QUALITY ASSESSMENT SCALE

QUALITY ASSESSMENT SCALE [posted as supplied by author]

The scale is adapted from the NHS Centre for Reviews and Dissemination guidelines for quality assessment of observational studies. It is divided into three parts, each with several items. A score is given in each item. These are summed to calculate an overall total for the scale (maximum of 24 points). The final score is converted in a proportion for an easier understanding of the quality score.

PART 1: Study Design

�	High quality (1.5)	Fair quality (1)	Poor quality (0.5)	Unknown (0)
Study design	Prospective	Cross-sectional	Retrospective	�
Inclusion and exclusion criteria	Detailed description of both & appropriate	Sufficient description & appropriate	Basic and insufficient information	�
Baseline characteristics	No imp. confounders	One important confounder	More than one imp. confounder	�
Generalisability of sample	Randomised & diverse sample	Some criteria that may indicate selectivity	Highly selected sample (place of care, wishes)	�
Sample size	More than 200	100-200	Less than 100	�
Assessment of place of death	Death certificates or files with clear and standard assessment	Files or reliable accounts (proximity of respondent, time distance)	Accounts poorly reliable	�
Assessment of the factor(s)	Standardized & appropriated for the variable(s)	Reliable accounts (proximity of respondent, time distance)	Non systematic, unreliable or inappropriate	�

Appropriateness of study design

High appropriateness

(1.5)

Appropriate (1)

Poor appropriateness

(0.5)

Not appropriate (0)

Can the study be generalized, can it offer a reliable account of reality and is it appropriate to the issue, having in mind ethical issues?

PART 2: Conduct of the Study

�

High quality (1.5)

Fair quality (1)

Poor quality (0.5)

Unknown (0)

Response rate *

80% or more

60-79

Less than 60

�

Reasons for non-response/

drop-out/missing data

Low & controlled or incontrollable

Acceptable & if reasons are not imp for results

High and reasons with predicted impact on results

�

*Flexible criteria

PART 3: Analysis

Description

�

Well reported (1)

Sufficiently reported (0.5)

Poorly reported or unknown (0)

Analysis procedures/

statistical tests

Tests used & significance criteria presented

General analysis & significance criteria (even if not presented, easily inferred)

�

Results

P values or CIs

One of the two

�

Quality

�	High quality (2)	Fair quality (1)	Poor quality (0)
Analysis procedures/statistical tests	Multivariate	Bivariate	�
Results	No errors in presentation and interpretation	Few minor errors	Important errors

Appropriateness

High appropriateness

(3)

Appropriate (2)

Poor appropriateness

(1)

Not appropriate (0)

Was the analysis appropriate, rigorously applied and correctly interpreted? Does it offer a reliable account of reality?

Table A Electronic databases and journals handsearched [posted as supplied by author]

Electronic searches			Journals handsearched
Databases	Keywords	MESH headings	Journals handsearched
MEDLINE (on OVID, 1966-3^rd Sep2004)	1. death adj2 (place or site or location or home$ or hospital$ or hospice$).mp. 2. dying adj2 (place or site or location or home$ or hospital$ or hospice$).mp. 3. 1 or 2 4. (factor$ or determinant$ or predict$ or affect$ or influence$ or associat$ or difference$ or relat$) adj3 3 5. terminal$ or palliative or end of life or advanced disease or hospice$ 6. 5 or (subject headings) 7. 4 and 6	Palliative care Terminal care Hospice care	Palliative Medicine (2003/4) Journal of Palliative Care (2003/4) Journal of Pain and Symptom Management (2003/4) Journal of Palliative Medicine (1998 to 2000; 2003/4) Supportive Care in Cancer (2003/4) BMC Palliative Care (2002 to 2004) -available online
PsycINFO (on OVID, 1972-3^rd Sep2004)		Palliative care Terminal cancer Terminally ill patients Hospice Death and Dying
CINAHL (on OVID, 1982-3^rd Sep2004)		Palliative care Terminal care Hospice and palliative nursing Hospices
ASSIA (1987-3^rd Sep2004)	*1. death within 2 (place or site or location or home or hospital* or hospice)* *2. dying within 2 (place or site or location or home or hospital* or hospice))* 3. 1 or 2 *4. (factor or determinant* or predict* or affect* or influence* or associat* or difference* or relat)* 5. 3 and 4 *6. terminal or palliative or end of life or advanced disease or hospice* or (subject headings) 7. 5 and 6**	Palliative care Terminal care Hospices

The journals selected for handsearching were the most frequent sources of studies included in the review or were suggested for further consultation by a library expert. Journal of Palliative Medicine was additionally handsearched for 1998-2000 as it is not indexed during this period in any of the used databases.

�

Table B High quality studies: multivariate analysis and quality score of 70% or greater (n studies=25; n patients=1,413,612) [posted as supplied by author]

Authors and reference

Year of publication (study period)

Country

Study design

Setting

Sample size

Data collection methods

Categories of

place of death

Analysis

Home death

Quality score

Burge et al. ^(w1)

2003

(1992-97)

Canada,

Nova Scotia

Retrospective cohort

General population

13936

2 health databases, including death certificate data

Hospital vs. out of hospital

Multivariate logistic regression

88%

Hunt et al. ^(w2)

1993

(1990)

South Australia

Retrospective cohort

General population

2715

Central cancer registry and death certificate data

Metropolitan public hospital, vs. home (among others)

Unconditional logistic regression

14%

88%

Jordhoy et al. ^(w17) *

Ahlner-Elmqvist et al. ^(w16)

2003

2000

(1995-97)

Norway, Trondheim

RCT

RCT: palliative vs conventional care

395

Clinical records

Home vs. elsewhere

Multivariate regression

20%

88%

77%

Bruera et al. ^(w3)

2003

(1996/98)

US,

Houston

Retrospective cohort

General population

11694

Death certificates

Home vs. hospital

Multivariate logistic regression

35%

85%

Costantini et al. ^(w4)

1993

(1986-90)

Italy,

Genoa

Retrospective cohort

General population

12343

Death certificates

Home vs. hospital

Multivariate logistic regression

28-33%

85%

Costantini et al. ^(w5)

2000

(1991)

Italy,

Liguria

and Toscana

Cross-sectional, with ecological analysis

General population

17597

Death certificates

Home vs. hospital

Multivariate logistic regression

31-73%

83%

Mor and Hiris ^(w6)

1983

(1981-82)

Cross-sectional

Hospice (home and inpatient)

3257

National Hospice Study: clinical records

Home vs. medical setting

Discriminant function analysis

48%

83%

Burge et al. ^(w7)

2003

(1992-97)

Canada,

Nova Scotia

Retrospective cohort

Patients with a minimum of 3 home visits

9714

4 health databases, including death certificate data

Hospital vs. out of hospital

Multivariate logistic analysis

81%

Cantwell et al. ^(w8)

2000

(10 months)

Canada, Edmonton

Cross-sectional

Palliative home care

Questionnaires

Home vs. hospital or hospice

Logistic regression

47%

81%

Gallo et al. ^(w9)

2001

(1994)

US,

Connecticut

Retrospective cohort

General population

4766

Tumour registry

Home vs. institutional

Multivariate relative risk regression

29%

81%

Higginson et al. ^(w10)*

Higginson et al. ^(w11)*

1999

1998

(1985-94)

UK, England

Ecological cross-sectional

General population

1.3 million

Death certificates

Home vs. elsewhere

Multiple regression analysis

27%

(5-75%)

81%

63%

Moinpour and Polissar ^(w12)

1989

(1980-85)

US,

Washington

Retrospective cohort

General population

26500

Cancer surveillance system and clinical records

Home vs. elsewhere

Logistic regression

33%

81%

Higginson and Costantini ^(w13)

2002

(6 months)

UK, Italy, Ireland

Prospective cohort

Palliative care teams (mainly community)

1326

Clinical records

Home vs. elsewhere

Multivariate logistic regression

45-72%

79%

Karlsen and Addington-Hall ^(w14)

1998

(1995-96)

UK,

London

Retrospective cohort survey

General population

229

Death certificates and questionnaires

Home vs. elsewhere

Logistic regression

21%

79%

McCusker ^(w15)

1983

(1976-78)

US,

New York

Retrospective cohort

General population

2989

Tumour registry, including death certificate data

Home vs. acute hospital or chronic care facility

Logit model

18%

79%

Bruera et al. ^(w18)

2002

(1997-98)

US,

Texas

Retrospective cohort

Cancer centre

1305

Tumour registry and death certificates

Hospital vs. home

Multivariate logistic regression

34%

77%

Hunt and McCaul ^(w19)

1996

(1990)

South Australia

Retrospective cohort

Hospice and palliative care services

2207

Cancer registry, death certificates

Metropolitan public hospital vs. private hospital, country hospital, nursing home or home

Polytomous logistic regression

14%

77%

Tang ^(w20)

2002

(1991-97)

Taiwan

Cross-sectional

Home care programme

264

Clinical records

Home vs. hospital

Logistic regression

69%

77%

Izquierdo-Porrera et al. ^(w21)

2001

(1995-97)

Spain,

Catalonia

Cross-sectional

Palliative care unit

199

Clinical records

General hospital vs. elsewhere

Multivariate Logistic regression

36%

75%

De Conno et al. ^(w22)

1996

(1989-91)

Italy,

Milan

Cross-sectional

Home care programme

348

Clinical records, routine interviews

Hospital vs. elsewhere

Multivariate Log-linear model

86%

73%

Edmonds et al. ^(w23)

2000

(1995)

Ireland

Cross-sectional

6 home care teams

508

Clinical records

Hospital vs. elsewhere

Logistic regression analysis

75%

73%

Fukui et al. ^(w24)

2004

(2002)

Japan

Retrospective cohort survey

241 home palliative care agencies

428

Questionnaires

Home vs. hospital

Multiple logistic regression

65%

73%

Bass et al. ^(w25)*

Bass et al. ^(w26)*

1984

1985

(1980-81)

US,

Ohio

Cross-sectional

Home hospice programme

146

Clinical records and interviews

Home vs. elsewhere

Regression analysis

57%

71%

Brazil et al. ^(w27)

2002

(1997-99)

Canada,

Ontario

Retrospective cohort survey

Home palliative care

151

Interviews

Home vs. elsewhere

Multivariate logistic regression

63%

71%

Fukui et al. ^(w28)

2003

(2001)

Japan

Retrospective cohort survey

259 home palliative care agencies

528

Questionnaires

Home vs. hospital

Multiple logistic regression

65%

71%

Studies with 2 papers reporting the same set of data.

�

Table C Medium quality studies: multivariate analysis and quality score less than 70% or univariate analysis and quality score of 60% or greater (n studies=17; n patients=77,828) [posted as supplied by author]

Authors and reference	Year of publication (study period)	Country	Study design	Setting	Sample size	Data collection methods	Categories of place of death	Analysis	Home death %	Quality score %
Mann et al. ^(w29)	1993 (1980-90)	US, New York	Retrospective cohort	Gynaecologic patients of tertiary care facility	454	Gynaecologic tumour registry	Hospital vs. home	Discriminant function analysis	22%	65%*
Gyllenhammar et al. ^(w30)	2003 (1999)	Sweden	Cross-sectional	5 palliative home care teams	221	Clinical records	Home vs. elsewhere	Logistic regression analysis	53%	56%*
Higginson and Wilkinson ^(w31)	2002 (1997-99)	UK	Cross-sectional	Marie Curie home nursing services	13311	Marie Curie NurseLink System	Home vs. elsewhere	Multivariate analysis	94%	52%*
Sims et al. ^(w32)	1997 (1995)	UK, Doncaster	Retrospective cohort	General population	820	Death certificates	Home, hospital, hospice, other	χ�	27%	77%
Grande et al. ^(w33)	1999 (?)	UK, Cambridge	RCT	RCT: hospital at home vs. standard care	229	Death certificates and unknown method	Home	χ�	58; 78%	73%
Addington-Hall et al. ^(w34)	1998 (1990)	UK	Retrospective cohort survey	General population	2051	Death certificates and interviews	Home, hospital, hospice, other	χ�	29%	69%
Grande et al. ^(w35)	2003 (16 months)	UK, Cambridge	Case control	Trial: hospital at home and cancer registry	327	Cancer registry, clinical records	Home, inpatient care	χ�, t-tests, Mann- Whitney U	39%	69%
Roder et al. ^(w36)	1987 (1981;85)	South Australia	Retrospective cohort	General population	1582	Death certificates	Hospital, hospice, nursing home, home	χ�	14%	69%
Davison et al. ^(w37)	2001 (1977; 87; 97)	Ireland, Belfast	Retrospective cohort	General population	1324	Death certificates	Home, hospital, hospice	χ�	35-28%	67%
Catalan-Fernandez et al. ^(w38)	1991 (1984-86)	Spain, Majorca	Retrospective cohort survey	General population	335	Tumour registry, clinical records and interviews	Home, hospital	χ�, Kruskal-Wallis test	46%	67%
Thorne et al. ^(w39)	1994 (1991-92)	UK, Exeter	Retrospective cohort	General population	1022	Death certificates	Home, community hospital, specialist services, nursing or residential home, hospice	χ�, t-tests	34%	67%
Flynn and Stewart ^(w40)	1979 (1957-74)	US, Ohio	Retrospective cohort	General population	55288	Death certificates	Hospital, nursing home, home	χ�	20%	65%
Goodwin et al. ^(w41)	2003 (8 months)	UK	Quasi-experimental comparative	Trial - palliative day care	173	Interviews	Home, hospital, hospice	χ�	31%	65%
Hinton ^(w42)	1994 (2 years)	UK, London	Cross-sectional	Hospice home care service	338	Clinical records	Home, inpatient	χ�	27%	65%
McWhinney et al. ^(w43)	1995 (1988-90)	Canada, Ontario	Case control	Palliative care home team	150	Clinical records	Home, hospital	χ� and t-tests	28%	65%
Axelsson and Christensen ^(w44)	1996 (1990)	Sweden, Jamtland	Cross-sectional	General population	203	Death certificates and clinical records	Home, nursing home, hospital	χ�	12%	63%
Higginson et al. ^(w45)	1994 (1988-92)	UK, London	Retrospective cohort	General population	?	Death certificates	Home	Spearman�s r	5-46%	63%

* - Studies that performed multivariate analysis but scored lower than 70% in the quality assessment.

�

Table D Low quality studies: univariate analysis and quality score less than 60% (n studies=16; n patients= 38,502) [posted as supplied by author]

Authors and references	Year of publication (study period)	Country	Study design	Setting	Sample size	Data collection methods	Categories of place of death	Analysis	Home death %	Quality score %
Hinton ^(w42)* Hinton ^(w46)*	1994 1998 (2 years)	UK, London	Prospective cohort	Hospice home care service	77	Interviews	Home, hospital, hospice	McNemar, χ�, t-tests	28%	58% 50%
Maida ^(w47)	2002 (1997-99)	Canada, Ontario	Retrospective cohort	Home palliative care	402	Clinical records	Home, hospital	χ�, t-tests, Mann-Whitney U	55%	58%
Woo et al. ^(w48)	1991 (1982-90)	Australia, Sydney	Retrospective cohort	Veterans hospital, patients with colorectal cancer	69	Colorectal cancer registry, clinical records and interviews with the patient�s doctor	Home, nursing home, hospice, hospital	χ�	12%	58%
Andershed and Ternestedt ^(w49)	1997 (1990-94)	Sweden, �rebro	Cross-sectional + retrospective cohort data	Hospital and hospital based home care	1464	Cancer registry and clinical records	Home, hospital, hospice, geriatric department, acute hospital	χ�	?	56%
Jimeno Aranda et al. ^(w50)	1993 (1980-89)	Spain, Zaragoza	Retrospective cohort	Patients with gastric cancer	539	Cancer registry	Home, hospital	χ�, t-tests	46%	54%
Dunphy and Amesbury ^(w51)	1990 (1988)	UK, London	Retrospective cohort	Hospice (inpatient and home care)	140	Clinical records	Home, hospice, hospital	χ�	61%	52%
Loven et al. ^(w52)	1990 (one year)	Israel, Petah Tiqva	Cross-sectional	Home care service	58	Clinical records	Home, general hospital, chronic hospital	Fisher�s exact test	29%	52%
Ward ^(w53)	1974 (1971)	UK, Sheffield	Retrospective cohort survey	General population	279	Death certificates and interviews	Home, hospital	χ�	44%	52%
Duke ^(w54)	1997 (1978, 89)	UK, Oxford	Comparison of 2 cross-sectional groups	Hospice (inpatient and home care)	467	Clinical records	Hospice, home, hospital, other	χ�	19%	50%
Gomas ^(w55)	1993 (1985-91)	France	Cross-sectional	Home care	171	Medical records?	Home, institution	?	37%	46%
Johnson et al. ^(w56)	1988 (1984-86)	UK, Sheffield	Cross-sectional	Home relative support scheme	171	?	Home, hospice, hospital	�analysis for trend�	60%	44%
Polissar et al. ^(w57)	1987 (1968-81)	US, Washington	Retrospective cohort	General population	33865	Death certificates and cancer surveillance system	Home, hospital, nursing home, other	�cross-tabulation and simple graphs�	15%	42%
Gilbar and Steiner ^(w58)	1996 (1992-93)	Israel, Haifa	Retrospective cohort	Home care unit	171	Clinical records	?	ANOVA analysis	?	40%
Malden et al. (w59)	1984 (1979-81)	Australia, Sydney	Case control	2 medical oncology units	225	Clinical records	Oncology unit, home, nursing care facility, others	?	18%	40%
Kane et al. ^(w60)	1984 (?)	US, Los Angeles	RCT	RCT � hospice vs. conventional care	247	Clinical records, interviews	Hospice, hospital, home	?	3; 7%	33%
Herd ^(w61)	1990 (1987)	UK, West Cumbria	Retrospective cohort survey	General population	157	Death certificates and interviews	Home, hospital	?	53%	33%