Monday, November 18, 2019

Mandated Queries of the Florida Prescription Drug Monitoring Program: A Three-Month Experience from a Cancer Center-Based Outpatient Palliative Medicine Clinic

by Chad Kollas (@ChadDKollas)

Background and Introduction

On July 2, 2018, Florida implemented a new law requiring all licensed physicians to query the state’s prescription drug monitoring program (PDMP), known as the Electronic Florida Online Reporting of Controlled Substances Evaluation (E-FORSCE), before prescribing controlled medications, including opioid analgesics. Like many other recent state and federal policies, this law was intended to reduce deaths from prescription opioid overdoses.
Our outpatient palliative medicine clinic initiated a quality improvement (QI) project to characterize the effects of the new and identify opportunities to improve palliative care in the setting of its implementation. We posted the results of the first month’s queries in an earlier Pallimed post, published in September 2018. This article represents the findings from the queries over the first three months’ queries and brings further clarity to our initial findings.

Methods

This quality improvement (QI) project was reviewed and approved by the Orlando Health/UFHealth Cancer Center Joint Oncology Committee for 2018-19. We began recording results of all E-FORSCE queries occurring after the law’s implementation of July 1, 2018 through September 30, 2018. We informed each patient that the PDMP query had become mandatory in Florida, and we discussed the results of each query with each patient. Each query examined the last 12 months of the patients’ controlled medication prescriptions.

We documented patients’ demographics, including each patient’s age, gender and limited identifying information, such as patient names and identification numbers, which were subsequently de-identified for statistical analyses. We recorded patients’ main diagnosis and pain symptoms, the number of prescribers listed by the PDMP as well as the dose of the patient’s opioid analgesia, measured in average daily morphine milligram equivalents, as provided by the state database. We also noted pain ratings with and without medications, including opioids and non-opioids, as well as the difference between those ratings. Beginning on September 6, 2018, E-FORSCE began displaying each queried patient’s “NarxCare Overdose Risk Score,” so we began record those scores as part of this QI project. According to its creator, Appriss Health, the NarxScore is an overdose risk score [that] is a predictive score for unintentional overdose death,” and that it is “designed to draw awareness to the presence of significant PMP data.”, although there are no studies in the medical literature assessing its validity as a predictive tool.

We classified the numbers of each patient’s prescribers binarily (one prescriber or more than one prescriber), and usage sage patterns were classified ordinally as medically legitimate, aberrant or medically illegitimate. Medically legitimate patterns reflected 1) cross-coverage of medication renewals, 2) evolving sites of care (such as a transitioning into oncologic care after treatment of cancer pain before establishing a cancer diagnosis), 3) prescribing of non-opioid controlled medications by primary care physicians (PCPS) and/or their cross-coverage and 4) prescribing within the policies and rules of the PMC.

We defined medically illegitimate patterns as involving 1) duplicative prescriptions (multiple prescriptions within close temporal proximity at multiple geographic site or multiple providers), 2) patterns unexplained by under-treatment and/or an inability to fill prescriptions in full quantity, or 3) behavior(s) prohibited by the policies and rules of the PMC. We used the category of aberrancy to describe situations which did not fall cleanly into the definitions of medically-legitimate or illegitimate usage.

Additionally, we classified opioid dosages ordinally, using the three planned categories for daily MME dose that used by the Centers for Medicare and Medicaid Services (CMS) in its opioid prescribing rules for 2019: less than 90 MME, 90-200 MME or greater than 200 MME. We also allowed for individualized comments regarding any unexpected or unusual findings when performing a PDMP query. The mean MME over the last 30 days was usually obtained from the PDMP, but was calculated based on prescribing instructions when not provided explicitly by that database.

Finally, we recorded patients’ self-reported pain ratings based on a 0-10/10 scale, with 0 indicating no pain and 10 indicating excruciating pain (or the most severe pain the patient had ever experienced). We recorded patients’ pain ratings with and without cancer pain medications, clarifying that we meant pain medications being prescribed medically for their cancer pain. In our clinic, the usual goal for cancer pain control is achieving a pain rating of 4-5 out of 10, not the achievement of a pain-free state, an approach that evolved from the American Pain Society’s recommendations for managing cancer pain.

Results

Demographics

We queried the Florida PDMP for a total of 359 outpatients from July 5, 2018 to October 4, 2018. The vast majority of these patients (90.3%) reported cancer-related pain, while the rest reported chronic pain from other conditions for which they received outpatient palliative care within the cancer center, including chronic pain from sickle cell disease (21 patients, 5.8% of total). The patients’ average age was 56.3 years old (range of 21 to 85) and 61.3% of them identified as female. Twenty-five of the 359 patients (7%) represented new consultations to the Palliative Medicine Clinic.

Prescribers and Usage Patterns

About 76% of the patients had more than one prescriber listed in the PDMP, while 21% had a single prescriber for controlled medications. Only two patients (0.6%) exhibited possible aberrancy: one had a query showing 15 prescribers, and the other was a healthcare professional who had self-prescribed renewals of anxiolytic medications (physicians in Florida are not permitted to prescribe to themselves, because Florida law specifies that a physician-patient relationship must involve two people; self-prescribing may be permissible in other states, albeit ethically questionable). The patient with 15 prescribers was ultimately not found to be using opioids illegitimately, but rather was participating in the healthcare system in a fragmented way, through two, parallel sets of third-party payers (health insurance and motor vehicle insurance). The self-prescribing physician already had received legitimate prescriptions for his medication from other physicians, but he had called in over-lapping renewals for the sake of geographic convenience; he stopped this behavior when reminded about its questionable nature. There were no patients whose prescribing pattern suggested otherwise illegal or medically-illegitimate behavior, such as “doctor shopping.”

In the first month of PDMP queries we had uncovered four cases in which the state PDMP was demonstrably incorrect, with an error rate of 3.2% in the database. At the completion of three months of queries, we identified 17 errors, which represented an error rate of 4.7% in the E-FORSCE database. We did not define as errors instances in which the E-FORSCE database misidentified a single prescriber as multiple prescribers, a very frequent occurrence within the database (for example, I was often identified as two prescribing physicians, under the names “Chad Kollas” and “Chad Kollas Kollas”), although the database does appear to use such duplication errors when calculating NarxScores. The most common errors were as follows: 1) no prescription listed in the PDMP when a prescription was clearly provided based on objective documentation (such as a filled medication bottle from a pharmacy listing the prescription date and prescriber’s name which were absent from the database), which accounted for 10 cases (2.8% of all cases); 2) grossly incorrect calculation of MME in 4 cases (1.1 % of all cases); and, 3) verifiably incorrect prescriber misattributed to the prescription, which occurred in 3 cases (0.8% of all cases).

Opioid Dose and MME Distribution

The mean opioid analgesic dose for all of the patients was 268 MME per day, but the median dose was 126 MME per day, with a very wide range of daily dosages (6 MME to 2880 MME; SD = 368.1). The distribution of Daily MME is shown in Table 1 (below).

Ten patients (2.8%) had received opioid analgesics within the last year, but were no longer taking them at the time of the PDMP query. Three patients (0.8%) died from their cancer during the QI project.

Pain Ratings and NarxCare Overdose Risk Scores


The patients’ mean pain rating without taking any pain medications was 8.9 out of 10 (SD = 1.2, Median = 9.0). The patient’s mean pain rating when taking their prescribed cancer pain medications was 3.9 out of 10 (SD 1.4, Median = 4.0), with an average reduction in pain rating was 5.0 points (SD = 1.5, Median = 5.0). This represented an average in reduction in pain of 56.3% with usage of the prescribed analgesic regimen. We recorded NarxCare Overdose Risk Scores for 144 of the 359 outpatients included in the QI project, as this measurement was first added to the E-FORSCE database in the third month of the project. The patients’ mean NarxCare Overdose Risk Score was 421 (median = 410, SD = 156.3). The E-FORSCE website recommends co-prescribing of naloxone to NarxCare Overdose Risk Scores greater than 450/999 and mentions that the 450/999 score correlates with an approximate CDC MED Equivalent of just 50 mg/day.

Discussion and Policy Implications

In theory, the Florida PDMP aspires to improve patient safety by embracing the concept that patients are best served by using a single prescriber for medically-appropriate controlled pain medications. Accordingly, each patient in our palliative medicine clinic receives written educational material describing the policies and procedures of the clinic, including a directive encouraging them to receive prescriptions for controlled pain medications from a single prescriber and use a single pharmacy to fill them. Our QI projected showed, however, that about three-quarters of outpatients in a palliative medicine clinic (PMC) housed within a regional cancer center had PDMP queries showing multiple prescribers for their controlled medications (even when account for errors in misidentifying a single prescriber as multiple prescribers). Only 21% of patient queries revealed a single prescribing physician. Fortunately, after careful investigation of each PDMP report, we learned that the vast majority of the patients were taking controlled medications in a medically legitimate way, with only two patients exhibiting medication aberrancy and no patients demonstrating medically inappropriate behavior. This observation seemed to reflect a frequently fragmented pattern of medical care rather than patient misbehavior. Based on patient comments during clinic encounters, the reluctance of physicians to prescribe opioid analgesics, even in a medically appropriate situation, often resulted in the provision of multiple, back-to-back short-term prescriptions, which led to an increased number of prescribers over a relatively short period of time.

Much of the reasoning behind Florida mandating a prescriber query of the PDMP was to uncover inappropriate patient behavior, which implicitly suggests the state believed that the rate of illegal behavior in patients was high. Our QI project suggests, however, that in a population of patients in an outpatient palliative medicine clinic, the risk of aberrancy is very low. Our QI project also suggests that the error rate of the Florida PDMP database (4.7%) was more than quadruple our patients’ aberrancy rate (0.6%) with no cases of medically inappropriate behavior identified. Given this, we express profound concern that Florida’s PDMP query requirements may worsen existing opioid stigma, making it even more difficult for palliative care patients to fill medically legitimate opioid prescriptions (Kollas CD, Boyer-Kollas B. Assessing Opioid Stigma in Oncology Outpatients Receiving Palliative Care for Cancer-Related Pain (S714). The Annual Assembly of the American Academy of Hospice and Palliative Medicine and the Hospice and Palliative Nurses Association: Education Schedule and Abstracts. Journal of Pain and Symptom Management February 2018;55(2):663-4.)

The introduction of the NarxScore onto the query result page of the Florida PDMP beginning in September 2019 added yet another variable for physicians to consider when prescribing controlled medications. Almost half of the patients for whom NarxScores were available in our QI project fell into a category for which Apriss recommends consideration of co-prescribed naloxone to reduce the risk of harm from an opioid overdose . None of the patients in our QI project experienced a need to use naloxone during the course of the project, but co-prescription of naloxone has gained support as a public safety measure to reduce opioid overdose deaths (AMA Task Force Reference). Our QI project does not allow an analysis of how the presence of the NarxScore itself might affect physician prescribing, but this would be an interesting area for future study. Based on our impressions of comments by fellow Florida physicians accessing E-FORSCE, we suspect that the NarxScore may introduce an additional “chilling effect” into the prescribing of controlled medications. Additionally, a very recent Florida law suggests that this chilling effect may be an intentional one. On June 25, 2019, Governor Ron DeSantis signed into law a requirement for Florida physicians to provide a written pamphlet entitled, “Information on Nonopioid Alternatives for the Treatment of Pain,” before prescribing any Schedule II analgesic to any patient (See here and here). Of note, the required pamphlet became available on June 28, 2019, just three days prior to the new law’s implementation date of July 1, 2019, and the sole version available as of this writing is written in English only. Moreover, this law also requires that prescribing physicians to inform patients explicitly that they have a right to refuse opioid analgesics, although patients have always possessed the right to refuse any unwanted treatments before implementation of the new law.

After our one-month assessment of this QI project, we posed a question that we hoped we could answer after gathering three months’ worth of information from PDMP queries: Is physicians’ time best spent querying a database with a higher error rate than the rate at which it identifies medically inappropriate behavior and which might lead to worsening opioid stigma and difficulty filling prescriptions for cancer pain medications? Unfortunately, we remain unable to answer this question definitively, although our current results affirmed that the error rate of E-FORSCE exceeds the rate at which patients exhibit medically inappropriate behavior. More importantly, in light of Florida’s most recent law and PDMP changes, we were probably asking the wrong question. We should have been asking who would be held accountable for the inaccuracies in the E-FORSCE database. That answer came in mid-June 2019, when the E-FORSCE website added a qualifying statement to each patient’s query, which reads as follows: “The Department of Health makes no claims, promises, or guarantees the accuracy, completeness, or adequacy of the contents of the information in the E-FORCSE® database, and expressly disclaims liability for errors and omissions in the contents. The records herein are based on information submitted by pharmacies and dispensing health care practitioners. Records on this report should be verified before any clinical decisions are made or actions are taken” (Italics are mine). This statement effectively passes the risk from errors within the Florida PDMP back onto physicians, essentially forcing them to independently “verify” the information produced by the query. So not only do physicians have to spend much more time when prescribing opioid analgesics, they also assume the potential risks of interpreting the Florida PDMP data, which may be incorrect about 5% of the time. These policy changes occurred despite announcements by the Centers for Disease Control and Food and Drug Administration that had begun to drive the pendulum of opioid policy back into a position of improved balance. While it remains unclear whether Florida physicians’ time is best spent querying a database, it seems clear that Florida’s politicians have doubled-down on a policy approach that bluntly endorses reductions in opioid prescribing, shifting risk for errors in the PDMP on the shoulders of Florida physicians and burdening physicians with mandated discussions that are clearly intended to discourage opioid prescribing, even in medically appropriate situations. Our patients deserve better.

Dr. Kollas has provided outpatient palliative care at Orlando Health UFHealth Cancer Center for over 17 years. He first wrote about his personal experiences with chronic illness and pain in a 1997 article in the “On Being a Patient” series in Annals of Internal Medicine. You can find him on Twitter at @ChadDKollas.

To cite this QI article: Kollas C. Mandated Queries of the Florida Prescription Drug Monitoring Program: A Three-Month Experience from a Cancer Center-Based Outpatient Palliative Medicine Clinic. Pallimed. November 2019. Available at: https://www.pallimed.org/2019/11/mandated-queries-of-florida.html

Pallimed | Blogger Template adapted from Mash2 by Bloggermint