Posted by ""Nathaniel Osgood"" <
osgood@cs.usask.ca>
Hi Scott,
Good to hear from you. As you and I discussed at one point, there
have been a number of past papers using system dynamics modeling to tobacco
and smoking issues. To my knowledge, there have been 5 primary threads of
work:
Work by Homer and Roberts in the early 1980s. The model included
lots of soft factors and feedbacks. I can dig out the reference for you if
you are interested.
Work by Tengs and myself and (later) Tengs and Sajjad Ahmad
examining the cost and health impacts (and, specifically, the cost
effectiveness) of state (CA) and national (US) tobacco control policies.
This work was primarily a Markovian formulation with tight reliance on hard
secondary data, as the lead investigator (Tengs) had strong reservations
about incorporating factors that were hard to precisely measure (including
the many important feedbacks operating in the tobacco space) The research
group that conducted this work is no longer in existence, but I'm including
several citations below.
Work by Mendez et. al. of the University of Michigan
(
dmendez@umich.edu). This work was supported at least in part by the
Substance Abuse Policy Research Program and the Robert Wood Johnson
Foundation. This work was ongoing within the past comouple of years, but I
am not sure of its current status. Citations to this are included below as
well. Prof. Mendez has published quite a lot in the tobacco area, and he
could probably point you towards the papers that have made particular use of
system dynamics methodology.
Unpublished consulting model constructed by HVM for (if I'm not
mistaken) a UK pharmaceutical company. This model incorporated lots of soft
factors but was only loosely calibrated. Contact for this work is
Joe Gitchell
4800 Mongomery Lane, Suite 1000
Bethesda, MD 20814-3433
(301) 718-8440
The work by Richardson that you mentioned.
As I have argued in my recent ISDC 2005 workshop presentation, my
work using system dynamics to investigate tobacco convinced me that
effective policy analysis would be better served by an agent-based modeling
approach than with an approach based on differential equations. There are
many reasons for this, but the central motivations were the importance to
policy effectiveness of both social networks (well documented by tobacco
sociologists), population heterogeneity, and the relatively strong amount of
data available on patterns of individual smoking behavior. (Although many
of my colleagues would probably disagree, I personally am comfortable
including ABM approaches as part of a system dynamicist's toolkit.) For the
past year, I have have been working on a (yet unfunded) agent-based model of
tobacco cost-effectiveness. While computationally expensive, this model can
quite comfortably capture aspects of population smoking dynamics that are
very difficult or awkward to capture using differential equation based
approach. Based on these preliminary results, I will be attempting to
secure funding for this work over the course of the coming year.
I know that I'm not alone in believing that this area remains ripe
for lots of good work using a system dynamics approach (interpreted
broadly). I'd be delighted to hear if you come across any more additional
work.
Sincerely,
Nate Osgood
-------------------References included below------------------
Homer and Roberts:
[Roberts, Homer et al] Roberts, E., Homer, J., Kasabian, A., and Varrell M.
A Systems View of the Smoking Problem: Perspective and Limitations of the
Role of Science in Decision-Making. Int. J. Bio-Medical Computing (13)
(1982) 69-86.
Literature References for Tengs et al:
The Cost-Effectiveness of Raising the Legal Smoking Age in California
Ahmad Med Decis Making.2005; 25: 330-340.
http://apha.confex.com/apha/133am/techp ... 113523.htm
Tengs, T., Osgood, N. and Lin, T. Public health impact of changes in
smoking behavior: results from the Tobacco Policy Model. Medical Care. 2001
Oct;39(10):1131-41.
Tengs, T., Osgood, N. and Chen, L. The Cost-Effectiveness of Intensive
National School-Based Anti-Tobacco Education: Results from the Tobacco
Policy Model. Preventive Medicine. 2001 Dec;33(6):558-70.
Osgood, N. The Tobacco Policy Model: An Introduction. Presented at the
System Dynamics Winter Camp, Austin, Texas, January 2000.
Osgood, N. Beyond Markov: A system dynamics model of national tobacco
policy. Poster presentation at the Society for Medical Decision Making
Annual Meeting, October, 1999
Grant/Literature References for Mendez et. al:
Substance Abuse Policy Research Program
The Impact of Smoking Control Policies on Future Smoking Prevalence and
Health Status: A System Dynamics Analysis, David Mendez, Ph.D.
http://www.saprp.org/grant_detail.cfm?AppID=784
www.rwjf.org/files/publications/books/1999/exh2-2.htm :
David Mendez, Ph.D. The University of Michigan The Impact of Smoking Control
Policies on Future Smoking Prevalence and Health Status: A System Dynamics
Analysis
Posted by ""Nathaniel Osgood"" <
osgood@cs.usask.ca>
posting date Thu, 4 Aug 2005 11:18:10 -0600