Opioid and stimulant co-prescribing: our efforts and remaining open questions—authors' reply

We thank Dr. Xu and colleagues, Dr. Palis and colleagues for providing the insightful comments regarding our manuscript titled “The recent trend of twin epidemic in the United States: a 10-year longitudinal cohort study of co-prescriptions of opioids and stimulants.”¹ We would like to take the opportunity to address issues they raised.

Xu and colleagues² argued about the stimulant exposure, we agree that longer enrollment could increase the likelihood of opioid exposure, and potentially increase the chance of observing co-occurrence of opioid and stimulant prescriptions, which might not be truly associated in clinical usage. In our original model design, including stimulant prescription as a time-varying covariate into trajectory model was the approach to account for time-dependent variations of stimulant exposures. This approach captures the time and frequency of stimulant use relative to opioid prescribing and helps reduce bias from different follow-up periods. To further address this concern, we conducted further analyses with shorter enrollment time intervals. Specifically, we identified a subcohort of 2,281,702 patients who received at least two opioid prescriptions within only a two-year observational period and re-estimated opioid dose trajectories. To evaluate the association between stimulant exposure and opioid dose trajectories, we employed a logistic regression model comparing two trajectory groups—the decreasing dose group and the increasing dose group with similar dosage ranges. We then estimated odds ratios (ORs), 95% confidence intervals (CIs), and p-values for two stimulant exposure definitions: (1) index stimulant prescriptions (Index STM), defined as the number of stimulant prescriptions within the one-year period prior to the initial opioid prescription, and (2) stimulant co-prescription (STM cnt), defined as the total number of stimulant prescriptions during the two-year study period.

As shown in Table 1, both stimulant exposure prior to opioid initiation (Index STM) and stimulant co-prescription during follow-up (STM cnt) remained significantly associated with increased odds of being in the increasing group compared to the decreasing group. These results support our original conclusion that the stimulant prescription has impacts on opioid dose trajectories, even within a much shorter follow-up period.

Table 1.

Odds ratio from a logistic regression model estimating trajectory membership for the decreasing dose group (reference) vs the increasing dose group.

	OR (CI)	p-value
Index STMa	5.442 (3.927, 7.542)	<0.0001
STM cntb	1.389 (1.330, 1.451)	<0.0001

OR, odds ratio; CI, confidence interval.

^aThe number of stimulant prescriptions within the one-year period prior to the initial opioid prescription.

^bThe total number of stimulant prescriptions during the two-year study period.

We fully acknowledge the beneficial effects of stimulant drugs as commented by Xu and colleagues. As of today, stimulant medications remain to be the primary treatment option of ADHD.³ However, we also need to face the fact that stimulant medications can be addictive when misused. Misuse of stimulants could lead to a cycle of escalating drug use, tolerance, and withdrawal symptoms, when reducing or stopping their use.⁴ During the past decade, the global prevalence of stimulant use has steadily increased and statistics from CDC showed that stimulant misuse has led to a significant increase in overdose deaths in the United States.5, 6, 7 Prescription stimulants is one of factors causing this issue, although it is often combined with illicit stimulants overdose deaths.

Studies have suggested the co-use of opioids and stimulants can increase the risk of overdose and other adverse health outcomes. In light of this, we were motivated to conduct this national-level study to better understand longitudinal co-prescription patterns of opioid and stimulant drugs across U.S. healthcare systems. Our study was mainly focused on pattern recognition and association analysis between opioid and stimulant prescriptions over an extended period. How to explain these patterns and its clinical applications was our next step of research.

Our results did not suggest the less importance of stimulant drugs as a critical therapeutic agent and did not recommend to reduce or stop stimulant prescriptions, as well as co-prescriptions. As one of first national investigations on this topic, we were aiming to provide a high-level analysis of the trend of co-prescriptions to research community and our finding suggests an urgent need for more investigations of specific clinical scenario of co-prescriptions associated with these patterns. How to differentiate the benefits and potential harmful effects of this prescription is a topic that requires further research.

We agree with Xu and colleagues that non-prescribed synthetic drugs are currently more concerning, and becoming more important risk factors. However, we cannot confirm that the current epidemic is purely driven by non-prescribed drugs. With that, we also fully agree with the point suggested by Palis and colleagues⁸ on the importance of harmful effects associated with prescribed opioids and stimulants.

According to 2023 NSDUH Annual National Report, physician prescriptions still account for almost 50% as the source of misused medicines and the vast majority of people who misused opioids were from prescription pain relievers.⁹ The impact of prescribed drugs is also reflected by significant number of overdose deaths where prescribed opioid or stimulants were involved. Furthermore, we should be aware of multiple symptoms associated with stimulant prescription misuse, such as anxiety, stress and internal restlessness,¹⁰ which we were not able to include in our current study due to the requirement of patient-level diagnosis records. Our study suggested that “co-prescriptions could serve as a persistent driving force” for twin epidemic based on patterns we observed in this large dataset.

Our study was limited by missing some important information in the dataset. For example, the dataset was de-identified and missed outpatient and illicit drug records, which restricted us to investigate detailed clinical context associated with opioid and stimulant prescriptions. But our results highlight the complexity of potential interactive contributions of opioid and stimulant drugs in healthcare system over the last ten years. We hope our study serves as one of first steps to better understand and address this significant healthcare issue.

Contributors

WS, SL and PZ conceptualized and drafted the response letter. All authors reviewed and approved the final version.

Declaration of interests

DWB reports grants and personal fees from EarlySense, personal fees from CDI Negev, equity from ValeraHealth, equity from Clew, equity from MDClone, personal fees and equity from AESOP, personal fees and equity from FeelBetter, and personal fees and equity from Guided Clinical Solutions, outside the submitted work. RDU serves as a secretary of the Society for Ambulatory Anesthesia (unpaid), a treasurer of the Association of Anesthesia Clinical Directors (unpaid), and a vice president of ERAS USA (unpaid), and reports fees from AcelRx. The other authors declare no competing interests.