Skip to main content

The Aftermath of a Drug Withdrawal: Modeling Spillover Effects Across Countries and Across Categories

Periodic Reporting for period 1 - WITHDRAWAL AFTERMATH (The Aftermath of a Drug Withdrawal: Modeling Spillover Effects Across Countries and Across Categories)

Reporting period: 2015-09-01 to 2017-08-31

The pharmaceutical market is one of the most important industries in the global economy. Projections for the next five years are that global spending on medicines will grow to 1.5 trillion US$ in 2021. Marketing is an important factor in this pharmaceutical industry. There are unique characteristics of pharmaceutical marketing (e.g. patient welfare optimization instead of profit maximization) that require industry-specific knowledge development. In this project I focused on an important and complex element of the pharmaceutical market, i.e. drug withdrawals. A drug withdrawal is when a drug is taken off the market because it is considered to be harmful. Usually, this withdrawal is prompted by unexpected adverse effects that were not detected during clinical trials. During the project, I looked in particular to the aftermath of drug withdrawals for the pharmaceutical industry, i.e. spillover effects. There are two types of spillover effects I considered in this project: (i) across drugs (i.e. do physicians start prescribing the competing drugs or is there a negative spillover effect on the category and/or focal firm?), and (ii) across countries (i.e. if a country lags in withdrawing a drug, does the withdrawal in leading countries spill over?).

Drug withdrawals show the flip side of pharmaceutical marketing. The growing pharmaceutical market increases the pressure for the regulators to approve new medicines and for firms the pressure to market their drugs to stay ahead of the competition. These forces contribute to the fact that some drugs were approved too early, and had to be withdrawn. In the years 2000-2010, there have been almost twice as many major withdrawals, compared to the decade before (respectively 28 and 15 withdrawals). On top of that, the pharmaceutical market has a sensitive nature, as it concerns the health of people. Together this gives pharmaceutical marketing a negative image. These aspects can give drug withdrawals disastrous consequences. The consequences are obvious for the focal drug, as sales drop to zero and the withdrawal often leads to a series of lawsuits.

However, it is unlikely that the effect of a withdrawal stops at the focal drug, as there are likely to be spillover effects to competing drugs in the same category, and to drugs of the same manufacturer. These spillover effects can be positive (e.g. additional sales from the withdrawn drug) or negative (e.g. less sales due to a negative image of the drug category or firm). There are some studies looking into the aftermath of a withdrawal from the perspective of the focal drug, but little research has been done on the spillover effects. In this project, I focused on these spillover effects. For example, a withdrawal (e.g. the statin Cerivastatin) in drug category A (i.e. statins) by firm X (i.e. Bayer) can have a spillover effect on the prescription behavior of doctors of other drugs in the same category (e.g. Lipitor, Crestor), but also related categories (e.g. fibrates), or other drugs made by the same firm (e.g. other cardiovascular drugs by Bayer).

Next to patients, doctors and firms there is another interest group that is important for the behavior before and after a drug withdrawal, namely the regulation agencies (e.g. FDA in US and EMA in Europe). Since drugs are often sold in multiple countries, it is interesting to see if, due to different regulatory systems, there are differences in withdrawals over countries. This leads to another form of spillover effects, i.e. across countries. For example, if a drug is withdrawn in the US, it is likely to affect sales/prescriptions in Europe, even if the drug is still on the market. An example is Cisapride, which was withdrawn in US, UK, and Germany, but was kept on the market in Australia and parts of Europe. The reason was that in these countries the regulations were stricter on the doses, and there were no alternative drugs available. This shows the heterogeneity over countrie
I started the project by looking at some high profile withdrawals and cleaning an existing dataset. These two actions gave insights showing that some countries are more often leading/lagging in withdrawing than others. However, in the dataset of sales there were too few withdrawals to generalize these cross-country findings. In the rest of the project I hence focused on the cross-category spillover, for which I collected additional data focusing on one country. The benefit is that this dataset is almost complete on the category dimension, allowing for generalization over multiple categories and multiple withdrawals. I developed a model to generalize category spillover effects. This model allows to model the dynamics of withdrawals (before as well as after the withdrawal). Further, this model can include spillover effects and the dynamics between sales and marketing.
The insights from the high profile withdrawals and data cleaning process show that some countries are more often leading/lagging in withdrawing than others. This is interesting from a regulator’s perspective. The category spillover model can add additional perspectives on the differences in spillover effect this has on related drugs. These findings are interesting from a managerial perspective. In a follow up of this project, I like to collect more data on a wider set of categories for multiple countries. With the additional data I can expand the model and link country and regulation characteristics to the differences in spillovers, where I can give insights in the balance between the increasing pressure to introduce new and better drugs, and on the other hand, avoiding drugs where the risks outweigh the benefits. The latter is of course of paramount importance for society as a whole.