = 0.57	ANCOVA (adjusted for age, race, education, relationship to patient and morale baseline score) Intervention (n = 32): M 1.75 (SD 21) Control (n = 27): M 1.58 (SD 0.32)
			6 months	Favours control Beta ‐0.12 t ‐2.30 P value = 0.03	ANCOVA (adjusted for age, race, education, relationship to patient and morale baseline score); controlling for survival days did not alter findings Intervention (n = 32): M 1.67 (SD 25) Control (n = 27): M 1.62 (SD 0.34)
Buckingham 1978 US	1) Symptom checklist‐90 (SCL‐90) measure of psychological problems; score range and interpretation not stated; caregiver report  2) Zuckerman and Lubin adjective checklist for the measurement of depression scores: range and interpretation not stated; caregiver report	Not stated	Not stated	Authors stated differences favouring intervention but statistical significance not stated	"Hospice patients and their primary care persons exhibited (…) lower levels of anxiety and depression than non‐hospice patients and their primary care persons" (Buckingham 1978); no data provided to support this statement
Greer 1986 (CBA)	1) Profile of Mood States (POMS) score: from 0 to 5; lower scores equal greater psychological distress; caregiver report   2) use of medication for anxiety or depression (yes/no; caregiver report)   3) report of increased drinking (yes/no; caregiver report)	Not stated	Not stated	POMS n.s.   Use of medication for anxiety or depression n.s.   Increased drinking n.s. "There were no significant differences among the three PCP groups on a modified mood state scale measuring anxiety and depression. Similarly, no significant differences among the groups were observed in use of medications for anxiety or depression, and there was no reported increased use of alcohol" (Greer 1986)	Adjusted estimatesa POMS Community‐based intervention: M 2.96 (SE 0.06) Hospital‐based intervention: M 2.98 (SE 0.08) Control (conventional care): M 3.01 (SE 0.13) Use of medication for anxiety or depression Community‐based intervention: 11% (SE 0.02) Hospital‐based intervention: 10% (SE 0.02) Control (conventional care): 14% (SE 0.04) Increased drinking Community‐based intervention: 4% (SE 0.01) Hospital‐based intervention: 5% (SE 0.01) Control (conventional care): 6% (SE 0.02)

ANCOVA: analysis of covariance; b: metric regression coefficient; CBA: controlled before and after study; M: mean; n.s.: non‐significant; PCP: primary care person; SD: standard deviation; SE: standard error.

^aEstimates adjusted for sample differences; standard errors based on the linear regression equation for continuous variables and on logistic regression equation for dichotomous variables.