ResearchPod Summary
As opioid and stimulant overdose deaths remain a significant public health concern, researchers sought to identify the demographic and clinical factors associated with the concurrent prescribing of these two drug classes. While both substances are monitored individually, there is limited evidence regarding the patterns that lead to patients receiving both simultaneously.
This retrospective study analyzed data from the South Carolina Department of Public Health (SCDPH) Prescription Drug Monitoring Program (PDMP) between 2016 and 2021. The researchers examined over 9.7 million opioid prescriptions to identify those that overlapped with a stimulant prescription for at least 30 days. They used generalized estimating equations to assess the association between coprescribing and variables including patient age, sex, opioid supply length, and daily morphine milligram equivalents (MME).
Of the opioid prescriptions analyzed, 1.65% involved a concurrent stimulant prescription. The study found that patients aged 50 and younger were 2.74 times more likely to receive both medications compared to older patients. The strongest predictor was the duration of the opioid prescription; patients with an opioid supply of 30 days or longer had 4.38 times the odds of receiving a concurrent stimulant. Additionally, female patients and those with a daily MME between 50 and 100 mg showed a statistically significant, though smaller, increased likelihood of receiving both drugs.
Understanding the factors that drive opioid-stimulant polypharmacy is essential for clinicians to identify patients at higher risk of adverse outcomes, such as fatal overdose. The findings suggest that younger patients and those on longer-term opioid therapy are particularly vulnerable to this combination. These results provide a foundation for future research to investigate whether specific medical diagnoses—such as chronic pain or obesity—or socioeconomic factors drive these prescribing trends, ultimately helping to inform safer clinical practices.
Alex: Welcome to another episode of ResearchPod. Today, we're looking at a study from the Southern Medical Journal that tackles a persistent, often hidden risk in clinical practice: the co-prescribing of opioids and stimulants.
Sam: That's right. These two drug classes are individually high-risk, but their combined use is particularly dangerous. And the central puzzle is that while we have plenty of data on overdose mortality, we've been largely blind to the prescribing architecture that leads to these combinations in the first place.
Alex: So is this study essentially trying to map the demographic and clinical markers that predict when a patient ends up on both?
Sam: Exactly. The authors drew on six years of data from the South Carolina Prescription Drug Monitoring Program to identify the factors driving this polypharmacy. Their argument is that the risk isn't simply about high-dose requirements—it's tied to identifiable patterns in patient age, gender, and the duration of opioid treatment itself.
Alex: If you can identify the "who" and the "how" behind these prescriptions, you might be able to intervene before the combination becomes a safety issue. How did they model these relationships?
Sam: They used Generalized Estimating Equations—GEE. The design choice matters here. Standard logistic regression assumes each prescription is an independent event, but a single patient typically generates multiple prescriptions over time. GEE clusters observations by patient, accounting for that within-person correlation so the standard errors stay valid and the odds ratios aren't inflated by individual-level dependence. It's the same logic as clustering students by teacher in an educational study.
Alex: You're partitioning out the patient-level variance before estimating the population-level effects. What did the model actually find?
Sam: The load-bearing finding is the association with prescription duration. Patients on opioids for thirty days or more had an odds ratio of around 4.4 for receiving a concurrent stimulant—a substantial signal. They also found that patients fifty and younger were nearly three times as likely to receive a concurrent stimulant compared to older cohorts.
Alex: And the gender and dosage variables—were those in the same range?
AI-generated third-party summary by ResearchPod. Not official content or an endorsement by the paper authors or affiliated organizations.
Sam: Significant, but with smaller effect sizes. Being female carried an odds ratio around 1.6, and landing in the moderate daily dose range—fifty to a hundred morphine milligram equivalents—added an odds ratio around 1.2. Statistically meaningful, but duration is clearly the dominant driver in this model.
Alex: That's worth sitting with. The longer a patient is on an opioid, the more likely the clinical trajectory shifts toward adding a stimulant—possibly to manage side effects like sedation. How stable is that finding?
Sam: They ran a sensitivity analysis comparing the full dataset against a restricted version that excluded records with implausible values—extreme ages, outlier MME figures. The odds ratios barely moved, which suggests the main findings aren't artifacts of noisy data at the margins.
Alex: But there's a significant gap in what the data can actually tell you. Without diagnostic codes, how do you distinguish inappropriate polypharmacy from a medically necessary combination?
Sam: That's the primary limitation, and the authors are candid about it. Without ICD-10 codes, you can't separate a stimulant prescribed to counteract opioid-induced sedation from one prescribed for an unrelated comorbid condition. The study gives you the pattern—the demographic and duration-based markers—but the clinical intent behind each prescription remains unobserved.
Alex: So the "what" is reasonably well-characterized, but the "why" is still a black box. Where does that leave a clinician trying to act on this?
Sam: It points toward better integration at the point of care. The real utility would be in clinical decision support systems that don't just surface PDMP data, but cross-reference it with EHR diagnostic codes in real time. That would allow for targeted alerts when a new prescription creates a high-risk profile—giving the prescriber the context to judge whether the combination is evidence-based or a signal worth investigating.
Alex: It's a meaningful shift in framing—moving from retrospective mortality analysis toward prospective, patient-level prescribing safety. Thanks for walking through the model mechanics, Sam.
Sam: Understanding the structural patterns is the necessary first step. The clinical guardrails can only be as good as the data architecture supporting them. Thanks for listening to ResearchPod.