Using EQ-5D to measure patient reported outcomes

Using EQ-5D to measure patient
reported outcomes
April 2017
Liz Vernon-Wilson
Outcomes Researcher
Tom Rudd Unit, Moorgreen Hospital
1
Introducing PROM data
EQ-5D-5L is a Patient Reported Outcome Measure or PROM.
A short questionnaire that the service user completes to describe their
health and wellbeing.
Collecting PROM data at multiple time points enables an individual’s
progress to be followed.
PROM responses can inform care plan goals.
Contextualised aggregated data from many similar service users can help
us understand how well our services perform from our service users’
perspective.
2
About the EQ-5D PROM
The EQ-5D-5L questionnaire has two parts.
First, 5 domain questions ask about specific
issues:
– mobility
– self-care
– usual activities
– pain or discomfort
– anxiety or depression.
The EQ-5D-5L uses 5 levels of
responsiveness to measure problems. The
range is; no problem - disabling/extreme
The domain items can analysed individually
or be conflated into one Index score.
3
The EQ-5D VAS
The second part of EQ-5D data is a global score.
This gives an overall impression of wellbeing
today.
Service users rate from 0 (worst health
imaginable) to 100 (best health) on a visual
analogue scale (VAS).
Collecting other information helps contextualise
the outcomes data. We can use
• Point of care (admission/discharge)
• Intervention
• Duration
• Diagnosis, Care cluster (mental health) or
other health resource group (HRG)
4
Working with PROs
Contextualised PROM responses can be used to compare interventions or
teams. They are a starting point for reflective practice; discussions about
what works well and why, help us improve.
It can also be helpful to put PRO (patient reported outcome) information
alongside clinician rated measures, or other quality or process indicators.
Different patient groups could start with quite di ssimilar PROM profiles
and achieve different outcomes over time.
5
Contextualising OPMH outcomes.
Mental health care clusters are a needs-based classification system.
Allocation to a particular care clusters depends on broad identification of the
nature of someone’s presenting problems and current severity of their
symptoms. Clusters fall into three super groups; non-psychotic, psychotic or
organic, and range from low needs to severe. Cluster choice acknowledges
risk, chronicity and likely duration of care episode too.
Service users in different clusters can start with different EQ-5D profiles and
achieve different outcomes.
6
Distinguishing Service User
Groups (VAS)
New referrals to OPMH can usually be distinguished by their EQ-5D-5L scores.
People entering memory services (cluster 18) often give very positive responses
to the health thermometer question. They typically describe few problems on
the domain questions.
This compares quite differently to people referred for functional MH problems
(cluster 3 & 4).
The graph demonstrates the
distribution of VAS/thermometer
scores collected for these two
groups at referral. High scores
indicated good levels of self-rated
wellbeing.
Index scores also follow a similar
pattern.
Cluster 18, n=1011; cluster 3&4 n=330
7
Distinguishing Service User
Groups (Index)
The index score represents a summary of domain questions. Service users
who respond “No problem” to all five questions return a domain score of 1,
whilst someone responding with the most extreme/disabling problem on all
five questions returns an index score of -0.594.
The graph shows people with
memory problems (cluster
18) return higher index
scores (fewer problems) than
people with functional
mental health problems
(cluster 3 & 4).
Possible range -0.594- 1.000
Observed range -0.4 – 1.000
Cluster 18 n=992; cluster 3&4 n=301
8
Expected Outcomes
Before examining outcomes data, it can be helpful to state what changes
in the data are expected.
For example, is it reasonable to anticipate older people who have good
wellbeing scores will remain well, or deteriorate over time ?
If people record a low score, what change is it reasonable to expect and by
when?
– Would this be in the holistic score or domain score?
– By review or by discharge?
Now you can test the data and your assumptions.
9
Different Outcomes revealed
Paired data reveals different patterns
for the two groups of service users
with different presentations
described.
Cluster 18 pairs lie around the
No change in score line, whilst the
points from service users in clusters 3
& 4 are more likely to fall above the
line (top left shows improvement).
There is statistical difference in the
before and after treatment scores for
service users in the cluster 3 & 4
group, but not for those in cluster 18.
This comparison principle could be
applied to treatments, wards etc .
10
What clinical changes occur?
Examining change in each of the 5 domains shows that the greatest
difference before and after treatment is observed in the A nxiety/depression
domain score.
Change is also noted in domain 3, Usual activities , demonstating how mental
health impacts on everyday functioning. Both of these domains demonstrate
statistically significant differences before and after treatment (p< 0.01).
n=36
11
Case Study; Albert
Case studies help us understand how particular interventions, interactions and
relationships support recovery. This service user’s name has been changed.
Albert’s wife of 50 years died in 2015, following a short illness. He has suffered
with a severe grief reaction. Albert has experienced lasting depression and
anxiety since his bereavement. Initially, his symptoms included diurnal
variation in mood, sleep disturbance, severe loss of appetite, psychomotor
agitation, guilt & self-blame. Albert also developed marked anxiety including
cario-respiratory features of panic.
He has a history of depression.
His interventions have included;
•
•
•
•
Anti-depressants
4 weeks admitted care
Bereavement counselling, music therapy
OT support
•Crisis plan for times of distress
•WRAP
•Coping strategies
•PTS
12
Albert’s PROM scores help
track his progress. They
demonstrate the impact of
interventions and transitions
in care.
Admission
CMHT support
reduced after
discharge
1st Anniversary
of wife’s death
Choosing to
disengage from
psychological
therapy
Discussing these scores and the case
history highlighted transitions between
care teams and professionals need careful
management. What would your team do
with this information?
Summary
Patient reported outcome measures, such as EQ-5D-5L, demonstrate our
service users’ view on changing health and wellbeing states as they travel
care pathways.
Contextualising PROM data with information about team, diagnosis,
intervention and duration of care episode offers insight into clinical
effectiveness.
Here, EQ-5D data has distinguished different patterns of change in health
state for older aged adults accessing integrated community mental health
services. Those referred with early cognitive impairment reported stability in
global wellbeing. Service users referred for non-psychotic mental health
problems including anxiety & depression reported improvement in this
domain, their usual activities and global wellbeing.
Comparison of PROM data with other information from the EPR, such as
clinician reported outcome and detail about intervention will enhance our
potential to use outcomes data effectively.
14
For more information on your team’s outcomes data, please contact:
Liz Vernon-Wilson at
[email protected]
Clinical Outcomes
Tom Rudd Unit,
Moorgreen Hospital
Southampton
SO30 3JB
023 8047 5152
15