Background & Aims

Prescription drug monitoring programs (PDMPs) are state-level interventions aimed at improving opioid prescribing, informing clinical practice and protecting patients at risk. California’s implementation of PDMP consists of a database of all schedule II-IV controlled substance prescriptions dispensed in California. Studying PDMP could resemble a retrospective database study to evaluate potential associations between exposure to certain doses, types or duration of opioids and selected outcomes (which often lacking in this kind of de-identified data) in a population large enough to provide sufficient precision, with nearly continuous follow up, and with only few exclusion criteria which often limit participation in clinical trials. This approach would improve the generalizability of the results and allow an evaluation of confounding factors as well as increasing the precision of the estimated association.

Methods

We plan to achieve our aim with a multi-step process beginning with data cleaning, formatting and anomaly detection. Next, we will standardize the prescription data by converting each opioid medication into morphine milligram equivalent (MME) per day per prescription, which allows us to then make comparisons between different prescriptions. Exploratory data analysis will be used to examine prescription patterns by time trends, patient demographics, location and prescriber specialty. Next, we will define calculated intermediate outcome variables based on available data such as sustained opioid use, MME >100 per day, escalating doses over time, opioid shopping (multiple prescribers) and initiation of medication assisted treatment programs such as methadone or buprenorphine. These intermediate outcomes will then be used to create machine learning models to predict the risk of developing opioid misuse/ abuse.

Results

State of California population was about 38-39 million during the study period 2012-2020 which staggering 31 million of them received opioid prescription less than a quarter (90 days). More than half, 17 million did not receive opioids beyond one quarter. Rate of patients with opioid prescription in multiple quarters over a decade continues to decline rapidly to about half million patients with prescriptions in 8 of 36 separate non-consequent quarters. Number of patients with opioid prescription in multiple quarters continues to decline as we look at higher number of quarters to about 30,000 patients with 34 non-subsequent quarters. For a reason that is not clear to us this number jumps up to 103,000 patients who have received opioid prescription all 36 quarters (duration of our study) without any gap.

Conclusions

To better understand the extent of prescription opioid use, misuse and abuse we created a methodology to study California PDMP database. This methodology will allow us to:

1. To evaluate the strength of prescription-related factors associated with chronic continuous use of
opioids.
2. To use this model to study chronic continuous use of opioid that could be derived from PDMP
records alone.
3. To reduce the model to the fewest and simplest possible data elements that could inform a
proactive alert aimed at prescribers.
4. To assess model sensitivity and specificity.
5. To assess model generalizability.

References

[1]Robert Dufour, PhD; Jack Mardekian, PhD; Margaret K. Pasquale, PhD; David Schaaf, MD; George A. Andrews, MD, MBA, CPE, FACP, FACC, FCCP; and Nick C. Patel, PharmD, PhD B. Understanding Predictors of Opioid Abuse: Predictive Model Development and Validation. Am J Pharm Benefits 2104;6:209–15.
[2]Geissert P, Hallvik S, Van Otterloo J, O?Kane N, Alley L, Carson J, et al. High-risk prescribing and opioid overdose. Pain 2018;159:150–6. doi:10.1097/j.pain.0000000000001078.
[3]Ferris LM, Saloner B, Krawczyk N, Schneider KE, Jarman MP, Jackson K, et al. Predicting Opioid Overdose Deaths Using Prescription Drug Monitoring Program Data. Am J Prev Med 2019;57:e211–7. doi:https://doi.org/10.1016/j.amepre.2019.07.026.
[4]Moons KGM, Altman DG, Reitsma JB, Ioannidis JPA, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): Explanation and Elaboration. Ann Intern Med 2015;162:W1. doi:10.7326/M14-0698.

Presenting Author

Siamak Rahman

Poster Authors

Siamak Rahman

MD

UCLA

Lead Author

Moon-Seong Jeong

UCLA

Lead Author

Boyang Fu

UCLA

Lead Author

Sriram Sankararaman

UCLA

Lead Author

Topics

  • Epidemiology