Is self management an effective and economically sustainable approach in the management of chronic a

1
Is self management an effective and economically
sustainable approach in the management of chronic
and complex health issues?John Petkov B.Sc.,
B.App.Sc., M.ScApplied Statistics
UnitUniversity of SA
2
ABS STATISTICS

• Mean age of males 84.2
• Mean age of females 87.7
• Set to rise in the next 10 years
• Largest growth is in the 65 age group
• Increase in age is due to
• Falling fertility
• Increased Longevity (less important)

3
Will health costs rise due to this ageing?

• Not sure-depends on who is asked.
• Evidence suggests that Australia may not feel
  the
• same impact as the USA as it is less dependent
  
• on private health insurance schemes.

4
Chronic conditions that are associated with older
people

• Diabetes
• Cardiovascular disease
• Arthritis
• Osteoporosis
• Respiratory disease
• Depression

5
Treatment of chronic diseases is more frequent
than acute care

• Management of chronic disease is important
  because of
• Decrease in health costs
• Quality of life (priceless)

6
Concept of self-management

• This is quite different to the present models
• Health care professional diagnoses and
  prescribes
• Patient complies
• Public Health models aim is prevention
• BUT.
• It doesnt help the increase of chronic disease
  due to ageing.

7
Self-management is

• Problem solving
• Resource utilisation
• Decision making
• Patient-health provider relationship
• Taking action

8
Sharing Health Care National Data(Stanford,2000)

• 258 people recruited in Whyalla as part of the
  overall Australia-wide program
• Each person had at least one of the chronic
  conditions associated with older people as
  outlined above

9
The support offered was

• Care/self-management planning
• Education/training
• Support

10
A subset of the cohort was also offered the Lorig
management program

• This consisted of a weekly 3-hour training and
  information session. This lasted for 6 weeks

11
The data was collected over four time frames

• Baseline
• 6 months
• 12 months
• 18 months

12
The Whyalla cohort also completed the Partners
in Health program

• The questionnaire asked the patient to assess
  their level of
• Knowledge
• Compliance
• Coping with the condition and new lifestyle

13
Check on responses

• The health provider also completed the same
  questionnaire for that patient at the same point
  in time. This is seen as a check

14
Other data

• The patient also reported
• Level of service use
• Assessment of his/her health over the time

15
Data Type

• So essentially we have longitudinal data gathered
  over four time frames.
• How do we analyse this data?

16
Method of Analysis

• Random Coefficient Regression (mixed modelling)
  was used to model the longitudinal data
• This is the best (and most modern) method

17
Mixed Modelling

• Mixed fixed and random effects
• Random effects are patient effects-each patient
  has an individual starting point and individual
  rate of change
• Correlations are also taken into account. A
  response at one point in time is almost certainly
  related to the response at the previous time
  point

18
But best of all !

• Missing data is not a problem due to the
  correlation specification. All the information is
  used.
• In other methods of analysis, one missing data
  point results in the loss of that subject.

19
Consider the following
20
Some Maths (sorry)

• Ordinary Least Squares
• For person i at time j we have

21
Random coefficient regression
22
Results Example of question What I know about
my illness is9 point scale-higher is
better(very significant over time- p0.000)
23
We can look at the intercepts (starting points)
24
And the slopes
25
And the correlation between intercepts and slopes
26
When we look at the Lorig versus others
27
I take my medication as asked by my doctorNo
change over time
28
Hospital UseThis is count data and a
Generalised Linear Poisson Mixed Model with a log
link is used
29
Hospital VisitsVery significant (p 0.001)
30
How good is the model?
31
Monte Carlo simulation is used (10,000
samples)The improvement is shown below
32
Level of Pain0-10 with high score indicating
more painVery significant (p 0.006)
33
General Health 1-5 and low score is
betterSignificant (p 0.02)
34
CONCLUSION

• There is good evidence to suggest that
  self-management leads to improved health outcomes
  and, as a result , less service use.
• Chronic disease sufferers can get by with a
  little help from their friends