Pharmacotherapy research landscape and knowledge gaps of opioids in maternal and pediatric populations

Abstract

The use and misuse of opioids has surged in the past decade, with nearly half of the users being female. Although opioid use is lower among pregnant women, trends mirror the general population. While pediatric exposures largely occur through prescriptions. This review presents a novel landscape analysis of pharmacology knowledge gaps in opioids in the maternal and pediatric populations. We queried PubMed for studies on 27 opioids, focusing on pharmacokinetics (PK), and pharmacoepidemiology (PE) or clinical trials (CT) in maternal and pediatric populations. English‐language publications were included, and data were synthesized to identify gaps. Additionally, MarketScan claims data and United States Food and Drug Administration (FDA) drug labels were analyzed to compare scientific evidence, opioid prescriptions/orders, and FDA recommendations. Morphine, fentanyl, methadone, and buprenorphine are the most researched opioids in PK and PE/CT literature in both populations, but hydrocodone, oxycodone, and codeine are the most prescribed. Nine opioids lack FDA labels, and four of the 18 labeled drugs lack any human data. Hydrocodone, oxycodone, and codeine labels include lactation‐focused PK information, with some pediatric clinical data for the latter two. Seven opioids lack PK and PE/CT studies in the maternal population, and PK research is absent for seven opioids, and PE/CT data is lacking for eight opioids in the pediatric population. PK studies often focus on labor, delivery, and lactation accompanied by neonatal data, whereas pregnancy research mainly occurs in PE studies. In pediatric populations, study types are evenly distributed among children, but PE studies focus more on adolescents. Drug concentration is the most reported parameter in PK studies, and neonatal opioid withdrawal syndrome (NOWS) is a key outcome in both PK and PE studies. NOWS is also researched more using real‐world data, whereas neurodevelopmental outcomes are often captured in prospective observational studies. There is substantial disparity between the most commonly researched and prescribed opioids. In particular, the opioid pharmacology knowledge gaps are larger in pregnant women and for the highly prescribed opioids hydrocodone and oxycodone. The limited human data in FDA labels underscores the need for additional studies. Studies using real‐world data can potentially help address these gaps.

1. INTRODUCTION

The National Survey on Drug Use and Health (NSDUH) in the United States provides annual population estimates on substance use based on a nationally representative sample. In 2022, 8.9 million respondents reported opioid misuse, 45% of whom were females. ¹ , ² Among females using any prescription opioids in the past year, 10.3% reported misusing them. ² Among pregnant women, the rates of opioid use increased from 0.46% to 0.84% between 2006 and 2015, reaching 0.9% in 2018. ³ , ⁴ Similarly, misuse rates in pregnant women rose from 0.17% to 0.39% between 1998 and 2011. ⁵ The prevalence of opioid misuse was as high as 21.2% in pregnant women, according to the 2019 Pregnancy Risk Assessment Monitoring System (PRAMS) survey conducted by the Centers for Disease Control and Prevention (CDC), while opioid use was lower at 6.6%. ⁶

Recent reviews have sought to characterize opioid use patterns, care management, and associated health outcomes during pregnancy but also highlight the limited research on adverse maternal and pediatric outcomes. ⁵ , ⁷ , ⁸ , ⁹ Some outcomes, such as neonatal opioid withdrawal syndrome (NOWS), are well‐documented and studied, while others remain underexplored. ¹⁰ , ¹¹ , ¹² Notably, studies on maternal morbidity and mortality are scarce and are often secondary to pediatric‐focused outcomes. ³ , ⁵ , ¹³ Although neonates are subject to prenatal exposures, children and adolescents are predominantly exposed through prescriptions. ¹⁴ Opioids in pediatric patients are commonly prescribed for surgeries or dental procedures, with a 2019 study reporting prescription rates as high as 61.4% based on longitudinal pharmacy data. ¹⁵

Although research in maternal and pediatric populations has increased in recent years, these groups have been historically understudied, particularly in clinical trials (CTs). ¹⁶ , ¹⁷ Physiological changes during pregnancy necessitate the evaluation of drug pharmacologic effects through pharmacokinetics (PK) studies, and existing PK research on some opioids provides useful data, but PK has inherent design constraints. ¹⁸ , ¹⁹ Pediatric PK opioid research has expanded as well but shows high variability across distinct age groups. ²⁰ These limitations underscore the need for innovative approaches to address the unique needs of these sensitive populations.

The use of real‐world data (RWD) like medical claims or electronic health records (EHR) has significantly facilitated the recent rise in research in these once excluded groups. ³ , ⁵ , ¹⁰ , ¹² , ²¹ These pharmacoepidemiologic (PE) studies have been instrumental in filling knowledge gaps and generating evidence that can be used to not only inform safe prescribing practices but also promote drug development in these demographics. ²² To this end, the goal of the Maternal and Pediatric Precision in Therapeutics Data and Model Knowledge and Research Coordination Center (MPRINT‐DMKRCC, https://mprint.org/) has been to create and develop an MPRINT Knowledgebase (KB) that will include curated pharmacologic data from PK, PE, and CT studies in maternal and pediatric populations. ²³

The aim of this review was to map the pharmacotherapy research landscape of PK and PE/CT studies on opioids in maternal and pediatric populations. To complement this analysis, we examined prescription/order rates using a sample of Merative MarketScan commercial claims data (hereafter referred to as MarketScan). ²⁴ Additionally, we analyzed United States Food and Drug Administration (FDA) labels, focusing on population‐specific sections to assess the availability of information and human data underlying their recommendations.

2. METHODS

2.1. Drug selection

Opioids were identified using the Anatomical Therapeutic Chemical (ATC) classification system ²⁵ and the codes: N02A (opioids), N01AH (opioid anesthetics), and R05DA (opium alkaloids and derivatives, cough suppressants). Levorphanol, the only opioid without an ATC code, was also included, while antagonists were excluded. The 27 opioids selected are listed in Table S1.

2.2. Subpopulation categories

The maternal population was divided into the following subsets. (i) pregnant/pregnancy – studies during pregnancy or reporting opioid use across any trimester, (ii) labor/delivery – any opioids that were used or given during the labor or delivery period, including the immediate postoperative period for women undergoing cesarean section, (iii) postpartum – studies covering 6–8 weeks post‐delivery, excluding lactation‐focused research, (iv) lactation – studies focused on nursing, breastfeeding, or lactation, and (v) fetal/fetus – considered jointly with maternal studies due to dependence on the mother.

The pediatric population was divided into the following subsets. (i) preterm – birth at ≤37 weeks of gestation, (ii) newborn/neonate – the first 0 to 28 days of life, (iii) infant – >28 days up to 1 year of life, (iv) children – >1 year up to 12 years old, and (v) adolescent – >12 up to 21 years old. ²⁶ For simplicity, the toddler and childhood stages were combined together into “children.”

2.3. Literature search strategy

PubMed was queried using the following search string ‐ “opioids AND [population type] AND [study type] NOT animals NOT reviews,”, limiting it to humans and original publications. Population terms included pregnancy, postpartum, lactation, neonate, and pediatric. Pharmacokinetics (PK) and pharmacoepidemiology/epidemiology (PE) were used to specify study type. The search was initially conducted in June 2022 followed by an updated search in September 2022. Additionally, non‐compartmental PK studies from an alternative modeling‐focused search were also included to ensure completeness, but PK modeling studies were not included in our review (details in Appendix S1). Titles and abstracts retrieved were screened by A.S.

2.3.1. Initial abstract screening and data extraction

Eligibility was determined based on the criteria specified in Table 1. Clinical trials were classified separately after the initial screen as they comprised approximately one‐fourth of the publications retrieved using PE search terms. The full text for the relevant publications was collected for data extraction after this initial review. Relevant information from each study type was extracted with the data elements specified in Table 2.

TABLE 1.

Inclusion/exclusion criteria for selection of publications related to opioids in the maternal and pediatric populations.

Criteria	Pharmacokinetics	Pharmacoepidemiology or clinical trials
Inclusion	Non‐combination drug studies Drug interaction studies Clinical Non‐compartmental Individual or summary‐level values for PK parameters are available	Drugs as the primary exposure or intervention Outcomes associated with the primary drug Prospective or retrospective Quantitative
Exclusion	In vitro studies Animal studies Physiologically based PK studies Population PK studies Other PK modeling or pharmacometrics studies	Studies where the drug was a secondary exposure Qualitative Medication frequency, drug prevalence, or patterns of drug use Case reports Patient/hospital usability studies Patient/provider engagement studies Data mining of registry or health records

TABLE 2.

Data elements extracted from the opioid publications based on study types.

Common data elements
Drug names/classes Population type Pregnancy stage or gestational age Age for pediatric population classification

Study specific data elements
Pharmacokinetics PK parameters Pharmacodynamic/clinical endpoints Specimen information: type of sample and sampling method	Pharmacoepidemiology or clinical trials Study information Study design Data sources Reported outcomes Outcome‐related parameters Population characteristics/risk factors Drug exposures Outcome definition

2.3.2. Statistical analysis

We performed a descriptive analysis to report publication frequencies for each of the opioid drugs as well as the different subpopulations with respect to PK and PE/CT studies. The initial analysis focused on estimating publication frequencies for each opioid by PK and PE/CT studies, respectively. The next stage incorporated subpopulation information to determine if any studies have been conducted for each opioid drug within each subpopulation for the PK and PE/CT studies.

The PK‐specific analyses included identifying and calculating the publication frequencies for the different PK parameters reported, clinical endpoints measured, and types of specimens sampled along with their sampling frequencies. Specimens in the PK studies were sampled at different frequencies and time points. Based on a combination of both, we categorized the sampling frequencies into four types: (i) intensive sampling with several samples collected within a couple of days, (ii) sparse sampling with one sample collected per subject, (iii) cross‐over referring to multiple samples per subject collected across different time periods, (iv) cross‐sectional referring to one sample per subject collected across different time periods, and (v) case study, indicating whether PK information is available at the individual or sample level.

The PE/CT‐specific analyses included identifying and calculating the publication frequencies for the type of study – PE or CT as well as PE study designs such as retrospective, prospective, and others (cross‐sectional, longitudinal, case–control). Next, risk factors associated with the outcomes and frequently reported in PE/CT studies were determined. Finally, relevant primary as well as secondary health outcomes reported in the PE/CT studies were identified and evaluated. Because of the innumerable and varied outcomes reported in the publications, we created broad categories that would encompass outcomes related to the different subpopulations, including maternal, birth, pediatric, opioid‐specific, and health care utilization outcomes.

Additionally, we differentiated the outcomes by data sources – those collected from administrative/EHR data versus non‐administrative/EHR data. Administrative/EHR data are also called RWD and are routinely collected in relation to a patient's health status or delivery of health care. ²⁷ On the other hand, PE/CT studies that involve active prospective data collection from the patients or enrollees of the study are considered non‐RWD. The total number of studies based on administrative/EHR data was 228, and the non‐administrative/EHR category included the remainder of the PE studies (n = 273). We did not include CT studies since the design and conduct of these trials are unlike PE studies. In this analysis, we investigated which health outcomes are more likely to be studied between the two data sources for all opioid drugs combined. The results are presented as the proportion of studies reporting one or more outcomes to the total number of studies within each data source. We tested the difference between the two using the z‐score and two‐tailed test for two proportions at a significance level of 0.05 using the NORM.S.DIST function in Microsoft Excel (Redmond, WA).

2.4. Opioid prescription/order rate analysis

The Merative MarketScan commercial claims database was utilized to determine prescription/order rates among pregnant and pediatric enrollees between 2016 and 2021. ²⁴ In brief, MarketScan is one of the largest and longest‐running US claims databases and captures detailed, de‐identified claims data on clinical utilization, expenditures, and enrollment. These data span inpatient, outpatient, prescription drug, and other healthcare services providing a comprehensive view of real‐world healthcare usage as well as clinical outcomes.

Our analysis included pregnant enrollees if they had at least one delivery and/or procedure code (Table S2). The date associated with the first delivery code was defined as the index date, and only those enrollees who were continuously enrolled for 270 days before and 180 days after the index date were included. Any opioid‐related prescriptions/orders between 270 days before and 180 days after the index date were counted. Pediatric enrollees were included if they were in the following age groups during the 5‐year period– neonates/infants (0–1 years), children (1–12 years), and adolescents (13–18 years). Opioid prescriptions/orders during those 5 years were counted cumulatively for each drug and age group. The proportion of patients given an opioid was then calculated for each drug by dividing it by the total number of patients enrolled during the 5‐year period in each subpopulation. Opioids prescribed in the outpatient setting are labeled “outpatient prescriptions,” and those ordered during an outpatient or inpatient procedure are labeled “outpatient/inpatient orders.” The rates for the prescriptions and orders were calculated separately.

2.5. FDA drug label analysis

Labels for the opioid drugs were downloaded from the FDA's website. ²⁸ The information on active ingredients was obtained from three different components: openfda_substance_name, active_ingredient, and spl_product_data_elements. Only those FDA labels with a single active ingredient were selected for further analysis. The labels with the individual opioids as core ingredients and the corresponding maternal and pediatric medication information were then extracted and organized to conduct descriptive analysis. The labels were considered to have information on maternal and pediatric populations if the sections under those categories had any text, regardless of the content presented in the text. Analysis to determine if data from human studies were present in the labels was conducted manually, with only a single version and formulation of the FDA label used for each of the opioids.

3. RESULTS

Table 3 presents the number of opioid publications initially retrieved and eventually retained for the maternal and pediatric populations and the PK and PE/CT studies, respectively. Statistics based on the final curated publications are presented below, separated by study types. Since studies may evaluate or report on more than one drug or population, the frequencies presented below may not add up to the total number of publications retained.

TABLE 3.

Number of opioid publications retrieved from PubMed and retained based on our inclusion/exclusion criteria.

Study and population types	Number of retrieved publications	Number of retained publications
Pharmacokinetic studies
Maternal	220	147
Pediatrics	399	135
Total	619	282
Pharmacoepidemiology studies or clinical trials
Maternal	1018	328
Pediatrics	1033	320
Total	2051	648

3.1. Pharmacokinetic (PK) studies

A total of 282 PK publications were curated. The rest of the publications were excluded after full‐text review because of studies with unrelated population types or drugs, studies focusing on assay development, ontogeny, binding or transporter enzymes, modeling/pharmacometrics, no PK parameter values, non‐English publications, reviews/opinions/commentaries/letters to the editor, or lack of access to full text.

Of the 27 opioids, 23 opioids had PK studies that were relevant to our populations and could be curated for PK parameters (Figure 1A). Among these, morphine (n = 78) was the most frequently studied drug, followed by fentanyl (n = 47), and methadone (n = 36). Drugs with only a single study reporting any PK data were found for oxymorphone and dihydrocodeine. Opioids with less than five studies (but more than one study) included butorphanol, ketobemidone, meptazinol, nalbuphine, pentazocine, piritramide, propoxyphene, and tapentadol. Drugs noted to have no PK studies were dextropropoxyphene, levorphanol, phenazocine, and phenoperidine. Among the published PK studies, 89.9% reported PK parameter information for the parent drug, with 35.6% of these studies also presenting information on one or more metabolites, while 4.4% of the studies focused on an analyte other than the parent drug. The rest presented data in the form of relevant figures such as concentration‐time curves.

FIGURE 1.

Number of pharmacokinetics (A) and pharmacoepidemiology/clinical trial publications (B) for each opioid.

When the individual opioid drugs are evaluated further by the maternal and pediatric subpopulations (Figure 2A), the distribution of the PK publications is varied. On the one hand, morphine, methadone, and fentanyl were studied across all maternal and pediatric subpopulations. In contrast, only one study reported PK parameters for meptazinol in the pregnant and neonatal populations, respectively. Similarly, dihydrocodeine concentrations were reported for a neonate and a 14‐year‐old in a single case report. Additionally, only eight different opioids have any PK studies in the pregnant and adolescent populations and six different opioids in the postpartum and preterm populations, respectively.

FIGURE 2.

Knowledge gaps in pharmacokinetics (A) and pharmacoepidemiology/clinical trials (B) research by population types for each opioid. White – no publications, dark gray – 1+ publications.

We noted that drug concentration (n = 223) was the most reported parameter in the PK studies, followed by clearance (n = 93) and half‐life (n = 83) (Figure 3A). PK parameters unique to the maternal and pediatric populations included fetal‐maternal ratio, milk‐to‐plasma ratio, and estimated infant dose from breast milk concentrations. Many PK studies also report clinical endpoints, with the top reported endpoint being the opioid‐specific NOWS (n = 54) among neonates, followed by opioid‐related side effects (n = 49) in either the maternal or pediatric populations (Figure 3B).

FIGURE 3.

Distribution of parameters (A) and clinical endpoints (B) reported in the opioid pharmacokinetics (PK) publications.

Figure 4A shows the number of publications that used the different sampling techniques in the two population types. Among these, the most noted was intensive sampling, followed by case study, then sparse sampling. All three were higher in the pediatric population compared to the maternal population. Cross‐over and cross‐sectional were the least common sampling methods but were noted more in the maternal than pediatric populations. Additionally, we also collected information on the type of specimen used, the most common being blood in both populations, either in the form of plasma or serum (pediatric = 45.7%, maternal = 36.2%), followed by urine (n = 13.5%), and then breast milk (n = 10.6%) (Figure 4B).

FIGURE 4.

Sampling frequencies (A) and specimen types (B) used in the opioid pharmacokinetics studies.

3.2. Pharmacoepidemiologic studies or clinical trials

A total of 648 PE/CT opioid publications were curated, with the rest of the publications excluded based on criteria similar to those listed under PK study results. PE/CT studies were distributed across 23 different opioid drugs and two additional categories, including any/all opioids and illicit opioids (Figure 1B). Among individual opioids, methadone (n = 183) was studied the most but was followed by publications that included any/all opioids (n = 160) in their studies. Some of the least studied drugs included ketobemidone (n = 3), butorphanol (n = 2), dextropropoxyphene (n = 1), and piritramide (n = 1). Other drugs with only four publications included alfentanil, dihydrocodeine, propoxyphene, and tapentadol. There are four opioid drugs that have no PE/CT publications, including levorphanol, meptazinol, phenazocine, and phenoperidine. The use of one or more illicit substances was also captured and represented about 15% of the PE/CT curated studies.

When we evaluate the publication distributions by the maternal and pediatric subpopulations (Figure 2B), the PE/CT knowledge gaps among opioids are very diverse. Although methadone, buprenorphine, fentanyl, and meperidine have been studied in all subpopulations, some drugs have been evaluated only in a single population, such as oxymorphone and propoxyphene in pregnant women, butorphanol during labor/delivery, dextropropoxyphene in newborns/neonates, and piritramide in children.

Among all the curated PE/CT studies across both populations, 22.5% were clinical trials and 77.5% were PE studies. Among PE studies, 65.1% were retrospective, followed by prospective (27.3%), with mixed/other study designs comprising the remaining 7.6%. Some other studies that specified their study designs and were curated as such included cross‐sectional (6.8%), longitudinal cohort (5.8%), and case–control (4.6%).

3.2.1. Risk factors

Table 4 presents publication frequencies for risk factors associated with outcomes in PE/CT studies. The most curated ones include concomitant drugs (n = 149), followed by comorbidities (n = 99) such as chronic hypertension, diabetes, infections, or other organ disease. Other risk factors captured include demographics (age, race/ethnicity, sex), tobacco/alcohol use, social determinants of health, time or trend analysis, and genetics.

TABLE 4.

Frequency of risk factors associated with pharmacoepidemiology/clinical trial outcomes in maternal and pediatric populations.

Risk factors	Publications, n
Concomitant drugs	149
Comorbidities	99
Age	81
Social determinants of health	54
Time period/trend analysis	49
Tobacco/alcohol use	47
Race/ethnicity	44
Sex	36
Genetics	14

3.2.2. Health outcomes

Table 5 presents the proportion of studies reporting one or more PE outcomes by their respective data source. Among maternal outcomes, all of the studies were reported similarly across RWD and non‐RWD. On the other hand, some birth outcomes were captured significantly differently between RWD and non‐RWD. NOWS, preterm birth, and small for gestational age or low birth weight outcomes were significantly greater in RWD studies, whereas Apgar score was reported more in the non‐RWD studies. Other outcomes like death or distress of the fetus or newborn, birth defects or congenital anomalies, and breastfeeding were reported similarly across both RWD and non‐RWD sources. Neurodevelopmental or behavioral conditions are pediatric outcomes that were more likely to be studied in non‐RWD, whereas circulatory effects of the opioids were more prevalent in RWD studies. All other pediatric outcomes, including respiratory, cardiac, or bleeding effects and sleep disorders, have been studied similarly using both RWD and non‐RWD.

TABLE 5.

Relative frequency of outcomes related to opioids in the maternal and pediatric populations using real‐world data in pharmacoepidemiology publications.

Outcomes	Administrative/EHR, %	Non‐administrative/EHR, %	p‐Value
Maternal outcomes
Maternal substance use/urine drug screen	11.8	10.6	0.667
Labor/delivery outcomes	7.5	5.9	0.472
Perinatal morbidity	7.0	5.9	0.596
Maternal mortality	6.6	4.0	0.201
Birth outcomes
Neonatal opioid withdrawal syndrome	45.2	35.9	0.035
Preterm birth	25.9	17.6	0.024
Small for gestational age/low birth weight	20.2	9.9	0.001
Breast feeding information	14.5	10.6	0.194
Apgar score	10.5	21.2	0.001
Birth defects/congenital anomalies	9.6	7.7	0.435
Neonatal/infant mortality	8.3	8.1	0.912
Miscarriage/stillborn/intrauterine growth restriction/demise	6.1	5.9	0.897
Respiratory distress syndrome	4.4	3.3	0.522
Fetal distress	3.1	2.2	0.542
Pediatric outcomes
Respiratory effects	11.0	8.1	0.267
Neurodevelopmental/behavioral outcomes	7.0	13.2	0.024
Cardiac effects	4.8	4.8	0.976
Circulatory effects	3.1	0.4	0.016
Bleeding effects	2.2	2.2	1.000
Sleep disorders	1.3	2.9	0.219
Opioid‐specific outcomes
Opioid dose information	16.2	24.5	0.023
Overdose	12.3	3.3	<0.001
Analgesia outcomes/satisfaction	10.1	13.6	0.234
Respiratory depression	7.5	10.3	0.276
Pruritus	3.9	4.8	0.659
Nausea or vomiting	2.6	7.7	0.012
Somnolence	2.6	2.6	0.960
Sedation	0.9	4.8	0.011
Healthcare utilization outcomes
Length of hospital stay	25.0	21.6	0.368
Admission/readmission	15.8	11.4	0.147
Hospital discharge	8.8	9.2	0.881
ED/hospital/clinic visit	8.8	4.0	0.029

Note: p‐Values < 0.05 are shown in bold.

Abbreviations: ED: Emergency Department; EHR: Electronic Health Records.

For opioid‐specific outcomes, overdose was reported significantly more often in RWD studies, whereas opioid dose information, nausea or vomiting, and sedation were reported significantly more often in non‐RWD studies. The remaining opioid‐specific side effects were also more likely to be reported in non‐RWD studies, although the difference between RWD and non‐RWD studies was not significant. Health care utilization outcomes, such as length of hospital stay, admission/readmission, and hospital discharge, were studied in both RWD and non‐RWD, whereas ED, hospital, or clinic visits were reported more often in RWD studies.

3.3. Opioid prescription/order rates

Table 6 presents the prescription/order rates for 18 opioids in the maternal and pediatric populations from the MarketScan commercial claims database between 2016 and 2021. The total number of enrollees in the maternal population from 2016 to 2021 was 729,524. The highest prescriptions were noted for oxycodone (22.8%) followed by hydrocodone (13.7%) in the postpartum period. Codeine and tramadol were two other drugs prescribed to >1% of the maternal population in the postpartum period. Hydrocodone was prescribed at a slightly higher rate (3.6%) than codeine (2.9%) or oxycodone (2.2%) during pregnancy, but all rates were lower compared to the postpartum period. Buprenorphine was prescribed to 1.1% of the maternal population during pregnancy.

TABLE 6.

Prescription/order rates (%) for 18 opioids in the maternal and pediatric populations using the Merative MarketScan commercial claims data from 2016 to 2021.

Drug name	Maternal						Pediatric
	Pregnancy		Labor/Delivery		Postpartum		Neonates/Infants		Children		Adolescents
	N = 729,524		N = 729,524		N = 729,524		N = 2,828,907		N = 22,172,458		N = 9,995,976
	Outpatient prescriptions	Outpatient /inpatient orders	Outpatient prescriptions	Outpatient /inpatient orders	Outpatient prescriptions	Outpatient /inpatient orders	Outpatient prescriptions	Outpatient /inpatient orders	Outpatient prescriptions	Outpatient /inpatient orders	Outpatient prescriptions	Outpatient /inpatient orders
Buprenorphine a	1.11	0.01	0.01	<0.01	0.12	<0.01	<0.01	0.0	<0.01	<0.01	<0.01	<0.01
Butorphanol a	0.03	0.35	<0.01	0.04	0.01	0.03	0.0	<0.01	<0.01	<0.01	<0.01	<0.01
Codeine a	2.87	0.0	0.03	0.0	3.38	0.0	0.02	0.0	0.38	<0.01	1.39	<0.01
Dihydrocodeine	<0.01	–	0.0	–	<0.01	–	0.0	–	<0.01	–	<0.01	–
Fentanyl a	0.03	0.83	<0.01	0.03	0.01	2.24	<0.01	0.60	<0.01	0.98	<0.01	1.35
Hydrocodone	3.60	–	0.08	–	13.73	–	0.16	–	0.65	–	4.11	–
Hydromorphone a	0.13	0.59	<0.01	0.01	0.79	1.05	<0.01	0.01	<0.01	0.06	0.02	0.38
Levorphanol	0.0	–	0.0	–	<0.01	–	0.0	–	0.0	–	<0.01	–
Meperidine a	0.02	0.24	<0.01	0.02	0.05	0.21	<0.01	0.01	0.01	0.04	0.01	0.09
Methadone a	0.03	0.33	<0.01	<0.01	<0.01	0.01	0.01	<0.01	<0.01	<0.01	<0.01	<0.01
Morphine a	0.07	1.16	<0.01	0.09	0.17	0.94	0.01	0.12	0.01	0.42	0.01	0.62
Nalbuphine a	<0.01	0.26	0.0	0.04	<0.01	0.03	0.0	<0.01	0.0	<0.01	0.0	0.01
Oxycodone	2.23	–	0.09	–	22.76	–	0.11	–	0.21	–	1.44	–
Oxymorphone a	<0.01	<0.01	0.0	<0.01	<0.01	0.0	0.0	0.0	<0.01	0.0	<0.01	0.0
Pentazocine a	<0.01	0.0	0.0	0.0	0.01	0.0	0.0	0.0	<0.01	0.0	<0.01	<0.01
Remifentanil	0.0	–	0.0	–	0.0	–	0.0	–	<0.01	–	0.0	–
Tapentadol	0.02	–	<0.01	–	0.01	–	0.0	–	<0.01	–	<0.01	–
Tramadol	0.84	–	0.01	–	1.61	–	<0.01	–	0.01	–	0.39	–

^a Indicates opioids with outpatient/inpatient orders.

The highest outpatient/inpatient orders were noted for fentanyl (2.2%) in the postpartum period, and postpartum outpatient/inpatient orders were approximately 1.0% for hydromorphone and morphine. Also, morphine (1.2%) orders were highest in the pregnant population. Whereas opioid prescription and order rates were lowest during labor/delivery, with many of the opioids being given to <0.1% of the maternal population.

The total number of pediatric enrollees from 2016 to 2021 in each of the age groups included 2.8 million neonates/infants, 22 million children, and 10 million adolescents. Opioid prescriptions were highest among adolescents in the pediatric population, with hydrocodone prescribed to 4.1% of adolescents. Codeine and oxycodone were the other two opioids prescribed to >1% of the adolescent population. Hydrocodone was also the most prescribed opioid among children (0.7%) and neonates/infants (0.2%).

The most outpatient/inpatient orders were noted for fentanyl and morphine across all pediatric populations, with the highest proportion of orders noted among adolescents, followed by children, and lastly the neonate/infant groups. Outpatient/inpatient fentanyl was ordered for 1.4% of the adolescents, approximately 1% of the children, and 0.6% of the neonates and infants. Whereas outpatient/inpatient morphine was ordered for 0.6% of the adolescents, 0.4% of the children, and 0.1% of the neonates/infants. Opioid prescription and order rates were lowest in the neonate/infant group.

Lastly, opioids with no or minimal (<0.01%) prescription or order rates across all subpopulations included dihydrocodeine, levorphanol, oxymorphone, and remifentanil.

3.4. FDA drug label findings

Table 7 presents information on text available within the FDA opioid labels for the maternal and pediatric populations, considering the single active ingredient and its various salt forms. Four out of the 18 opioids have more than 50 drug labels, and eight out of the 18 opioids have more than 30 labels. If we consider only the main ingredient without the salt form, multiple versions of the drug labels are available for only six of the 18 opioids, including buprenorphine, fentanyl, morphine, oxycodone, sufentanil, and tramadol. The nine opioids without any FDA labels associated with them include dextropropoxyphene, dihydrocodeine, ketobemidone, meptazinol, pentazocine, phenazocine, phenoperidine, piritramide, and propoxyphene. Table 8 presents findings from our analysis of FDA labels to determine the presence of data from human studies. Information based on human studies is lacking in labels for alfentanil, hydromorphone, levorphanol, and tapentadol.

TABLE 7.

United States Food and Drug Administration labels for 18 opioids with any text in the maternal and pediatric sections (Single active ingredient and their salt forms).

Drug name	FDA labels, n	Maternal			Pediatric
		Pregnancy	Labor/delivery	Lactation/nursing mothers
Alfentanil	1	1	0	0	1
Buprenorphine	33	29	0	6	29
Butorphanol	5	3	2	3	3
Codeine	8	2	0	0	2
Fentanyl	33	32	18	18	32
Hydrocodone	3	3	2	2	3
Hydromorphone	30	29	6	6	29
Levorphanol	7	7	3	4	7
Meperidine	3	3	0	0	3
Methadone	39	38	17	17	38
Morphine	78	73	13	12	72
Nalbuphine	8	3	2	1	3
Oxycodone	80	65	22	22	67
Oxymorphone	10	6	2	2	6
Remifentanil	6	6	0	0	6
Sufentanil	3	3	1	0	3
Tapentadol	2	2	0	0	2
Tramadol	135	107	27	28	107

TABLE 8.

Availability of animal or human data for each opioid in the United States Food and Drug Administration labels.

Drug and salt form	Formulation	Subsection	Any information (1 = yes)	Animal data (1 = yes)	Human data (1 = yes)	Human PK data (1 = yes)	Human CT data (1 = yes)	Human PE/clinical data (1 = yes)
Alfentanil HCl	Injection	Pregnancy	1	1
Alfentanil HCl	Injection	Labor/Delivery	1 a
Alfentanil HCl	Injection	Lactation	1	1
Alfentanil HCl	Injection	Pediatric
Buprenorphine	Tablet	Pregnancy	1	1	1		1	1
Buprenorphine	Tablet	Labor/Delivery
Buprenorphine	Tablet	Lactation	1	1	1	1
Buprenorphine	Tablet	Pediatric	1 b		1		1
Butorphanol tartrate	Injection	Pregnancy	1	1
Butorphanol tartrate	Injection	Labor/Delivery	1		1		1	1
Butorphanol tartrate	Injection	Lactation	1		1	1
Butorphanol tartrate	Injection	Pediatric	1 c		1			1
Codeine sulfate	Tablet	Pregnancy	1	1	1
Codeine sulfate	Tablet	Labor/Delivery	1 a
Codeine sulfate	Tablet	Lactation	1	1	1	1		1
Codeine sulfate	Tablet	Pediatric	1		1			1
Fentanyl citrate	Injection	Pregnancy	1	1	1 d
Fentanyl citrate	Injection	Labor/Delivery	1 a
Fentanyl citrate	Injection	Lactation	1		1 d
Fentanyl citrate	Injection	Pediatric	1		1			1
Hydrocodone bitartrate	Capsule	Pregnancy	1	1	1 d
Hydrocodone bitartrate	Capsule	Labor/Delivery	1 a
Hydrocodone bitartrate	Capsule	Lactation	1	1	1	1
Hydrocodone bitartrate	Capsule	Pediatric
Hydromorphone HCl	Tablet/Solution	Pregnancy	1	1
Hydromorphone HCl	Tablet/Solution	Labor/Delivery	1 a
Hydromorphone HCl	Tablet/Solution	Lactation	1	1
Hydromorphone HCl	Tablet/Solution	Pediatric
Levorphanol tartrate	Injection	Pregnancy	1	1	1 d
Levorphanol tartrate	Injection	Labor/Delivery	1 a
Levorphanol tartrate	Injection	Lactation	1 a
Levorphanol tartrate	Injection	Pediatric
Meperidine HCl	Tablet/Solution	Pregnancy	1	1 e	1 d
Meperidine HCl	Tablet/Solution	Labor/Delivery	1 a
Meperidine HCl	Tablet/Solution	Lactation	1		1 d
Meperidine HCl	Tablet/Solution	Pediatric	1		1	1
Methadone HCl	Tablet	Pregnancy	1	1 e	1		1	1
Methadone HCl	Tablet	Labor/Delivery	1 a
Methadone HCl	Tablet	Lactation	1	1	1	1
Methadone HCl	Tablet	Pediatric	1 b		1		1
Morphine sulfate	Tablet	Pregnancy	1	1 e	1			1
Morphine sulfate	Tablet	Labor/Delivery	1 a
Morphine sulfate	Tablet	Lactation	1	1	1	1
Morphine sulfate	Tablet	Pediatric
Nalbuphine HCl	Injection	Pregnancy	1	1
Nalbuphine HCl	Injection	Labor/Delivery	1		1	1		1
Nalbuphine HCl	Injection	Lactation	1	1	1 d	1
Nalbuphine HCl	Injection	Pediatric	1 c		1			1
Oxycodone HCl	Tablet	Pregnancy	1	1
Oxycodone HCl	Tablet	Labor/Delivery	1 a
Oxycodone HCl	Tablet	Lactation	1	1	1	1
Oxycodone HCl	Tablet	Pediatric	1		1		1
Oxymorphone	Tablet	Pregnancy	1	1
Oxymorphone	Tablet	Labor/Delivery	1		1		1
Oxymorphone	Tablet	Lactation	1	1
Oxymorphone	Tablet	Pediatric	1		1	1	1
Remifentanil HCl	Injection	Pregnancy	1	1	1	1	1
Remifentanil HCl	Injection	Labor/Delivery	1 a
Remifentanil HCl	Injection	Lactation	1	1
Remifentanil HCl	Injection	Pediatric	1		1	1	1
Sufentanil citrate	Injection	Pregnancy	1	1
Sufentanil citrate	Injection	Labor/Delivery	1		1	1	1
Sufentanil citrate	Injection	Lactation	1	1
Sufentanil citrate	Injection	Pediatric	1		1	1		1
Tapentadol HCl	Tablet	Pregnancy	1	1	1 d
Tapentadol HCl	Tablet	Labor/Delivery	1 a
Tapentadol HCl	Tablet	Lactation	1	1 d	1 d
Tapentadol HCl	Tablet	Pediatric
Tramadol HCl	Capsule	Pregnancy	1	1	1 d			1
Tramadol HCl	Capsule	Labor/Delivery	1		1	1
Tramadol HCl	Capsule	Lactation	1	1	1	1
Tramadol HCl	Capsule	Pediatric	1		1			1

Note: A value of 1 indicates that information is available in the label for the category specified in a column. Blank space denotes that no data was available for that specific opioid, subpopulation, human or animal, and/or study type(s).

^a If only any information = 1, it indicates that some guidance based on the general opioid drug class is provided.

^b Pediatric human data are based on prenatal exposures.

^c Pediatric human data are based on exposures during labor or delivery.

^d Risk summary mentions insufficient data from studies (e.g., in pregnant women) without providing details.

^e No formal animal reproductive and developmental toxicology studies.

4. DISCUSSION

In this review, we provide a broad analysis of the research landscape on opioids in the maternal and pediatric populations. We have synthesized evidence from PubMed on opioid PK and PE studies as well as CT. We found that buprenorphine, fentanyl, methadone, and morphine are the most studied opioids in PK and PE/CT literature across all the maternal and pediatric subpopulations. In contrast, the MarketScan findings indicate that hydrocodone, oxycodone, and codeine are the most prescribed opioids, especially in the postpartum period and in adolescents. Although sections in the FDA labels for these three opioids contain some text, relevant data from human studies is sparse. Hydrocodone, ²⁹ oxycodone, ³⁰ and codeine ³¹ have lactation‐focused PK information in the labels, and the latter two have some information for the pediatric population as well. Of the nine opioids that did not have FDA drug labels, ketobemidone ³² and piritramide ³³ have not been approved in the United States, and dextro−/propoxyphene was withdrawn from the US market in 2010. ³⁴ Although dihydrocodeine was approved as a combination product, all current formulations are evidently discontinued by the FDA, based on their marketing status on Drugs@FDA. ³⁵ Similarly, single‐ingredient formulations of pentazocine were also previously approved by the FDA but are currently discontinued. Lastly, meptazinol, ³⁶ phenazocine, ³⁷ and phenoperidine ³⁸ are considered experimental.

4.1. Three‐way comparison of literature evidence, prescription/order rates, and FDA drug labels for opioids

Findings for a three‐way comparison of the literature evidence, prescription/order rates, and FDA drug labels on opioids are only available for pregnancy and labor/delivery in the maternal population and the overall pediatric population (Figure 5A–C). The joint analysis highlights differences across the three based on the prescribed/ordered opioids in these subpopulations. The findings from the evaluation of human data in the FDA labels are presented in Table 8 and expanded below.

FIGURE 5.

Knowledge gaps across pharmacoepidemiology/clinical trials and pharmacokinetics research compared to Merative MarketScan prescription/order rates and availability of the United States Food and Drug Administration (FDA) label information for pregnancy (A), labor/delivery (B), and pediatric populations (C). The dashed red line highlights the differences in highly prescribed or ordered opioids versus those commonly researched.

4.1.1. Pregnancy (Figure 5A)

Among opioids prescribed or ordered during pregnancy in MarketScan enrollees, butorphanol, hydromorphone, and pentazocine have no published PK or PE/CT evidence. Additionally, no PK studies were identified for oxymorphone, tapentadol, tramadol, or dihydrocodeine, while nalbuphine lacked PE/CT studies. Hydrocodone and oxycodone were prescribed at rates of 3.6% and 2.2%, respectively, yet lack PK studies or have fewer than five PE/CT publications each. Similarly, codeine, with a prescription rate of 2.9%, has limited literature, with fewer than five publications in both PK and PE/CT domains. Furthermore, FDA drug labels for these three opioids rely primarily on animal data, citing insufficient evidence in pregnant women for codeine and hydrocodone. To determine the need for further PK or prospective clinical studies in pregnancy, RWD studies could help validate existing findings and identify overlooked health care patterns across different states in the United States. ²¹ , ³⁹ , ⁴⁰ , ⁴¹

In contrast, methadone (0.3%) and buprenorphine (1.1%) were prescribed or ordered at lower rates among pregnant MarketScan enrollees but have substantial PK and PE/CT evidence, along with human data in FDA labels. Despite increased use through opioid agonist treatment programs, their utilization remains low in pregnancy, aligning with reported rates of 0.4% for methadone and 0.9% for buprenorphine among 2.5 million pregnant Medicaid enrollees. ⁴² Among pregnant and postpartum Medicaid enrollees with clinically documented opioid use disorder (OUD), only half received any OUD medication across 39 U.S. states between 2017 and 2018. ⁴³ Multiple barriers limit access to OUD therapy in pregnancy, contributing to low utilization. ⁴⁴ Methadone is available exclusively through specialty opioid treatment programs or inpatient settings, making initiation and maintenance difficult for pregnant individuals. ⁴⁴ , ⁴⁵ Although buprenorphine can be prescribed in outpatient settings, access to providers remains more challenging than access to opioid treatment programs. ⁴⁴ , ⁴⁶

4.1.2. Labor or delivery (Figure 5B)

Among opioids that were prescribed or ordered during labor or delivery in MarketScan, tapentadol lacked both PK and PE/CT studies, and oxymorphone had no PE/CT evidence and only one PK study conducted during delivery. ⁴⁷ FDA labels with human data included nalbuphine, sufentanil, and tramadol (PK data); butorphanol and nalbuphine (PE data); and butorphanol, oxymorphone, and sufentanil (CT data). Although no clinical trials for oxymorphone were identified in our primary search, a post‐hoc PubMed query found three efficacy and safety studies, ⁴⁸ , ⁴⁹ , ⁵⁰ with only one conducted during labor and delivery. ⁵⁰ Although opioid prescription patterns during the postpartum period are well documented, evidence on their use during labor and delivery remains scarce. ⁵¹ , ⁵² , ⁵³ Our claims‐based analysis, using hospital and procedural codes within a 48‐h window before and during delivery, captured opioid prescriptions/orders in less than 0.1% of parturients, but no directly comparable literature was found. However, extensive research exists on pain management via epidural or spinal analgesia during labor. ⁵⁴ , ⁵⁵ A 2008 CDC report across 27 U.S. states found that 61.0% of 1.8 million pregnant individuals with singleton vaginal deliveries received epidural or spinal anesthesia. ⁵⁶ However, this report has limitations, such as the exclusion of cesarean deliveries, reliance on birth certificates for data collection, and convenience sampling. Additionally, opioids may or may not be used in these procedures, making their specific utilization challenging to determine.

4.1.3. Pediatric (Figure 5C)

In pediatric patients, butorphanol and oxymorphone were the only prescribed or ordered opioids without PK or PE/CT literature evidence. Additionally, tapentadol had fewer than five PK studies and no PE/CT publications. Opioid prescriptions were highest among adolescents, with hydrocodone at 4.1% and oxycodone, codeine, and fentanyl reaching 1.4% each. Among these, only codeine and oxycodone had human data in FDA labels for adolescents. Although the oxymorphone FDA label includes human data for children over 12 years of age, ⁵⁷ a web search retrieves only an abstract matching the study details in the label. ⁵⁸ Although certain oxymorphone brands and formulations have been discontinued, findings from this study remain in the FDA label of a currently marketed formulation, raising concerns about the reliance on limited data despite the absence of a formal recommendation for its use. ⁵⁹

4.2. Two‐way comparison of literature evidence and FDA drug labels or prescription/order rates for opioids

4.2.1. Postpartum or lactation

Opioid prescription rates were highest during the early postpartum period among MarketScan‐enrolled females, with oxycodone at 22.8% and hydrocodone at 13.7%. A proportion of these individuals were likely breastfeeding, but differentiation was not possible due to the limited availability of lactation‐specific codes. FDA drug labels primarily focus on lactation, with human study data available for several opioids, particularly regarding exposure through human milk. Opioids with PK information in FDA labels include buprenorphine, butorphanol, hydrocodone, methadone, morphine, nalbuphine, oxycodone, and tramadol, and codeine includes both PK and PE data. However, lactation‐related literature remains sparse, with only single studies reporting PK data in breast milk for hydrocodone, oxycodone, and codeine. ⁶⁰ , ⁶¹ , ⁶² Breastfeeding as a primary outcome is rarely examined in PE studies, with fewer than five publications in our analysis addressing breastfeeding‐related outcomes. ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ The presence of lactation‐related information in FDA labels highlights a critical need for further research in both PK and PE/CT domains.

4.3. Real‐world evidence on opioids

Our analysis of RWD revealed significant differences in reported health outcomes. Frequently captured outcomes in RWD included NOWS, preterm births, low birth weight, opioid overdose, and health care visits. In contrast, non‐RWD sources, primarily prospective or longitudinal studies with extended follow‐up, commonly reported APGAR scores, neurodevelopmental or behavioral outcomes, and opioid‐related dosage or side effects.

For example, neurodevelopmental outcomes were derived from studies such as Ethanol, NeuRodevelopment, Infant and Child Health (ENRICH) birth cohort, the Boston Birth Cohort, and the Maternal Life Study (MLS). ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ Conversely, NOWS was identified using International Classification of Diseases (ICD‐9/10) codes from sources like the National Inpatient Sample (NIS), Medicaid claims databases, and the Kids' Inpatient Database (KID). ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ These examples illustrate the diversity of RWD and non‐RWD sources used in outcome evaluation. Our findings demonstrate that certain outcomes are readily captured using RWD, underscoring its value for evidence generation while reducing the need for direct participant involvement in these populations.

This study has several innovations and strengths but also some limitations. It is the first to review multiple opioids in both maternal and pediatric populations, highlighting knowledge gaps in PK and PE/CT literature together with drug label and prescription/order data. The review covers diverse subpopulations and study types with certain exclusions such as PK modeling studies and reviews. We used PubMed for literature retrieval, focusing on English‐language publications for accessibility and data curation within the MPRINT‐KB. Our inclusion and exclusion criteria were designed to prioritize human clinical studies, which may have inadvertently excluded other relevant literature. Additionally, our prescription/orders analysis is based on a 5‐year window of US claims data, which may not represent all maternal and pediatric populations across care settings, countries, or time frames. Although our primary FDA label analysis focused on the presence or absence of text in the relevant sections, manual evaluation for the existence of human data was performed for each opioid. However, this was only feasible with a single version and formulation of the FDA label available for each of the opioids. Lastly, publications after 2022 were excluded due to logistic constraints, but the KB aims to incorporate the latest research in future updates.

Our research landscape and utilization analyses reveal a disparity between commonly researched and prescribed or ordered opioids, especially in specific maternal and pediatric subpopulations, thus highlighting significant knowledge gaps. Hydrocodone, oxycodone, and codeine are the most prescribed opioids postpartum and among adolescents, yet FDA label recommendations for these drugs are based on limited human data. Real‐world data will be indispensable for bridging these gaps and instrumental in evidence generation if aligned with the FDA's Framework for Real‐World Evidence program. ²⁷ Through MPRINT‐KB, we aim to consolidate existing data on pharmacotherapies in these populations to support and accelerate future research. We anticipate MPRINT‐KB to provide evidence that will aid the development of regulatory policies, including the adaptation of medication labels to more accurately reflect the efficacy and safety of opioids in the maternal and pediatric populations.

FUNDING INFORMATION

This study has been funded by the National Institute of Child Health and Human Development (NICHD) grants P30HD106451 and P50HD106446.

CONFLICT OF INTEREST STATEMENT

Sara Quinney is a member of the Pharmacotherapy editorial board. All other authors declare no conflicts of interest.

Supporting information

Appendix S1

Shendre A, Liu X, Chiang CW, et al. Pharmacotherapy research landscape and knowledge gaps of opioids in maternal and pediatric populations. Pharmacotherapy. 2025;45:367‐385. doi: 10.1002/phar.70024

DATA AVAILABILITY STATEMENT

The authors confirm that the data supporting the findings of this study are available within the article (and/or) its Appendix S1. The Merative MarketScan data that support the findings of this study are available from Merative MarketScan. Restrictions apply to the availability of these data, which were used under license for this study.