Skip to main content

Table 4 Studies addressing issues in Statement 5: Study characteristics, intervention content and outcomes reported

From: UK quality statements on end of life care in dementia: a systematic review of research evidence

Study & country

Design

Population

Intervention content, analysisa and number with a statement about future care (SFC) or related document

Results: Impact on care outcomes of having or not having advanced directives or related documentsa

Nicholas 2014 et al. [39] USA

Retrospective study which controls for confounding variables Data from 3876 participants of the Health & Retirement Study (HRS) who died between 1998 and 2007. Study exit interview (after death) by knowledgeable informant (KI)

Age 65 years and over, consented to share Medicare claims data for last 6 months of life. Includes nursing home (NH) (n = 1812) and community dwelling people (n = 2064). 21.7 % severe dementia, 43.1 % mild dementia or cognitive impairment, 35.2 % normal cognition (as assessed at last HRS interview –mean 436 days before death – using validated cognition measures). Results only reported for severe dementia.

Comparison of end-of-life care with those that had and those that didn’t have a written advance directive (AD) or “living will” as stated by proxy respondent in post-mortem interview. General liner model and logistic regression to generate predicted spending and probabilities of use in last six months of life. Adjustments were made for patient characteristics, including cognitive functioning, an AD and their interaction, stratified by NH use in last 6 months. 36.4 % of people with severe dementia had a treatment limiting AD (40 % of NH, 27.4 % of community dwelling).

REGRESSION: results presented for subgroup with severe dementia only

Medicare spending ($1,000)

 

Community-dwelling

NH

No SFC

32.2

24.7

SCF

20.7***

22.5

Hospital death (%)

 

Community-dwelling

NH

No SFC

31.8

20.6

SFC

13.9***

14.6**

Life sustaining treatment (%)

 

Community-dwelling

NH

No SFC

19.8

11.6

SFC

10.6

9.8

ICU use (%)

 

Community-dwelling

NH

No SFC

19.6

10.8

SFC

10.2***

7.1

Difference in results of AD and no AD not significant except where **/*** P value ** < 0.05, ***p < 0.01

Silveira 2010 et al. [40] USA

Retrospective study which controlled for confounding variables. Data from 3746 participants in HRS Study who died between 2000 and 2006. Study exit interview (after death) by knowledgeable informant (KI) – a relative most commonly as adult child

Age 60 years and over, died 2000–2006, 46 % had cognitive impairment prior to death (but not dementia specific) as stated by KI

Comparison of end-of-life care as stated by KI in post-mortem interview with those that had and those that didn’t have a written SFC or Power of Attorney (PoA).

REGRESSION: Adjusted odds ratio and (95 % confidence interval) for presence versus absence of a living will or a PoAa

 

SFC

PoA

Multiple logistic regression. Adjustments made for socio-demographic and clinical characteristics. 42.5 % required decision-making at end of life, 70 % of whom lacked capacity, 67.6 % of whom had SFC. 83 % who requested limited care and 97 % who requested comfort care received care consistent with preferences

Hospital death

0.71 (0.47, 1.07)

0.72 (0.55, 0.93)a

All care

0.33 (0.19, 0.56)a

0.54 (0.34, 0.86)a

Limited Care

1.79 (1.28, 2.50)a

1.18 (0.75, 1.85)

Comfort Care

2.59 (1.06, 6.31)a

2.01 (0.89, 4.52)

All care is care under any circumstances to prolong life, Limited care is care in certain circumstances, this is as opposed to all care possible in order to prolong life, Comfort care is comfortable and pain-free while forgoing extensive measures to prolong life.

aDifference statistically significant

Vandervoort 2012 et al. [41] Belgium

Retrospective study which controls for confounding variables

NH resident.

Presence of a written doctors orders to withhold treatment, such as do-not-hospitalise or do-not-resuscitate, or advance patient directive (SFC), such as a “living will” as stated by nurse in post-mortem questionnaire. This questionnaire also captured care outcomes, Comparison of outcomes dependent on presence of SFC. Multivariate logistic regression model. Outcomes only on SFC presented. (The others are not relevant to this review).

REGRESSION: Adjusted odds ratio and 95 % confidence interval presented on SFC and patient directives only

Study of deaths in NH of residents with dementia. All NH (594) invited to participate. Post death questionnaire completed by nurse.

Diagnosis of dementia, and severity, as stated by nurse.

Symptoms in the last week of life using Edmonton Symptom Assessment Scale

Died in 2-month study period in 2006. 72 % female, 78 % 80 years and above. 63 % severe and 37 % moderate dementia. 764 deaths in 345 NH (58 % response rate).

Pain

 

0.72 (0.37, 1.40)

Tiredness

 

1.08 (0.54,2.15)

Nausea

 

0.55 (0.20,1.51)

Depression

 

1.42 (0.73,2.77)

Anxiety

 

1.38 (0.73,2.58)

Drowsiness

 

0.78 (0.40,1.51)

Appetite

 

1.14 (0.55,2.40)

Shortness of breath

 

1.19 (0.63, 2.23)

QOL last week of life

 

1.14 (0.58,2.24)

Mildness of death

 

1.70 (0.70, 4.12)

Death in hospital or palliative care unit

 

2.09 (0.92, 4.72)

All odds ratios not significant

Vandervoort 2014 et al. [42] Belgium

Retrospective study which controls for confounding variables. Study of NH of residents with dementia who died within a 3-month period in 2010 (representative sample using random cluster-sampling). Post death questionnaire completed by nurse, GP, family member or friend & NH administrator.

Diagnosis of dementia as stated by nurse or GP. 58 % female, 84 % 80 years and above. 51 % very severe dementia, 25 % severe dementia, 24 % moderate or mild dementia

Comparison of scores on the Comfort Assessment in Dying in Dementia Scale (CAD-EOLD).

REGRESSION: Mean (standard deviation) score in total and subscales of the CAD-EOLDa

Relative available to complete questionnaire. 101 deaths in 69 NH (58 % response rate).

Multivariate logistic regression adjusting for age, gender, level of dementia and sentinel events.

 

SFC

No SFC

17.5 % had a written advanced directive, 56.7 % had GP orders

Total

31.9 (7.1)

29.1 (6.3)

Physical distress

8.8 (2.9)

8.2 (2.2)

Dying symptoms

8.6 (2.4)

8.0 (2.6)

Emotional distress

10.2 (2.3)**

9.0 (2.3)

Wellbeing

6.2 (1.9)

5.9 (1.9)

 

DNH

No DNH

Total

32.8 (6.4)

29.1 (6.4)

Physical distress

8.8 (2.9)

8.2 (2.2)

Dying symptoms

9.1 (2.2)

8.0 (2.6)

Emotional distress

10.4 (1.8)***

9.0 (2.4)

Wellbeing

6.4 (2.0)

5.9 (1.9)

 

DNR

No DNR

Total

32.9 (5.6)

29.1 (6.5)

Physical distress

9.3 (2.7)

8.1 (2.7)

Dying symptoms

8.6 (2.3)

8.1 (2.7)

Emotional distress

10.6 (1.7)****

9.0 (2.4)

Wellbeing

6.5 (1.6)

5.9 (1.9)

 

PDMA

No PDMA

Total

29.1 (7.6)

29.4 (6.3)

Physical symptoms

8.2 (2.6)

8.1 (2.3)

Dying symptoms

7.8 (3.0)

8.0 (2.6)

Emotional distress

9.4 (3.3)

9.1 (2.3)

Wellbeing

5.6 (0.9)

5.9 (1.9)

aAll adjusted odds ratios reported as not significant apart from: ** 2.99 (CI 1.1, 8.3) ***2.54 (0.8, 7,7), ****3.45 (CI 1.1,11).

  1. aAll trials presented the outcomes reported here as a primary outcome. Results presented are the most complete reported in the published paper. CAD-EOLD = Comfort Assessment in Dying in Dementia Scale, DNH = do-not-hospitalise, DNR = do not resuscitate, KI = knowledgeable informant, GP = general practitioner/primary care physician, HRS = Health & Retirement Study, NH = nursing home, PDMA = proxy decision maker assigned, PoA = Power of Attorney. QOL = quality of life, SFC = statement about future care