Years of life lost due to disability YLD from Ischemic Heart Disease and Stroke, Thailand 2004: Data

1
Years of life lost due to disability (YLD) from
Ischemic Heart Disease and Stroke, Thailand 2004
Data sources, methods and preliminary results
2
Burden of disease

• Expressed in disability adjusted life years
  (DALYs)
• Two components
• Years of life lost (YLL)
• Number of deaths due to the condition of interest
  multiplied by standard remaining life expectancy
• Years lost due to disability (YLD)
• Number of new cases of the condition multipled by
  the duration of the condition and associated
  disability weight
• YLL YLD DALYs
• Today, focus on YLD estimation

3
Ischemic heart disease (IHD)

• 3 health states for YLD estimation
• Myocardial infarction surviving acute phase
  (28-days)
• Angina (pre- and post-MI)
• Heart failure (post-MI only)
• Due to lack of data, estimate YLD for each state
  independently
• e.g. Estimate angina pre- and post-AMI together
• Incidence x duration x disability weight

4
Acute-myocardial infarction

• Incidence
• Rely on the no. of admissions discharged alive
  from the DRG dataset
• AMI ICD-10 code I21
• Adjusted for less than 100 coverage using
  self-reported hospital admissions from Health
  Welfare Survey, 2005

5
Adjustment factors

• DRG covers 30 baht (UC) and civil servant scheme
• Excludes private health insurance and
  employer/employee funded insurance
• We determined the ratio of self-reported hospital
  admissions in the last year from HWS, 2005 to
  observed admissions in DRG, 2004, by subgroups
  of
• Age (0-14, 15-44, 45-69, 70)
• Sex
• Type of hospital
• Community/other
• General / Regional / University
• Private

6
Adjustment factors (calculated)
Adjustment factors (final)
7
No. of total admissions by source
8
(No Transcript)
9
Acute-myocardial infarction

• Duration
• 3 months (0.25 year) used in Australia BOD
• Local experts suggest shorter period of
  disability (2 months)

10
Disability weights GBD Dutch
Full health
0
eczema
mild heart failure 0.06
uncomplicated diabetes
mild-moderate angina 0.08
mild depression
0.2
ankle fracture
moderate heart failure 0.35
0.4
severe vision loss
Acute myocardial infarction (treated) 0.395
Acute myocardial infarction (untreated) 0.491
burns 20 body surface
AIDS
severe angina 0.57
0.6
moderate stroke 0.63
severe heart failure 0.65
severe depression
0.8
severe stroke 0.92
severe dementia
1
Worst possible health
11
Acute-myocardial infarction

• Disability weight
• Treated Untreated
• Australia 2003 100 treated
• Thailand 1999 50 treated, 50 untreated
• Discussion with experts assume 100 treated for
  Thailand 2004.

12
DisMod
Dead from other causes
Healthy
All other mortality
Incidence
Remission
Dead from disease
Diseased
Case fatality
13
DisMod Inputs

• At least 3 inputs required to model duration
• Incidence
• Prevalence
• Remission
• Case-fatality

14
Angina pectoris

• DisMod model using the following inputs
• 1. Incidence
• Admission rate discharged alive from the DRG
  dataset
• Angina pectoris ICD-10 code I20 with no I25
  (chronic IHD) code
• 2. Remission
• Based on revascularization rate from the DRG
  dataset
• 3. Case-fatality rate from AusBOD adjusted
  upwards by 10

15
(No Transcript)
16
first-ever
17
Annual case-fatality rate for AP
18
Distribution of severity

• Angina Pectoris
• Mild/moderate vs Severe
• Australia 95 mild moderate, 5 Severe
• Thailand 1999 All mild initially, severe for
  last 6-months
• Local expert 90 mild/moderate, 10 severe (Dr
  Sukit)

19
Congestive heart failure

• DRG data for congestive heart failure is
  problematic
• Higher probability of repeat admissions in a
  single year ? double counting
• Cannot attribute to IHD alone due to multiple
  aetiology
• Analysis of CHF admissions (I50) with
  corresponding IHD code indicate implausibly low
  due to IHD, even when limited to larger hospitals
• Less than 20
• Epidemiological studies suggest 60

20
Congestive heart failure

• DisMod model using the following inputs
• 1. Incidence
• 15 of new, non-fatal AMI results in congestive
  heart failure
• 2. Remission of 0
• 3. Case-fatality rate from US-BOD adjusted
  upwards by 10
• As per GBD 2000

21
Distribution of severity

• Heart failure
• Mild vs Moderate vs Severe
• Australia 60 Mild, 35 moderate, 5 Severe
• Thailand 1999 60 Mild, 30 moderate, 10 Severe
• Assume the same for 2004

22
YLD due to IHD
23
Stroke

• 2 health states for YLD estimation
• Stroke, surviving acute phase (first 28-days),
  with short term disability only (within 1st year)
• Stroke, surviving acute phase, with permanent
  long-term disability
• Assume 50 (males), 65 (females) of non-fatal
  stroke have long-term disability (Bonita et al,
  1984)
• Other estimates may be available from TES study

24
Stroke

• DisMod model using the following inputs
• 1. Incidence
• Non-fatal admissions for stroke (I60-64) from DRG
  dataset
• Adjusted for estimated proportion of first-ever
  from China MONICA study
• 2. Remission of 0
• 3. Prevalence of ever having had a stroke from
  the National Health Examination Survey (NHES),
  2004

25
(No Transcript)
26
first-ever non-fatal stroke
27
Prevalence of stroke
28
Distribution of severity

• Stroke
• Short term cases
• Thailand 1999 Mild disability only
• Long-term cases
• Thailand 1999 60 Mild, 30 moderate, 10 Severe
• As for 2004

29
YLD due to stroke
30
Limitations

• DRG counts admissions not individuals
• Inflation factors assume that the distribution of
  admissions by cause in the DRG dataset (UC, CSMB)
  is the same in those not in the DRG dataset (SSS,
  Private) within hospital types, age and sex
  groups
• More problematic for private hospitals
• Self-reported admissions from HWS may
  underestimate total admissions
• Selection bias alive individuals from household
  sample only
• Recall bias
• Difficult to determine first-ever admissions vs
  both first-ever recurrent admissions
• Assumption that all cases are hospitalized

31
Summary

• DRG data provides reasonable data on overall
  admission rate for IHD and stroke
• Difficult to determine first-ever events vs
  first-ever and recurrent
• Approach assumes all cases hospitalized
• Some data on prevalence and remission
• No robust data on IHD prevalence
• Very little data on case-fatality, disability and
  distribution of severity
• Key areas for future research