Skip to main content

The Opioid Epidemic: information cascades gone wrong

The Cascade of Care framework was first proposed as a policy solution to the HIV/AIDS epidemic in the 1980s by public health officials. More recently, this framework has been revisited and reintroduced as a means of controlling the opioid epidemic that has decimated, in particular, rural white communities across the United States.

Unlike infectious diseases like HIV, the opioid epidemic is characterized by addiction, not contagion. Since the opioid epidemic is not contagious in the way that many epidemics are, but still fits the CDC’s definition of an epidemic, I wanted to explore whether or not the crisis fits the branching process model we discussed in class.

The branching process model defines a simple formula for seeing whether or not a disease or phenomenon will persist beyond fringe cases by multiplying the probability that you get sick upon contact with an infected individual and the amount of people they are exposed to. Logically, a drug addiction epidemic would not fit into this mold because there should be no positive correlation between exposed individuals and infection.

Curiously, we actually see that in the case of opioids, a higher k can actually contribute to better outcomes for affected individuals. This is because when others see the effect of opioid and the dependency that they cause, they are steered away from initially taking them. Furthermore, lack of information about using, particularly if an individual’s opioid usage escalates to drugs like heroine or other intravenous opioids, can expedite fatality due to secondary bacterial infections like blood-transferred Hepatitis.

We actually could observe that for initial outbreak of the opioid crisis, where information was largely concentrated amongst pharmaceutical companies, a more accurate model would be that of information cascades. Physicians and patients initially both did not have enough information to determine that opioid usage could spiral, so we saw pharmaceutical companies’ market false safety information through a sustained campaign of misinformation to medical professionals. Since such a large network of doctors did not have correct information and policy surrounding pharmaceuticals is relatively lax in the US, doctors and their patients were not able to form correct low and high signals. Their knowledge of the drug’s safety was a high signal, meaning that the expected value of the product was positive, despite the drug being bad. In this case, if purchasers computed the Bayes computation of what is the probability the product is Good given it’s high signal, they would get a probability higher than p or probability of being good because of manipulation by pharmaceutical marketing forces.




Leave a Reply

Blogging Calendar

December 2019