How frequently are pharmaceuticals used off label? Perhaps more than you think. Although these figures are a bit dated, Tabarrok (2000) details the extent of off-label prescribing in the U.S. as follows:
According to a study by the U.S. General Accounting Office, 56 percent of cancer patients have been given non-FDA-approved prescriptions, and 33 percent of all prescriptions in cancer treatment were off-label (General Accounting Office [GAO] 1991). Another survey, of AIDS patients, found that 81 percent of patients received at least one drug off-label, and 40 percent of all reported drug use was off-label (Brosgart and others 1996). Experts have estimated that nearly all pediatric patients (80 to 90 percent) are prescribed drugs off-label (Jaffe 1994; Kauffman 1996; Goldberg 1996).
If on-label prescribing has the best available evidence, why would a physician use off-label prescribing? Tabarrok gives three reasons:
- Evolving Evidence: Best practices may change the standard of care faster than the FDA approves new uses for existing drugs.
- Efficacy may not match individual effectiveness. The treatment currently deemed as “best practice” may still fail to stem a given patient’s disease. Further, “patients with terminal diseases may rationally demand ‘experimental’ treatment.”
- Cost. The FDA approval process for new indications is expensive and lengthy
Tabarrok’s article also highlights the important need to trade off Type 1 and Type 2 errors when regulating pharmaceuticals. He posits that FDA is likely biased towards avoiding Type 2 errors (the approval of drugs that are actually harmful) and is less concerned with Type 1 errors (not approving drugs that are helpful). However, Type 1 errors are clearly costly although receive less attention.
If the FDA approved a drug that killed thousands of people, that story would make the front page of every newspaper in the nation. Congressional hearings would certainly he held, the head of the FDA would probably lose his or her job, and the agency would be reorganized. But if the FDA rejected a drug that could save thousands of people, who would complain? When a drug kills a patient, that person is identifiable, and family and friends may learn the cause of the death. In contrast, the patient who would have lived, had new drugs been available, is identifiable only in a statistical sense.
Clearly, off-label prescribing is helpful in some cases and harmful in others. The key is to maximize the frequency of the former and decreases the frequency of the latter; which is easier said than done.