In this issue of Pediatrics, Qato et al1 present a detailed analysis of the prevalence of prescription drug use by children in the United States based on the NHANES. With their results, they point out not only the substantial number of children who are on a prescription drug at any given time (between 1 in 7 and 1 in 5 depending on age and sex) but also the number who are concurrently on >1 prescription drug and are potentially at risk for drug–drug interactions (DDIs). They conclude, reasonably, that preventive efforts to improve the safe use of medications by children should promote public and health professional awareness of the increased risks associated with the concurrent use of prescription and over-the-counter medications.1 

However, raising professional awareness is unlikely to be enough. In fact, professional education results in only small changes in clinical practice.2 Moreover, as a general pediatrician, I rarely (if ever) prescribe many of the medications involved in the DDI examples the authors discovered, although my patients may well take them. This makes the task of ensuring that the antibiotic I want to prescribe is safe in the presence of my patients’ atypical antipsychotic medications particularly daunting. The need for an automated process, perhaps embedded in an electronic health record (EHR), to alert the prescriber to dangerous DDIs before a prescription is written comes to mind.

Such software exists, of course. The authors used 1 such product (Micromedex; Truven Health Analytics, Ann Arbor, MI) to identify the potential for serious DDIs among the children in their study. Such software alerts have long been known to reduce medication errors, such as concurrent prescriptions that might cause DDIs.3 Yet, effectively alerting clinicians to potential DDIs is not as straightforward as it might seem. First, children often seek health care from multiple providers. The psychiatrist who prescribed an antipsychotic may well use a different EHR than the primary care pediatrician prescribing an antibiotic. To provide truly reliable alerts to potentially dangerous DDIs, EHRs must exchange data with other EHRs, a so-called health information exchange. Although more and more EHRs are able to exchange data with other EHRs, it is still a minority of office-based physicians who use a health information exchange.4 As a result, an EHR may fail to alert the prescribing physician to a problem. Of course, medication reconciliation is a quality improvement strategy intended to fix this, but medication reconciliation has had limited effectiveness.5 

A second significant barrier works in the opposite direction: overalerting the physician. The prevalence of children with the potential for serious DDIs is only ∼0.6%, according to Qato et al,1 and these are only potential DDIs. The rate at which actual clinical consequences occurred was, as the authors point out, beyond the scope of their study. However, the rate at which clinical consequences actually happen is important because an EHR that interrupts the clinician too frequently with false alarms is soon ignored. This phenomenon, known as alert fatigue, is well known in the informatics world and can irritate clinicians and render clinical decision support useless.6 

The dilemma is that providing too few alerts will result in an occasional DDI; too many and the physicians begin to ignore all alerts. The solution is to select alerts that strike the right balance between false-positive alerts, which annoy clinicians and lead to alert fatigue, and false-negative alerts, which represent a missed opportunity to prevent a bad outcome. One approach, known as the threshold approach to clinical decision-making, has existed for nearly 40 years.7 It uses decision analytic techniques to identify the risk threshold at which an action (showing an alert) should be taken. It requires knowledge of the false-positive and false-negative rates and the costs of false-positives and false-negatives. In this case, costs can be financial or clinical (for example, missing a rare serious DDI versus the risk of “training” clinicians to ignore alerts).

Information technology offers a promising strategy to avoid potentially serious DDIs, but the approach must be thoughtful, weighing the real risks and benefits of interrupting the clinician with each possible DDI. Understanding these risks and benefits will be an important guide for extending this important work by Qato et al1 so that we know how often potential DDIs lead to clinical consequences and how serious those consequences are.

     
  • DDI

    drug–drug interaction

  •  
  • EHR

    electronic health record

Opinions expressed in these commentaries are those of the author and not necessarily those of the American Academy of Pediatrics or its Committees.

FUNDING: No external funding.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2018-1042.

1
Qato
DM
,
Alexander
GC
,
Guadamuz
JS
, et al
.
Prescription medications among children and adolescents in the United States.
Pediatrics
.
2018
;
142
(
3
):
e20181042
2
Forsetlund
L
,
Bjørndal
A
,
Rashidian
A
, et al
.
Continuing education meetings and workshops: effects on professional practice and health care outcomes.
Cochrane Database Syst Rev
.
2009
;(
2
):
CD003030
[PubMed]
3
Kaushal
R
,
Shojania
KG
,
Bates
DW
.
Effects of computerized physician order entry and clinical decision support systems on medication safety: a systematic review.
Arch Intern Med
.
2003
;
163
(
12
):
1409
1416
[PubMed]
4
Hersh
W
,
Totten
A
,
Eden
K
, et al
. Health Information Exchange: Evidence Reports/ Technology Assessments, No. 220. Rockville, MD: Agency for Healthcare Research and Quality. Available at: https://www.ncbi.nlm.nih.gov/books/NBK343580/. Accessed June 26, 2018
5
Bayoumi
I
,
Howard
M
,
Holbrook
AM
,
Schabort
I
.
Interventions to improve medication reconciliation in primary care.
Ann Pharmacother
.
2009
;
43
(
10
):
1667
1675
[PubMed]
6
Wright
A
,
Aaron
S
,
Seger
DL
,
Samal
L
,
Schiff
GD
,
Bates
DW
.
Reduced effectiveness of interruptive drug-drug interaction alerts after conversion to a commercial electronic health record [published online ahead of print May 15, 2018].
J Gen Intern Med
. doi:
[PubMed]
7
Pauker
SG
,
Kassirer
JP
.
The threshold approach to clinical decision making.
N Engl J Med
.
1980
;
302
(
20
):
1109
1117
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The author has indicated he has no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The author has indicated he has no financial relationships relevant to this article to disclose.