Prescription Drug Monitoring Programs to Curb Prescription Drug Abuse: Examining the Components of Program Efficacy

By Joshua J. Timmons
2014, Vol. 6 No. 05 | pg. 3/4 |

Information Sharing Between States

As state adoption of PDMPs become commonplace, the need for interstate data sharing becomes vital. It is not enough to monitor the doctor shopping habits of a population in one state, when all it takes to avoid detection is the crossing of state lines. A survey of Kentucky’s PDMP data by Hopkins (2012) showed that people were picking up drugs at Kentucky pharmacies with prescriptions from every other state in the country. For this reason, the potential exists for Kentucky residents to simply doctor shop in other states. Furthering the example, almost 6% of Kentucky’s filled prescriptions from neighboring states alone (COE, 2011). Doctor shopping in adjacent states has emerged as an issue of heated discussion and is now a concern for every unconnected PDMP in the country.

To prevent prescription drug abusers from crossing state lines to circumvent the triggering of red flags in their home state, PDMPs need to adapt to share prescription drug information. A PDMP will only be effective if it can take into account the more ambitious of doctor shoppers. An early PDMP review from 2002 demonstrated there was a “spillover effect” from Kentucky instituting an aggressive PDMP: after its creation, rates of OxyContin diversion worsened in three neighboring states (USGAO, 2007). The nationwide solution to spillover has been a network of hubs that share PDMP data between states. In this way, if a patient sees a doctor in California on vacation, his primary care physician in Ohio will be aware of any prescription he may have received while away. Moving forward, it is imperative that every state links with the emerging information exchange.

In the early discussions of PDMPs sharing data, the problem arose that every state created their program independently. Some states used private tech firms at the same time that others were creating theirs in house (United States Department of Justice, n.d.) For obvious reasons, the variation in technology and software resulted in PDMPs that were, more often than not, incompatible. In order to get around these inherent technological discrepancies, a nationwide standard needed to be created. To this end, the Bureau of Justice Assistance created the Prescription Monitoring Information Exchange (PMIX) architecture. Software architecture can be defined, by the IEEE Computer Society (2000), as, “the fundamental organization of a system embodied in its components, their relationships to each other, and to the environment, and the principles guiding its design and evolution.” By creating this architecture, the federal government laid the groundwork for a common system to bridge all states.

At its core, the PMIX architecture was created with the intention of making data sharing free and open to states that wish to use it, common with its formatting, secure through double sided encryption, and not limiting in states’ choice of hub provider. Altogether, the PMIX includes specifications on state to state and hub to hub service, security standards, and system governance (Vogt, 2013). Despite strict guidelines, the PMIX remains flexible and grants states the greatest degree of freedom possible; as such, states may create their own data hubs using whatever technology they choose, and with whichever other states they decide (so long as it conforms to the architecture specifications).

Advertisement

To address the need for uniform data collection, the American Society for Automation in Pharmacy (ASAP) created a nationwide standard. The ASAP standard is, for all practical purposes, a specific type of data formatting. It defines what information prescription providers record; payment method being one example. Although subject to updates, the current version is ASAP 4.2. Every PDMP has some form of ASAP standard, but only 23 have 4.2; 17 have 4.1 and the rest have an earlier version (American Society for Automation in Pharmacy, 2014). While it is important for every PDMP to maintain the most recent version, this should only become a concern after other precursory steps, necessary to join the PMIX, have been taken. These include building a hub infrastructure and coordinating with other states through legal channels. Despite technical challenges, PMIX use has been strong thus far. Between 2011 and 2012, with only 9 states on the system, 265,000 data requests were processed; as of October 2013, two and half million requests have been processed for 16 states– with more to join soon (National Association of Boards of Pharmacy, 2012).

Breaking away from system configuration, another topic encompassed in interstate data sharing is the threshold for “questionable” behavior. Some states may determine that prescriptions from 5 different physicians are a red flag, while others say 7. This becomes especially important if an abuser has prescriptions from 6 different prescribers but is living in a state that requires 7 to constitute “questionable” behavior. Should the individual be blacklisted in states with lower thresholds, or is he/she granted immunity by living in a state that makes his/her doctor shopping actions permissible? The remedy for this is a national standardization for what constitutes doctor shopping. If states are going to collaborate in the exchange of PDMP data, it is important for the line in the sand to remain uniform among all states.

An effective threshold would need to be based upon statistical findings from emerging PDMP data. The bar was set by McDonald and Carlson (2013) at a rather high 32 different opioid prescriptions from 10 different prescribers. This number was the extreme and constituted only 0.7% of all opioid purchasers; a national threshold would need to be more inclusive. Recall that an estimated 4.6% of all Americans abused painkillers between 2010 and 2012 (SAMHSA, 2013).

A more widely referenced model was created by White, Birnbaum, Schiller, Tang, & Katz (2009). In their research– the goal being to create a tool capable of predicting nonmedical use– they found that over a three month period, the following factors were strongly correlated with what is medically considered abuse: “age 18 to 34 years, male sex, 4 or more opioid prescriptions, opioid prescriptions from 2 or more pharmacies, early prescription opioid refills, escalating morphine sulfate dosages, and opioid prescriptions from 2 or more physicians.” The researchers went on to link 12 or more prescriptions, 3 or more pharmacies, being male, being between the ages 18 and 24, having psychiatric outpatient visits, and getting early prescription opioid refills– over a year long period– with abuse. A model such as this could allow prescribers to focus on instances that warrant attention, and prevent PDMPs from wasting resources on instances that do not.

A concern that may be raised is whether some groups, particularly men between 18 and 34, will be targeted by PDMPs. Perhaps some patients in need of prescription medication will lose access to invaluable treatment due solely to an algorithm. Objections may be raised that PDMPs will turn a blind eye to women– an understandable concern given that opioid abuse is strongly on the rise among women (CDC, 2013b). The response of a policy maker should be that the line needs to be drawn somewhere. The groups at highest risk to use prescriptions nonmedically should warrant extra attention and raise PDMP “red flags” most easily. In effect, a national threshold needs to be established– to unify interstate efforts to curb doctor shopping– and the most effective route to making one will be the use of a statistically based model.

Although the issues of system framework, ASAP version, and doctor shopping thresholds are significant, the utility of a cross-state prescription drug firewall makes the initial investment– of both capital and discussion– worthwhile. The states that have yet to build the infrastructure to join the PMIX and adopt the most recent ASAP standard should consider it a valuable component to their PDMP if they are committed to cutting down on drug diversion. Additionally, state PDMP leaders need to meet in order to create a threshold for what constitutes doctor shopping. All other PDMP components aside, a monitoring program incapable of stopping shoppers that cross state lines is hardly a monitoring program at all. As every state completes the installation and enacting of PDMP legislation, the inclusion of PMIX compatibility becomes all the more critical.

Data Collection Timeliness

The prescribing and picking-up of prescriptions are the two largest areas of focus for a PDMP. When the prescription was made, who prescribed it, when the prescription was filled, and where it was filled all constitute important PDMP data. Congruently they make up a set of knowledge that gives clinicians, pharmacists, and law enforcement the relevant information that they need to do their job effectively. Understandably, the sooner that these points of data are available the more valuable the data becomes. As an illustration, imagine the example of a doctor shopper. The quicker information regarding the shopper enters the state’s PDMP, the more rapidly relevant prescribers and providers can respond. On the other hand, a PDMP with a sizable delay is by all means handicapped. In essence: the faster the better.

In the effort for timely data entry, Oklahoma is a standout leader. Throughout 2009, Oklahoma ranked the highest among all states for nonmedical consumption of pain relievers (Oklahoma Department of Mental Health and Substance Abuse Services, n.d.). Given the extent of their issue, they were the first state interested in switching to a real time, or “point of sale” system. In 2012, they set out with the audacious goal of making their PDMP responsive to within just 5 minutes from the moment of payment at a pharmacy. Fortunately, support for real-time updating was included within ASAP 4.1, the data standard that Oklahoma’s PDMP was already using, but there were still some hurdles the state faced before it could get the system down to a 5 minute response time (Vogt, 2012). Retail drug chains, regulatory agencies, independent retailers, and software providers were all active in the transition and held to a negotiated deadline.

Despite the complication of getting all relative parties on board, the cost of conversion was small: an advanced estimate put the cost at just twenty one thousand dollars (Vogt, 2012). Today, Oklahoma exists as one of two states with a real time PDMP. The fact that Oklahoma was able to accomplish the transition to point of purchase– at a cost just over twenty thousand dollars– shows that other states should be able to upgrade their response systems as well. Although the lasting impact of Oklahoma’s switch has yet to be thoroughly studied, due to recentness, one Canadian province studied the effect of its own switching to point of purchase.

In British Columbia, their “centralized prescription network” became real time all the way back in 1995. Using old prescription records from between 1993 and 1997, Dormuth, Miller, Huang, Mamdani, and Juurlink (2012) were able to analyze the effect of real time data collection by comparing nonmedical opioid use before and after the change. For their analysis, “nonmedical use” was defined as prescription of opioids from two different physicians, at two different pharmacies, within a week of one another; study participants were grouped into either an “appropriate” or “inappropriate” category.

Advertisement

During the 30 months before the institution of real time monitoring, 3.2% of opioid filled prescriptions were deemed inappropriate (among a large population of residents receiving social assistance). Once the real time system was implemented, there was a sharp drop in inappropriate opioid use followed by a steady fall throughout the following 30 months; overall, there was a 32.8% decline. (Dormuth et al., 2012) The lesson learned was that not only is there a noticeable correlation between data updating timeliness and nonmedical prescription use, but a rather sizable one.

There is another appeal to point of purchase systems that comes from an issue that transcends the fight against doctor shopping and pill mills. Between 2004 and late 2005, there were an approximated seven hundred thousand visits to ERs from adverse prescription reactions (Budnitz, 2006). Of these ER visits, 32.1% were from accidental overdoses and 28.6% were from adverse effects. For emergency physicians, a recurring question is “did you take any medication?” To have the best clinical judgment, physicians need up to date information of what drugs their patient was prescribed, when the prescription was last filled, and who prescribed it to them. All of these facts can aid an emergency department– especially is the patient is unresponsive, as nearly 3% of ER visits are (Weisberg, Strub & Garcia, 1983) – but only if the information is accurate and up to date. A delay of even a week could prove catastrophic if, for example, a patient begins to experience complications after taking their medication for the first time.

Emerging research, such as Lammers, Adler-Milstein, and Kocher (2013), has suggested that the more information available to emergency physicians, the better decisions they tend to make: when emergency departments have access to health information exchanges, they tend to avoid duplicate testing and make sound medical decisions for their patients. In a similar vein, real time PDMPs would augment the medically actionable knowledge available in emergency situations. Using PDMPs as an additional tool, ER doctors would be able to know not only whether their patient is in the ER because of drug seeking behavior, but also what medication may have led to adverse reactions. Real time updating is also relevant when considering that one of the biggest issues among doctors reluctant to use their PDMP is the lack of updated information (Doland, 2012). For this reason, the widespread adoption of point of purchase could bring in a large pool of clinicians that would otherwise remain on the fence.

Currently, only New York and Oklahoma have adopted real time or point of purchase systems; this is especially disappointing since altogether 40 states have ASAP 4.1 or 4.2, and both support real time data collection. More surprising still is that five states– Alaska, New Mexico, Pennsylvania, Rhode Island, and South Carolina– only update once per month (PDMP TTAC, n.d.). The implication is that doctor shoppers in those states have an entire month to procure prescription medications before any prescriber can catch on. As demonstrated by British Columbia, the usefulness of a PDMP can, to a large extent, be linked to the speed of its data collection. Among existing PDMPs, the usefulness of real time data collection has flown under the radar thus far, but as PDMPs continue to evolve, smaller data collection intervals should become a top priority. Ideally, within the next decade, all PDMPs will be real time and accurate to within just a few minutes.

RELATED ARTICLES