Providers turning to health IT to combat the opioid crisis

Organizations such as Geisinger Health System are exploring innovative uses of data to limit prescribing and prevent addiction

Providers have begun to harness technology to improve their management of opioids and avoid misuse.

It’s not a moment too soon. The Centers for Disease Control and Prevention’s most recent reports, released in November, confirm that the opioid crisis is worsening. The number of drug overdose deaths in the United states in 2017 was 9.6 percent higher than in 2016. The rate of drug overdose deaths involving natural and semisynthetic opioids like oxycodone and hydrocodone has risen steadily since 1999. The rate of drug overdose deaths involving synthetic opioids such as fentanyl, which individuals frequently turn to when they can no longer obtain prescribed opioids, increased 45 percent between 2016 and 2017.

One of the key components to reversing these trends is to keep people from becoming addicted in the first place, says Jim Turnbull, CIO of University of Utah Health and Co-chair of the College of Healthcare Information Management Executives (CHIME) Opioid Task Force. Providers are uniquely positioned to address this problem because they are the ones prescribing the drugs.

“We want to make sure we’re not contributing to addiction. We don’t want to be the source of people getting addicted,” Turnbull adds.

CHIME’s Opioid Task Force was launched in early 2018 to leverage the knowledge and expertise of its members to find and share IT-based solutions to the opioid crisis. It is compiling examples of data-driven initiatives and raising funds for its Health IT Action Center, a web-based repository for healthcare organizations seeking resources. The Task Force has also partially completed a playbook for CIOs, says Turnbull.

However, using technology to combat the crisis is still relatively new. A KLAS report on opioid management published this past September found that most provider organizations relied mainly on their EHR vendor to help them with opioid stewardship.

Some enterprising providers are taking a more hands-on approach and directly applying health IT to address opioid prescribing in their organizations. Many of these initiatives are homegrown, using an entity’s own data, and are so recent that their effects have yet to be determined. But even those projects in their infancy look promising.

“We realized we needed to see how data and IT plays into this. We’re using data as a platform to make cultural change,” says Alexander Garza, MD, chief quality officer at St. Louis based SSM Health.

Mining the EHR

Danville, Pa.-based Geisinger Health System is one of the first provider organizations to spearhead the use of data to assess its opioid prescribing habits. Pennsylvania has one of the highest death rates from opioid overdoses in the nation, according to the CDC.

“About three to four years ago, we realized that we were prescribing more opioids than we should. So we started using analytics to see how much was being prescribed and compare prescribers to others in the network,” says Geisinger’s Senior Vice President and CIO John Kravitz.

The health system created a provider dashboard linked to its EHR, using Tableau’s platform, to display the volume of opioid prescribing. The dashboard identified prescribing patterns among the providers and flagged several “heavy” prescribers. Geisinger then used the information to focus on these prescribing outliers and instill best practices in prescribing overall.

Opioid prescriptions have declined from 60,000 per month across the system to 22,000 a month, and the number continues to drop. “No one wants to be on the high end of opioid prescribing. The doctors are very aware of the numbers in the dashboard,” says Kravitz.

The dashboard was implemented in tandem with Geisinger’s redesign of its surgical program to reduce opioid use and improve the patient experience, which includes presurgical consults alerting patients to the pain levels they’ll have and the provision of alternative pain medications. That pilot program, launched in June 2017 and called ProvenRecovery, drove an 18 percent decrease in opioid usage across the organization, says Kravitz.

On November 15, Geisinger announced that the program will be rolled out across 42 surgical procedures impacting approximately 15,000 surgery cases a year, with the goal of reaching 100 surgical specialties by the end of 2019.

“We’ve started closing the spigot on getting new people addicted. It will be instrumental to have treatment programs to help [those already addicted] to help them get out of it,” says Kravitz.

SSM Health is taking a similar approach, extracting the data from its EHR and creating a heat map colored coded by diagnosis and dose to flag outlier opioid prescribers. “The heat method was used to develop a performance improvement tool. It will become part of a dashboard going forward,” says Garza. SSM is now working on how best to share the results with prescribers.

Kravitz recommends that other providers dive into this data.

“This is not rocket science. We all have the capability to produce a dashboard, pull the data and see what providers are ordering and how big a problem you have. When we saw our volume, our eyes were wide open. We had no idea,” he says.

Data analytics predict risk

Health IT can also be used to pinpoint which patients are more likely to become addicted.

For example, researchers at the University of Colorado Anschutz Medical Campus are developing a statistical model to better predict which hospitalized patients prescribed opioids are more at risk of progressing to chronic opioid use after discharge.

They devised a study, using data from the EHR used by Denver Health Medical Center, an urban safety net hospital. They identified patient-specific variables which were highly associated with the progression to chronic opioid therapy (COT), such as high opioid requirements during hospitalization, receipt of an opioid at hospital discharge or a history of substance use disorder.

The model correctly predicted future COT in 79 percent of the patients and no COT correctly in 78 percent of the patients. The health system is now in the process of validating the statistical model in a university hospital setting, according to Susan Calcaterra, MD, a fellow in addiction medicine at the CU School of Medicine and lead author of the study.

“It’s pretty clear we’re in an epidemic and prescribing is part of that. There are ways to change your practice that’s beneficial to the patient,” says Calcaterra.

After the model is validated, it can be integrated into the health system’s EHR. “It runs in the background. As one puts the order for an opioid in the computer, information is provided to the doctor if the patient is at high risk of COT. It presents information so that you can act on it, such as tapering opioids before discharge, incorporating pain management to decrease the opioids required, and talking to the patient about the risks,” she says.

Calcaterra recommends that other healthcare entities adopt this kind of data analytics.

“Harnessing EHR data is a new area, especially when using machine learning to predict things. [But] this information is already captured in the EHR, so it doesn’t take extra work on the part of the physician. The goal is not to increase the workload of those already busy,” she says.

Changing the EHR itself

Some health systems are tweaking their EHRs to limit opioid prescribing. University Utah Health developed an algorithm using Fast Healthcare Interoperability Resource (FHIR) standards that automatically measures the morphine equivalent daily dose (MEDD) of an opioid being prescribed and calculates if the dose exceeds the CDC recommended amount, says Turnbull.

Previously this measurement was performed manually, which takes longer and is more prone to error. If an order exceeds the MEDD, the algorithm will send the clinician an alert showing the clinician the calculation and measure.

Most EHR vendors have been automating this function within the last few months, but are not using FHIR, which makes it transportable from EHR to EHR, says Turnbull.

Since this change has just been deployed, there isn’t enough data yet to determine whether or how the alert has affected prescribing, but the initial reviews are positive. “Doctors have said it’s great,” says Turnbull.

SSM Health has established a team to review the 5,000 order sets in its EHR and identify those that have opioid orders baked into them. Once these have been identified, the team will assess from a clinical standpoint whether an order set should be revised so that it won’t necessarily default to an opioid prescription, such as after a Caesarian section.

“Back in medical school pain was seen as a fifth vital sign. But now we know that pain is manageable and you don’t necessarily need opioids,” says Garza. The affected order sets will be rearranged so that opioids are a last resort. “We’re organizing the EHR to give clinicians better resources regarding pain,” he says.

Another strategy is to improve EHR functionality, as planned by the Addiction Institute of Mount Sinai and Center on Addiction, a new alliance announced in October focused on younger people with opioid addiction.

“Many substance use programs are archaic technologically. We should be advancing technological strategies, such as looking for biomarkers in the medical records and to use genetics to improve dosing,” says Yasmin Hurd, Ward-Coleman Chair of Translational Neuroscience at the Icahn School of Medicine at Mount Sinai and Director of the Addiction Institute of Mount Sinai.

Digital pills to monitor opioids

Not all digital data originates in the patient’s medical record. Boston based Brigham and Women’s Hospital piloted a study in 2017 to use “digital pills” to track patterns of opioid use of patients who received a prescription to take oxycodone as needed after treatment for acute fractures.

“We wanted a more realistic way to see how people use and take the pills,” says study senior author Edward Boyer, MD, of the division of medical toxicology in the Hospital’s Department of Emergency Medicine.

The researchers used the eTectRx ID-Cap system; each pill has a unique radiofrequency emitter and a gelatin capsule containing an oxycodone tablet. When the capsule dissolves, the medication is released and the emitter energized. The radio signal is read by a device worn by the patient and transmitted via a smartphone app.

The digital tracking system revealed that the patients self-administered significantly fewer opioids than expected to manage pain, indicating that they had been prescribed too many. On average the patients ingested only six pills even though they were given 21, a seven-day supply. Most doses were ingested within the first three days.

“Opioid analgesics are useful in extremely limited doses and an extremely limited time. There are no guidelines on PRN dosing. Our approach was to interrupt the narrative that people needed opioids for a long time. [Studies like these can help us] better assess people’s response to acute pain,” says Boyer.

Brigham and Women’s Hospital is now working to use the digital pills in studies with HIV patients, who often receive opioids for pain. “If one has opioids and HIV there’s not only a risk of more severe opioid use but also of depression, decreased affect and poor HIV adherence. The digital pills are an indisputable measure of ingestion. We’re doing it to figure out the intersection [of opioid use] with HIV outcomes,” says Boyer.

Beefed up electronic prescribing

Electronic prescribing of controlled substances (EPCS) is an increasingly popular use of health IT to reduce some of the misuse of opioids, such as drug diversion, prescription forgery and doctor shopping. Geisinger, one of the early adopters of EPCS, enabled such electronic prescribing in August 2017, with two factor authentication via cellphone and Bluetooth and an encrypted solution for opioids. Eighty percent of controlled substances are now electronically prescribed, says Kravitz.

Most of the benefits in using health IT to combat opioids are “soft” savings, says Kravitz. “We’re more concerned about patients not getting addicted,” he points out.

However, moving to EPCS sped up processing from about 15-20 minutes per prescription to less than three minutes. The increased efficiencies resulted in a savings of about $1 million, he estimates.

Enhanced analytics for PDMPs

State prescription drug monitoring programs (PDMPs) are becoming more effective, thanks in part to advances in health IT that integrate state PDMP data into the work flow of an EHR or pharmacy management system. This eliminates the need for users to manually log into the PDMP website to make a separate query about a patient’s opioid use.

“The [integration] software is a scalable, efficient and modern way to run these data bases,” says Rob Cohen, president of Louisville, Ky based Appriss Health, which provides the software planform for 43 of the nation’s 53 PDMPs.

However, basic PDMPs only list a patient’s prescriptions; they don’t capture a patient’s complete opioid history or analyze the data in the PDMP. “So we need to shift the response to that, as well,” says Cohen.

One of the ways to do so is by using a more advanced PDMP solution with additional data to make it more useful. For example, Appriss Health has developed a substance use disorder platform called NarxCare that includes analytics on additional data, such as whether a patient has ever received Narcan to reverse an opioid overdose, information supplied by the state.

“Without the additional information, say a patient showed up in the ED. His PDMP history might be zero, but he may have had overdoses from illicit opioid use and the physician [unwittingly] writes a prescription for opioids,” Cohen warns.

The NarxCare platform also provides a series of clinical risk indictors, or scores, built on an algorithm and based on a patient’s history in the PDMP, such as the number of overlapping prescriptions. The scores serve as awareness tools, so a doctor can quickly surmise whether she needs to look more deeply at the data base or ask the patient more questions, says Cohen. NarxCare will also help prescribers connect patients with additional resources if needed, such as medication assisted treatment.

Nine states have integrated NarxCare into their PDMP platform; eight more are planning on doing so in the near future, says Cohen.

North Carolina replaced its homegrown PDMP with Appriss Health’s online portal and added NarxCare to it at the urging of several health systems, according to Kody Kinsley, Deputy Secretary for Behavioral Health and Intellectual Disabilities at the North Carolina Department of Health and Human Services in Raleigh. “The providers were nipping at my heels for this. They recognize the value,” he says.

The state has had 40,000 users of NarxCare since it went live on September 19, 2018; half of those users have done so daily.

Homegrown patient-centric apps

Health systems are also creating their own apps to assess and reduce the number of opioids patients take. Geisinger is integrating data from a pain app it developed that measures patient-reported pain and other data into the dashboard and the patient’s medical record.

“It’s a communication between the patient and provider, a way of getting away from using opioids for pain management,” says Kravitz.

Brigham and Women’s Hospital is piloting a study to see whether music delivered through an app can help limit pain and thus reduce the need for pain medication. The app measures a patient’s heart rate and alters the music to relax the patient more, which in turns reduces the patient’s anxiety. A patient may then not take a scheduled opioid pill, says Boyer. The app has been tested on 60 people; the researchers are about to analyze the data.

Data part of the arsenal

Health IT by itself won’t halt the opioid crisis. “Technology is a tool, one part of the puzzle,” says Turnbull.

However, when used together and with other tools – such as education – the data can make an impact. Turnbull recommends that providers continue to maximize health IT in this manner.

“Get involved. CIOs need to take a leadership role, stay on top of what’s going on in the industry and champion use of these tools,” he says.

Others agree. “We’ve thought too narrowly in addressing the epidemic. We need new ideas and to go big,” says Hurd.

More for you

Loading data for hdm_tax_topic #better-outcomes...