Stimulants and Pediatric Cardiovascular Risk

Brief regulatory history of the stimulant class

Before the recent interest in pediatric stimulant safety emerged in the late 1990s, adult experience with prescribed anorexigenic agents, for example, dextroamphetamine, for weight control in adult women in the 1950s was widespread. History provides pharmacologic class evidence of amphetamine tolerance leading to increased doses and dependency, while lacking evidence of effectiveness for sustained weight loss (Brandon and Smith 1962). In 1971, to reduce inappropriate use of stimulants as “diet pills,” the federal government restricted access by reclassifying stimulants as controlled substances (Schedule II of the Controlled Substances Act), establishing production quotas, and the diet pill industry quickly vanished (Rasmussen 2008). Thus, serious adverse events led to policy and practice change.

Pediatric use of stimulants has had a distinctly different course for methylphenidate, a less potent amphetamine analog, as the main product promoted for attention-deficit/hyperactivity disorder (ADHD) from its introduction in 1955. Forty years later, in 1996, mixed amphetamine salts (marketed as Adderall^©) entered the market and thanks to rapid growth in use in a relatively short time frame, by 2004, mixed amphetamine salts comprised ∼43.9% of the market (Kornfield et al. 2013). In 2005, in the wake of a review of deaths in 20 international reports of individuals who had taken the medication, Adderall XR, regulators in Canada requested market withdrawal (New York Times 2005). At that time, most clinical guidelines implied that stimulant use in youth was by-and-large safe at therapeutic doses and serious CV outcomes were rare and not greater than the occurrence of sudden death in the general population, although analyses often lumped methylphenidate and amphetamine salts together, ignoring the much longer (40 years) experience with methylphenidate than with the mixed amphetamine salt products (9 years). The manufacturer argued that the data were not sufficient to warrant withdrawal from the Canadian market. A few months later, Health Canada withdrew its objections and restored Adderall XR to the market (Food and Drug Administration 2014c).

Subsequently, the Food and Drug Administration (FDA) reviewed U.S. adverse event data in the 5-year period (1999–2003) and identified 12 pediatric cases of sudden unexplained death with prior exposure to mixed amphetamine salt products, 5 with cardiac structural abnormalities, and several cases showing an unexplained accumulation of drugs to toxic levels at therapeutic doses (Food and Drug Administration 2014c).

Fresh concerns and publicity created renewed interest in the CV safety of stimulants, and the FDA held a public hearing asking an expert drug safety advisory committee to review the empirical data summarizing the risks (Food and Drug Administration 2014a). The committee voted for a boxed warning on CV risk with stimulants used to treat ADHD. In addition, a medication guide to assist patients in understanding these risks was recommended (American Academy of Child and Adolescent Psychiatry and American Psychiatric Association 2013). Shortly after, cardiology expert, Steven Nissen, published a perspective detailing the events of that meeting (Nissen 2006), which brought much broader public and professional attention to the issue. In a subsequent pediatric advisory committee meeting (March 2006), the committee did not support a black box warning but argued for a lower level warning, which FDA eventually adopted (Food and Drug Administration 2006). FDA addressed the need for additional study of CV risk and stimulant use by supporting population-based risk assessment studies and these are detailed below. Currently, stimulant labels include a warning on serious CV risks in children and adolescents with structural cardiac abnormalities or other serious heart problems.

CV risk findings from the MTA trial

A recent systematic review of outcomes research on pediatric ADHD concluded that there is a paucity of literature following individuals over time to assure that benefits of treatment over time are clear (Parker et al. 2013). Partly, this conclusion stems from some of the uncertainties that derive from analysis of the long-term findings of the MTA study (The MTA Cooperative Group 1999). Over the past 15 years, the NIMH-funded study of combined type ADHD in 7–9-year olds resulted in published findings of an initial 14-month randomized assessment and naturalistic follow-up outcomes among the 370 available at 3 years (out of the original randomized 547 youth) (Jensen et al. 2007). At year 8, there were 89 youth and at year 10, there were only 18 youth currently on stimulants (Molina et al. 2009; Vitiello et al. 2012). In addition, a “local normative comparison group” of 288 gender- and age-matched children was recruited from the same classrooms as the initial sample. A large body of literature has been published on these extensively analyzed and interpreted findings. Articles have mainly corroborated or debated the strengths, weaknesses, and implications of this large clinical trial, which had the advantage of expert investigators, public funding, and state of the art medication management.

The CV effects of methylphenidate in hyperactive children, for example, increased blood pressure and heart rate, have been known since the 1970s (Ballard et al. 1976). Exploring the mechanism of action of methylphenidate in humans using positron emission tomography scans resulted in correlations of methylphenidate dose with increased dopamine in brain and epinephrine in plasma accounting for clinical cardiac effects (Volkow et al. 2003). Although serious adverse effects of methylphenidate are more often thought to be a consequence of abusive use or in cases of overdose (Klein-Schwartz 2002), CV effects have been observed in youth at therapeutic doses (Stowe et al. 2002).

The question about the extent of CV risk with methylphenidate was taken up in a naturalistic follow-up analyzing CV data from the MTA clinical trial population. Vitiello et al. (2012) conducted a secondary analysis of blood pressure and heart rate across the interval from 14 months to 10 years. Hypertension (outcome) was categorized as normal, prehypertension, stage 1 or stage 2 hypertension. Exposure (methylphenidate) was categorized as never/current/former users for study subjects and matched normative controls. They concluded a lack of association between increased blood pressure and stimulant use. However, the data are difficult to interpret since the number of youth currently on stimulants decreased sharply over the years, and the design did not account for carryover of pharmacological effects due to switching from one treatment group to another treatment group. At 14 months, the heart rate was significantly higher in methylphenidate-treated groups (mean = 84.2 bpm [SD = 12.4] on medication alone and mean = 84.6 bpm [SD = 12.2] on medication plus behavioral therapy) compared with behavioral treatment only (mean = 79.1 bpm [SD = 12.0]) or community care (mean = 78.9 bpm [SD = 12.9]). Cumulative methylphenidate exposure at 3 years (p < 0.019) and 8 years (p < 0.001) (Molina et al. 2009) but not at 10 years was associated with increased heart rate (Vitiello et al. 2012).

Despite considerable research study over a decade, there are notable design limitations in follow-up analyses, namely, diminishing number of subjects at subsequent evaluations, probable differential loss to follow-up and carryover effects. This brief summary of the MTA clinical trial findings on CV safety of methylphenidate treatment of ADHD was presented as a starting point for safety data on short-term (14 months) and 2–10-year follow-up periods. Nevertheless, the limitations of finding rare events (e.g., small sample sizes, high dropout rates) in a clinical trial population suggest that long-term risk assessment of stimulants requires moving beyond clinical trials.

Objectives of the review

This article presents a review of population-based studies on the pediatric CV risk findings of stimulants for the treatment of ADHD. The main objective of this review is to summarize findings and to evaluate the strengths and weaknesses of population-based studies assessing the risk of CV events in relation to stimulant use in youth. In the discussion, we focus on the implications of revised FDA product labeling for possibly new practice standards on clinical monitoring of cardiac status in stimulant users. We also suggest future population-based psychopharmacologic outcomes research to advance drug safety on pediatric mental health treatments in “real-world” clinical settings. The article aims to assist clinicians in understanding empirical safety research as it may inform individual prescriber decisions about monitoring for long-term stimulant safety.

Population-based studies on stimulant use and the risk of rare CV events

Beyond clinical trials, the availability of computerized administrative claims data has provided opportunities to assess person-specific medication exposures summarized in large community-treated populations. Despite their potential limitations, for example, incomplete clinical data, claims data provide a fresh approach with adequate statistical power to enable rare and long-term adverse outcomes to be assessed while avoiding volunteer bias in sample selection. Table 1 summarizes key elements in nine retrospective population-based studies assessing the risk of serious CV events in stimulant-treated youth. Patient cohorts (aged 2–24 years) were derived from administrative claims of U.S. public and private insurance sources and international data from the United Kingdom (McCarthy et al. 2009) and Denmark (Dalsgaard et al. 2014). In addition, deaths were identified from state and federal death registries. Analyses included large retrospective cohort studies and, uniquely, a “classic case–control” study (Gould et al. 2009).

In the Gould et al. study, the frequency of stimulant exposure before sudden death among 7–19-year olds was compared with the frequency of stimulant exposure among age-matched youths who were passengers in fatal car accidents. Stimulant use was assessed by informants or medical examiner reports, toxicology, or death certificates. The authors observed greater odds (six- to sevenfold) of stimulant use in cases diagnosed with sudden unexplained death (n = 10/564, 1.8%) than among matched controls, that is, motor vehicle passenger fatalities (n = 2/564, 0.44%).

Several studies analyzing administrative claims from large retrospective youth cohorts chose sudden unexplained death as the primary outcome or a composite outcome of serious CV events, that is, cardiac deaths, acute myocardial infarction, or stroke, and found no association with current stimulant exposure (Winterstein et al. 2007, 2009, 2012; McCarthy et al. 2009; Cooper et al. 2011; Schelleman et al. 2011). In a UK practice research database, McCarthy et al. (2009) found no association of sudden death with ADHD drug exposures (methylphenidate, dextro-amphetamine or atomoxetine). Notably, across studies, the number of stimulant-associated deaths was extremely low (Winterstein et al. 2007; Gould et al. 2009). Schelleman et al. analyzed public (Medicaid) and private claims (HealthCore) and found no increased risk of sudden death. However, death was ascertained from hospitalization or emergency department (ED) visit data, a less comprehensive source than death certificates (Schelleman et al. 2011).

In contrast to the other study data sources described in Table 1, Denmark maintains national health registries for comprehensive tracking of exposure and outcome, which permitted analysis of a birth cohort (1990–1997) in those with and without stimulant exposure (Dalsgaard et al. 2014). Stimulant exposure was typically measured from the first available dispensing date (or first available medical visit with an ADHD diagnosis). There was a >80% increased risk of inpatient/outpatient or ED cardiac events among stimulant-exposed than nonstimulant-exposed youth. A secondary analysis of stimulant use according to dosage exposure was achieved by looking backward for stimulant use from the CV event, and at 3 and 12 months before the event. Analysis of these time windows led to the authors' hypothesis that early adverse experience could account for dose reductions or discontinuation and supports the need to consider dose in various exposure windows before assessing dose-related adverse event risk. The main implication of this analysis concerns the design of population-based studies, in which change in dose patterns over time can influence study outcomes by retrospectively selecting the most recent dose as representative of the patient exposure.

In the 2007 study of Medicaid-insured youth, Winterstein et al. (2007) assessed the risk of a cardiac ED visit among youth newly diagnosed with ADHD and found a 20% increased risk of cardiac-related ED visits for current stimulant users compared with nonusers. In a subsequent study, the authors assessed the risk of a cardiac ED visit among new users of methylphenidate compared with new users of mixed amphetamine salts, finding no difference. However, among all stimulant new users, concurrent use of an antidepressant or atypical antipsychotic increased the risk of a cardiac ED visit by 67% and 90%, respectively (Winterstein et al. 2009). Olfson et al. examined privately insured youth for inpatient or ED visits for cardiac events and for cardiac-related symptoms such as syncope. Neither endpoint was significant (Olfson et al. 2012).

These epidemiologic approaches to population-based risk are a welcome advance beyond clinical trials, case reports, and case series. Specifically, they expand methodology beyond the FDA Adverse Event Reporting System (FAERS), which is a voluntary passive surveillance system lacking data to assess the incidence of adverse events (Food and Drug Administration 2014b).

Limitations

A number of limitations of U.S. administrative data sources should be acknowledged: (1) limited infrastructure (e.g., registries) to monitor U.S. residents prospectively regardless of health insurer; (2) limited information on the accuracy of exposure in terms of consumed daily dosage. Overrepresentation of the actual consumed medication can bias toward the null. In addition, duration of use and concomitant psychotropic use subject the study to potential misclassification bias (Schneeweiss 2014). Most of the population-based studies of youth presented in Table 1 did not consider the effect of long-term cumulative exposure to stimulants on the risk of CV events. Population-based studies in youth are needed, which account for lifetime cumulative exposure to these agents; (3) paucity of information on the accuracy of exposure and reasons for discontinuation, which could contribute information on “real-world” clinical tolerability.

Clinical monitoring of stimulants for safety in community-treated youths

The article by Vetter et al. (2008) presented the initial American Heart Association's position in support of baseline electrocardiography (ECG) monitoring and introduced a lively debate on whether ECGs should be required at baseline to establish a cardiac profile before stimulant initiation (Perrin et al. 2008). Previously, in 2006, FDA mandated revisions to product label information and required manufacturers to develop medication guides intended to communicate benefit and risk to families and other caretakers. Specifically, stimulant product labels were revised in the “Warnings and Precautions” section. For example, the product labeling for Concerta^© (branded methylphenidate extended release), now includes statements on 10 serious adverse events with CV events leading the list. A clear directive is given to monitor patients for changes in heart rate and blood pressure (Shire 2013). Implementation of such monitoring, particularly in psychiatric office settings where physical assessment has not been traditional, is not known (Bange et al. 2014). Clinicians' reliance on parents to assess and monitor such subtle changes in health status may not be adequate as CV changes may be gradual, modest, and cumulative (Vitiello et al. 2012) with long-term impact essentially unknown.

Also challenging in terms of service delivery is the integration of primary care and psychiatry, which could help to assure physical health monitoring before initiating medication and subsequently as a safety priority. However, national survey data on pediatricians' use of baseline ECG are not encouraging. It appears that many are uncomfortable without expert interpretation of ECGs and, furthermore, some are reluctant to discuss cardiac risk with families for fear of losing their acceptance of and adherence to stimulant treatment (Leslie et al. 2012). Nevertheless, encouraging signs of uptake of ECG screening have been observed (Chen et al. 2015). Some argue that the absence of cost–effectiveness of baseline ECG for stimulant users is a barrier to changing the practice. By contrast, in the area of drug safety for athletes, the cost–effectiveness of screening ECG has been demonstrated (Wheeler et al. 2010). Apparently, the slowly evolving change in monitoring practices for ADHD in youth will influence those clinicians who are concerned about malpractice or who have sufficient concern about the uncertainty of risk based on the current evidence. More certain is the reality that the increasingly detailed risk information on product labels leaves practicing clinicians more accountable than in the past when labels were silent on the risk of such adverse events. With the promise of more comprehensive electronic medical record keeping, integrated collaborative monitoring by primary care and child psychiatry may be feasible.

In the latest (2007) ADHD practice parameter of the American Academy of Child and Adolescent Psychiatry, there is a brief discussion of the 2006 FDA communications related to the risk of sudden death in stimulant-treated youth compared with a general population estimate, emphasizing the rarity and absence of a difference for medicated and nonmedicated populations (Pliszka and AACAP Work Group on Quality Issues 2007). The guideline did not endorse routine cardiac evaluation unless preexisting heart disease or symptoms suggest significant CV disease. The recently created AACAP Medication Guide (American Academy of Child and Adolescent Psychiatry and American Psychiatric Association 2013) discusses common adverse events (side effects), for example, appetite and weight loss. It also provides information on serious, rare events, for example, CV effects, including deaths. The discussion continues with a recommendation to parents to review the family's cardiac history and to alert the doctor if it is suggestive of a risk. In the AACAP practice parameter, undiagnosed heart defects are noted but questioned as a risk factor, in contrast to the American Academy of Pediatrics' clear warning of a risk for stimulant use in youth with previously undetected cardiac abnormalities (Perrin et al. 2008).

In terms of improving drug safety outcomes, there is much to be done and creating an agenda for research on drug safety in community populations tops the list. Drug safety is too important to leave to the occasional class action lawsuit. In 2008, in a review of off-label use and monitoring of psychotropic drugs, we advocated for a comprehensive assessment of health status (excepting emergency, short-term use) before introducing a psychotropic medication. Baseline health status measures should include physical measures such as pulse, respiration rate, and blood pressure, growth over time using standardized growth charts, including height, weight, and body mass index, and a laboratory panel, including complete blood count, urinalysis, blood urea nitrogen, serum electrolytes, and liver function tests (Vitiello 2008; Zito et al. 2008). An ECG before initiating a stimulant, although not endorsed by the American Academy of Pediatrics and the American Academy of Child and Adolescent Psychiatry (except in youth with high-risk conditions), may be most useful in anticipation of the possibility of complex drug combinations that would collectively increase the burden on the CV system, for example, when a regimen of stimulant, antidepressant, and antipsychotic are used together. Such combinations were associated with increased cardiac-related ED visits (Winterstein et al. 2009).

In addition to baseline screening for CV risk, stimulant use alone, or in combination with other potent psychotropic medications such as antipsychotics and antidepressants, suggests assessment of cardiac status at regular intervals. This may be warranted particularly for those identified at baseline as high-risk patients and may result in a collaborative monitoring by psychiatry with CV experts.

Future directions for psychopharmacologic safety outcomes research in U.S. youth

In the strategic plan of NIMH, strategy 3.2 calls for expanding and deepening a personalized approach, that is, precision medicine, to mental health intervention research. In this strategy, interventions should emphasize outcomes research in terms of measuring functioning, presence of side effects, and adherence to treatment in an effort to assess long-term benefit/risk (National Institute of Mental Health 2014). This objective is consistent with the review in this article since the MTA findings support further clarification of individualized, that is, “personalized” stimulant benefits and risks, reflecting “real-world” conditions (e.g., socioeconomic, family, and other caregivers, and residential and academic environments) in the decision to initiate medication and monitor medication use and outcomes over time.

In addition to new product development with costly clinical trials that do not shed light on long-term effectiveness or safety, the child mental health research agenda could embrace robust outcomes research in community-treated youth in both pediatric and psychiatric settings. With electronic medical records becoming widely available, there is a potential for large community cohorts in longitudinal analytic models to be followed over several years (Castillo et al. 2015). With further development of infrastructure and epidemiologic methods in the United States, new approaches can address some of the weaknesses of past studies. Novel approaches, for example, large simple (pragmatic) trials, may provide knowledge of outcomes among populations typically seen in usual practice settings in contrast to the more restricted clinical trial population (Schneeweiss 2014; Vitiello 2015).

Vitiello summarized the state of the art of psychopharmacology in 2007, calling for “proactive, coordinated, multipronged approaches to study the safety of psychotropic use in childhood, with special attention to distal outcomes, and to serious, though rare, adverse effects” (Vitiello 2007). The present review aimed to show that safety outcomes research is an evolving process that must move beyond initial clinical trials and perhaps beyond retrospective analysis of administrative claims into prospective community cohort studies, including large simple trials. Renewed clinical and research attention to the CV risk question seems appropriate, particularly for mixed amphetamine salt products, which is the other major stimulant product widely prescribed. Amphetamines have greater potency than methylphenidate and despite shorter time on the market have raised early safety signals (Food and Drug Administration 2014c). Cohort studies could be facilitated by electronic medical record protocol-driven evaluations of risks and tolerability of medications in the postmarketing phase of clinical management in community populations to calibrate benefit–risk assessment. In that way, postmarketing surveillance could truly become “active” surveillance.