Tuesday, June 14, 2016

Prescription Drug Shortages: Data Limitations In An Era Of Big Data

Blog_medicine nurse

In 2012 hundreds of prescription drugs were in short supply in the United States. Over 80 percent of these drugs were generic and many were commonly used as anti-infective, anesthesia, and chemotherapy agents (ASPE 2011; Woodcock and Wosinska 2012). That same year, the Food and Drug Administration (FDA) was given new and expanded powers to monitor and resolve shortages under the Food and Drug Administration Safety and Innovation Act (FDASIA).

A nice recent piece in Health Affairs evaluates trends in drug shortages before and after the passage of FDASIA and by an acute versus non-acute care drug categorization. Using 2001-2014 data from the University of Utah's Drug Information System (UUDIS), which contains the universe of shortages reported to the American Society of Health-System Pharmacists and is the primary data used in numerous government reports, the paper demonstrates an increase in the number and duration of acute care drug shortages since 2012, despite a decline in the number of new drug shortages over the same period.

But what do data from the FDA, the agency that regulates prescription drugs, show? Do they suggest the same patterns? As the authors note, the FDA and UUDIS track shortages for different purposes and in different ways. While the FDA aims to capture market-wide disruptions, UUDIS is intended to support practicing pharmacists and clinicians. The authors do not analyze the FDA data because, until recently, they were not readily available and consistently reported.

Source of Shortage

The data source policymakers and regulators should use moving forward is less clear. A comparison of the current FDA and UUDIS drug shortage list indicates that this point is not simply academic (see Exhibits 1 and 2). Of the 149 drugs on the UUDIS list as of May 5, 2016, only 52 of them (34 percent) were also on the FDA's list. In terms of the acute versus non-acute care drug categorization, 69 percent of those currently on both the UUDIS and the FDA drug shortage lists are acute care drugs, about 10 percentage points higher than the percent on either the full FDA list or the full UUDIS list.

Figure 1. Overlap in UUDIS and FDA Current Drug Shortage Lists

Mireille_Jacobson-Exhib1

Figure 2. Acute vs. Non-Acute Drugs by Source of Shortage

Mireille_Jacobson-Exhib2

Source: Data for Figures 1 and 2 compiled from the UUDIS and FDA.

Among the many important issues that cannot be addressed by comparing shortages lists is which drug shortages have the potential to impact clinical practice or patient welfare. In some cases, for example, shortages affect only a limited form of a drug and clinicians may, with little additional hassle or cost, be able to safely and effectively manage treatment using other forms of the drug. Alternatively, a substitute drug of similar quality and cost may be readily available also limiting the impact of the shortage.

In these cases, the public may need not be alarmed by shortages, although policymakers may still be interested in their underlying causes. At the other extreme, some shortages affect drugs with no comparable clinical substitute and with direct impacts on survival.

Geographic Distribution

A somewhat related issue is the geographic distribution of shortages. As a map of (normalized) Google Trends data for the term "drug shortage" suggests (see Exhibit 3), not all areas may be equally affected by shortages. Indeed, a medical director from our own institution shared with us the experience of fielding calls from colleagues in other parts of the country on how to handle drug shortages that were simply nonexistent here. Regulators and policymakers should know not just when a drug is in shortage but where that shortage is affecting patients or providers. Such information seems crucial for assessing the impact of the shortage and effectively mitigating any negative effects.

Figure 3. Geographic Variation in Google Search for "Drug Shortage"

 Mireille_Jacobson-Exhib3

Notes: Data are from Google trends and cover searches for "drug shortage" from 2004 to the present. Search counts are normalized to DC (100). Map was made at illuminations.nctm.org

This post is not meant to argue that we should not or cannot use existing data or even that one data source is superior to the other. Rather, we have to wonder if we can't do better at little additional cost.

In an era of electronic ordering systems, social media, big data, and predictive analytics, can we build on existing data sources to create a richer, more complete understanding of current drug shortages? What if those with a Drug Enforcement Administration number, a National Provider Identifier, or some other prescription-related credential could text or upload to a website a report on the difficulty of obtaining a specific drug. While such reporting would be voluntary, it could—assuming enough providers participate—serve as an early warning system, alerting regulators and the public not only to what drugs may be in short supply but where those bottlenecks may be occurring, i.e., in what locations and settings.

This is just one of many ways in which we can begin to get a better understanding of where to put our efforts. Without a clear picture of where and when shortages occur, we are severely constrained in determining the causes of shortages. And without understanding the root causes of shortages, a long-term solution to persistent drug shortages seems less likely.



from Health Affairs BlogHealth Affairs Blog http://ift.tt/1WMvf0D

No comments:

Post a Comment