Author: Stanton Glantz
Position: Professor of Medicine
Institution: University of California San Francisco
E-mail: glantz@medicine.ucsf.edu
Additional Authors: James Lightwood
Submitted Date: August 29, 2008
Published Date: September 1, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Our study estimates the observed cost savings attributable to reduced smoking in a specific population with a given history of changes in smoking behavior. The estimated differences in state per capita health care expenditures attributable to changes in smoking will depend on the kind of changes that occurred (e.g., rates of smoking cessation, uptake, consumption per current smoker, and population prevalence of passive smoking) over the sample period. Our results are not, and are not intended to be, equivalent to existing cross-sectional or life-cycle estimates of the cost of an average current smoker.

Estimates of actual historical savings in a specific population (our study), and theoretical model-based estimates of cross-sectional and life-cycle costs (Barendregt’s study), can play complementary roles. The results of analyses based on aggregate population based data can provide the information needed for better validation of the model-based estimates; the model-based estimates can generate testable hypotheses for statistical analyses, and suggest promising strategies for estimation. The resulting validated model-based estimates can provide detailed analysis required for planning and program evaluation that may never be feasible using statistical analysis of aggregate data.

We disagree, however, that the results of Barendregt’s model based study, or any other model based study, can be considered a sure guide to how our results will evolve as future data become available. Reviews of existing life-cycle studies reveal a range of results, that predict that smoking will be associated with increasing or decreasing life-cycle health expenditures [1, 2]. Different results may be correct, in their relevant setting for the countries and sample periods providing the data, or may be due to differences in modeling assumptions that are difficult to validate [3].

One could make a similar case for most interventions that reduce disease and extend life. The fact that childhood immunization prevents infants from dying and therefore allows them to live longer and develop diseases of old age is not a reason to stop immunization programs.

Barendregt misinterprets our results when he links the 15-year life of the California Tobacco Control Program to the 15 year time horizon in his study for the transition from reduced to increased costs following a change in smoking behavior. In addition, our study took into account other factors that reduced smoking in California in addition to state tobacco control efforts: price changes and a pre-existing secular trend in California smoking behavior. Smoking behavior of the California population began to diverge from that of most other states before the beginning of the California tobacco control program.

As a practical matter, most policymakers are interested in short and intermediate term economic effects in part because long-term life-cycle effects are very uncertain. To the extent that policy makers are interested in reducing health care costs in the short term, our results provide direct empirical evidence that large scale tobacco control programs provide large and growing returns.

REFERENCES:

[1] Warner KE. The economics of tobacco: myths and realities. Tob. Control 2000;9;78-89.
[2] Sloan F, Ostermann J, Picone G, Conover C, and Taylor Jr D. The Price of Smoking, MIT Press, 2004: Cambridge(MA).
[3] Warner K, Hodgson T and Carroll C. Medical costs of smoking in the United States: estimates, their validity, and their implications. Tobacco Control 1999;8:290–300).