Transcript Download

MEASURING HEALTH RELATED QUALITY OF LIFE IN DIABETES:
THE VALIDATION OF THE DIABETES HEALTH PROFILE (DHP-18) AND THE
DEVELOPMENT OF A BRIEF MEASURE OF DIABETES RELATED HRQL (DHP-12)
Brendan Mulhern1 & Keith Meadows2
1
Health Economics and Decision Science, University of Sheffield; 2 DHP Research and Consultancy Ltd
Introduction
Results
There is a need to validate existing patient reported outcome measures (PROMS) to
provide evidence on their validity. Subsequently, short measures can be developed from
validated PROMS to lessen respondent burden. The aim of this study was to:
•Validate the three domain conceptual framework of the Diabetes Health Profile (DHP-18)
•Apply Rasch analysis to investigate the performance of the DHP-18 and develop a short
12 item version (DHP-12).
The Diabetes Health Profile (www.diabeteshealthprofile.com)
The DHP-18 (Meadows et al., 2000):
•Assesses psychosocial functioning in Type 2 diabetes across three domains:
Psychological Distress; Barriers to Activity; Disinhibited Eating (scored on a 0-100 scale)
•Is the diabetes-specific outcome measure selected for the UK Department of Health
Patient Reported Outcome Measures Pilot for Long Term Conditions in Primary Care.
Methods
DHP-18 validation:
•Exploratory factor analysis was used to investigate the factor structure of the DHP-18
Rasch analysis:
•Rasch analysis was used to investigate the performance of the DHP-18 and select items
for the DHP-12. Rasch assesses the performance of individual items in relation to the
underlying trait measured by the dimension level model.
•Rasch assesses:
•The ordering of item response categories (disordered items are rescored )
•Differential item functioning (DIF) across different background characteristics
•The fit of each item to the overall model
•The position of each item on the underlying severity (latent) scale (negative
values cover the severe end of the scale, zero is the average, positive values
cover the mild end of the scale
Rasch was applied to each of the DHP-18 dimensions across two data subsets
DHP-18/DHP-12 comparison:
•The correlations between the measures were assessed
•Internal consistency for each domain was assessed using Cronbach’s alpha
•A separate comparison sample was used for this analysis
Factor and Rasch analysis
Sample:
S1: Development
n
Aged above 65
Gender
Male
Comparison sample
S2: Validation
771
810
2953
57.6%
57.9%
52.42%
59.4%
62.2%
53.90%
DHP-18 scores
Psychological Distress
19.07
18.21
19.26
Barriers to Activity
22.33
21.11
23.30
Factor and Rasch analysis (DHP-12 items highlighted in green)
Factor loading
Rasch analysis
Item
PD
Psychological distress domain
6. Lose temper if people nag
15. Lose temper or shout due to diabetes
14. Depression due to diabetes
16. Lose temper over small things
17. Moody about diabetes
18. More arguments at due to diabetes
Barriers to activity domain
2. Difficult staying out late
3. Days tied to meal times
12. Worry about busy shops
1. Food control life
4. Avoid going out if sugars low
10. Worry about colds/flu
13. Edgy when out and nowhere to eat
Disinhibited eating domain
11. Wish not so many nice things to eat
5. Eat to cheer yourself up
7. Hard to say no to food you like
8. Easy to stop eating
9. Eat extra when bored/fed up
BA
DE
Ordered?
1
2
0.54
0.86
0.62
0.8
0.69
0.68
0.62
0.72
0.44
0.47
0.63
0.44
0.56
0.50
0.77
0.75
0.63
0.67
No DIF?
1
2
1
Fit?
2
Location
1
DHP-12?
Scoring
2



√
√
√






√
√

√
√
√
√
√

√
√
√


√
√
√
√


√
√
√
√
n/a
n/a
-0.96
-0.52
0.5
0.97
n/a
n/a
-0.73
-0.19
0.14
0.79


√
√
√
√
n/a
n/a
0-2
0-2
0-2
0-2
√
√






√

√



√
√

√
√
√
√
√
√

√
√
√
√


√
√
√
√
√


√
√
√
√
√
n/a
n/a
n/a
-0.96
0.24
-0.07
0.26
n/a
n/a
n/a
-1.03
0.16
0.02
0.26



√
√
√
√
n/a
n/a
n/a
0-2
0-2
0-2
0-2
√


√
√

√
√
√
√
√
√
√
√
√
√
√
√
√
√

√
√
√
√

√
√
√
√
n/a
0.87
-0.04
0.33
-1.61
n/a
0.35
0.08
0.57
-1.00

√
√
√
√
n/a
0-2
0-2
0-3
0-3
DHP-12
•Four items from each domain have been selected.
•Responses to 10 items are disordered so have been rescored (raw score range is 0-26).
•One item (depression due to diabetes) displays DIF by gender. This item has been retained
as depression in diabetes has been found to be higher in females (Lustman et al, 1988).
DHP-18/DHP-12 comparison:
PD factor
DHP-18
DHP-12
BA factor
DHP-18
DHP-12
DE factor
DHP-18
DHP-12
Mean (sd)
Range
19.26 (18.69)
27.19 (23.93)
0-100
0-100
Correlation
0.95
% floor
% ceiling
Alpha
17.6
20.3
0.2
1.2
0.83
0.76
8.8
16.1
0.1
0.1
0.68
0.48
10.3
14.5
1.0
2.3
0.76
0.71
0.86
23.30 (17.94)
24.68 (18.68)
0-100
0-100
33.45 (23.85)
38.11 (26.90)
0-100
0-100
0.96
Discussion
•The three domain structure of the Diabetes Health Profile was confirmed.
•Rasch analysis indicates that the DHP-18 has satisfactory performance overall. This means
that the measure provides a valid basis for the development of a new short diabetes specific
PROM.
•A variety of Rasch analysis criteria have been used to develop a12 item version of the
Diabetes Health Profile (DHP-12) that includes four items in each domain.
•Initial psychometric analysis demonstrates evidence for the comparability of the DHP-12 and
DHP-18, but the internal consistency of the Barriers to Activity domain is low. Further analysis