Pediatric ADHD Medication Errors Reported to United States Poison Centers, 2000 to 2021

This study retrospectively analyzed data from the NPDS, which is maintained by America’s Poison Centers (formerly the American Association of Poison Control Centers) and includes information from calls made to US PCs.^14,15  Data were also obtained from the US Census Bureau to calculate population-based exposure rates. This study was determined to be exempt from approval by the institutional review board at the authors’ institution.

This study was limited to exposures designated by the NPDS as “unintentional therapeutic errors” involving ADHD medications, which included amphetamines and related compounds, methylphenidate, clonidine, guanfacine, modafinil, and atomoxetine. Data were requested using product-specific codes and generic substance category codes; however, exposures involving modafinil and atomoxetine were unable to be separated from each other in the data provided by America’s Poison Centers, so data for these 2 medications were combined during analyses. Only reports from PCs in the 50 states and District of Columbia were included, which allowed calculation of population-based rates. Exposures involving illicit stimulants (like cocaine) or other stimulants (like caffeine), which are not prescribed for ADHD, were excluded. Exposures categorized as “confirmed nonexposure” or “unrelated exposure” (n = 1441) or therapeutic errors that occurred in a health care facility (HCF) (n = 218) were also excluded.

Individuals were grouped by age for analysis (<6 years, 6–12 years, and 13–19 years). Exposure site was categorized as residence (own or other), school, other (public area, workplace, or restaurant), or unknown. Type of exposure was categorized as multiple-substance or single-substance. ADHD medications were grouped into (1) amphetamines and related compounds, (2) methylphenidate, (3) guanfacine, (4) clonidine, and (5) modafinil or atomoxetine.¹⁶  There were NPDS coding changes before 2004 that affected the modafinil or atomoxetine medication category, so trend analyses for that category started in 2004.

NPDS categorized medical outcomes as (1) no effect, (2) minor effect (minimal symptoms that generally resolve rapidly), (3) moderate effect (more pronounced, prolonged, or systemic than minor effect), (4) major effect (symptoms are life-threatening or result in significant disability or disfigurement), (5) death, (6) not followed (minimal clinical effects possible), and (7) unable to follow (judged as potentially toxic exposure).¹⁷  During analyses, “unable to follow (judged as a potentially toxic exposure)” was considered as unknown. Moderate effect, major effect, and death were grouped as serious medical outcomes during analyses. Level of health care received was defined by the NPDS as (1) no HCF treatment received, (2) treated or evaluated and released, (3) admitted to a critical care unit (CCU), (4) admitted to a non-CCU, (5) admitted to a psychiatric facility, (6) patient refused referral or did not arrive at HCF, and (7) patient lost to follow-up, left against medical advice, or unknown.¹⁸  The “no HCF treatment received” category was calculated as the sum of the “managed on-site (not in a HCF)” and “other” categories of the NPDS management site variable. If the management site was “unknown,” the case was included in the “lost to follow-up, left against medical advice, or unknown” category of the level of health care received variable. During analyses, lost to follow-up, left against medical advice, or unknown cases were considered as unknown. Admitted to a CCU, admitted to a non-CCU, and admitted to a psychiatric facility were grouped as “admitted” during analyses.

Data analyses were conducted using IBM SPSS 28.0 (IBM Corporation, Armonk, NY) and SAS 9.4 (SAS Institute, Inc. Cary, NC). National exposure rates were calculated using US census estimates from 2000 through 2021. Except for analyses of trends over time, all analyses of ADHD medication-related therapeutic errors in this study were limited to exposures in which the medication was the first-ranked substance. A first-ranked substance is the substance judged by a Certified Specialist in Poison Information at the PC to be most likely responsible for the observed clinical effects. First-ranked ADHD medication exposures include all single-substance exposures involving an ADHD medication as well as multiple-substance exposures with an ADHD medication as the first-ranked substance. During analyses of trends over time, all ADHD medication-related therapeutic errors (regardless of whether the ADHD medication was first-ranked or nonfirst-ranked) were included. Statistical significance of trends over time was evaluated using simple or piecewise linear regression, as appropriate, with a significance level of α = .05. Break points for piecewise regression models were based on scatter plots of annual exposure rates. Odds ratios (ORs) were calculated with 95% confidence intervals (CIs).

From 2000 through 2021, 87 691 therapeutic error exposures involving ADHD medications as the first-ranked substance among youth <20 years old were reported to US PCs, yielding an average of 3985 exposures annually (Table 1). There were 5235 therapeutic errors reported in 2021, averaging 1 exposure every 1.6 hours (data not shown in table). The 6 to 12-year-old age group accounted for two-thirds (66.6%) of exposures, and approximately three-fourths (76.4%) of exposures were to males. Most (92.9%) exposures occurred in residences, whereas 5.5% occurred at school and 1.6% occurred in another location (public area, workplace, or restaurant) (data not shown in table). Most (82.7%) exposures did not receive treatment in a HCF, however 2.3% were associated with HCF admission, including 0.8% to a CCU. In addition, 4.2% (n = 3561) of exposures were associated with a serious medical outcome; however, most of these were moderate effect cases (3479, 4.1%) and only 82 (0.1%) major effect cases and no deaths were reported (Table 1). Children <6 years old were more likely to experience a serious medical outcome (OR = 2.1; 95% CI: 1.9–2.3) or be admitted to a HCF (OR = 3.4; 95% CI: 3.0–3.7), including admission to a CCU (OR = 3.7; 95% CI: 3.1–4.4), than 6 to 19-year-olds.

Most (79.7%) therapeutic error-related exposures involving ADHD medications as the first-ranked substance were single-substance exposures. Among these single-substance exposures, 4.1% were associated with a serious medical outcome and 2.1% were associated with a HCF admission. Similarly, 4.1% of multiple-substance exposures were associated with a serious medical outcome and 2.8% a HCF admission (data not shown in table). The percentage of individuals with single-substance exposures decreased with increasing age group, whereas the opposite was true for individuals with multiple-substance exposures (Table 1). Statistically, single-substance and multiple-substance exposures were equally likely to be associated with a serious medical outcome (OR = 1.0; 95% CI: 0.9–1.1), whereas multiple-substance exposures were more likely to be associated with HCF admission (OR = 1.4; 95% CI: 1.2–1.5), including admission to a CCU (OR = 1.3; 95% CI: 1.1–1.6), than single-substance exposures.

Amphetamines and related compounds accounted for approximately half (50.5%) of therapeutic errors involving ADHD medications as the first-ranked substance, followed by guanfacine (23.1%) and methylphenidate (14.7%) (Table 2). Amphetamines and related compounds were the most common ADHD medication category among both single- and multiple-substance exposures (53.9% and 37.2%, respectively), followed by guanfacine (21.8% and 28.4%, respectively). All (n = 763) first-ranked clonidine exposures were multiple-substance exposures (data not shown in table). Therapeutic errors involving guanfacine as the first-ranked substance were associated with 42.7% of serious medical outcomes, whereas therapeutic errors involving amphetamine and related compounds were associated with 42.4% (Table 2). Therapeutic errors involving guanfacine were more likely to be associated with a serious medical outcome (OR = 2.3, 95% CI: 2.1–2.5) or admission to a HCF (OR = 5.4, 95% CI: 4.8–5.9) compared with those involving amphetamine and related compounds.

Among therapeutic errors involving ADHD medications as the first-ranked substance, the most common scenarios were “inadvertently taken or given medication twice” (53.9%), followed by “inadvertently taken or given someone else’s medication” (13.4%) and “wrong medication taken or given” (12.9%) (Table 3).

Trend analyses included all 124 383 therapeutic error-related exposures involving ADHD medications (ie, analyses were not limited to first-ranked substances). There were 1906 of these exposures reported to US PCs among youth <20 years old in 2000, which increased by 299.0% to 7603 exposures in 2021. The rate of exposure per 100 000 US population increased unevenly over the study period, increasing significantly from 0.68 in 2000 to 1.51 in 2004 (P = .0011), then decreasing nonsignificantly to 1.23 in 2007 (P = .6517), followed by a significant increase to 2.54 in 2019 (P < .0001), and then significantly decreasing to 2.29 in 2021 (P = .0105) (Fig 1). The overall trend pattern was driven by males and 6 to 12-year-olds (Figs 1 and 2).

Among youth <20 years old, the rates per 100 000 US population of serious medical outcome or admission to a HCF associated with therapeutic errors involving ADHD medications increased significantly from 2000 to 2009 (P = .0006 and P = .0016, respectively), followed by a more rapid significant increase from 2009 to 2013 (P < .0001 and P < .0001, respectively), and then a nonsignificant change from 2013 to 2021 (P = .3640 and P = .9669, respectively) (Fig 3). The overall trend pattern was driven by amphetamine and related compounds until 2009 and by guanfacine after 2009 (Figs 4 and 5), primarily among 6 to 12-year-olds, who experienced the highest rates of serious medical outcomes and HCF admission (data not shown).

The rate of therapeutic errors associated with amphetamines and related compounds per 100 000 US population increased significantly from 0.39 in 2000 to 0.55 in 2007 (P = .0002), followed by a more rapid significant increase to 0.91 in 2012 (P < .0001), then a significant decrease to 0.66 in 2021 (P < .0001). Therapeutic errors involving guanfacine increased significantly from 0.07 in 2000 to 0.16 in 2009 (P = .0008), followed by a rapid significant increase to 0.58 in 2012 (P < .0001), and then a less rapid significant increase to 0.62 in 2021 (P = .0328). Therapeutic errors involving methylphenidate increased significantly from 0.08 in 2000 to 0.10 in 2010 (P = .0210), followed by a more rapid significant increase to 0.38 in 2021 (P < .0001). Therapeutic errors involving the modafinil or atomoxetine category decreased significantly from 0.47 in 2004 to 0.12 in 2021 (P < .0001) (Fig 6).

The frequency of first-ranked, out-of-hospital ADHD medication-related therapeutic errors among youth <20 years old reported to US PCs increased by almost 300% during the 22-year study period to 7609 exposures in 2021. This increase may be attributable, in part, to an increased diagnosis of ADHD and an associated increased use of ADHD medications among US children. Previous studies found a 42% and 83% increase in the diagnosis of ADHD from 2003 to 2011 and 2001 to 2010, respectively.^5,18,19  Another study using data from the National Health Interview Survey reported a steady increase in the estimated prevalence of ADHD diagnosed among US children and adolescents from 1997 to 2016, which was corroborated by a report from the Centers of Disease Control and Prevention using the same data source along with data from the National Survey of Children’s Health.^20,21  Additionally, the Diagnostic and Statistical Manual of Mental Disorders was updated to the fifth edition in 2013, leading to changes in the diagnostic criteria of ADHD, which may also have contributed to an increase in ADHD diagnoses.^5,22–24  Regardless of the cause of the increased diagnosis of ADHD among US children and adolescents, the increase in ADHD medication errors parallels the increase in ADHD diagnoses. We are unaware of data showing that health care providers became more likely to prescribe medications to a child with the diagnosis of ADHD during the past 2 decades. Therefore, we are unable to comment on whether changes in prescribing practices may have contributed to the observed increase in ADHD medication errors.

As illustrated in Fig 1, there was a statistically significant decrease in the rate of therapeutic errors from 2019 to 2021, which temporally corresponds with the coronavirus disease 2019 pandemic. This may have been partly attributable to a decrease in use of ADHD medications during the pandemic. One study demonstrated that almost half of students did not continue their pre-existing ADHD medications during the pandemic.²⁵  Another study reported a 10.8% decrease in pediatric ADHD medication prescription rates during the pandemic.²⁶  Prior studies have identified a seasonal pattern in ADHD medication prescribing, associated with medication use only during the school year.^27,28  Because many students did not attend school in-person at the beginning of the pandemic, this may have contributed to a decrease in ADHD medication use and associated therapeutic errors.

Approximately three-fourths of reported therapeutic errors involving ADHD medications as the first-ranked substance occurred among males. This is likely attributable to ADHD being more commonly diagnosed and treated among males, with estimates of the male-to-female ratio ranging from 2:1 to 10:1.^15,29–33  This ratio varies by age group and narrows among adults.^{13,15,28,34,35} 

Stimulant medications are a first-line treatment of ADHD because of their ability to reduce ADHD core symptoms.^36–39  Amphetamine and related compounds were associated with half of the therapeutic errors involving ADHD medications as the first-ranked substance, and methylphenidate accounted for 15%. In addition, amphetamine and related compounds had the highest rate of therapeutic errors throughout the study period. These findings are consistent with those of other studies reporting an increase in the production and use of prescription stimulants in the United States during the study period.^5,13  An expansion in stimulant formulation options also occurred during this time, including long-acting formulations and novel delivery systems.^37,40 

Nonstimulant medications, such as atomoxetine, clonidine, and guanfacine, have also proven efficacious in treating ADHD, however their effect sizes in ADHD symptom resolution are smaller compared with stimulant medications. As such, they are generally recommended as a second-line treatment option.³⁷  Atomoxetine was approved by the US Food and Drug Administration (FDA) for pediatric and adult ADHD treatment in 2002.⁴¹  ADHD frequently co-occurs with other conditions, including tic disorder or Tourette syndrome,^4,41,42  and treatment of ADHD with comorbid Tourette syndrome poses a challenge because stimulant use can precipitate or exacerbate tics.^41,43,44  This added to the anticipation of atomoxetine as a nonstimulant ADHD treatment option for patients with comorbid Tourette syndrome. However, in 2004 several case reports indicated atomoxetine may be associated with the onset of tics,^41,45,46  although subsequent studies found no association with tic exacerbation.^41,42  Additionally, a black box warning was issued by the FDA in 2005 after atomoxetine was found to be associated with increased suicidal ideation.^37,47,48  These events were likely related to the peak in the atomoxetine prescription rate in 2004 with a steady decline thereafter.⁴⁹  This is also consistent with the observed decrease in the rate of therapeutic errors associated with atomoxetine and modafinil combined in this study after 2004.

Clonidine is an antihypertensive medication that was occasionally used “off-label” for ADHD before its FDA approval for this indication in 2010.⁵⁰  FDA approval likely led to an increase in clonidine prescriptions for ADHD, which may explain the statistically significant increase in clonidine-related therapeutic errors from 2009 to 2012. In addition, all first-ranked clonidine-related therapeutic errors were multiple-substance exposures in this study. This supports a previous report of clonidine utilization patterns from 2003 to 2008, which found that among children prescribed clonidine for ADHD, 95% were also prescribed additional stimulant or nonstimulant medications.⁵⁰ 

Like clonidine, guanfacine is a centrally acting alpha2-adrenergic receptor agonist initially approved for treatment of hypertension among adults. It was first described as an off-label treatment of ADHD in 1995⁵¹  and was subsequently approved for this indication in 2009.^52,53  The marked increase in guanfacine-related therapeutic errors observed in our study beginning in 2009 coincides with its FDA approval for ADHD treatment. In addition, the increases in serious medical outcomes and HCF admissions associated with ADHD medication-related therapeutic errors starting in 2009 were driven by guanfacine-related events. Therapeutic errors involving guanfacine were more likely to be clinically significant; guanfacine was more than twice as likely to be associated with a serious medical outcome and more than 5 times as likely to be associated with admission to a HCF than amphetamine and related compounds. The higher admission rate may be, in part, attributable to the need for clinical observation because of guanfacine’s extended-release formulation combined with a narrow therapeutic window between a therapeutic and toxic dose. Although case reports of guanfacine-related overdose deaths have been reported,^52,54  there were no deaths associated with therapeutic errors involving this or any of the ADHD medications in this study.

In this study, most therapeutic errors were attributable to “inadvertently took or given medication twice,” followed by “inadvertently took or given someone else’s medication.” Strategies to decrease inadvertently taking or giving a medication twice or taking or giving someone else’s medication include patient and caregiver education about maintaining a record of when medications are taken or the use of child-resistant pill organizers.¹¹  Another strategy may be a transition from pill bottles to unit-dose packaging, like blister packs, which may aid in remembering whether a medication has already been taken or given.^7,55  Development of improved child-resistant medication dispensing and tracking systems may help prevent these types of therapeutic errors as well as others, such as wrong medication taken or given or medication doses taken or given too close together. New interventions should be evaluated for effectiveness in reducing the incidence of therapeutic errors.

Although prevention efforts should focus on the home setting (where approximately 93% of exposures occurred), additional attention should be given to schools and other settings where children and adolescents spend time and receive medication. Continued efforts to increase awareness among health professionals about prevention of these therapeutic errors also seems warranted, given that there were 1714 iatrogenic cases (examples may include pharmacist or school nurse dispensing errors).

This study has several limitations. It underestimates the frequency of pediatric ADHD medication-related therapeutic errors because not all events are reported to PCs. Reporting bias may occur, such as more serious exposures being more likely to be reported. Miscategorization may occur when determining which substance should be first-ranked in multiple-substance exposures. The medications included in this study are used to treat medical conditions other than ADHD; however, this is more common among adults. NPDS data are based on self-reports from callers and cannot be completely verified by the PC or America’s Poison Centers. Additionally, not every exposure represents an overdose or poisoning.

The frequency of cases reported to PCs of pediatric out-of-hospital therapeutic errors related to ADHD medications increased by almost 300% during the 22-year study period and is likely attributable to increased prescribing of these medications. Although most exposures were associated with no to minimal consequences, more than 2% were associated with admission to a HCF and 4.2% had a serious medical outcome. Because therapeutic errors are preventable, more attention should be given to patient and caregiver education and development of improved child-resistant medication dispensing and tracking systems.

CCU

critical care unit

FDA

US Food and Drug Administration

HCF

health care facility

NPDS

National Poison Data System

PC

poison center