California established the first PDMP in the United States in 1939. California’s PDMP, the Controlled Substance Utilization, Review, and Evaluation System (CURES), introduced a searchable client-facing component in 2009 allowing prescribers and pharmacists to search the database and generate reports of patients’ prescription histories. In 2013, funds were provided from the state for CURES to streamline the application process. At the same time, the California Department of Justice implemented further improvements, including automatic patient safety alerts and improvements to patient reports generated in the system. This bundle of updates, known as CURES 2.0, was implemented on January 1, 2016. Mandatory registration went into effect on July 1, 2016 for all clinicians, defined as any prescriber or pharmacist licensed to prescribe or dispense controlled substances in California.

This study used the following data from CURES: prescription records for all Schedule II, III, and IV controlled substances dispensed by California pharmacies; registration dates for all prescribers and pharmacists registered to use the PDMP; and records of user-generated reports of patient prescription histories between 2010 and 2017. Data in the PDMP begin in 2009 and track physicians, pharmacists, dentists, nurse practitioners, physician assistants, and others. A full list can be seen in Appendix A (available as a supplement to the online version of this article at http://www.ajph.org). The prescription records reported by pharmacies included drug information, prescriber and pharmacy name, address, and the prescriber DEA number associated with that prescription; prescriber license numbers are not reported. PDMP registration data included prescriber name, DEA number, prescriber license numbers, and clinician registration date. We used PDMP utilization data to calculate the monthly number of patient prescription reports created by each PDMP user. We performed all analyses using R version 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria).

Registration for both prescribers and pharmacists accelerated as the mandatory registration deadline approached. From January to December 2016, the number of registered prescribers increased 174%, from 44∣413 to 121∣895. Over the same period, the number of registered pharmacists increased 63%, from 23∣752 to 38∣789.

We observed a corresponding increase in the percentage of prescribers and pharmacists registered for the PDMP (Figure 1). The percentage of registered eligible pharmacists increased from 54.2% in January 2016 to 87.1% in March 2017, with an increase of 14% in the month before mandatory registration was implemented. The percentage of prescribers who were registered increased from 29.5% to 67.2% during the same period, with an increase of 16% in the month before mandatory registration was implemented. The percentage of registered prescribers and pharmacists leveled off immediately following implementation of mandatory registration.

The monthly number of patient prescription reports created in the PDMP increased substantially in 2016 following a trend similar to that seen for registration. The monthly number of patient reports generated by prescribers increased 96% (from 167∣793 reports per month to 328∣339 reports per month) for prescribers and 39% (from 342∣140 reports per month to 475∣264 reports per month) for pharmacists from January 2016 to January 2017. Most of this increase occurred in the 6 months between the launch of CURES 2.0 and the mandatory registration deadline in June 2016. Full results are shown in Figure A2 of online Appendix C.

Figure 2 compares the forecasted versus actual counts of registered and active PDMP users and shows that the gap between the number of clinicians registered in the PDMP and the number of active PDMP users (users creating ≥∣1 patient report in a given month) widened considerably around the implementation of mandatory registration. The upper (gray) line in each figure represents the number of registered prescribers and pharmacists. The lower (black) line represents active users. Although the ratio of registrants to active users has steadily increased since 2010, this ratio increased rapidly leading up to mandatory registration. In January 2016, 28% of registered prescribers were active users. By July 2016, only 14% of registered prescribers were active users. We observed a similar drop for pharmacists, from 60% to 35%.

Table 1 shows the estimated effect of CURES 2.0 and mandatory registration. We estimated that these changes increased the number of registrants by 101.1% for prescribers and 26.1% for pharmacists and increased the number of active users of the PDMP by 53.5% for prescribers and 17.9% for pharmacists. The time series for active pharmacist users required a log transformation to satisfy model assumptions, resulting in a wider confidence interval. With a much larger increase in registration than in active users, we observed a decrease of over 25% in the proportion of registrants that are also active users. This decrease is also represented in Figure 2 as the widening shaded region.

Table 2 contrasts the usage behavior of early versus late adopters of the PDMP in the 15 months following mandatory registration. We observed large differences between the 2 populations. For both prescribers and pharmacists, early adopters were approximately 4 times more likely than late adopters to be active PDMP users. In the 15 months following mandatory registration, prescriber early adopters generated 56% more reports than late adopters and pharmacist early adopters generated 689% more reports per user than late adopters. Approximately 81.5% of prescribers registered for the PDMP prescribed a controlled substance at least once in the 15 months following mandatory registration.

The sensitivity analysis of the prescriber registration compliance estimator found only small differences in the percentage registered among the method used in our primary analysis and the 2 alternative methods. The average difference between the highest and lowest estimate over all months was 1.3%, and the maximum difference between any 2 estimates was 8.3%. We also found that only 0.1% of prescriptions in the PDMP were prescribed by using a clinic-based DEA number. Thus, estimates of registration based on DEA numbers were not greatly affected by the use of clinic-based DEA numbers. Complete results of this analysis are provided in online Appendix A.

Our evaluation of CURES 2.0 and California’s mandatory PDMP registration law found that registration increased by 176% for prescribers and 64% for pharmacists from January 2016 to January 2017. Furthermore, the average number of monthly patient reports created increased by 96% for prescribers and 39% for pharmacists during the same period. These results indicate that the law achieved its goal of increasing PDMP registration and use; however, registration was not universal. We estimated that changes to the PDMP were associated with a 100% increase in registrants for prescribers, but only a 53.5% increase in the number of active users. As a result of registration outpacing the number of active users, the proportion of registrants that used the PDMP at least once in a given month fell by 27.6%. We observed a similar result for pharmacists.

Despite the mandate, registration plateaued at about 67% for prescribers and 87% for pharmacists at the end of our observation period, so there is still potential to increase the percentage registered, especially for prescribers. Following mandatory registration, the percentage of clinicians plateaued, leaving a large percentage of clinicians unregistered for an undetermined reason. Clinicians who remained unregistered may have been retired or not engaged in active practice for some other reason, practicing outside of California, or unaware of the PDMP registration requirement. In a survey of California physicians and pharmacists conducted after implementation of mandatory registration, only 29% of unregistered clinicians reported being aware that registration for the PDMP was mandatory.¹³

Compared with early adopters of the PDMP, late adopters (those who registered within 8 months of the implementation of mandatory registration) were significantly less likely to use the PDMP during a given month and created fewer patient reports per month. This difference persisted for over 1 year after mandatory registration was implemented. The proportion of clinicians creating at least 1 patient report in a given month was about 4 times larger for early adopters of the PDMP than for late adopters, who were more likely to have registered in response to the mandate. One possibility is that many late adopters did not prescribe or dispense controlled substances regularly and so perceived little need to use the PDMP; however, we did find cases of late adopters who prescribed controlled substances and were not active PDMP users.

Our study did not identify the optimal proportion of registered clinicians who should be active PDMP users, but we would not expect all clinicians to use the PDMP regularly. Clinicians who are not in active clinical practice, who do not prescribe or dispense outpatient controlled substances in their practice, or who practice outside of California may rarely have a reason to use the PDMP. Our finding that 18.5% of prescribers who were registered for the PDMP did not prescribe a controlled substance in the 15 months following mandatory registration is consistent with the finding from survey data that 30% of physicians and 24% of pharmacists in California reported that the PDMP was not relevant to their practice, often because they were retired, practiced outside California,¹⁴ or practiced only in inpatient settings.

The problem of estimating percentage registration for prescribers is not unique to California’s PDMP and remains a methodological hurdle to assessing and enforcing mandatory PDMP registration in states that do not maintain a comprehensive database that links state prescribing license and DEA license numbers. We calculated the percentage of prescribers with DEA numbers who were registered for the PDMP using 3 different methods and data sources and showed that all 3 methods produced comparable results. As all 3 methods are likely to approximate the true percentage of registered prescribers, we recommend that other states use the method presented in our primary analysis because it can be calculated using only PDMP prescription data and aligns with existing recommendations.⁹ Although this method is likely to be accurate for evaluating statewide compliance with mandatory registration, assessing compliance at the individual level will require case-by-case analyses when comprehensive databases that reliably link DEA and license numbers do not exist. The California PDMP is currently implementing the collection of state prescribing license number with PDMP prescription data to help address this problem.

Our study had several limitations. Mandatory registration was enacted 6 months after the release of CURES 2.0. One of the main improvements implemented in CURES 2.0 was electronic registration, which removed a major barrier to PDMP usage. Because of this short time interval, we cannot completely separate the effect of this update from the effect of mandatory registration. However, we did follow registration and use rates for 15 months after implementation of mandatory registration, so our estimates of the effects on PDMP use can be considered an upper limit of the potential effect of mandatory registration on PDMP use.

Another limitation is that the PDMP prescription data do not include controlled substances dispensed from federally regulated pharmacies, such as those under the jurisdiction of the Department of Defense and Indian Health Services. Finally, the scope of the study limited our ability to make definitive causal inferences based on the observed association between the mandate and changes in PDMP use patterns. Although it is highly likely that the new registration mandate explains the dramatic increase in PDMP registration immediately prior to that law’s implementation, we did not account for other policy changes or programs that may also have increased PDMP usage rates, such as the Centers for Disease Control and Prevention’s opioid prescribing guidelines (published March 2016) or contemporaneous efforts in California to promote safe prescribing and decrease opioid-related overdoses. Our study did not examine the effect of mandatory PDMP registration on controlled substance prescribing or opioid misuse; these effects should be investigated in future studies.

This study suggests that, although CURES 2.0 and mandatory registration provide a boost to the number of active PDMP users and an increase in the overall number of PDMP reports created, most clinicians who register shortly before the implementation of mandatory registration will subsequently use the PDMP less than once per month. This finding indicates that mandatory PDMP registration alone is not sufficient to maximize PDMP usage rates. Policymakers pursuing this goal will also need to consider other efforts, such as increased clinician education, or even mandating PDMP use prior to controlled substance prescribing.^7,14 California requires PDMP use effective October 2, 2018. Although evidence concerning the effect of PDMP policies on overdose death rates is mixed, 2 recent studies found low-strength evidence that mandatory PDMP use was associated with reductions in fatal overdose¹⁵ and reductions in prescribing.¹⁶ This suggests the possibility that the potential benefit of PDMPs is tied not only to access to the PDMP, but also to widespread PDMP usage by clinicians.