Trends in Opioid Prescribing for Adolescents and Young Adults in Ambulatory Care Settings

We conducted a retrospective analysis of clinical care data from ambulatory settings across the United States using the National Hospital Ambulatory Medical Care Survey (NHAMCS) and National Ambulatory Medical Care Survey (NAMCS). Data were queried from January 1, 2005, to December 31, 2015 (latest year of available data).

The NHAMCS and NAMCS databases are annual, national probability samples of ambulatory care visits throughout the United States. The NHAMCS collects data on visits to hospital-based EDs, and the NAMCS collects data on visits to office-based practices. The surveys are conducted annually by the National Center for Health Statistics (NCHS), a division of the Centers for Disease Control and Prevention.¹⁵ 

The NHAMCS uses a 4-stage probability design, in which data are obtained from samples of geographically defined areas, hospitals within areas, EDs within hospitals, and patient visits within EDs. National estimates are derived from data collected on ∼30 000 visits annually in ∼600 hospital EDs. A random 4-week reporting period is assigned to each emergency service area, during which a systematic, random sample of patient visits is surveyed by using a standard patient record form. Hospital staff, trained by US Census Bureau workers, are responsible for data collection at each site. Proportional weights are applied to allow for extrapolation to national figures.

The NAMCS collects data from patient visits to offices of nonfederally employed physicians classified as “office-based, patient care.” Data are collected by the individual physician, physician’s staff, or US Census Bureau field representatives. Physicians are assigned a random reporting week, with the sampling procedures being designed to obtain ∼30 records from each physician. The NAMCS uses a multistage probability design using primary sampling units, physician practices within primary sampling units, and patient visits within practices. Data are collected on ∼40 000 patient visits annually, and proportional weights are applied to allow for extrapolation to national estimates. For both the NHAMCS and NAMCS, data processing occurs at a central facility with quality assurance performed by computerized algorithms and manual review to ensure validity and consistency.

We included visits by patients aged 13 to 22 years with an outpatient visit to an ED (ie, the patient was discharged at the end of the visit) or any visit to an outpatient clinic. We chose 2005 as the beginning of the study period because it was the first survey year to differentiate between medications given during a visit versus medications given as a prescription. Adolescents were classified as ages 13 to 17 years, and young adults were classified as ages 18 to 22 years. We studied patients up to 22 years of age because these patients frequently continue to receive care from pediatric providers and have similar risk profiles to older adolescents in terms of opioid and other prescription-drug abuse.^16,17 Demographic information, including age, sex, race, geographic region, and year of visit, were analyzed for each visit. Missing values of age, sex, and race were imputed by the NCHS.¹⁸ Clinic visits were classified by primary specialty type according to the specialties available in the NAMCS data set. All clinic specialty types with a sampled visit count >30 were included.¹⁹ 

The NHAMCS and NAMCS provide information on up to 8 medication prescriptions provided at a visit depending on the year analyzed. Opioids were identified by using the National Drug Code Directory for 2005 (codes 1720 and 1721) and the Multum Lexicon Drug Database for 2006 through 2015 (codes 060 and 191) because these were the medication classification systems used by the NHAMCS and NAMCS during those survey years.²⁰ 

Each visit within the NHAMCS and NAMCS includes up to 3 diagnoses, with the first being considered the primary diagnosis.²¹ Diagnoses are coded by using the International Classification of Diseases, Ninth Revision (ICD-9). We selected the primary diagnosis for each visit.

Data were analyzed by using the sampled visit weight that is the product of the corresponding sampling fractions at each stage in the sample design. The sampling weights are adjusted by the NCHS for survey nonresponse within time of year, geographic region, urban or rural location, and ownership designations, yielding an unbiased national estimate of occurrences and characteristics. Because of the complex sample design, sampling errors were determined accounting for the clustered nature of the sample. When generating estimates from survey data, we considered the survey design by specifying the primary sampling units and the patient-visit sampling weights provided by the NHAMCS and NAMCS (denoting the inverse of the probability that the observation is included) to ensure accurate estimates. As specified by the NCHS, if the sampled visit count was <30 or the relative SE was <30%, the calculated estimates may not be stable and are specifically indicated.¹⁹ All analyses were performed in Stata version 14 (Stata Corp, College Station, TX) by using the suite of estimation commands for survey data (svyset and svy).

We calculated descriptive statistics for the demographics of patients receiving an opioid prescription both overall and stratified by age group and setting (ie, ED or clinic). Rates of opioid prescribing were calculated and reported with 95% confidence intervals (CIs). We calculated the rate of visits associated with an opioid prescription by year for both age groups in each ambulatory setting as well as for the different outpatient clinic specialties and most common diagnoses treated with opioid medications in the ED. For each year, we calculated rates of prescribing by specific opioid. We also examined rates of opioid prescribing for the most common diagnoses treated with opioids in the ED stratified by geographic region. To test for linear trends in opioid prescriptions over time, we used logistic regression models with opioid prescriptions as the dependent variable and time (measured in years) as the independent variable, reporting odds ratios (ORs) with 95% CIs.

There were 78 430 visits by adolescents and young adults over the study period in the raw sample, with 47 159 visits to an ED and 31 271 visits to an outpatient clinic. In the weighted sample, this extrapolated to 197 million visits to an ED and 801 million visits to an outpatient clinic. Frequency of visits did not differ significantly by year for either care setting.

Nearly 52 million visits (5.2%; 95% CI 4.9% to 5.5%) were associated with an opioid prescription, of which 29 440 874 (56.5%) were provided in EDs and 22 652 315 (43.5%) were provided in outpatient clinics. The rate of opioid prescribing was 14.9% (95% CI 14.4% to 15.6%) among ED visits and 2.8% (95% CI 2.5% to 3.1%) for outpatient clinic visits.

Among patients 13 to 17 years of age, 2.9% (95% CI 2.6% to 3.2%) of visits included an opioid prescription compared with 7.5% (95% CI 7.0% to 8.1%) of visits among patients 18 to 22 years of age (Table 1). For both age groups, rates of opioid prescribing were significantly higher among ED visits compared with clinic visits, with a sixfold increase seen among adolescents and a fourfold increase seen among young adults. Rates of opioid prescribing were the highest for ED visits by young adults, with 17.9% (95% CI 17.1% to 18.6%) of visits being associated with an opioid prescription.

There was no significant change in the rates of opioid prescribing over the study period for either age group. However, when stratified by ambulatory setting, there was a small but significant decrease in the rate of opioid prescriptions among ED visits (OR 0.96; 95% CI 0.95 to 0.98), which was seen both in visits by adolescent patients (OR 0.95; 95% CI 0.92 to 0.97) and young adults (OR 0.98; 95% CI 0.96 to 0.99; Fig 1A). In outpatient clinics, there was no change in the rate of visits associated with an opioid prescription over time, either overall (OR 1.02; 95% CI 0.98 to 1.06) or by age group (Fig 1B).

For adolescents, outpatient clinics with the greatest number of visits associated with an opioid prescription were family practices and orthopedic surgery clinics (Table 2). The highest rates of opioid prescribing were for visits to general surgery clinics (8.9%; 95% CI 5.7% to 12.1%) and orthopedic surgery clinics (7.9%; 95% CI 6.2% to 9.6%). Psychiatry and pediatric clinics had the lowest rates of opioid prescribing (0.1% [95% CI 0.0% to 0.3%] and 0.4% [95% CI 0.3% to 0.6%], respectively).

Among young adults, the greatest number of visits with an opioid prescription were to family-practice clinics and clinics classified as “other specialties” (eg, gastroenterology, hematology, and vascular medicine), whereas the clinics with the highest opioid-prescribing rates were orthopedic surgery (11.5%; 95% CI 8.3% to 14.6%) and other specialties (8.4%; 95% CI 5.7% to 11.2%). The lowest rates of prescribing were for visits to pediatric clinics and obstetrics and gynecology clinics (0.8% [95% CI, 0.2% to 1.5%] and 1.7% [95% CI 1.0% to 2.3%], respectively).

The greatest number of opioid prescriptions dispensed to adolescents in the ED were for visits related to acute injuries (eg, ankle sprain and metacarpal fracture) as well as abdominal pain, acute pharyngitis, and dental disorders (Table 3). The highest rates of opioid prescribing were for dental disorders (59.7%; 95% CI 41.0% to 78.4%) and clavicle fractures (47.0%; 95% CI 29.8% to 64.3%). Among young adults, diagnoses most commonly resulting in an opioid prescription were similar and also included urinary tract infection, cellulitis, backache, and low-back pain (Table 4). Among this population, the highest rates of opioid prescribing were for visits for dental disorders (57.9%; 95% CI 49.4% to 66.5%) and low-back pain (38.0%; 95% CI 29.4% to 46.6%). Among the 5 most common complaints resulting in an opioid prescription in the ED, opioid prescribing tended to be lower in the Northeast and higher in the West regions of the United States (Supplemental Table 5).

We also examined relative opioid prescribing for individual opioids, which included hydrocodone, oxycodone, codeine, and tramadol (Fig 2). Other opioids, including suboxone, hydromorphone, fentanyl, morphine, and nalbuphine, were prescribed for a cumulative of 1.9% of all outpatient visits and included too few numbers to examine trends (<30 cases in each year of the study period). Hydrocodone was the most frequently prescribed opioid over the entire study period, although rates decreased relative to other opioids over the study period (OR 0.94; 95% CI 0.91 to 0.98). Oxycodone prescriptions remained relatively stable (OR 1.0; 95% CI 0.96 to 1.1), as did codeine prescriptions (OR 1.03, 95% CI 0.96 to 1.1). Tramadol had too few cases to analyze until 2012, when it comprised 15.1% of all opioid prescriptions and was prescribed at a stable rate through 2015 (OR 1.0; 95% CI 0.86 to 1.3). The relative contributions of each opioid for the most recent year of data available are shown in Supplemental Table 6.

In this large sample of national ambulatory care visits, opioids were prescribed in nearly 15% of ED visits by adolescents and young adults. Rates were much lower in outpatient clinics, where opioid prescriptions were dispensed in 3% of visits, although there was substantial variation across clinical specialties. We also found wide variation in prescribing rates for different conditions treated in EDs, with certain conditions being associated with particularly high rates of opioid prescriptions. Visits for dental disorders, for example, resulted in an opioid prescription in ∼60% of visits for both adolescents and young adults, whereas approximately half of adolescents with a clavicle fracture were prescribed an opioid. These findings serve to inform initiatives aiming to reduce excessive opioid prescribing, especially as we seek to further define and address the opioid epidemic in our younger patients.

Although the focus of opioid-prescribing guidelines and educational campaigns has been primarily on pain management in adult patients and those with chronic conditions,^11,22 adolescents and young adults are at high risk for opioid misuse and abuse after exposure from medical treatment, and a number of studies indicate that the opioid epidemic disproportionately affects young people. Legitimate use of prescription opioids among high school students has been associated with a 33% increase in the risk of future opioid misuse among young adults.²³ Data from a national survey conducted in 2016 showed that 7.3% of people between the ages of 18 and 25 years have misused opioids in the past year compared with only 4.0% among those ≥26 years old.²⁴ In 1 study evaluating opioid-naïve patients receiving an opioid prescription after surgery, those 15 to 24 years of age had the highest rates of filling the prescription as well as the highest rates of opioid misuse after the initial prescription.²⁵ Overall, >40% of adolescents who abuse a prescription opioid will progress to heroin use.²⁶ These findings highlight the tremendous risks associated with opioid use among adolescents and young adults when compared with older adults.

Our findings indicate that opioid prescribing in ambulatory care visits is particularly high in the ED setting and that certain diagnoses appear to be routinely treated with an opioid. Overall, higher prescribing rates in the ED compared with other ambulatory settings are expected given the nature of many ED visits. Dental disorders and acute injuries were among the conditions with the highest rates of opioid prescribing in the ED, potentially reflecting long-standing practices of ED physicians in treating patients with these conditions. However, when considering the total volume of opioid prescriptions dispensed, certain common conditions, including abdominal pain, acute pharyngitis, urinary tract infection, and headache, contributed large numbers of prescriptions as well. In identifying areas for intervention, opioid-prescribing patterns should be assessed both in terms of the rate and numbers of opioids prescribed.

Our results are comparable to those of other studies examining ED prescribing patterns in adults. One large study of Medicare and Medicaid beneficiaries revealed that 15% of patients were prescribed an opioid during ED visits.²⁷ Another study using NHAMCS data revealed that 18.7% of all discharges from EDs had an opioid prescription, and a multicenter analysis of ED care found that 17.0% of patients discharged from the ED were prescribed an opioid.^28,29 These figures also indicate that ED physicians prescribe opioids to adolescents and young adults at similar rates as they do to older patient groups.

Little information is available on the epidemiology of opioid prescribing among outpatient clinics. Previous work using national prescription data revealed that dentists, nonpediatrician general practitioners, and orthopedic surgeons dispensed the greatest number of opioid prescriptions to patients 10 to 19 years of age in outpatient clinics.¹⁴ One study evaluating opioid prescribing by primary care physicians to patients with back pain revealed that 61% of patients received an opioid prescription.³⁰ In another study, as many as 94% of adult postoperative patients received an opioid prescription after common elective procedures.³¹ However, to our knowledge, there have been no assessments focused on delineating opioid-prescribing rates among adolescents and young adults receiving care in outpatient clinics.

Policies and guidelines have been implemented (many at the state level) to curb opioid prescribing, including prescription-monitoring programs (PMPs), mandatory prescriber training, and pain-management clinic regulation. Overall, these interventions have been found to have variable impact on opioid prescribing. One study using a national data set of outpatient clinics in 24 states over 10 years reported a sustained reduction in the prescribing of schedule II narcotics after the implementation of a PMP.³² Another analysis focused on EDs in Ohio found that the initiation of a PMP was associated with a 12% reduction in opioid prescriptions.³³ ED providers appear to adhere to Centers for Disease Control and Prevention recommendations regarding opioid prescribing, which may account for the reductions in ED opioid prescribing.³⁴ However, other studies have shown a lack of effect of PMPs on prescription-opioid abuse, misuse, or dependence^35,36 and a minimal reduction in opioid prescriptions among ED patients.³⁷ Further study is needed to assess the impact of these interventions on adolescents and young adults, specifically, to ensure that programs are addressing the unique requirements of this patient population.

Our study has a number of strengths and limitations. The large number of sampled visits in the NHAMCS and NAMCS allows for detailed analyses of specific patient populations and treatment practices. Additionally, the availability of multiple years of survey data enables evaluations of changing practices over time, which is particularly relevant for opioids as policies are implemented to reduce excessive prescribing. The NHAMCS and NAMCS also sample visits from multiple types of ambulatory settings, allowing for assessment of ambulatory care visits with the highest rates of opioid prescribing. One of the limitations of the data sets is that they are based on visits and not individual patients, precluding longitudinal follow-up to ascertain opioid prescribing and use at the patient level. We were also unable to assess opioid prescriptions in greater detail, including the duration and quantity of medication dispensed, as well as patient adherence to prescription instructions. Estimates in the NHAMCS and NAMCS are adjusted for nonresponse for only a subset of variables, resulting in potential nonresponse bias.³⁸ 

Finally, data are only available through 2015 and do not capture more recent changes in opioid prescribing related to increased public awareness and more robust policy interventions.

During ambulatory care visits in the United States, opioids are prescribed to adolescents and young adults in <3% of outpatient clinic visits but in almost 1 in 6 visits to EDs. Certain conditions appear to be routinely treated with opioids among ED patients, with prescribing rates exceeding 40% for certain diagnoses. These findings inform targeted interventions and educational programs aiming to ensure judicious use of opioids in adolescents and young adults.

CI

confidence interval

ED

emergency department

ICD-9

International Classification of Diseases, Ninth Revision

NAMCS

National Ambulatory Medical Care Survey

NCHS

National Center for Health Statistics

NHAMCS

National Hospital and Ambulatory Medical Care Survey

OR

odds ratio

PMP

prescription-monitoring program