Prescription drug monitoring program policy reform: human and veterinary practitioner prescribing in West Virginia, 2008–2020

Abstract

No study has examined Prescription Drug Monitoring Program (PDMP) data for West Virginia or among specialty practices, such as veterinary medicine. The objectives of this study were to conduct time series modeling to describe the PDMP policy reform impact on prescribing rates for human and veterinary providers. We obtained data from the WV PDMP for 2008 through 2020 for the number of opioid prescriptions filled and providers. We estimated prescribing rates for human and veterinary providers separately based upon the top five opioids prescribed by veterinarians. We estimated temporal effects using a Bayesian log-normal time series model for humans and veterinarians separately. Throughout the study period prescribing rates increased significantly for veterinarians, and decreased for human providers, particularly during 2018 after implementation of the Opioid Reduction Act. Findings provide novel insight into the differential impact of policy on specialty practices and highlight decreasing human opioid prescribing observed elsewhere.

Introduction

Misuse of prescription opioids is a serious public health concern, particularly for Appalachia [1]. In response to the burgeoning opioid epidemic in the United States (U.S.), many states have made legislative changes to address opioid over-prescribing, including the enactment of Prescription Drug Monitoring Programs (PDMPs) [2]. PDMPs require healthcare providers to report any patient who receives a prescription for controlled substance to a state monitoring board. Their purpose is to provide real time patient prescription histories to curtail potential abuse and limit drug diversion [2–5]. While PDMP’s can provide critical information to health care providers, they are state-level databases, susceptible to limitations given each state’s reporting requirements, and operations [6].

In West Virginia (WV), Senate Bills 437 and 273 amended PDMP legislation in 2012 and 2018, respectively. Senate Bill 437 amended the PDMP to require reporting of controlled substance dispensations within 24 h, and at least annual patient medication history consultations with the WV Board of Pharmacy [7]. Senate Bill 237 (Opioid Reduction Act) limited the day supply of initial opioid prescriptions, particularly for specialties, such as veterinarians; these were subsequently limited to 7-day supply [8]. Despite apparent active reform, efficacy of PDMPs in reducing opioid misuse or abuse has been mixed due in part to state-level differences in functionality and adoption of mandatory versus voluntary reporting among healthcare providers [4, 9–11]. Additionally, limited attention has been given to the impact of PDMP policy reform on specialty providers. Opioids are also a critical drug class for treating acute pain and managing chronic pain in animals [12]. In 2015, the American Animal Hospital Association (AAHA) and American Association of Feline Practitioners (AAFP) released guidelines for pain management in dogs and cats. In general, these guidelines recommend that opioids as a drug class be used routinely as preoperative and postoperative medications, and at the discretion of the provider for management of cancer pain and for quality of life care [12].

Recent media reports have suggested that individuals desperate to receive opioid prescriptions may seek medications from veterinarians. Specific instances of drug seeking behavior involving veterinary practices include: pet owners harming their animals, robberies involving firearms, and pet owners changing animal prescriptions before submitting orders to online retailers [13–15]. These reports highlight phenobarbital, tramadol, and ketamine as potential targets of drug misuse [13, 14]. These initial reports motivated studies from Pennsylvania and nationally to increase awareness around drug diversion and opioid misuse in veterinary medicine. A Pennsylvania study by Clark et al. among 136 providers at an academic veterinary hospital found a 41.2% increase in prescribed Morphine Milligram Equivalents (MME) between 2007 and 2017 [16]. This increased veterinary use was in contrast to decreasing statewide trends in opioid prescribing for humans [16]. Mason et al. surveyed 189 Colorado veterinarians and found 13% reported knowledge of intentional injury or illness to a pet in an effort to obtain opioids, and 44% were aware of opioid misuse by clients or practice staff [15].

West Virginia is an ideal state to investigate disparities in opioid prescribing practices among medical providers and specialties, as it has led the country in overdose deaths since 2014 [17]. Yet to our knowledge there are no published research studies using WV PDMP data. The objectives of this study are to examine temporal shifts in prescribing rates for human and veterinary providers from 2008 to June 2020. Findings are expected to close significant gaps in knowledge surrounding the impact of PDMP policy reform on prescribing practices for humans and animals in West Virginia, the only state completely contained within Appalachia.

Methods

Study population

We obtained data from West Virginia’s PDMP from January 2008 to June 2020. As of 2019 the WV Board of Veterinary Medicine reported 163 stationary veterinary practices in WV [18]. Of these, 69 (42%) reported data to the PDMP during at least one year. For human providers, there were a total 12,367 unique prescribers reporting data for at least one year. To limit annual participation bias, statistical analyses focused on prescribing rates for veterinary and human providers. We estimated prescribing rates by diving the number of prescriptions written by the total providers within the same time frame. Data for the number of prescriptions written reflect only those prescriptions written for WV residents.

Data management and analysis

We obtained numbers of prescriptions filled in six-month increments for the five most prescribed opioids among veterinarians for veterinary and human providers from the West Virginia Board of Pharmacy’s PDMP for 2008-June 2020. The top five drugs prescribed among veterinarians were Tramadol, Codeine, Fentanyl, Lortab (Hydrocodone/acetaminophen), and Morphine. We requested human data for these same drugs for comparison during the study period. We conducted separate Bayesian multivariable log-normal time series models to examine trends for veterinary and human providers, respectively, using the INLA package in R [19, 20]. The dependent variable for both models was total opioids prescribed, created by summing across the five different drugs types. Each model included an offset for log of the number of providers contributing prescribing data for the opioids of interest during the study period. Fixed effects included year as a categorical variable, and a random effect for six-month interval within a year to capture intra-year variation in prescribing. Random temporal effects were investigated using a random walk model, as has been done in previous public health Bayesian modeling studies [21]. Parameter estimates from human and veterinary models were transformed through exponentiation to obtain incidence rate ratios (IRR). An IRR above one indicated that the prescribing rate for a given year is higher than expected given prescribing in 2008. Significant effects of independent variables were assessed using 95% credible intervals (95% Crl). An interval including zero indicates the corresponding covariate is not influential. Models were conducted separately for human and veterinary providers due to the relative difference in sample size among veterinary providers contributing data to the WV PDMP.

Results

Among the 69 veterinary and 12,367 human providers, there were a total of 7919 veterinary prescriptions and 13,350,645 human prescriptions for Tramadol, Codeine, Fentanyl, Lortab, and Morphine from 2008 to June 2020. For veterinary providers there were 7633 Tramadol, 107 Codeine, 82 Fentanyl, 77 Lortab, and 20 Morphine prescriptions, respectively, over the study period. Unique veterinary providers reporting to the WV PDMP differed by six-month increment ranging from 1 provider in 2008 to 61 providers in 2019. Among veterinarians reporting to the PDMP, number reporting only once each year ranged from 0 to 25% prior to senate bill 437 enactment in 2012, and from 0 to 7.4% after. For human providers, there were 1,960,976 Tramadol, 804,485 Codeine, 387,936 Fentanyl, 9,675,339 Lortab, and 521,909 Morphine prescriptions over the study period. Unique human providers reporting to the WV PDMP differed by six-month increments ranging from 436 in 2008 to 46,537 in 2015. Time trends by opioid and provider type are displayed in Fig. 1.

Fig. 1.

Average number of prescriptions per provider by provider type for the five most prescribed drugs among veterinarians from January 2008 to June 2020. Both graphs have the same legend

Temporal trends in prescribing rate for veterinary and human providers relative to Senate Bills 437 and 273 are displayed in Fig. 2. Regression analyses identified increasing opioid prescribing trends among veterinarians across the study period. However, the increasing patterns in opioid prescribing were not statistically different from 2008 until 2019 (IRR = 3.37, 95% Crl = 1.266, 9.006). Alternatively, human regression analyses indicated decreasing prescribing trends among the top five opioid drugs prescribed by veterinarians. This decreasing trend in human prescribing was not statistically different from 2008 until 2018 (IRR = 0.636, 95% CI 0.429, 0.945); after which each year indicated statistically significant decreasing trends for 2019 (IRR = 0.581, 95% Crl = 0.391, 0.865) and 2020 (IRR = 0.561, 95% Crl = 0.347, 0.94), respectively. Yearly parameter estimates for veterinary and human prescriber models from 2008 to June 2020 are displayed in Table 1.

Fig. 2.

Prescribing rate for Human and Veterinary providers relative to PDMP reform in West Virginia, 2008-June 2020. Prescribing rate is calculated for human and veterinary providers within six-month increments using top 5 opioid drugs prescribed among veterinarians

Table 1.

Incidence rate ratio (IRR) and 95% credible intervals (Crl) from the human and veterinary prescriber time series models

Year	Incidence rate ratio	95% Crl
Veterinary provider model
2008	Reference	Reference
2009	0.356	(0.135, 0.936)
2010	0.284	(0.108, 0.748)
2011	0.288	(0.109, 0.759)
2012	0.228	(0.086, 0.601)
2013	0.240	(0.090, 0.633)
2014	0.601	(0.226, 1.588)
2015	1.332	(0.502, 3.532)
2016	1.291	(0.486, 3.428)
2017	1.118	(0.420, 2.974)
2018	1.716	(0.644, 4.567)
2019	3.377	(1.266, 9.006)a
2020	3.238	(0.978, 10.69)
Human provider model
2008	Reference	Reference
2009	1.114	(0.766, 1.617)
2010	1.145	(0.785, 1.666)
2011	1.078	(0.738, 1.574)
2012	1.150	(0.785, 1.682)
2013	1.178	(0.802, 1.728)
2014	1.146	(0.779, 1.687)
2015	0.836	(0.567, 1.232)
2016	0.778	(0.526, 1.149)
2017	0.731	(0.494, 1.084)
2018	0.636	(0.429, 0.945)a
2019	0.581	(0.391, 0.865)a
2020	0.561	(0.347, 0.904)a

^aIndicates statistical significance at 0.05 level

Discussion

Significant gaps remain in the PDMP and opioid policy reform literature regarding controlled substance prescribing in West Virginia and for specialty practices, such as veterinarians. This study reports temporal trends in opioid prescribing among the 69 veterinary and 12,367 human providers reporting to the PDMP. Overall, findings indicated increasing opioid prescribing among veterinary providers, and decreasing opioid prescribing among human providers, particularly around Senate Bill 273 enacted in 2018. Increasing trends in controlled substance prescribing among veterinarians and decreasing trends in human opioid prescribing are consistent with previous research [10, 16, 22]. To our knowledge this is the first study to assess time trends in prescribing between human and veterinary providers using regression modeling.

Temporal trends in the prescribing rate displayed in Fig. 2 appeared to be largely driven by prescribing patterns of specific substances for both veterinary and human providers (Fig. 1). For veterinary providers, introduction of Tramadol as a controlled substance in 2014, resulted in a sharp increase in prescriptions per provider. For human providers, increasing mandates in the PDMP through Senate Bills 437 and 273 resulted in a decline in Lortab prescriptions which lowered overall opioid prescribing trends observed for the top 5 opioid drugs prescribed by veterinarians. Caution is taken to infer that trends for prescribing rate among human providers in our study mimic overall opioid prescribing trends. An important consideration of this study was to compare veterinary and human prescribing behaviors on the basis of the top five opioids prescribed among veterinarians.

Limitations to our approach exist. PDMP data were available only for the veterinary providers who utilized pharmacies to fill pet medications, whereas human PDMP data were available for all prescribers. Lack of consideration for other important drivers of drug misuse, such as economic disparities, could also differentially bias observed associations between prescribing rate over time and provider type. Other potential sources of bias include standardizing by total number of unique providers in-place of unique patients receiving or picking up medications from the pharmacies. Unfortunately, this information is not currently captured reliably in veterinary PDMP data where prescriptions are often filled under the pet’s name instead of the owner’s and in cases where owners may have multiple pets per household.

Despite increased awareness, the American Veterinary Medical Association’s (AVMA) requested that veterinary practices be exempt from PDMP reporting where possible, due to limited research and insufficient resources compared to human medical practices [23]. Veterinarians have been reluctant to participate in state prescription drug monitoring programs due to concern in viewing human medical records, lack of oversight into how to distinguish between human and pet prescriptions in the PDMP, and incompatible veterinary electronic medical record systems [24–26]. Despite valid logistical concerns, the National Alliance for Model State Drug Laws (NAMSDL), supports mandatory reporting for all human and pet providers prescribing opioids [27]. Only 20 states, however, including West Virginia, require veterinarians to report their opioid prescribing [22]. Our findings support the notion that veterinarians may have a significant role in the ongoing opioid epidemic in West Virginia. However, this role is still uncertain as trends in veterinary opioid prescribing rates were motivated primarily by Tramadol, which has been shown to have a low rate of abuse [28, 29]. Given these uncertainties and lack of formal best practices for use of PDMPs for clinician decision making, it’s difficult to make recommendations as to whether veterinarians should be required to report to state PDMPs [30]. While a formal decision maybe premature, its clear that improved registration and database linkage capabilities would increase efficacy of PDMP data for evaluating the role veterinarians and other specialty practices have in the ongoing drug epidemic. More research is needed to determine if there is a link between veterinary opioid prescribing and patterns of human substance misuse and inform future legislative changes to reduce the burden of the current drug epidemic on providers in the United States.