Impact of the 2017 FDA Drug Safety Communication on Codeine and Tramadol Dispensing to Children

On April 20, 2017, the Food and Drug Administration (FDA) issued a Drug Safety Communication (DSC) announcing a contraindication against codeine and tramadol use in children aged <12 years and against tramadol use in children aged <18 years after tonsillectomy and/or adenoidectomy. It also included a warning against codeine and tramadol use in adolescents with obesity or conditions that may increase the risk of respiratory depression. This DSC followed an FDA review identifying risks of respiratory depression and death associated with pediatric use of these medications and expanded upon the 2013 boxed warning against codeine use after tonsillectomy and/or adenoidectomy in children aged <18 years.¹

Codeine prescribing decreased after the 2013 boxed warning.^2,3 The national impact of the 2017 DSC remains unknown. We examine the change in codeine and tramadol dispensing for treatment of pain to children aged <18 years after the 2017 FDA DSC.

Methods

We extracted 2014–2019 opioid-dispensing data for patients aged <18 years from the IQVIA Longitudinal Prescription Database, which covers 86% to 92% of US retail pharmacy prescriptions.⁴ Prescriptions for cough/cold formulations, from veterinarians, or obtained by mail-order, were not included.

We conducted interrupted time series analysis using segmented regression to evaluate the association between the DSC and the monthly number of patients dispensed ≥1 codeine or tramadol prescription, by age (<12 years [“children”]; 12–17 years [“adolescents”]) per 100 000 persons in the United States population in each age group.⁵ Because of outcome overdispersion, we used a quasi-Poisson model using population size as the offset, including terms for: continuous month, an indicator for May 2017 or after (level change), and their interaction (slope change). We included quarterly fixed effects to address seasonality. After controlling for seasonality, there was no evidence of first-order autocorrelation. To account for secular trends and evaluate any substitution effects, we also investigated opioids other than codeine and tramadol (“other opioids”).

We conducted 2 sensitivity analyses. The first excluded April 2017 to June 2017 as an implementation period. The second excluded 2019 tramadol data since an abrupt, unexplained, and sustained decrease in tramadol-dispensing to children occurred in January 2019.

Analyses were conducted in R version 4.1.1. This study did not qualify as human subjects research.

Results

From 2014 to 2019, there were 1 920 947 codeine prescriptions dispensed to 1 618 520 children and 2 396 972 prescriptions dispensed to 1 991 403 adolescents. A total of 514 090 tramadol prescriptions were dispensed to 190 405 children and 1 055 458 prescriptions were dispensed to 674 545 adolescents. Codeine and tramadol dispensing decreased between the pre- and post-DSC periods (Table 1).

TABLE 1.

Absolute Median Dispensing Rates by Opioid Type Before and After the FDA DSC and Segmented Regression Results, by Age Group, 2014–2019

	Absolute Median Monthly Rates		Segmented Regression Analyses
	Absolute Median Monthly Rate of Dispensing Before DSCa	Absolute Median Monthly Rate of Dispensing After DSCa	Continuous Monthb IRR (95% CI)	Level Changeb IRR (95% CI)	Slope Changeb IRR (95% CI)
Codeine
<12 y	88.45	15.59	0.983c (0.981–0.985)	0.742c (0.684–0.806)	0.950c (0.944–0.955)
12–17 y	163.53	85.99	0.996c (0.993–0.998)	0.829c (0.762–0.898)	0.982c (0.978–0.986)
Tramadol
<12 y	18.96	11.49	0.986c (0.983–0.989)	1.223c (1.068–1.401)	0.970c (0.963–0.978)
12–17 y	70.98	40.47	0.990c (0.987–0.990)	0.954 (0.898–1.012)	0.993c (0.990–0.996)
Other opioidsd
<12 y	144.26	94.51	0.990c (0.988–0.992)	1.065 (0.990–1.145)	0.989c (0.986–0.993)
12–17 y	545.38	377.66	0.991c (0.987–0.994)	0.998 (0.879–1.132)	0.998 (0.992–1.005)

Because of differences in labeling changes by age group, analyses are stratified by age group: <12 years and 12 to 17 years. Opioid-dispensing data are from IQVIA Longitudinal Prescription Database, 2014 to 2019, which represents 86% to 92% of all retail transactions in the United States. It excludes all cough/cold formulations.

^aMedian monthly rate of unique patients dispensed an opioid per 100 000 persons of that age stratum nationwide.

^bInterrupted time series with a segmented quasi-Poisson regression controls for seasonality of dispensing to children and adolescents using fixed quarterly effects. Continuous month is the change in outcome associated with time-unit increase throughout the study period (total = 72 months). Level change is the immediate changes after the DSC (May 2017). Slope change is the time in months after the DSC (total = 32 months).

^cDenotes statistical significance at the α = 0.05 level.

^dOther opioids include opioids dispensed that were not included in the DSC, such as hydrocodone, oxycodone, morphine, hydromorphone, fentanyl, and other less-commonly dispensed opioids.

For codeine, among children, the DSC was associated with an immediate 26% decrease in dispensing rates (incidence rate ratio [IRR] 0.742, 95% confidence interval [CI]: 0.684–0.806), followed by a sustained 5% decrease (IRR 0.950, 95% CI: 0.944–0.955). Among adolescents, the DSC was associated with an immediate 17% decrease in codeine-dispensing rates (IRR 0.829, 95% CI: 0.762–0.898), followed by a sustained 2% decrease (IRR 0.982, 95% CI: 0.978–0.986) (Fig 1, Table 1).

FIGURE 1.

Monthly time series of codeine- and tramadol-dispensing rates per 100 000 persons, by age group, 2014 to 2019.

For tramadol, among children, the DSC was associated with an immediate 22% increase in dispensing rates (IRR 1.223, 95% CI: 1.068–1.401), followed by a sustained 3% decrease (IRR 0.970, 95% CI: 0.963–0.978). Among adolescents, the DSC was not associated with any immediate change; there was a <1% sustained decrease in tramadol-dispensing rates (IRR 0.993, 95% CI: 0.990–0.996) (Fig 1, Table 1).

For other opioids, the DSC was not associated with an immediate change, but was associated with a sustained decrease among children. The DSC was not associated with any change among adolescents (Table 1).

Accounting for an implementation period revealed similar results. After excluding 2019 data, the DSC was not associated with a slope or level change in tramadol dispensing among children; for adolescents, the results were similar to our primary analysis (Supplemental Table 2 and Supplemental Fig 2).

Discussion

The 2017 FDA DSC was associated with immediate and sustained decreases in codeine dispensing to children and adolescents. Notably, codeine continues to be dispensed despite known safety risks, though at substantially lower rates in recent years.

Our findings align with studies evaluating the 2013 boxed warning against codeine, but differ from a recent study finding no change in prescriptions for codeine cough and cold medications (CCMs) after this DSC.^2,3,6 The difference may be because of lower baseline prescribing of codeine CCMs compared with codeine pain medications and earlier efforts to decrease use of codeine as an antitussive agent and CCMs more broadly.^7,8

When including data through 2019, the DSC was associated with a level increase in tramadol dispensing to children and a slope decrease. However, the DSC was not associated with either a level or slope change when excluding 2019 data. This discrepancy may be because of the abrupt decrease in tramadol dispensing starting in January 2019, which may have been driven by other factors, potentially including data error. Research using other data sources is needed to confirm the effects of the DSC on tramadol dispensing to children.

Despite decreased codeine and tramadol dispensing, there were no increases in dispensing for other opioids. Consistent with data showing decreasing overall pediatric opioid dispensing, these findings make substitution of other opioids for codeine or tramadol less likely.⁹

IQVIA data does not contain patient diagnoses, so we could not determine if decreased adolescent codeine dispensing was specific to the population targeted by the DSC. Future research should investigate specific prescribing practices, including prescriber specialties and linked patient diagnoses or procedures. Despite the DSC, codeine and tramadol continue to be dispensed to children, suggesting further interventions are needed to ensure safe pain medication prescribing to children.

Glossary

CCMs

cough and cold medications

CI

confidence interval

DSC

Drug Safety Communication

FDA

Food and Drug Administration

IRR

incidence rate ratio