Professional Documents
Culture Documents
Pediatric Emergency Care • Volume 00, Number 00, Month 2017 www.pec-online.com 1
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
DePhillips et al Pediatric Emergency Care • Volume 00, Number 00, Month 2017
hospital EDs, 1 urban and 1 suburban, and 2 children’s UC sites, sprains, burns, and dental and eye injuries. The most common
both suburban. Table 1 shows the number of visits and patient de- noninjury diagnoses included abscesses, ear and throat pain, viral
mographics by site. The urban ED is staffed by pediatric emer- illness, stomatitis, and abdominal pain. The majority of these co-
gency medicine (PEM)–trained physicians and fellows and the variates were chosen based on previous literature, which showed
pediatric, emergency medicine (EM), and family practice resi- that opioids were more frequently prescribed for patients who
dents they supervise, as well as a few EM-trained physicians. were female, adolescent, and white and for those with private in-
The suburban ED is staffed by PEM physicians, general pediatri- surance or written from an urban ED.10,24
cians, and nurse practitioners. Pediatric emergency medicine phy- The primary outcome of interest was the opioid prescription
sicians work only in the children’s EDs in our system. The UCs duration in days. As the majority of prescriptions were written for
are staffed by general pediatricians and nurse practitioners. Visits a specific dispense quantity and did not include length of treat-
for all children 18 years or younger who were discharged from the ment, the number of days was calculated. This was achieved by di-
ED or UC from June 1, 2012, to May 31, 2013, and who received viding the total dose prescribed in milligrams (dispense quantity
an outpatient prescription for an oral opioid narcotic were in- in tablets or milliliters strength of medication) by the total daily
cluded in this study. Visits were identified using the electronic dose (number of doses per day prescribed dose). The visit was
central discharge database, which captured all visits during the excluded from the analysis if information needed to make this cal-
study period. culation was missing.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care • Volume 00, Number 00, Month 2017 Opioid Prescribing Practices
DISCUSSION
This retrospective study of opioid prescribing over 1 year in
the pediatric EDs and UCs of a Midwestern children’s hospital
system found that, although the proportion of pediatric ED visits
FIGURE 1. Data abstraction.
in which an opioid was prescribed was relatively low (3%), the
proportion of those visits for which more than a 5-day supply
was prescribed was nearly 1 in 5. There is little data from other
hypothesized that injury diagnoses and English speakers would re- children’s hospitals available, so it is not known how these num-
ceive more opioids, and PEM physicians would prescribe less, so bers compare with other sites. One study looked at pediatric ED
these were also chosen as reference groups. All data were analyzed visits where an opioid was given or prescribed and found that this
in SPSS version 18 (IBM Corporation, Armonk, NY). was most frequent in the Western region and less frequent but sim-
ilar in the Northeast, Midwest, and South.24 However, they did not
RESULTS separate opioids given in the ED and prescribed for home, so we
There were a total of 4606 ED and UC visits identified by the cannot extrapolate based on these data. Factors associated with
electronic medical record as having an outpatient prescription for larger supplies prescribed were young patient age, UC setting, pre-
an opioid narcotic written during the study period. Four hundred scriptions by a resident physician, and noninjury diagnosis.
forty-eight were excluded because they were discontinued. There We defined our outcome as more than 5 days of opioid pre-
were a number of visits where several prescriptions were written scribed, which was consistent with existing ED guidelines that rec-
for the same encounter, and it could not be determined which pre- ommend between 3 to 7 days, depending on the state. The existing
scription was given; these 83 prescriptions were not included. This
left a total of 4075 available for analysis, or approximately 3% of all
ED and UC visits during the year (Fig. 1). Eighty percent originated TABLE 2. Demographics of the Study Population (total
in the ED and 20% from the UC. Demographic characteristics of N = 4075)
patients in the study mirrored our general patient population across
the included ED and UC sites with respect to language but were Variable n (%)
more likely to be older white males and commercially insured
(Tables 1 and 2). Pediatric emergency medicine physicians Sex
prescribed 16% of opioids. The pediatric and rotating (EM and Male 2306 (57)
family practice) residents they supervised were responsible for Female 1769 (43)
24% and 11% of the prescriptions, respectively. General Primary language
pediatricians wrote for 49%. Injury complaints accounted for 65% English 3816 (93)
of diagnoses, and noninjury, 35%. More specifically, the most Spanish 225 (6)
common diagnoses receiving an opioid prescription were fractures Other 34 (1)
(40%); skin complaints including burns, wounds, abscesses, and Race/ethnicity
skin infections (16%); and other orthopedic diagnoses including
White 2560 (63)
sprains, dislocations, and crush injuries (9%). Few prescriptions
(<1%) were written for children with medical conditions associated Black 775 (19)
with chronic pain of which the majority were for children with Hispanic 441 (11)
hemoglobinopathies (sickle cell disease, spherocytosis, etc). Other 299 (7)
Prescription Information Payment source
Government insurance 1776 (44)
Oxycodone and oxycodone-containing products accounted
for 93% of the prescriptions. The remaining 7% were composed Commercial insurance 1978 (48)
of hydrocodone, codeine, and products containing those opioids. Self-pay/other 321 (8)
There were no refills given on any of the prescriptions. Of the Age
4075 prescriptions, 84 were excluded for further analysis because <1 y 83 (2)
of insufficient information needed to calculate the number of days 1–4 y 992 (24)
supplied. Of the 3991 prescriptions in the analysis, the median 5–9 y 1180 (29)
number of days prescribed was 3.3 with an interquartile range of 10–14 y 1226 (30)
2.5 days and a total range of 1 to 44 days. A total of 721 (18%) >14 y 594 (15)
were written for more than a 5-day supply (Fig. 2).
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
DePhillips et al Pediatric Emergency Care • Volume 00, Number 00, Month 2017
TABLE 3. Patient and Clinical Characteristics Associated With Opioid Prescriptions of More Than 5 Days’ Duration
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care • Volume 00, Number 00, Month 2017 Opioid Prescribing Practices
TABLE 4. Patient and Clinical Characteristics Associated With Opioid Prescriptions of More Than 5 Days’ Duration, Adjusted Odds
Ratio
Variable 0–5 d (n = 3270), n (%) >5 d (n = 721), n (%) Adjusted OR* (95% CI)
Age
<1 y 40 (1) 41 (6) 12.3 (7.3–21.0)
1–4 y 597 (18) 371 (51) 7.7 (5.5–10.8)
5–9 y 973 (30) 180 (25) 2.4 (1.7–3.4)
10–14 y 1123 (34) 84 (12) 0.9 (0.7–1.4)
>14 y 537 (17) 45 (6) REF
Location
ED 2635 (81) 543 (75) REF
UC 636 (19) 178 (25) 1.4 (1.1–1.7)
Prescriber
PEM physician 538 (16) 98 (14) REF
General pediatrician 1565 (48) 374 (52) 1.1 (0.9–1.5)
Resident physician 1167 (36) 249 (34) 1.4 (1.1–1.8)
Diagnosis
Injury 2299 (70) 416 (58) REF
Noninjury 971 (30) 305 (42) 1.4 (1.2–1.7)
*Adjusted for all other variables in the table.
REF indicates reference group.
acute care guidelines in New York City (NYC), Ohio, and Oklahoma significant predictor. We believe that this is still a good focus for
suggest limiting the prescription to a 3-day supply.15–17 Those in education because pediatricians also do not receive sufficient ed-
Arizona and Washington suggest prescribing no more than ucation on opioid prescribing.
30 doses (5–7 days).18,19 Using 5 days may have resulted in a con- Interestingly, patients with noninjury diagnoses were more
servative estimate of the prevalence of prescriptions written for often prescribed more than 5 days of opioid analgesics compared
prolonged duration. If we had defined prolonged duration as more with those with injury diagnoses. The most common noninjury di-
than 3 days per the Ohio, Oklahoma, and NYC guidelines, 56% of agnoses included abscesses, ear and throat pain, viral illness, stoma-
prescriptions in this study would have met this definition. titis, and abdominal pain. One possible reason is that providers may
In our study, the younger the patient, the more likely a greater view pain associated with these diagnoses as longer lasting than in-
than 5-day supply of opioids was prescribed. Initially, we thought jury diagnoses. There is minimal literature looking at how long pain
this may be the result of a default dispense quantity (120 mL) for lasts in noninjury diagnoses, and therefore they may be prescribed
liquid medications specific to our electronic prescribing system. more than necessary. Further research is needed.
However, after removing all prescriptions written for 120 mL There was a large variability in the number of days of opioid
and recalculating ORs, we did not find a difference in results or prescribed, which is consistent with the fact that guidelines for
significance from our original analysis. Perhaps some physicians opioid prescribing for acute care settings are few. Prescribing
prescribe more liquid to account for spillage or spitting out of guidelines exist for treatment of chronic, nonterminal pain includ-
medication. In addition, providers may be aware that adolescents ing those of the CDC, but few for acute pain.28 Examples of recent
are at higher risk of misuse of opioids and therefore prescribe a guidelines for acute pain include those in NYC, Ohio, Oklahoma,
shorter duration to older children.25,26 Arizona, and Washington.15–19 The American College of Emer-
There was an increased likelihood of a patient receiving more gency Physicians released a clinical policy in 2012 addressing pre-
than 5 days of opioids if the prescription was written by a resident scribing of opioids for adults in the ED.29 The policy supports that
physician. A possible explanation for this is that compared with long-acting opioids should not be used for acute pain, and opioid
PEM physicians resident physicians do not receive in-depth train- prescriptions should be written for less than 1 week. Missouri and
ing or education on writing opioid prescriptions to children during Kansas (the setting of this study) currently have no guidelines.
medical school or residency and may be unaware of updates on
the status of prescription opioid abuse, which circulate in EM
Listservs and in the EM literature.27 Also, resident physicians Limitations
can independently prescribe controlled substances in this state. One limitation of our study is the use of a deidentified ad-
Our univariate analysis showed a significant relationship be- ministrative data set, which precluded us from examining other
tween general pediatricians and opioid prescriptions for more than clinical factors likely related to the amount of opioid prescribed
5 days (P = 0.03). However, this was found to not be significant in such as the extent, severity, and nature of the injury/illness, as well
our multivariate logistic regression (OR, 1.1; 95% CI, 0.9–1.5). as specific prescriber information. Another limitation is generaliz-
Although the cause of this is uncertain, we could postulate that ability. This study included 1 hospital system in the Midwest, so
general pediatricians are more commonly prescribing opioids for the findings may not be applicable to other parts of the country.
noninjury diagnoses as they see less injuries than do PEM pro- All of the EDs and UCs studied are part of a children’s hospital
viders, and by controlling for injury versus noninjury diagnosis system, and all are staffed by pediatricians or PEM physicians.
in the multivariate analysis, this was no longer found to be a There are other acute care settings where both adult and pediatric
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
DePhillips et al Pediatric Emergency Care • Volume 00, Number 00, Month 2017
patients are cared for by EM or primary care physicians, and our 13. Office of National Drug Control Policy. Summit on Opioids. Presented
results may not be applicable because they may be more or less likely on June 19, 2014. Available at: www.whitehouse.gov/blog/2014/06/16/
to prescribe prolonged durations of opioids. Finally, in Missouri and join-discussion-summit-opioids-june-19. Accessed August 21, 2014.
Kansas, it is legal for residents to write opioid prescriptions. This may 14. Utah Department of Health. Utah Clinical Guidelines on Prescribing
bias the results toward prolonged duration compared with states Opioids. Salt Lake City, UT: Utah Department of Health; 2008.
where it is not legal for residents to prescribe opioids. Available at: http://health.utah.gov/prescription/pdf/guidelines/final.04.
09opioidGuidlines.pdf. Accessed November 18, 2014.
CONCLUSIONS 15. Hartocollis A. New York City to restrict prescription painkillers in public
Nearly 1 in 5 opioid prescriptions written from our pediatric hospitals’ emergency rooms. New York Times. January 11, 2013:A16.
EDs and UCs were for a prolonged duration, and prolonged pre- 16. Ohio Department of Health. Ohio Emergency and Acute Care Facility
scriptions were more frequently prescribed for younger patients, Opioids and Other Controlled Substances (OOCS) Prescribing Guidelines.
written from the UC or by resident physicians, or written for non- Columbus, OH: Ohio Department of Health; 2012. Available at: http://
injury diagnoses. These findings can help suggest areas where www.healthy.ohio.gov/~/media/HealthyOhio/ASSETS/Files/edguidelines/
prescribing practices can be improved. Guidelines for opioid pre- EGs%20no%20poster.ashx. Accessed November 18, 2014.
scribing should be developed for reference in the pediatric ED and
17. Oklahoma State Department of Health. Oklahoma Emergency Department
UC. Educational interventions including mandatory courses or mod- (ED) and Urgent Care Clinic (UCC) Opioid Prescribing Guidelines.
ules on appropriate opioid prescribing in medical school and resi- Oklahoma City, OK: Oklahoma State Department of Health; 2013.
dency, as well as continuing medical education for all physicians, Available at: http://www.ok.gov/health2/documents/UP_Oklahoma_
should be required. Finally, we need to further evaluate our prescrib- ED-UCC_Guidelines.pdf. Accessed November 18, 2014.
ing practices for common diagnoses. More studies are needed evalu-
18. Arizona Department of Health Services. Arizona Opioid Prescribing
ating the use of opioids at home for specific painful diagnoses to help
Guidelines. Phoenix, AZ: Arizona Department of Health Services; 2014.
us tailor the dispense amount to only what is needed. By improving
Available at: http://www.azdhs.gov/clinicians/documents/
our prescribing practices, we hope to decrease the amount of opioids
clinical-guidelines-recommendations/prescribing-guidelines/
available for nonmedical use and abuse in the home and community. draft-opioid-prescribing-guidelines.pdf. Accessed November 18, 2014.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.