Rising Opioid Prescribing in Adult U.S. Emergency Department Visits: 2001–2010

Authors


  • Presented at the Society for Academic Emergency Medicine Annual Meeting, Atlanta, GA, May 2013.
  • The authors have no financial relationships relevant to this article or conflicts of interest to disclose. Dr. Pines, an associate editor for this journal, had no role in the peer review or publication decision for this paper.

Abstract

Objectives

The objective was to describe trends in opioid and nonopioid analgesia prescribing for adults in U.S. emergency departments (EDs) over the past decade.

Methods

Data from the National Hospital Ambulatory Medical Care Survey (NHAMCS) 2001 through 2010 were analyzed. ED visits for adult patients (≥18 years of age) during which an analgesic was prescribed were included. Trends in the use of six commonly prescribed opioids, stratified by Drug Enforcement Agency (DEA) schedule, as well as nonopioid analgesics were explored, along with the frequency of pain-related ED visits. For 2005 through 2010, data were further divided by whether the opioid was administered in the ED versus prescribed at discharge.

Results

Between 2001 and 2010, the percentage of overall ED visits (pain-related and non–pain-related) where any opioid analgesic was prescribed increased from 20.8% to 31.0%, an absolute increase of 10.2% (95% confidence interval [CI] = 7.0% to 13.4%) and a relative increase of 49.0%. Use of DEA schedule II analgesics increased from 7.6% in 2001 to 14.5% in 2010, an absolute increase of 6.9% (95% CI = 5.2% to 8.5%) and a relative increase of 90.8%. Use of schedule III through V agents increased from 12.6% in 2001 to 15.6% in 2010, an absolute increase of 3.0% (95% CI = 2.0% to 5.7%) and a relative increase of 23.8%. Prescribing of hydrocodone, hydromorphone, morphine, and oxycodone all increased significantly, while codeine and meperidine use declined. Prescribing of nonopioid analgesics was unchanged, 26.2% in 2001 and 27.3% in 2010 (95% CI = –1.0% to 3.4%). Hydromorphone and oxycodone had the greatest increase in ED administration between 2005 and 2010, while oxycodone and hydrocodone had the greatest increases in discharge prescriptions. There was no difference in discharge prescriptions for nonopioid analgesics. The percentage of visits for painful conditions during the period increased from 47.1% in 2001 to 51.1% in 2010, an absolute increase of 4.0% (95% CI = 2.3% to 5.8%).

Conclusions

There has been a dramatic increase in prescribing of opioid analgesics in U.S. EDs in the past decade, coupled with a modest increase in pain-related complaints. Prescribing of nonopioid analgesics did not significantly change.

Resumen

Objetivo

Describir las tendencias en la prescripción a adultos de analgésicos opiáceos y no opiáceos en los servicios de urgencias (SU) de Estados Unidos durante la pasada década.

Metodología

Se analizaron los datos de la National Hospital Ambulatory Medical Care Survey (NHAMCS) desde 2001 hasta 2010. Se incluyeron las visitas al SU de pacientes adultos (≥ 18 años de edad) en las que se prescribió un analgésico. Se exploraron las tendencias en el uso de los seis opiáceos más frecuentemente prescritos – estratificados por la lista de la Drug Enforcement Agency (DEA) - así como de los analgésicos no opiáceos, junto con la frecuencia de visitas al SU por dolor. Desde 2005 hasta 2010, los datos se dividieron también en función de si el opioide se administró en el SU o se prescribió al alta.

Resultados

Entre 2001 y 2010, el porcentaje de todas las visitas al SU relacionadas con el dolor y no relacionadas con el dolor donde se prescribió cualquier analgésico opiáceo se incrementó de un 20,8% a 31,0%, un incremento absoluto de un 10,2% (IC 95% = 7,0% a 13,4%), y un incremento relativo de un 49.0%. El uso de analgésicos de la lista II según la DEA se incrementó de un 7.6% en 2001 a un 14,5% en 2010, un incremento absoluto de un 6,9% (IC 95% = 5,2% a 8,5%) y relativo de un 90,8%. Los agentes de las listas III a V se incrementaron de un 12,6% en 2001 a un 15,6% en 2010, un incremento absoluto de un 3,0% (IC 95% = 2,0% a 5,7%) y relativo de un 23,8%. La prescripción de hidrocodona, hidromorfona, morfina y oxicodona se incrementó significativamente, mientras que el uso de codeína y meperidina descendió. La prescripción de analgésicos no opiáceos no varió, un 26,2% en 2001 y un 27,3% en 2010 (IC 95% = –1.0% a 3,4%). La hidromorfona y la oxicodona tuvieron el mayor incremento en su administración entre 2005 y 2010, mientras que la oxicodona y la hidrocodona tuvieron los mayores aumentos de prescripción al alta. No hubo diferencias en las prescripciones al alta para los analgésicos no opiáceos. El porcentaje de visitas por patologías dolorosas durante el periodo se incrementó de un 47,1% en 2001 a un 51,1% en 2010, un incremento absoluto de un 4,0% (IC 95% = 2,3% a 5,8%).

Conclusiones

Ha habido un gran incremento en la prescripción de analgésicos opiáceos en los SU de Estados Unidos en la pasada década, acompañado de un aumento modesto en las patologías relacionadas con dolor. La prescripción de analgésicos no opiáceos no cambió de forma significativa.

In 2011, a total of 238 million opioid prescriptions were filled by U.S. pharmacies, up from 174 million in 2000.[1-4] Rising prescription rates have been paralleled by even more dramatic increases in opioid abuse and associated fatalities from both medical and nonmedical use. In 2011, 6.1 million Americans reported using prescription medications for nonmedical purposes within the previous month.[3] In addition, the majority of these pharmaceuticals used for nonmedical purposes were obtained from prescriptions, usually from friends or relatives.[3] According to the Centers for Disease Control and Prevention (CDC), 14,800 poisoning deaths in 2008 involved opioid analgesics, compared to 4,000 in 1999.[4]

With the Joint Commission's emphasis on the identification and treatment of pain,[5] emergency departments (EDs) and hospitals have become a major source of opioid prescriptions.[6, 7] One study noted a 1.5-fold increase, from 23% to 37%, in opioid analgesic prescribing in U.S. EDs between 1993 and 2005 for painful conditions.[6] The American College of Emergency Physicians (ACEP) issued a clinical policy for opioid prescribing in 2012, recommending prescription of short-acting opioid analgesics when necessary to treat painful conditions.[6] No studies to our knowledge have investigated recent trends in ED opioid prescriptions, nor explored which specific preparations may be contributing to higher rates of opioid use in the ED.

In this study, we describe trends in opioid and nonopioid prescribing in U.S. EDs from 2001 through 2010 and also whether pain-related visits are increasing. We specifically explored which demographic groups, medications, and ED reasons for visit may be accounting for more of the change in the prescribing rates of opioid and nonopioid analgesics.

Methods

Study Design

This was a retrospective review of publicly available data from the National Hospital Ambulatory Care Survey (NHAMCS) from 2001 to 2010. This study used publicly available data without identifiers and was not considered human subjects research.

Study Setting and Population

The NHAMCS is an annual multistage probabilistic sample of visits made to U.S. EDs conducted by the CDC's National Center for Health Statistics. The survey is designed to make national estimates about the use and provision of services in hospital-based EDs. The NHAMCS survey instrument is a patient record form, which is completed by trained staff for a random sample of visits during randomly assigned reporting periods. Specific data included are patient demographics, payment source, reason for visit, recent admissions or ED visits, diagnosis, services provided, type of provider seen, cause of injury, medications prescribed, procedures performed, and patient disposition. Data on facility characteristics are also included, such as geographic region, metropolitan status, and funding source. Data are available for public use and can be downloaded from the CDC Web site.[8]

Study Protocol

All opioids in the database were identified and then categorized based on Drug Enforcement Agency (DEA) schedule and grouped together as either schedule II controlled substances or schedules III, IV, and V. We then further focused the analysis on six opioid analgesics commonly prescribed in the ED, some with high abuse potential: hydromorphone, meperidine, morphine, and oxycodone (all DEA schedule II) or hydrocodone and codeine (schedules III–V). All single-agent and combination opioid preparations (for example, oxycodone/acetaminophen) were included. For morphine and oxycodone, we also specifically examined sustained-release products. Nonopioid analgesics such as acetaminophen and nonsteroidal anti-inflammatory drugs were evaluated to determine if there were changes in prescribing of these medications over the study period. A single author (MMA), who is board-certified in emergency medicine and medical toxicology and is also a registered pharmacist, performed all medication coding. Another author (PMM) coded pain-related visits. Pain-related visits were defined as reasons for visit in which pain was explicitly mentioned. Examples of common pain-related reasons for visit included back pain, headache, and chest pain.

Some data included in NHAMCS changed over the course of the study period. For example, the number of medications recorded per visit increased from six to eight in 2003. To ensure comparability across the study period, we limited the analysis from 2003 to the first six medications coded. The distinction between medications administered in the ED and those prescribed at discharge was not delineated in NHAMCS until 2005. We explored trends in overall prescribing for 10 years and performed a second analysis focusing on trends in administration in the ED versus prescribing at discharge for the years 2005 through 2010.

The sample was restricted to visits involving patients 18 years of age or older. Additional patient demographic data analyzed included age older than 65 years, sex, race, and source of payment (Medicare, Medicaid, self-pay, private, other). Hospital characteristics such as geographic region, location, and hospital type were also explored.

Data Analysis

We tabulated the proportion of visits (for any reason) where patients received one or more of each drug class by year. We then calculated the absolute change in the proportion of patient visits in which each class of drug was prescribed using survey-weighted linear combinations of estimators and calculated 95% confidence intervals (CIs). The relative percentage change in prescribing between 2001 and 2010 for each subcategory was calculated. We also assessed if there were any changes over time in “pain-related” or “non–pain-related” visits. In addition, we explored trends in opioid prescribing for several common pain-related primary reasons for visit: abdominal pain; back pain; chest pain; headache; musculoskeletal pain; and tooth, mouth, and ear pain. A comprehensive list of pain-related reasons for visit and medications by category is shown in Data Supplement S1 (available as supporting information in the online version of this paper).

To ensure that sample sizes were sufficient to demonstrate differences, we used survey-weighted effect sizes to calculate unweighted sample size estimates for each analysis conducted. Actual sample sizes for tested and reported subgroups (e.g., specific commonly prescribed medications, broad chief complaints, demographic groups, hospital types) were larger than necessary to detect significant effects between 2001 and 2010. Due to small sample sizes for some reasons for visit, analyses of pain-related reasons for visit were grouped into two 2-year time periods (2001–2002 and 2009–2010). However, for some individual medications (i.e., meperidine and long-acting opioids), there were insufficient observations (fewer than 30 per cell) to make stable estimates in both 2001 and 2010; therefore, these data are not reported. A p-value of <0.05 was considered significant for assessing whether trends were significant in all analyses. All analyses were performed using Stata, version 12 (StataCorp, College Station, TX).

Results

Frequency of Opioid Utilization

Between 2001 and 2010, the percentage of ED visits where opioid analgesics were prescribed increased from 20.8% to 31.0%, an absolute increase of 10.2% (95% CI = 7.0% to 13.4%) and a relative increase of 49.0%. Visits in which a DEA schedule II analgesic was prescribed increased from 7.6% in 2001 to 14.5% in 2010, an absolute increase of 6.9% (95% CI = 5.2% to 8.5%) and a relative increase of 90.8%. Prescriptions for DEA schedules III and V increased from 12.6% to 15.6%, a 3.0% absolute difference (95% CI = 2.0% to 5.7%) and a 23.8% relative increase.

Demographic Differences

There was a significant increase in opioid prescribing across all age groups, including those over 65 years (Table 1). In addition, there were increases in opioid use in both blacks and whites; however, blacks were consistently prescribed fewer opioids than whites throughout the study period. There were significant increases in opioid use for ED visits in all categories of payer.

Table 1. Demographic Characteristics of ED Visits During Which an Opioid Analgesic Was Administered/Prescribed
Characteristic2001, % (n)2010, % (n)Absolute Change, %95% CIRelative Change, %
Patient
Age > 65 yr12.2 (619)18.9 (961)6.64.4 to 8.854.9
Sex
Male20.7 (2,290)30.9 (3,350)10.26.7 to 13.849.3
Female20.8 (2,871)31.0 (4,436)10.27.0 to 13.349.0
Race
White22.1 (4,162)31.8 (5,980)9.76.2 to 13.143.9
Black15.5 (852)28.4 (1,519)12.99.8 to 16.183.2
Other19.5 (147)27.6 (287)8.11.0 to 15.241.5
Source of payment
Private insurance 23.4 (2,177)30.9 (2,907)7.53.4 to 11.532.1
Medicaid21.7 (692)30.7 (1,989)9.04.6 to 13.341.5
Medicare14.9 (722)23.1 (1,356)8.25.8 to 10.555.0
Self-pay22.0 (903)33.4 (1,733)11.46.1 to 14.951.8
Other18.3 (588)33.9 (429)15.69.4 to 21.885.2
Hospital
Geographic region
Northeast15.8 (925)23.8 (1,466)8.02.7 to 13.250.6
Midwest18.8 (1,109)30.7 (1,692)11.98.4 to 15.463.3
South21.3 (1,904)32.1 (2,937)10.86.8 to 14.950.7
West27.7 (1,223)36.1 (1,691)8.43.0 to 13.930.3
Location
Urban20.9 (4,479)32.1 (6,881)11.27.6 to 14.853.6
Nonurban20.2 (682)25.3 (905)5.1–0.4 to 10.725.2
Type
Nonprofit20.6 (3,719)31.3 (5,815)10.77.2 to 14.351.9
Government18.7 (788)29.2 (1,261)10.55.2 to 15.856.1
For-profit25.6 (654)30.6 (710)5.0–0.5 to 10.519.5

Although opioid utilization increased across the United States, there was significant regional variation in opioid prescribing patterns. The Midwestern states saw the largest proportional increase in opioid prescriptions, from 18.8% to 30.7% (95% CI = 8.4% to 15.4%). The Western states had the highest overall frequency of opioids prescribed throughout the study period, ranging from 27.7% in 2001 to 36.1% in 2010. The Northeast had the lowest rates of opioid utilization throughout the study period. Opioids were more commonly prescribed in urban EDs and in nonprofit hospitals (Table 1).

Utilization Trends for Specific Analgesics

Prescribing rates for all of the opioid analgesics except for codeine and meperidine increased over the study period. The greatest relative increases were noted in ED use of hydromorphone and morphine (668.2 and 330.1%, respectively). Hydrocodone was the most commonly prescribed opioid, increasing from prescription in 10.4% of ED visits to 14.0% of ED visits (95% CI = 1.1% to 6.1%). Codeine prescribing decreased slightly from 1.6% to 1.3% of ED visits (95% CI = –0.8% to 0.2%) and meperidine prescribing decreased from 4.9% to 1.0% (95% CI = –4.8% to –3.0%). There were no changes in the prescribing of nonopioid analgesics, including acetaminophen and nonsteroidal anti-inflammatories: 26.2% in 2001 to 27.3% in 2010 (95% CI = –1.0% to 3.4%; Table 2).

Table 2. Utilization of Individual Opioids and Nonopioid Analgesics
MedicationUtilization, % (n)Absolute Change, %95% CIRelative Change, %
20012010
Codeine1.6 (456)1.3 (330)–0.3–0.8 to 0.2–16.8
Hydrocodone10.4 (2,518)14.0 (3,532)3.61.1 to 6.134.7
Hydromorphone0.9 (217)6.8 (1,670)5.94.9 to 6.8668.2
Meperidine4.9 (1,174)1.0 (245)–3.9–4.8 to –3.0–79.6
Morphine1.6 (413)6.7 (1,674)5.14.1 to 6.1313.6
Oxycodone3.1 (805)7.1 (1,780)4.12.7 to 5.5133.4
Nonopioids26.2 (6,902)27.3 (7,347)1.2–1.0 to 3.44.5

The percentage of visits for “painful conditions” (identified by the first reason for visit) increased from 47.1% in 2001 to 51.1% in 2010, an absolute increase of 4.0% (95% CI = 2.3% to 5.8%). When examined by common reasons for visit, there was a larger relative increase in hydromorphone prescribing compared to the other opioids for all diagnoses except tooth, mouth, and ear pain. The most notable relative increase in hydromorphone prescribing was in patient visits related to chest pain, which increased 798.2% from 2001 to 2010. There were also significant increases in morphine and oxycodone prescribing for most reasons for visit (Table 3). Depending on the agent involved and reason for visit, there was not adequate sample size to make accurate comparisons for a few scenarios, including sustained-release opioids.

Table 3. Changes in Medication Utilization by Reason for Visit, 2001–2002 to 2009–2010
Chief ComplaintMedicationUtilization, % (n)Absolute Change, %95% CIRelative Change, %
2001–20022009–2010
  1. NSAID = nonsteroidal anti-inflammatory drug.

  2. *Opioids not included did not have adequate sample size for comparison.

Abdominal painCodeine1.5 (67)1.1 (70)–0.4–1.1 to 0.4–25.0
Hydrocodone10.3 (464)14.7 (758)4.41.9 to 6.842.0
Hydromorphone2.9 (143)17.5 (965)14.612.2 to 17.1507.3
Meperidine11.9 (534)2.1 (124)–9.7–11.7 to –7.8–81.9
Morphine4.9 (228)15.5 (857)10.68.5 to 12.7215.9
Oxycodone3.0 (160)7.1 (423)4.22.8 to 5.5140.9
Acetaminophen5.3 (290)4.8 (324)–0.4–1.5 to 0.6–8.2
NSAIDs18.1 (892)18.0 (1,115)–0.1–2.5 to 2.2–0.6
Back painCodeine1.9 (88)1.9 (68)0.0–0.8 to 0.80.0
Hydrocodone23.6 (815)25.7 (964)2.1–2.2 to 6.38.7
Hydromorphone3.0 (104)12.2 (470)9.27.2 to 11.3312.9
Meperidine11.6 (357)2.5 (89)–9.2–10.9 to 7.4–78.9
Morphine4.1 (138)8.7 (323)4.63.0 to 6.2110.9
Oxycodone8.0 (300)15.2 (550)7.24.4 to 10.190.1
Acetaminophen5.8 (243)5.2 (218)–0.7–0.2 to 0.7–11.3
NSAIDs43.9 (1,635)41.5 (1,620)–2.4–6.3 to 1.6–5.4
Chest painCodeine0.8 (42)0.8 (37)0.0–0.5 to 0.4–2.4
Hydrocodone5.9 (256)7.7 (318)1.8–0.1 to 3.730.0
Hydromorphone0.6 (32)5.0 (201)4.53.5 to 5.5798.2
Morphine4.4 (195)11.4 (438)7.05.3 to 8.7160.2
Oxycodone1.6 (72)2.8 (136)1.30.5 to 2.078.6
Acetaminophen4.9 (250)6.6 (293)1.60.2 to 3.032.9
NSAIDs35.9 (1,574)41.2 (1,731)5.32.1 to 8.514.8
HeadacheCodeine2.1 (64)1.6 (45)–0.5–1.5 to 0.5–22.6
Hydrocodone13.6 (394)14.6 (442)1.1–2.2 to 4.37.9
Hydromorphone2.7 (89)10.1 (339)7.45.4 to 9.4277.2
Meperidine14.8 (422)2.6 (92)–12.2–14.7 to –9.7–82.3
Morphine2.4 (72)7.7 (226)5.33.6 to 7.0224.5
Oxycodone3.3 (123)5.5 (179)2.10.5 to 3.864.0
Acetaminophen9.6 (339)10.0 (349)0.3–1.9 to 2.64.1
NSAIDs27.0 (874)30.7 (980)3.70.4 to 6.913.5
Musculoskeletal painCodeine2.7 (186)1.8 (121)–0.9–1.6 to –0.1–31.6
Hydrocodone21.4 (1,190)25.3 (1,497)3.90.2 to 7.718.4
Hydromorphone1.0 (66)6.7 (403)5.84.7 to 6.8600.0
Meperidine5.8 (310)1.4 (78)–4.4–5.3 to –3.5–76.2
Morphine2.3 (127)7.7 (419)5.44.2 to 6.7238.6
Oxycodone6.5 (404)12.5 (728)6.03.6 to 8.291.0
Acetaminophen6.7 (444)5.7 (397)–1.0–2.4 to 0.4–14.9
NSAIDs41.1 (2,532)38.2 (2,362)–2.9–5.9 to 0.1–7.1
Tooth/mouth/ear painCodeine4.1 (77)4.2 (57)0.1–2.0 to 2.23.4
Hydrocodone23.9 (346)27.4 (489)3.6–1.8 to 8.914.9
Oxycodone8.1 (123)12.6 (205)4.50.6 to 8.555.8
Acetaminophen7.1 (144)6.6 (131)–0.4–2.8 to 2.0–6.2
NSAIDs20.9 (352)26.0 (462)5.20.6 to 9.724.7

Prescribing Trends 2005–2010: ED Administration Versus Discharge Prescription

From 2005 to 2010, hydromorphone and oxycodone had the greatest relative increases for medications administered in the ED, with 102.5 and 54.2% relative increases respectively over the 5-year period. For discharge prescriptions, oxycodone (79.6%) and hydrocodone (15.4%) had the greatest relative increases between 2005 and 2010. There were no significant increases in discharge prescriptions for nonopioid analgesics (Table 4). Because visits during which sustained-release morphine and oxycodone products were prescribed were very rare, there was insufficient sample size to make accurate estimates or comparisons.

Table 4. Utilization of Individual Opioids and Nonopioids in the ED and Prescribed at Discharge, 2005 Versus 2010
Medication20052010Absolute Change, %95% CIRelative Change, %
In ED, % (n)
  1. *Inadequate sample size for comparison denoted by (—).

Codeine0.3 (63)0.3 (79)0.0–0.2 to 0.20.0
Hydrocodone2.6 (683)3.0 (758)0.4–0.1 to 1.017.4
Hydromorphone3.2 (753)6.5 (1,609)3.32.4 to 4.1102.5
Meperidine
Morphine5.0 (1,180)6.5 (1,622)1.50.6 to 2.530.6
Oxycodone1.3 (330)2.0 (527)0.70.3 to 1.154.2
Nonopioids15.2 (3,799)17.4 (4,473)2.21.0 to 3.414.2
 At Discharge, % (n)   
Codeine0.9 (198)0.8 (196)–0.1–0.4 to 0.3–5.8
Hydrocodone7.9 (1,934)9.1 (2,333)1.20.1 to 2.315.4
Hydromorphone
Meperidine
Morphine
Oxycodone2.2 (496)3.9 (952)1.71.0 to 2.579.6
Nonopioids10.2 (2,608)10.1 (2,795)0.1–1.3 to 1.1–1.1

Discussion

We found a near doubling in the prescribing of opioid analgesics in U.S. EDs over the past decade, coupled with a modest increase in nonopioid use and pain-related complaints. Based on the relatively steady levels of nonopioid analgesia prescriptions, it appears that opioids are being prescribed in addition to (rather than instead of) nonopioids. Codeine and meperidine were the only opioids to see lower prescription rates, possibly in response to concerns about safety and efficacy of these medications.[9, 10] Certain reasons for visit seem to be contributing more to trends in opioid prescribing. Specifically, opioids are being prescribed more frequently for chest pain, abdominal pain, and headaches. Increases in opioid prescribing for certain conditions, such as abdominal pain, may be due to changes in practice patterns where treating pain is now known not to affect diagnostic precision as previously thought; reasons for these trends in other specific complaints are less clear.

Many factors likely influence the overall trend toward more aggressive pain management. Generally, the trend toward opioid prescribing likely reflects a response to concerns by providers, patient advocacy groups, and accrediting bodies that pain is undertreated.[1, 5, 6, 11] In many institutions, ED provider performance is now being measured and reimbursements tied to patient experience data, factors that might alter prescribing patterns.[12] Recent analysis has shown that physicians err on the side of prescribing opioids in an effort to avoid withholding analgesia from patients who might be in pain.[13] One potential concern is that the well-intentioned desire to treat pain has resulted in use of prescription of opioids for conditions where it may not be the ideal agent. For example, we found significant increases in the prescribing of opioids for headache, a practice that has been discouraged by current clinical practice guidelines.[14, 15] Another contributor to ED opioid prescribing could be outpatient provider hesitancy to prescribe opioids. This phenomenon could in turn drive referrals to the ED for pain control and might explain the slight increase in pain-related visits.

Our findings demonstrate there were significant differences in the opioids administered in the ED versus those prescribed at discharge. Increased prescribing of hydromorphone and morphine were major drivers of opioid use in the ED. These agents are generally administered parenterally, and thus are not as easily diverted as oral formulations. Appropriate hospital-based parenteral administration of opioid analgesics in the acute setting has not been associated with subsequent addiction.[16-18] At the same time, data regarding how acute parenteral administration of opioids in the setting of preexisting chronic opioid use (medical or nonmedical) affects addiction behavior are lacking; this presents an area for future research. Of note, there was an increase in discharge prescribing of oxycodone and hydrocodone, which have been frequently implicated in prescription drug abuse. The increase in outpatient discharge prescriptions for agents with high abuse potential is concerning, as some of these prescriptions may ultimately be diverted and abused.

Another clinical challenge that may become more commonly encountered in the ED setting is opioid-induced hyperalgesia, a phenomenon where opioid use may actually increase pain perception. It has been described primarily in patients on long-standing opioid therapy. Although the true clinical effect of opioid-induced hyperalgesia has yet to be defined, increased pain perception on the part of patients may drive opioid prescribing patterns in patients already on chronic opioid therapy.[19, 20]

Opioid prescriptions in ED visits by black patients increased by 84%, but despite this larger relative increase, black patients are still less likely to be prescribed opioids than white patients. Interestingly, the magnitude of the disparity in opioid prescribing between whites and blacks decreased significantly over the 10-year period. The reasons for such continued disparities are complex and are still not entirely clear.[7, 21] Some have suggested that racial differences might represent overprescribing of opioids to white patients.[7] Nonmedical use of opioids is more common in white patients, and providers might be less likely to identify signs of abuse in white patients.[22, 23] Provider bias, provider–patient communication, and provider decision-making are complex and difficult to study, but important to understand if the pain of minority patients is to be managed equitably. Future studies should focus on investigating factors that contribute to differential opioid prescribing based on race.

The expansion of opioid prescribing and its repercussions have prompted regulators and providers to initiate efforts aimed at standardizing opioid administration and preventing diversion and abuse. Long-acting opioid analgesics have been incorporated into the FDA's Risk Evaluation and Mitigation Strategies program, which requires patient medication guides to be dispensed with prescriptions and expanded provider education.[24] Our study did not find a significant number of visits during which sustained-release opioid products were prescribed, indicating that these agents are not commonly prescribed in EDs. ACEP and two states (Washington and New York) have developed standardized opioid prescribing guidelines.[4, 7, 25, 26] Several states have instituted prescription drug monitoring programs to alert providers to “doctor shopping” in an effort to reduce improper prescribing and diversion. Preliminary data suggest that these programs have a positive effect on prescribing practices and drug diversion; however, the true effects and optimal organization of these programs have yet to be determined.[27-29] The geographic differences described in our study and elsewhere[30-32] highlight the degree to which regional standards of care and different regulatory environments influence prescribing practice. Finally, the pharmaceutical industry has begun developing abuse-deterrent formulations of high-risk products.[23]

Our findings suggest the need for further research, provider education, and evidence-based guidelines to assist clinicians who prescribe analgesia in the ED. The ACEP guidelines are an initial step; however, they do not completely address the many facets and complexities of opioid use. Efforts to examine dose equivalence and appropriate indications for opioids are needed. On the provider level, educational programs to help clinicians identify drug-seeking behavior and feedback on one's own prescribing practices might be helpful. Finally, while identifying and addressing pain is an important therapeutic goal, it is important to educate patients about the risks of broad dissemination of opioids.

Limitations

The main limitations of this study reflect the changing NHAMCS database over the study period and the use of survey administrative data for this sort of analysis.[33] Although the multistage sampling of the NHAMCS database is designed to make the sample representative, these estimates may contain minor inaccuracies; however, the consistency and rigor of the NHAMCS methodology should protect against major inaccuracies.[8] Additionally, we did not adjust for potential family-wise Type I error rate inflation, which may arise due to multiple comparisons.

The NHAMCS is a national-level database that includes information pertaining to visits, not specific patients. As such, it is not possible to review individual patient charts for further detail or to follow patients longitudinally. In addition, the same patients could have been evaluated and received analgesic multiple times during the sampling period. There can also be inaccuracies in specific variables. For example, determining race is challenging because race is typically recorded by an unrelated observer rather than by the patient and therefore may be misclassified. Other variables, such as pain scores, are poorly populated in NHAMCS.[8, 33]

There were some changes in medication coding over the study period. Prior to 2003, the NHAMCS survey form recorded only six medications per visit. The number of medications recorded was increased to eight in 2003.[8] There is not a specific order in which medications are recorded onto the form, making it possible that some opioid medications were discounted. While we restricted analyses to the first six medications in each year in the study period to ensure comparability, this may have resulted in an underestimate of recent opioid utilization. We could not differentiate between medications administered in the ED versus those prescribed at discharge for the early part of the study period (2001–2004), as these data were not divided by disposition during that time. It was not possible to examine the route of administration because this information is not recorded on the survey form. There may have been inaccuracies in distinguishing between different product preparations, such as immediate- versus sustained-release products, because for some medications listed, this information is not specified. Data regarding dosing schedule and duration of therapy are not included in NHAMCS.

Another major limitation of using a large database such as NHAMCS is that it is not possible to review the chart to assess the appropriateness of opioid administration or to evaluate the outcomes associated with opioid administration.[8, 33] In addition, we cannot directly link prescribing patterns with subsequent diversion or the nonmedical use of opioid analgesics.

Conclusions

There has been a significant increase in opioid utilization in visits over the past decade, with the most profound rises found in hydromorphone use. Although we cannot directly link ED prescribing with subsequent abuse, this trend requires further investigation given the concomitant rise in prescription opioid abuse and related fatalities. ED providers must be vigilant in addressing pain while minimizing the adverse effects associated with opioid use. Ultimately, a multifaceted approach will likely be required to ensure appropriate use of opioid analgesics in the ED and to mitigate the morbidity and mortality caused by these medications.

Ancillary