Opioid Fills in Children Undergoing Surgery From 2011 to 2014

A Retrospective Analysis of Relationships Among Age, Initial Days Supplied, and Refills

Martha Wetzel, MSPH; Jason M. Hockenberry, PhD; Mehul V. Raval, MD, MS


Annals of Surgery. 2021;274(2):e174-e180. 

In This Article


Examining a pediatric surgical cohort drawn from a large population of privately insured individuals, we found a significant relationship between days supplied of the initial prescription and probability of getting a refill for 3 of the 5 procedures studied. This relationship was most evident for tonsillectomy and posterior spinal fusion. In contrast, refills for ACL repair remain relatively common regardless of days initially supplied. The contradictory relationships by procedure illustrate the importance of procedure-specific guidelines. In addition, further work should explore the effectiveness of opioids versus nonopioid analgesics for each surgery and the distribution of duration of pain, as nonopioid analgesic regimens have been shown to be effective at reducing the quantity of opioids needed to control pain in select situations.[23,24]

In general, procedures with higher refill rates also had higher initial fill rates. Our initial fill and refill rates were in line with procedure-specific literature on ACL repair, appendectomy, and posterior spinal fusion.[4,25,26] Orchiopexy stood out as a procedure with a relatively high initial postoperative opioid fill rate of 70.5% in conjunction with a very low refill rate. Further research on the pain levels and opioid usefulness for orchiopexy is warranted.

In agreement with prior research,[20,26] we found that older age was associated with a higher use of opioid medications after common children's surgical procedures. We found that older age was associated with higher perioperative opioid fill rates for most procedures and with higher refill rates for all procedures examined.

Perioperative opioid prescription fills were related to age for all procedures except for ACL repair and posterior spinal fusion, which had fill rates of 88.9% and 84.6%, respectively. Prior literature has shown about 15% of opioid prescriptions for postoperative pain in the pediatric population go unfilled,[27] which suggests that nearly all patients received opioid prescriptions for these 2 surgeries. The patient factors that affect the doctor's decision to write a script may differ from the factors that affect the patient's decision to fill the script. Thus, 1 explanation of our results on the relationship between age and perioperative opioid fills is that age tends to be a factor in clinicians' decisions to prescribe, but has a murkier relationship with patient fill decisions.

Our results indicate that expectations and standards for refills must be set on a procedure-by-procedure basis that accounts for patient age. We demonstrate in this paper that days supplied is a significant predictor of refills for some procedures even after controlling for MME, and therefore must be included along with MME in any recommendations. As a first step in the development of prescribing guidelines, we show procedure-specific days' supply associated with a 20% refill probability by age in Table 4. The different requirements by age are important to note, given that our data show that actual practice is not accounting for the influence of age on days supplied requirements. We also provide graphs showing the probability of a refill as a function of days supplied. This information facilitates tailoring practice to different cut-offs for acceptable refill rates.

Prescribing for humerus fracture and tonsillectomy, in particular, can be tailored to the patient's age. For humerus fracture, the moderate rate of perioperative opioid fills (47.1%) in conjunction with the relatively low number of days supplied required to meet patient needs indicates a potential area of overprescribing for young patients. In contrast, given the high quantities of opioids necessary to limit refill requests for posterior spinal fusion surgery, a 20% refill rate appears to be too ambitious of a goal. For this surgery, clinicians may need to accept and plan for a higher rate of refills. Prior research has shown high variability in pill consumption among spinal fusion patients,[26] indicating that across the board increases in days supplied to address refills would have considerable consequences in terms of leftover medication.

In addition to facilitating the fine-tuning of prescribing by procedure and age, these results indicate the unavoidability of a certain percentage of patients needing refills for select high-pain procedures. Given the likelihood of refill requests, it may be prudent to explain the logistics of acquiring a refill for a scheduled substance to patients and caregivers while still emphasizing that the importance of using the least amount of opioids needed. Furthermore, additional emphasis on the effectiveness of over-the-counter analgesics may also be necessary. These findings also demonstrate that further research on the use of opioids for cholecystectomy and ACL repair is warranted, and that new modalities of pain control are needed for posterior spinal fusion in particular.


There are several limitations to this analysis. First, it was not possible to distinguish between refills for postsurgical pain and opioids prescribed for nonsurgical pain. However, we believe it is unlikely that a high percentage of the apparent refills would be for conditions unrelated to the surgery, because we excluded individuals with ongoing opioid prescriptions in the year leading up to surgery. In addition, between 7% and 21% of patients filled an opioid prescription in the 30 days before the procedure. Some of these patients likely received a prescription for preoperative pain with the expectation of another opioid fill for postoperative pain if needed, whereas some of them likely filled their prescription for postoperative pain in advance of the procedure. To avoid biasing the results by inadvertently counting expected second prescriptions as refills, we limited our population for the refill analysis to patients who filled a script on the day of or day after surgery. As a sensitivity test, we repeated the refill analysis including patients who received a preoperative opioid prescription. The results did not change our conclusions.

Patient weight was not available in our data, necessitating the use of median weight by patient age and sex. Given that patients undergoing spinal fusion or cholecystectomy tend to be below and above the median weights, respectively, we performed an additional check for these procedures using the 25th and 75th percentile weights instead of the medians. The results of this check were nearly identical to the original analysis. Furthermore, we did not have access to information on the use of over-the-counter analgesics. Lastly, we were unable to account for patients who experience uncontrolled pain, but did not receive a refill.

It is possible that some of the declining probability of receiving a refill is due to clinician reluctance to refill a prescription several weeks after surgery. However, the fact that the refill rates for both cholecystectomy and ACL repair remain relatively similar regardless of the number of days supplied indicates that this is not a consistent phenomenon across procedures.

Because these data are limited to commercially insured individuals, our findings may not apply to the Medicaid population. However, given that the HCCI data represent approximately 1 quarter of commercially insured individuals, our findings are applicable to a wide swath of commercially insured pediatric patients.