Commentary on Perry et al.: New means, new measures—without discarding all the previous ones!

Multi-dimensional analyses of complementary data sources are needed to capture the different facets of a multi-factorial and hidden phenomenon such as prescription drug abuse. Starting from the observation that there is little agreement in the literature about how to best operationalize prescription drug-seeking (PDS), Perry et al. compared the performances of multiple PDS indicators during 10 years of the opioid epidemic in the United States [1]. The authors illustrated the value of a network centrality measure (PageRank) that they had introduced recently [2]. The method uses patterns of co-prescription ties between patients and prescribers to identify irregular and excessive prescribing activity that cluster around central actors. While adding one more indicator may at first sound counterproductive, the new indicator proposed by Perry et al. fulfils a gap in the available methods by considering PDS from a ‘social’ perspective. This enables the identification of PDS using information that shows which, rather than how many, prescribers are visited (e.g. those who prescribe high doses of opioids, those who operate pill mills or those who do not monitor PDMP data). PageRank should, however, not be considered as a universal indicator that could replace all its predecessors. It should be considered according to what makes its added value at its root: an innovative indicator addressing PDS from a complementary perspective. It is clear from the testing of a range of so-called traditional indicators presented in the article how difficult it is to quantify PDS. In order to face the challenges of monitoring prescription drug abuse, several authors and health authorities advocate for a proactive post-marketing monitoring using multi-dimensional detection systems and complementary data sources [3-5]. The new indicator proposed by Perry et al. fits into this logic by focusing upon the social aspects behind PDS. Some of the traditional indicators that previously demonstrated their benefits also remain relevant. Among these, overlapping prescriptions from multiple prescribers (i.e. doctor-shopping) and use of high doses are at the core of the safety concern from a medical and pharmacological perspective. Overlapping prescription relies upon circumventing the optimal one-to-one patient–prescriber relationship, resulting in a lack of medical management, because one given prescriber does not know that other prescribers are also prescribing the same drug. The lack of medical management, as well as the use of high doses, increases the risks for harmful outcomes, such as high-risk use, overdose and death [6-9]. In fact, what makes PDS that hard to monitor is that it is most often a hidden behaviour. A direct but often forgotten implication is that there cannot be a gold standard measure that would enable definitive validation of a new indicator. As a result, there is a need for an empirical validation process before using PDS as a proxy for prescription drug abuse. The process must assess the external validity of the proxy for several pharmacological classes, accounting for the specificities of the health system of interest (e.g. availability of prescription drugs and illicit alternatives, cost of prescription drugs and visits, prescription and control methods and risks involved for fraud). As an illustration, we recently provided a review of two decades of development and validation of a doctor shopping indicator used as a complementary component of the multi-dimensional monitoring conducted by the French Addictovigilance Network [10]. The validation process relied upon linking doctor-shopping to numerous determinants of prescription drug abuse, such as pharmacological data (e.g. formulations and doses [11]), chronological data (e.g. impact of measures [12]), geographical data (e.g. comparison between regions [13]) and health outcome data (e.g. profiles of patients [14]). Finally, COVID-19 outstandingly illustrated how multiples indicators are needed for monitoring a pandemic, which is what prescription drug abuse is. There cannot be a universal indicator to assess such a multi-factorial public health issue. Complementary indicators are needed to capture the different facets of this multi-factorial phenomenon and overcome the potential limitations of any proxy of PDS taken individually. None. Thomas Soeiro: Conceptualization. Joelle Micallef: Conceptualization.