Clinical studies of specific new drugs have shown that these drugs increase survival rates. Here are three examples:
• Stenestrand et al studied the impact on survival of statin treatment following acute myocardial infarction. They found that 1-year mortality was 9.3% in the no-statin group and 4.0% in the statin treatment group.
• Grier et al found that adding two experimental drugs to the standard four-drug chemotherapy regimen has significantly improved survival in patients with non-metastatic Ewing’s sarcoma, a highly malignant bone cancer of children and young adults. The overall survival rate increased from 61 percent to 72 percent for Ewing’s sarcoma patients with localized disease who underwent the experimental six-drug chemotherapy.
• The journal U.S. Pharmacist reported that patients suffering from advanced metastatic melanoma who were treated with a combination of an investigational agent, Ceplene, and interleukin-2 (IL-2) had twice the survival rate as patients who were treated with IL-2 only. The patients were enrolled in a three-year study. The study also showed that the Ceplene/IL-2 combination significantly increased survival in a subpopulation group of advanced metastatic melanoma patients with liver metastases. The rate of survival in this group was six times that of the group given IL-2 only.
Also, I have performed several studies using aggregate data (Lichtenberg ) that indicated that the introduction of new drugs has increased longevity. The objective of the present study is to examine the impact of the vintage (original FDA approval year) of drugs used to treat a patient on the patient’s probability of survival, using micro data on virtually all drugs and diseases from Puerto Rico’s Medicaid program, which covers about 1.5 million people.
To determine the effect of the vintage distribution of a person’s prescribed medicines on probability of death, conditional on demographic characteristics (age, sex, and region), utilization of medical services, and the nature and complexity of illness, I will estimate the following model:
Suppose individual A consumed only medicines approved in 1985. For that individual, P0ST70 = P0ST80 = 1, and P0ST90 = 0. Hence E(DIEDa | Za) = P1970 + P1980. Suppose individual B consumed only medicines approved in 1995. For that individual, P0ST70 = P0ST80 = P0ST90 = 1. Hence E(DIEDb | Zb) = P1970 + P1980 + P1990, and E(DIEDB | ZB) – E(DIEDa | ZA) = P1990. The parameter P1990 may be interpreted as the difference between the death probability of people consuming only post-1990 medicines and that of people consuming only pre-1991 medicines. More generally: