Working Paper: NBER ID: w15880
Authors: Frank R. Lichtenberg
Abstract: I analyze the effects of four types of medical innovation and cancer incidence on U.S. cancer mortality rates during the period 2000-2009, by estimating difference-in-differences models using longitudinal (annual) data on about 60 cancer sites (breast, colon, etc.). The outcome measure used is not subject to lead-time bias. I control for mean age at diagnosis, the stage distribution of patients at time of diagnosis, and the sex and race of diagnosed patients. \n\nUnder the assumption that there were no pre‐dated factors that drove both innovation and mortality and that there would have been parallel trends in mortality in the absence of innovation, the estimates indicate that there were three major sources of the 13.8% decline of the age-adjusted cancer mortality rate during 2000-2009. Drug innovation and imaging innovation are estimated to have reduced the cancer mortality rate by 8.0% and 4.0%, respectively. The decline in incidence is estimated to have reduced the cancer mortality rate by 1.2%. The social value of the reductions in cancer mortality attributable to medical innovations has been enormous, and much greater than the cost of these innovations.
Keywords: medical innovation; cancer mortality; difference-in-differences; econometrics
JEL Codes: C23; C33; I12; J1; L64; L65; O33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Drug Innovation (O32) | Cancer Mortality Rate (I12) |
| Imaging Innovation (O35) | Cancer Mortality Rate (I12) |
| Decline in Cancer Incidence (I12) | Cancer Mortality Rate (I12) |
| Overall Decline in Cancer Mortality Rate (2000-2009) (I12) | Cancer Mortality Rate (I12) |