HOW CLINICIANS USE PRESCRIPTION DRUG MONITORING PROGRAMS: A QUALITATIVE INQUIRY

Abstract

Objectives

Prescription drug monitoring programs (PDMP) are now active in most states to assist clinicians in identifying potential controlled drug misuse, diversion or excessive prescribing. Little is still known about the ways in which they are incorporated into workflow and clinical decision making, what barriers continue to exist, and how clinicians are sharing PDMP results with their patients.

Design

Qualitative data were collected through online focus groups and telephone interviews

Setting

Clinicians from pain management, emergency and family medicine, psychiatry/behavioral health, rehabilitation medicine, internal medicine and dentistry.

Subjects

35 clinicians from 9 states participated.

Methods

We conducted two online focus groups and seven telephone interviews. A multidisciplinary team then used a grounded theory approach coupled with an immersion-crystallization strategy for identifying key themes in the resulting transcripts.

Results

Some participants, mainly from pain clinics, reported checking the PDMP with every patient, every time. Others checked only for new patients, for new opioid prescriptions, or for patients for whom they suspected abuse. Participants described varied approaches to sharing PDMP information with patients, including openly discussing potential addiction or safety concerns; avoiding discussion altogether; and approaching discussion confrontationally. Participants described patient anger or denial as a common response and noted the role of patient satisfaction surveys as an influence on prescribing.

Conclusion

Routines for accessing PDMP data and how clinicians respond to it vary widely. As PDMP use becomes more widespread, it will be important to understand what approaches are most effective for identifying and addressing unsafe medication use.

INTRODUCTION

Prescription drug overdoses have reached epidemic proportions.¹ Federal and state governments are eager to identify strategies to reduce misuse, abuse and diversion. The Office of National Drug Control Policy (ONDCP) advocates prescription drug monitoring programs (PDMPs) as one strategy for reducing prescription drug abuse.²

PDMPs have been implemented in nearly every state to monitor for controlled substances; these statewide databases collect information on varying schedules of controlled substances dispensed to patients. PDMPs were originally developed with law enforcement in mind to identify patterns of drug misuse, diversion or excessive prescribing (pill mills). However, PDMPs are increasingly seen as tools for improving healthcare, and may help clinicians identify patients with a need for mental health or addiction services or those receiving unsafe doses or drug combinations.

Although most states now have active PDMPs, there are variations in the ways these programs are designed, how they function and who can access them.³ Patterns of PDMP use may vary state-to-state depending on whether the PDMP operates under law enforcement, public health agencies, boards of pharmacy or other state agencies. An important limitation is that federal healthcare facilities (e.g. Department of Veterans Affairs, Department of Defense and Indian Health Service) are not required and do not generally report to state PDMPs. While some PDMPs such as California’s and New York’s are longstanding, most are still relatively new. A few studies have examined clinical use of PDMPs, focusing on how PDMPs influence prescribing, or on cursory information about how clinicians utilize and respond to PDMP information.^3–6 However, little is known about how clinicians integrate PDMP use into clinical workflow and few studies have tried to identify strategies clinicians use when discussing PDMP reports with patients. Similarly, little is known about clinicians’ experience with, perceptions of, or attitudes toward PDMP systems. Such information could help identify “best practices” regarding PDMPs and potential enhancements to improve their utility. Further, it could lead to recommendations or guidelines for using PDMP data in clinical decision-making, incorporating PDMP data into clinical workflow, and discussing concerns about the data with patients.

As a foundation for studying these issues, we conducted online focus groups with clinicians to identify how they use PDMP data in clinical-decision making, how they integrate the data into clinical workflow (including communication with the patient)and how they perceive the PDMP systems.

METHODS

Oregon State Public Health Division and Oregon Health & Science University Institutional Review Boards approved this study. In addition, we obtained a Certificate of Confidentiality from the National Institutes of Health intended to protect participant confidentiality.

Focus Group Recruitment

We identified potential clinician participants for the online focus groups through state PDMP administrators. We contacted administrators in two ways: by personal contact at the Harold Rogers PDMP national meeting in Washington DC in June, 2012 and by email. Of the 43 state administrators we contacted, 9 assisted with the recruitment. Table 1 lists the states participating in this study.

Table 1.

Characteristics of the 35 focus group participants

Characteristic	Frequency
Female / Male	8 / 27
States Represented	9 (WY, LA, FL, OH, UT, MI, WA, MN, NV)
Credential
Physician	26
Nurse practitioner	7
Physician Assistant	1
Dentist	1
Clinical Specialty of Physicians (n=26)
Pain Medicine	7
Emergency Medicine	6
Family Medicine	5
Psychiatry/Behavioral Health	5
Rehabilitation Medicine	2
Internal Medicine	1
Duration of PMP user (self-report)
Less than 1 year	7
1–5 years	22
More than 5 years	6
Frequency of PMP use/month (self-report)
Fewer than 5 times	3
5–9 times	7
10 or more times	25
Settings
Emergency department	7
Small private office <5 practitioners	7
Hospital outpatient clinic	4
Hospital inpatient	2
College-based health clinic	1
Academic practice	2
Solo practice	1
Large private practice > 5 practitioners	5
Community mental health center	3
Stand-alone urgent care center	1
Safety-net, low-income clinic	2

State administrators were the initial liaison between clinicians and the research study; they made first contact with potential participants, providing them with a letter explaining the study, and inviting clinicians to contact the research team via email. The administrators then provided a list of interested clinicians to the research team. A master list of 78 clinicians was created.

We used a purposive sampling approach to ensure variation on key attributes, including professional credentials (e.g.; medical doctors, nurse practitioners and dentists) and clinical practice settings (e.g.; pain clinics, emergency departments, primary care, inpatient psychiatry). Clinicians were contacted by phone by a physician study team member who confirmed their interest, answered questions and obtained verbal consent.

Thirty-five clinicians agreed to participate. We provided contact information for consenting clinicians to an experienced focus group moderator who confirmed clinicians’ participation, assigned each clinician an alias and password, and provided information about using QualBoard, a proprietary software product owned by 20/20 Research, Nashville, TN for online focus groups.

Focus group procedure

We conducted two focus groups, held approximately one month apart. Clinicians participated in one focus group, and were asked to spend approximately 15 minutes per day for three days, responding to questions posed by the facilitator, and to other participants’ posts. We designed each focus group to span three days to reduce the daily burden on participants. The placement of the questions followed this sequence: Day 1: introductions, demographics, credentials and frequency of PDMP use, advantages and drawbacks; Day 2: clinical uses of PDMP and how it is incorporated into workflow; and Day 3: PDMP discussions with patients, patient responses to PDMP reports and ongoing training. Each day the focus group moderator posted questions, including clarifying and probing questions to encourage participants to respond in more depth, as needed. The give and take among focus group participants stimulated an online conversation similar to an in-person focus group. As in many face-to-face focus groups, discussion often ranged widely across topics, regardless of the day. Additionally, clinician participation varied, with some logging in once a day and others logging in multiple times throughout the day to respond to other participant comments. Participants received a financial incentive to compensate for their time.

Telephone Interviews

All clinicians were asked if they would be willing to participate in a follow-up interview. Among those volunteers for the follow up interview, we selected seven clinicians deliberately to obtain variation by specialty and geographic location, to gain additional information regarding unique aspects of PDMP use, including specific protocols for discussing PDMP data with patients.

Data Management

Text from the online focus groups was downloaded into a Word document. Telephone interviews were digitally recorded and professionally transcribed. All data were de-identified and securely stored on HIPAA compliant, password-protected servers for use in ATLAS.ti™(version 7.1.3), a program for management and analysis of qualitative data.

Analysis

A multidisciplinary team composed of two primary care physicians, prescription drug monitoring program administrator, pharmacist, addiction therapist, communication scientist and qualitative research expert, and quality improvement expert analyzed the data. We used a grounded theory approach⁷ and engaged in two immersion-crystallization⁸ cycles to analyze data and identify findings. In the first cycle, our team read data aloud as a group to gain insight into participants’ experiences and use of PDMP (immersion). Through this process we identified important segments of text and developed a preliminary code list (crystallization).We refined this list and two team members analyzed the remaining data. Preliminary codes and new codes that emerged were discussed with a third research team member as well as with expert panel members to further refine codes and identify new findings. In a second immersion-crystallization cycle we examined data to identify patterns that emerged across focus group participants and that might be related to clinician specialty and practice setting. In this phase, we read text within each code (immersion) to identify important features, characteristics and patterns related to PDMP use; assess how use varied by clinician specialty; examine more closely how clinicians shared PDMP data with patients; and further identify the barriers and facilitators to PDMP use.

RESULTS

Thirty-five clinicians from nine states participated in the two focus groups. Participants represented a broad range of specialties and clinical practice settings (Table 1).

Clinicians’ Use of the PDMP

Clinicians described using the PDMP mainly for clinical purposes. However, participants noted that, at times, they used the PDMP for administrative uses, such as making sure no false prescriptions were written under their names.

Clinical uses of the PDMP included verifying current prescriptions or prescription fill history. As a Psychiatric Nurse Practitioner working in the hospital setting reported:

I use it to track prescription drug use on any patient who is admitted to my unit with a positive urine drug screen for benzodiazepines, opiates or barbiturates, drug overdoses, chronic pain managed by prescription pain killers, muscle relaxants, barbiturates or benzodiazepines, suspicion of drug dealing, unusual behavior suggesting drug withdrawal, suspicion of “doctor shopping,” and incongruence between history and clinical exam (Participant 1, Nurse Practitioner, Inpatient Psychiatry).

Participants also reported utilizing the PDMP to coordinate care with other clinicians. Some PDMPs provide information about other prescribers, and participants would contact these clinicians to discuss treatment and prescribing plans and to obtain additional information when prescribing practices appeared inappropriate. For example, a clinician wrote:“It is not uncommon for me to contact the patient’s primary care provider to discuss what my plans are for prescribing” (Participant 12, Nurse Practitioner, Pain Clinic). Another reported that“…in a handful of instances in using the PDMP I have contacted prescribing physicians in situations in which I feel their prescribing practices have been inappropriate”(Participant 14, Physician, Emergency Room).

Participants identified pharmacists, who also have access to the PDMP programs, as important partners in handling PDMP information. Clinicians reported that pharmacists verified data in the PDMP report and were valued for proactively calling clinicians if they noticed a patient using multiple prescribers for controlled medications, a questionable prescription dose, or suspected alteration.

I communicate with pharmacies mainly when the PDMP doesn’t match up with what the patient is saying, such as the dates that the prescriptions were filled. Also, when a patient says that they filled a prescription at the pharmacy and it hasn’t shown up on the PDMP, we will typically call the pharmacy for verification (Participant 8, Physician, Pain Clinic).

How clinicians integrated PDMP into workflow varied, and prescribers from different specialties described different approaches. Clinicians who practice in settings where substance abuse issues and use of opioids to treat pain were common and who had a more continuous relationship with patients (e.g., psychiatry and pain clinics) described a consistent and rigorous process for using the PDMP. For instance, a pain specialist wrote, “PDMP reports are run the business day before a visit during our chart prep. Of course another can be run the day of the visit if needed…Responses are received within seconds via email and a report is printed for the physician to review during the patient visit”(Participant 2, Physician, Pain Clinic)

In contrast, participants working in a setting where they treat patients on an episodic basis (e.g., emergency rooms) reported accessing the PDMP with less frequency or only when a red flag emerged during the visit:

Our approach is usually to address the primary reason for the ED visit, then obtain the PDMP report if there is concern or patient behavior that’s inconsistent. For example, patients demanding specific pain medications, especially when no allergies are indicated on triage, but allergies are reported when initial medications are ordered that patient doesn’t like or want. Sometimes patients will say a certain medication “doesn’t ever work” (Participant 3, Physician, Emergency Room).

Sharing PDMP Information with Patients

Participants reported sharing PDMP data with patients in variety of ways. For example, clinicians reported discussing PDMP data with patients to understand information in the report and identify issues of potential addiction or medication safety (e.g., potentially risky doses).Providers reported using language to normalize (italicized below) checking data from the PDMP to frame this investigation into the patient’s medication use as a practice routine.

If the PDMP is not problematic then I don’t bring it up. If it is problematic, I get the patient history then bring into the discussion a standard report that we routinely pull when controlled substances are a consideration…I use it to raise questions about missed information. For example: “You have been to 6 practices over the past 3 months for pain medication. This is unusual. Can you help me understand this?”(Participant 4, Nurse Practitioner, Family Medicine).

When discussing PDMP data with patients, participants reported seeing this as an opportunity to address issues of potential addiction or medication safety. For example a psychiatrist noted:

I usually start by stating, “I review the state’s PDMP to make sure I have an understanding of everything you are taking. Something concerns me about what is in your report. Can you help me understand ________?” The blank could be: Why are you getting medications from so many different doctors? Why are you paying cash to get some of your prescriptions and use insurance for others? Why do you use different pharmacies? Then I would follow up with, “These things make me concerned…Have you ever thought you might have problems with this medication?” I usually end with, “because of what is in the report it is not safe for me to prescribe a controlled drug. I can treat your problem with…” This usually works pretty well. Sometimes patients will become defensive in the process and might accuse me of not trusting them (Participant 5, Psychiatrist, Community Clinic).

Clinicians also reported completely avoiding the discussion of PDMP information by coaxing patients to leave the office quietly. For example:

Most of the time I do a quick old NSAID script that a lot of them don’t’ recognize. I tell the nurse that I’m working with, “Hey, I’m giving this guy a script for an old NSAID, if he says anything just play dumb and come get me”. You know, if they ask what it is then I go and have “the talk.” In about 95% of the cases, they never say anything, and …I don’t know if they don’t’ know what it is, or they didn’t see what it was until they go out to the pharmacy or until they got out in the parking lot. But they’re quietly and quickly walking out the discharge doors and we are turning over that room to get it cleaned. The nurse is happy. I’m happy. Let’s go onto the next patient. So, that’s 95% of my encountered situations (Participant 9, Emergency Room Physician).

Clinicians also discussed confronting patients with information obtained in the PDMP, sometimes with the aim of “catching patients in a lie.”

Usually I’ve asked them a few times have you seen other doctors, have you gotten any further prescriptions…If it does not seem legitimate…I will leave the room to get something and pull it [PDMP report] and then look at it. I may confront them if it seems like they’re lying and say, “Well here’s what I got here. It seems like you haven’t been very honest with me and so I’m not going to provide you with prescriptions(Participant 10, Emergency Room Physician).

Patient Responses to PDMP Information

Once a clinician decides to share PDMP information, patients may respond in a range of ways. Participants reported that patients’ responses include denial, justification or rationalization of prescription history. This participant’s description exemplifies a number of these responses:

The most common response is indignation. “That’s not me!” or “There must be some mistake!” or “Who do I talk to about this report, since it is obviously wrong.” Occasionally, they try to deny that they have seen the providers on the list or that they have actually visited the pharmacies (Participant 11, Emergency Room Physician).

PDMP information can also result inpatients acknowledging a problem and requesting help. For instance, a family medicine physician reported, “They often start out defensive, but later ask for help. I usually come across as quite non-judgmental and they ask for help a number of times before the interaction ends” (Participant 10, Family Medicine Physician).

Participants also reported that patient responses might have an emotional element ranging from anger to indignation to guilt and embarrassment. The following excerpts illustrate the range of patients’ emotional responses that participants described:

I share the PDMP information at the end of the exam. I explain that I am willing to help them with physical therapy, injections, membrane stabilizers, etc., just not with controlled medications. They aren't usually angry with me, but rather feel guilty and embarrassed. (Participant 8, Physician, Pain Clinic).

When I show them the actual list of recent activities [on the PDMP] they sort of wilt… After the initial indignation, they are quiet. I always tell them I am happy to provide care and non-narcotic prescriptions, but that based on the data I am unable to provide narcotic prescriptions (Participant 11, Emergency Room Physician).

Barriers, Recommendations and Training for PDMP Use

One important barrier to using the PDMP data to change prescribing practices is patient satisfaction ratings, such as Press-Ganey scores. Some organizations take these scores very seriously (e.g., align clinicians’ financial incentives with such scores) and clinicians perceive that withholding narcotic prescriptions and taking the extra time to review PDMP data can worsen scores. For instance, one emergency room physician reported, “ED wait times are a big driver of customer satisfaction, and something that the hospital keeps an eye on. Thus, it is much easier for a couple of doctors to just write for Vicod in, as opposed to sitting down to discuss the PDMP report with the patient and deal with an ensuing argument” (Participant9, Emergency Room Physician). Another clinician noted:

Pain is so subjective so often you just have to give out narcotics when the patient states they are 10/10 pain. But the environment that you work in makes a difference. I have worked in settings where the Press-Ganey scores are more important than patient safety or even staff safety (Participant 13, Physician’s Assistant, Emergency Room).

Participants also described barriers related to training, noting that training for the PDMP was limited to how to access the system. They did not receive guidance on how to interpret findings, integrate the PDMP into workflow, or talk with patients about the results. For instance, one Nurse Practitioner working in a pain clinic reported, “I had no formal training in how to use the PDMP or how to communicate with patients when there was an issue. For me, it was learn as you go” (Participant 12, Nurse Practitioner, Pain Medicine). Participants’desire for additional training varied, with some reporting a desire for additional training, and others reporting that their training on the PDMP was sufficient, stating that they know how to talk with patients about these types of issues.

Additional barriers to using the PDMP included difficulty accessing and navigating the PDMP (e.g., time to run a patient query), and difficulty interpreting PDMP data for use in patient care. Delays in pharmacy reporting, data errors, and data gaps were among the concerns. (Table 2).

Table 2.

Barriers to PDMP use identified by focus group participants and recommendations for facilitating use

Accessing the System

Clinicians having to register on multiple computers Not having 24 hour access to the PDMP in some states Frequent changes of passwords

Using the System

Lag time from when a prescription is dispensed to when it shows up in the PDMP The time it takes to log into the system, run a query and receive the report Data entry errors – wrong providers listed

Interpreting the Information

Missing data in the PDMP (information not reported by certain pharmacies, Indian Health Service, VA system) Not all clinicians use the system, so patients may obtain prescriptions from clinicians who are unaware of other prescribers No consistent recommendations on when to check PDMP; no financial incentive to do so

Participants also made recommendations for addressing these barriers. (Table 3).In particular, participants wanted access to PDMP data from other states (i.e., inter-state data sharing). A pain physician noted, “I have several patients who work out of state for weeks at a time throughout the year. There is no way of me knowing if they are seeing other doctors out of state. Having a nationwide PDMP would be helpful” (Participant 8, Physician, Pain Medicine).

Table 3.

Recommendations by participants that would optimize the use of PDMP’s

General Recommendations
More detailed information in the PDMP reports (number of days dispensed, whether drug is long acting or short acting) Delegated access in states where unavailable (although clinician opinion varied) Ability to access neighboring states or have national PDMP database System or policy changes that would optimize the use of the PDMP, (e.g., pro-active alerts, mandatory review under certain circumstances, integration into electronic health record, inclusion of other drugs (e.g., Tramadol, Cyclobenzaprine)
Training Recommendations
How to talk with patients about PDMP findings What to do with patients when the PDMP suggestions misuse, abuse or diversion Education to increase the use of the PDMP by clinicians

DISCUSSION

This focus group study identified clinicians’ strategies for using PDMP data. These included not only making prescribing decisions, but also coordinating care and identifying falsified information. Some clinicians accessed the PDMP with every patient, while others described using it only when suspicious of drug abuse. Discussion with patients regarding PDMP findings ranged from non-judgmental discussion to avoidance to confrontation. Our participants generally found the PDMP to be useful, but identified some barriers to use and a need for additional training on how to manage patients with a worrisome PDMP profile.

Some of our findings echo those of a survey of toxicologists, many of whom practiced emergency medicine.⁹ For example, that study also identified time constraints and navigation challenges as important barriers. However, our inclusion of clinicians from multiple specialties and settings uncovered a wider range of approaches to accessing and using PDMP data, and new information on the range of resulting conversations that occur between doctor and patient. As it was described by Perrone, the Drug Enforcement Agency’s (DEA) concept of ideal PDMPs mirrors the findings from our focus group participants. That concept includes: ease of access, real time updates, mandatory pharmacy reporting and interstate accessibility.¹⁰

Although some participants reported relying on subjective patient impressions in deciding when to access the PDMP, provider impressions of potential misuse or diversion do not always correspond with more structured data.^11,12 Thus, steps may be needed to encourage more consistent PDMP use, even in settings with a lower volume of pain care and opioid prescribing. This may not be for every patient at every visit, but for certain routine situations, such as every new patient, or whenever considering a prescription for a controlled drug. Future research should explore these complexities, as optimal patterns of use will likely vary with setting and specialty.

Participants described three general ways of sharing worrisome PDMP data with patients: non-judgmental discussion; avoiding discussions entirely; and using a more confrontational approach if patients do not appear to be candid. While clinicians did not make an explicit connection between the strategies they use to discuss PDMP data and how patients responded (e.g., embarrassment, indignation, anger), such a connection seems likely. If patients need addiction treatment or mental health care, certain approaches may be more successful than others. Thus, the different communication styles may lead to different patient responses, as well as a range of immediate, short and longer term outcomes. Likely, there is not a “one size fits all” approach to individual patients, and clinicians will need to consider the goal of the interaction and patient needs (e.g. treating addiction; keeping the patient alive) and balance this with time constraints, organizational demands, and available resources.

Examining the various approaches to discussing PDMP data and associated patient responses is a fruitful area for future research. For example, strategies used for alcohol abuse such as Screening, Brief Intervention, and Referral to Treatment (SBIRT)¹³ could potentially be adapted for opioid and other prescription drug use. Motivational Interviewing, which facilitates positive behavior changes, is typically part of SBIRT, and has been helpful in reducing substance use and promoting positive behavior changes in other chronic health conditions.^14,15 Future studies should aim to identify the most, and least, productive ways of communicating with patients about PDMP data to optimize clinician-patient interactions and patient outcomes.

Our study has several strengths, such as varied clinician specialties and practice settings, and the use of in-depth qualitative methods. However, there are also important limitations. Although participants represented nine states and a range of credentials and settings, the sample and data collection are from a relatively small group. We purposively selected participants because they reported substantial use or interest in their PDMP systems. This was important for ensuring we would be conducting focus groups with clinicians who had used the PDMP system, but their experiences may not generalize to the general clinician population, who may use the system less. Additionally, PDMP policies are evolving. Some states mandate PDMP checks in certain circumstances, and this may influence clinicians’ use of the PDMP. In February 2014, 16 states had legislation mandating use of the PDMP in certain circumstances, such as suspicion that a patient is seeking a controlled substance for non-medical reasons.¹⁶ Such rules may increase the consistency of PDMP use, but the actual impact on clinician behaviors is an important area for future research. Further, we did not collect data on pharmacists’ use of PDMPs. Although our focus in this study was on active prescribers who use PDMPs to make clinical decisions about whether to prescribe, pharmacists are an important user group of interest in future research. Controversies over recent rules proposed by some pharmacy chains highlight the importance of optimizing the complementary roles of clinician and pharmacist.^17,18

Other limitations include the potential biases inherent in all research, such as influences of our own predispositions and beliefs. We mitigated these by involving a multi-person, multidisciplinary team in developing focus group and interview questions and avoiding leading questions. In addition, this team was engaged throughout the analysis process. While we identified a wide range of experiences in using the PDMP, we cannot assess the frequency of the occurrences. Nonetheless, our findings may be useful for developing future large scale quantitative surveys designed to determine the prevalence of experiences and behaviors regarding the PDMP. Finally, this study is based on clinician self-report. There may be discrepancies between what people say and what they actually do, and we did not directly observe PDMP use or patient-clinician interactions.

Our study identified a range of strategies for deciding when to access PDMP data and how to respond to it. As use of PDMP data becomes more ubiquitous, it will be important to learn which of these strategies constitute “best practices” that optimize patient outcomes. Clinicians appear to be receptive to such information and, along with improving the ease of use, such information seems essential if PDMPs are to achieve their potential. Further research on how PDMPs can best be used in practice will help to maximize them as tools to improve quality of care and patient outcomes.