Drug shortages are a potential threat to public health. According to the World Health Organization, drug shortages are becoming, if not are already, a global problem (Gray and Manasse, 2012; WHO, 2016). Countries all over the world have reported drug shortages and subsequent problems that affect both developing and developed countries (Lynas, 2013; Quilty, 2014; Chen et al., 2016; Rosa et al., 2016; De Weerdt et al., 2017). Shortages have resulted in problems with medication such as errors, rationing, delayed supply, and denial of treatment in some cases (Fox et al., 2014; Vail et al., 2017). Shortages not only impact patients; health care professionals are confronted with lack of information on shortages and need to find proper solutions, causing pressure on primary and hospital care (McCartney, 2015), as well as an increase in costs (Iacobucci, 2017). Reliable data on drug shortages is limited. Scientific publications are mainly based on narrative reviews (De Weerdt et al., 2017), survey results (Pauwels et al., 2015), or anecdotal evidence from studies of specific drugs (Malone et al., 2016). The growing problems due to shortages have been addressed in public media as well (Koba, 2015; Fink, 2016). The media has especially focused on the harmful effects of shortages for patients.

When shortages occur, authorities struggle with whether and how to intervene. In the United States, the Food and Drug Administration (FDA) has received new and expanded regulatory powers to respond to national shortages by the Food and Drug Administration Safety and Innovation Act in 2011 (McCarthy, 2013). According to this act, a marketing authorization holder (MAH) has to notify the FDA of a discontinuation or interruption when it will lead to a meaningful supply interruption of a drug that is life-supporting, life-sustaining, or intended for use in the prevention or treatment of a debilitating disease or condition. They have to notify the FDA at least six months in advance or as soon as possible, but not later than five business days after the start of discontinuation or interruption. If an MAH fails to submit this information, letters are made public on the website of the FDA (Food and Drug Food Drug Administration, 2015). It seems that these measures have had impact. The number of shortages in the United States decreased by 24% in 2012 and by 60% in 2013, respectively (Fox et al., 2014). European legislation states that “the holder of a marketing authorization for a medicinal product … shall, within the limits of his responsibilities, ensure appropriate and continued supplies of that medicinal product to pharmacies” (The European Parliament the Council of the European Union, 2011). The European Medicines Agency has been involved in several drug shortages, in case the shortage was linked to a safety concern and multiple Member States were affected. However, most drug shortages are handled at a national level by national competent authorities (European Medicines Agency, 2013). In France, the French legislator chose to introduce new measures to the law in January 2016 to fight drug shortages. MAHs are now legally bound to ensure sustained supply and stocks of drugs they market in France to meet needs of patients (Bocquet et al., 2017). In the Netherlands, the Medicines Act of 2007 states that “a marketing authorization holder takes care that a medicinal product with a marketing authorization is in stock in sufficient amount for wholesalers and pharmacists” and the MAH “must notify the authorities of interruption of a medicine. Foreseen shortages must be reported 2 months in advance and unforeseen as soon as possible” (Dutch Government, 2007). Despite this legislation, pharmacies face drug shortages regularly, which makes it important to monitor these shortages at a national level and to advise on potential solutions. The Royal Dutch Pharmacists Association (KNMP) launched a public website to inform pharmacists about new shortages and about the status of ongoing shortages in 2004. Similar initiatives have been launched inside (Bochenek et al., 2017) and outside the EU. The University of Utah's Drug Information Service (UUDIS), for example, has a comparable website which is hosted by the American Society of Health-System Pharmacists (ASHP) (Fox et al., 2009).

All these initiatives to monitor and manage drug shortages, legally bounded or not, show the critical importance of timely and high-quality information on potential shortages throughout the supply chain. Only when signals on potential drug shortages are tagged in an efficient and adequate manner, authorities and health care professionals can act accordingly. This study aims to evaluate the added value of signals on potential shortages derived from two different sources: authorities and pharmacy practice. Therefore, we examined the extent and nature of potential drug shortages signaled by authorities and pharmacy practice in the Netherlands.

Authorities detected signals on 4,154 potential shortages; 2,128 (51%) signals were related to products marketed in the Netherlands. Most signals (n = 2,991; 72%) were on permanent shortages. The overall number of signals to authorities decreased during the study period from 1,366 to 880 per year (Figure 1), despite an increase in the number of signals on temporary shortages. Pharmacy practice detected signals on 1,577 potential shortages (Figure 2). Most signals (n = 1,140; 72%) were for temporary shortages. The overall number of signals to pharmacy practice increased during the study period from 340 to 495 per year due to an increase in signals on temporary shortages (Figure 1). During the study period, authorities detected in total 2.6 times more signals on potential shortages than pharmacy practice.

Of a total of 5,731 signals on potential shortages detected by both authorities and pharmacy practice during the study period, only 438 (8%) overlapped (Figure 2). Overlapping signals were for temporary shortages (n = 251; 57%) and for permanent shortages (n = 187, 43%). The median difference in reporting dates for permanent shortages between authorities and pharmacy practice was −180 days (interquartile range [IQR]: −8 to −405 days), meaning that pharmacy practice detected the overlapping signals earlier than authorities (Figure 3). Overlapping signals on temporary shortages were also detected by pharmacy practice before they were detected by authorities, with a median difference of −1 day (interquartile range [IQR]: 23 to −203 days).

Our results show totally different perspectives on potential drug shortages when signals from authorities and pharmacy practice are compared. Comparing both sources, we found a very small overlap in signals (8%), different types and characteristics of the affected products and a differential timing of signals. Whereas authorities faced a decrease in number of signals over time, pharmacy practice faced an increase in number of signals. For authorities, a typical signal as identified in this study would be the permanent shortage of a rectal product or an anti-infective for systemic use. For pharmacy practice, a typical signal would be a temporary shortage of an ocular product or an anti-infective for systemic use. Signals to authorities were likely detected by pharmacy practice as well, but the latter detected it earlier. Signals to pharmacy practice may have been detected by authorities as well, but this would not be very likely.

Authorities detected signals on permanent shortages much later than pharmacy practice. Although signals on temporary shortages were detected with a median difference of 1 day, results show a larger variation toward the delay to authorities (Q₃ = −203 days) than delay to pharmacy practice (Q₁ = −23 days). Since early signaling to authorities is considered critical in preventing and mitigating a drug shortage (Fox et al., 2014), authorities were insufficiently able to intervene during problems in pharmacy practice. Combining both databases gives much more insight and potential to timely action.

The major strength of this study is the direct comparison between signals on potential shortages to authorities and to pharmacy practice for a study period of several years. Authorities in many countries may have data on drug shortages but only few countries collect data from pharmacy practice at a national level and have historical data. It could be argued that data from pharmacy practice mainly relate to local shortages. However, 90% of these signals in our study were confirmed by the MAH as a shortage at a national level. The best available data on drug shortages originate from the American UUDIS/ASHP on pharmacy practice and from the FDA authority (Fox et al., 2009, 2014). These data are comparable to the datasets in the Netherlands since the Dutch as well as the American data report shortages at a national level which have been confirmed by MAHs (American Society of Health-System Pharmacists, 2018; Food Drug Administration, 2018). A direct comparison between the American databases on shortages was only made for a short study period, January 2013 through March 2013. FDA and UUDIS identified a different number of shortages; 17 and 39 respectively. Only 8 shortages overlapped and 5 of these 8 shortages were signaled 6 to 50 days earlier by FDA than UUDIS (US Government Accountablility Office, 2016). UUDIS and FDA have different perspectives on the definition of a shortage, organizations' objectives and the data at their disposal, resulting in different numbers (Wosinska and Jensen, 2015; US Government Accountablility Office, 2016). The UUDIS found (Chen et al., 2016) that 55% of their shortages affected parenteral products and the most common classes of drugs were central nervous system agents (17%) and anti-infective agents (14%). These latter results differ from our findings, which may be explained by the fact that we studied signals on potential shortages, whereas UUDIS studied shortages. Also, shortages reported to the UUDIS are published on a website which is hosted by the ASHP which represents pharmacists who serve in hospital settings. As expected UUDIS signals relatively more hospital products. KNMP detected signals from hospital-based and community-based pharmacists.

This study clearly shows a difference in signals on potential shortages detected by authorities and pharmacy practice. Half of this gap (2,026 signals; 51%) is due to signals on products which were never marketed in the Netherlands and therefore could not affect pharmacy practice. Another part of this gap is probably explained by the different sources of signals, perspective of authorities and pharmacy practice and their definition of a shortage. Authorities detect signals from MAHs whereas pharmacy practice detects signals from pharmacists. MAHs may be reluctant to signal potential drug shortages due to logistical and marketing reasons, since potential shortages might disrupt market shares and could be regarded as a negative message to communicate. Not reporting of mandatory signals has not been sanctioned by authorities so far, since they have been focusing primarily on finding solutions for shortages and not on reporting (Dutch Government, Ministry of Health, Welfare and Sport, 2017). Pharmacists may be reluctant to signal potential drug shortages in view of the voluntary nature of the system. Pharmacists probably signaled potential shortages with a large practical impact (to be solved by replacing with a different active substance) more frequently. Potential shortages with less impact (to be solved by replacing with a product with the same therapeutic active substance) were probably signaled less frequently.

Whereas authorities focus on shortages with patient impact at the population level, pharmacy practice handles continuity of care at the patient level. It is known that the definition of a drug shortage depends on the type of organization (De Weerdt et al., 2015) and can be interpreted differently (Barlas, 2014). In addition, not all signals become a shortage, and the impact of a shortage may differ between products and between stakeholders. Further research is needed to provide more insight into which signals have led to a shortage as well as the impact of these shortages at population and individual levels.

Two recent developments may narrow the observed information gap. First, the definition of a drug shortage recently became more uniform for MAHs. At an European level, MAHs together with all other organizations in the pharmaceutical supply chain agreed in 2016 on the definition of a suspected drug shortage as: “the inability of a community or hospital pharmacy, as a result of factors beyond their control, to supply a medicinal product to a patient within a defined period, for example 72 h” (AESGP et al., 2017. Also, Dutch authorities agreed on a broader definition of a shortage in 2017: “an interruption of availability of a medicinal product that is a burden for patients and where the burden is greater than is the case for regular (generic) substitution” (Dutch Government, Ministry of Health, Welfare and Sport, 2017). Both agreements may lead to more similarity in signals to authorities and to pharmacy practice. Second, MAHs can report to the newly established Medicine Shortages and Defects Notification Center of the Dutch authorities since 2017. This center enables MAHs to report at a central point, therefore providing a quicker and more complete overview to authorities. Future research is needed to examine the impact of these developments.

Authorities and pharmacy practice were informed on drug shortages differently with little overlap. In order to gain maximum insight in drug shortages at a national level combining data from both authorities and pharmacy practice seems to be necessary. No single source can give a complete picture. Moreover, the finding that authorities were informed later than pharmacy practice, raises concerns in terms of opportunities for authorities to assist practice to sort out solutions for drug shortages.

The datasets for this manuscript are not publicly available, since data is related to specific products and manufacturers, which is confidential information. No additional data available.

All authors were involved in the design of the original protocol and amending the protocol. Data entry from authorities was carried out by the Inspectorate and MEB and from pharmacy practice by KNMP. DP extracted and cleaned the data and analyzed the data with input from CG and PDS. DP wrote the first draft of the manuscript with AM. All authors contributed to subsequent and final drafts. AM is guarantor of the paper.