Transcript Insights from Economic-Epidemiology

Insights from economicepidemiology
Ramanan Laxminarayan
Resources for the Future, Washington DC
Economic Epidemiology
Mathematical conceptualization of the
interplay between economics, human
behavior and disease ecology to
improve our understanding of
– the emergence, persistence and spread of
infectious agents
– optimal strategies and policies to control
their spread
Economics

Measuring health outcomes
 Provides a single metric to compare costs and
benefits both contemporaneously and over time.

Incorporating behavior
 Can alter conclusions reached by purely
epidemiological models by incorporating
behavior.

Comparing public policies
 Increases relevance for application in the real
world.
Overview



Individual response and disease
Incentives of institutions (to invest in
hospital infection control)
Optimal allocation of resources between
two regions (or hospitals)
Individual response and disease

Vaccinations
– Insufficient incentives to vaccinate prevent
attainment of herd immunity thresholds

Drug resistance
– Insufficient incentives to make appropriate use
leads to ineffective drugs and increasing
prevalence

Testing
– Private testing behavior adds to public information
on disease prevalence
Rational epidemics

Prevalence response elasticity
– Hazard rate into infection of susceptibles is
a decreasing function of prevalence
(opposite of epidemiological model
predictions)
– Application to HIV
– Application to Measles
Geoffard and Philipson, Int. Econ. Rev., 1996
Blower et al, Science, 2000
Blower et al, Science, 2000
When should governments
intervene?



Disease prevalence increases adoption
of public programs
Impact of public program may be zero if
prevalence has already reached an
individual’s threshold prevalence
Paradoxically, the role of government
subsidies is lowest when prevalence is
highest since individuals will protect
themselves regardless
Philipson, NBER, 1999
Public price subsidies

Can price subsidies or mandatory
programs achieve eradication?
– Increase in demand by folks covered by
the program lowers the incentives to
vaccinate for those outside the program

Do monopolistic vaccine manufacturers
have an incentive to eradicate disease?
– Market for their product would disappear
with eradication
Geoffard and Philipson, Int Econ Rev, 1997
Disease Complementarities


Incentive to invest in prevention against
one cause of death depends positively
on probability of dying from other
causes
Intervening to prevent mortality not only
prevents a death but also increases
incentives for the family to fight other
diseases
Dow et al, Am Econ Rev, 1999
Does the theory fit the facts?



Do individuals actually observe
prevalence?
Why don’t we see prevalence
responsiveness at work everywhere?
Importance of observational learning
(herd behavior)?
Stoneburner and Low-Beer, Science, 2004
Stoneburner and Low-Beer, Science, 2004
Stoneburner and Low-Beer, Science, 2004
NNIS Data, 2004
Optimal infection control
Infection dynamics are given by
X
c
X
1 X

X 
Equilibrium prevalence is given by
X

c
S
c
1
S
c
1 2 
4 S
c
2S
c
Smith, Levin, Laxminarayan (PNAS, 2005)
Objective
Minimize costs of infection control and
infections
c DX

c
Local minima, if they exist, are solutions to

1 DX

ĉ0
Smith, Levin, Laxminarayan (PNAS, 2005)
Smith, Levin, Laxminarayan (PNAS, 2005)
Strategic interactions with other hospitals
X  
c
X
1 XX 
X Z
Y  
c

Y
1 YY 
Y Z
Ż r
X/n 
n 1
Y/n ZZ
The focal institution decides how much to invest
in HIC by minimizing the net present value of discounted
costs of HIC and hospitalization:

0 
c DX
t, c;n, c

e t dt
Smith, Levin, Laxminarayan (PNAS, 2005)
Implications for policy




Dutch experience: frequency of MRSA
infections is < 0.5% after an intensive
‘‘search-and-destroy’’ campaign, compared
with 50% in some areas
In Siouxland (Iowa, Nebraska, S. Dakota), an
epidemic of VRE was reversed
Regionally coordinated response to epidemic
Does this explain higher prevalence of ARB in
areas with high concentration of health care
institutions?
Rowthorn, Laxminarayan, Gilligan Submitted
Rowthorn, Laxminarayan, Gilligan Submitted
Rowthorn, Laxminarayan, Gilligan Submitted
Rowthorn, Laxminarayan, Gilligan Submitted
Allocating resources
Expenditure on drugs is subject to the
budget constraint cF1 F2 M
Finance is not transferable through
time.
Problem is to choose F₁and F₂ so as to
minimise the following integral
 rt
e 
I1
0
V 
Rowthorn, Laxminarayan, Gilligan Submitted
I 2 dt
Optimal allocation

At low levels of infection in both populations
– Preferentially treat population with higher
transmission coefficient because of greater
economic value associated with greater potential
to prevent secondary infections

At high levels of infection
– Preferentially treat population with lower levels of
infection since the higher probability of re-infection
in high infection populations reduces the economic
value of treatment
Rowthorn, Laxminarayan, Gilligan Submitted
0.5
0.0
0
0.10
35
70
Time
Region 1
Region 2
0.00
0
10
20
0.15
0.0
0
0.10
40
50
60
0
10
20
0.5
0
35
70
Time
0.05
50
40
50
60
0.2
0.5
0
35
0
35
70
Time
0.2
0.0
10
20
30
40
50
60
10
20
30
40
Time
Proportion infected
0.50
0.25
0.00
0.00
70
i
Worst path
0.8
50
60
70
0.25
0.50
0.75
Region 1
Proportion infected
0.75
0.6
1.0
0.4
0.2
0.5
0.0
0
35
0.50
0.25
70
Time
0.0
0
0.18
70
Time
Treated
0.0
Proportion infected
Treated
0.5
0.12
0.0
0
h
1.0
0.4
0.75
Time
0.8
0.6
f
1.0
0.4
0.06
Region 1
0.6
70
Best path
1.0
0.10
0.00
0.00
70
Region 2
30
60
Worst path
0.8
Time
Proportion infected
40
0.0
0.00
g
30
Treated
e
1.0
0.10
20
0.15
Time
0.0
10
Proportion infected
0.05
0.05
70
Proportion infected
Treated
Proportion infected
0.15
0
70
0.00
30
Best path
0.20
35
Time
Time
d
0.5
Region 2
0.05
0.20
1.0
Region 2
Proportion infected
0.15
c
Worst path
0.20
1.0
Treated
Proportion infected
b
Best path
0.20
Treated
a
0
10
20
30
40
Time
Rowthorn, Laxminarayan, Gilligan Submitted
50
60
70
0.00
0.00
0.25
0.50
Region 1
0.75
Closing thoughts


Epidemiological models take little or no
account of economic constraints or
incentives faced by individuals or
institutions
Economic models mostly ignore the
spatial and temporal dynamics of
disease.