Planned and unplanned disruptions in production of drugs are behind an overwhelming share of drug shortages. Stockpiling the finished product has been proposed as a solution for preventing future shortages. The general idea is that the federal government would purchase and stockpile select drugs, just as it does with vaccines and anti-bioterrorism agents. This would counterbalance the relatively low inventory levels currently in distribution channels. In contrast to how Medicare participates in the sterile injectable drug market, where it pays providers rather than manufacturers, here the government would purchase the drugs directly from manufacturers.
Presumably a stockpile could permanently lift the demand for these drugs if the government were to pay prices higher than the going market rate and/or the government were to buy additional drugs above and beyond those purchased by healthcare providers and patients. By increasing demand for the target drugs, a stockpile would theoretically increase incentives for manufacturers to expand production capacity, thereby reducing the likelihood of a shortage. And if a drug shortage were to occur, the federal government would release the stockpiled drug for use by the general public.
This concept is appealing for its presumed simplicity and also for the fact that it builds on an existing and well-accepted blueprint with the existing federal government’s Strategic National Stockpile of vaccines and bioterrorism agents. However, setting up a stockpile that will effectively help address future drug shortages is not feasible in the current fiscal and political environment. Here’s why.
The Problems Of When, And For What Drugs, To Develop Stockpiles
First, a stockpile has to be developed at a time when industry’s capacity is not constrained. For this reason a stockpile is not a solution to the current drug shortage problems with sterile injectables, which are driven by a sharp drop in available production capacity across the entire industry. Until there is ample spare production capacity, diversion of product into a stockpile would only compound the problem. While capacity remains limited, shortages could be triggered not only for drugs being stockpiled but also for other drugs that are produced on the same product lines. Yet once sterile injectable producers expand and improve their existing capacity as currently planned, the risk of these drugs going into shortage will drop, lowering the need for a stockpile in the first place.
Second, it is difficult to predict which drugs are likely to experience a shortage and therefore need to be stockpiled. The scope of products historically affected by shortages is very wide, and adding this range of products to the stockpile would be very expensive. To lower the price tag, it would be paramount to narrow down the variety of products held in the stockpile, but then the likelihood of having the product on hand in the event of a shortage would also decline.
Both of these problems are difficult to solve. For example, one way to reduce the number of drugs that would need to be held in the stockpile is to focus on drugs that are currently most vulnerable to shortages: sterile injectables. Narrowing the pool to sterile injectables yields 569 distinct molecules, some of which would likely need to be stocked in different doses. Some have suggested limiting the stockpile to a specific therapeutic class, such as oncology. Determining which therapeutic areas are in greatest need of a stockpile would provide an interesting challenge of assigning more value to one area of therapy than another.
On a more pragmatic level, historically shortages have ‘travelled’ across therapeutic classes. For example, central nervous system drugs dominated shortages in 2008-2009 with oncology drug shortages rising in prominence only in 2010. This is a reflection of the fact that manufacturing problems pop up in one facility one year and then one next year.
But could we not forecast and then stockpile what is likely to be in short supply? The problem here is that, in order to avoid worsening the shortage, this selection process would have to take place well before a supply disruption occurs. Unfortunately, it is difficult to predict supply disruptions long in advance because key indicators for increased risk of supply disruption, such as shrinking production capacity or manufacturing problems, reveal themselves too late to divert quality product into a stockpile without disrupting the market. If we try to forecast shortages far in advance by looking at stable leading indicators, such as the existence of only a small number of manufacturing lines producing a product, we would not narrow the list of stockpiled drugs sufficiently. For example, 342 sterile injectable molecules are produced in markets where the top producer has over 90 percent market share, often produced on one or two production lines.
The Troublesome Cost-Access Trade-Off
Last, but not least, there is an additional trade-off between cost and access. A large inventory of any drug will increase the likelihood that the supplies are sufficient to make up a shortage for that particular drug, but it will require both a higher price tag and the stocking will have to be done at a time when capacity is plentiful. If the government does not stock for the worst-case scenario, it will need to set new standards for allocating the drug were a shortage to occur. However, existing priority use standards for already stockpiled vaccines and bioterrorism agents (healthcare workers, then pregnant women, children and elderly for vaccines) are not easily applied to other drugs. Would today’s public have the appetite for government deciding which cancer patients get treatment?
The bottom line? For current shortages, building a stockpile would only further exacerbate the problem. Looking into the future, the ability to prevent and relieve future drug shortages must be traded off against its cost. A broad-based approach that includes a broad range of drugs is prohibitively expensive and inefficient, while a targeted approach will miss many drug shortages.