Not Too Rare to Matter: The Incidence of Neuroleptic Malignant Syndrome in Children and Adolescents Treated with Antipsychotics

Neuroleptic malignant syndrome (NMS) was initially described by Jean Delay and colleagues in 1960 as a severe adverse reaction to antipsychotic drugs characterized by hyperthermia and severe muscular rigidity (Delay et al., 1960). Although additional clinical features of NMS have been since described (Caroff & Mann, 1993), diagnostic criteria generally converge on four core symptoms: altered mental status, muscular rigidity, hyperthermia, and autonomic dysfunction (Pileggi & Cook, 2016; Strawn et al., 2007). Exposures to first-generation antipsychotics, high doses, injectable formulations, and rapid dose increases are commonly cited risk factors; however, NMS can occur with any antipsychotic drug, at any dose, and with oral and injectable forms (Pileggi & Cook, 2016; Strawn et al., 2007). The estimated incidence of NMS based on meta-analytic evidence is approximately 0.99 cases per 1000 treated patients (Gurrera et al., 2007), although even lower incidences have been reported more recently (Murri et al., 2015, Kane et al., 2019; Nielsen et al., 2012; Stubner et al., 2004)—perhaps reflecting greater representation of outpatients among populations at risk and the supplantation of first-generation antipsychotics by atypical antipsychotics. NMS is a potentially deadly neuropsychiatric emergency, with recent mortality estimates as high as 5.6% (Pileggi & Cook, 2016). Early recognition and treatment prevent the onset of severe complications and facilitate a full recovery for most patients (Strawn et al., 2007).

As compared with adults, far less is known about the clinical presentation, course, and risk markers of NMS in children and adolescents. Comprehensive reviews of mainly case literature suggest that the clinical presentation and course of NMS in children and adolescents may be similar to that of adults, including the risk of severe complications and potential lethality (Neuhut et al., 2009; Silva et al., 1999). Despite these propensities, NMS has rarely been considered within the broader debate regarding the benefits relative to risks of antipsychotics in children and youth with nonpsychotic indications (Daviss et al., 2016; Schneider et al., 2014). This may be attributed, in part, to the unknown (but presumably very low) incidence of NMS in youth treated with antipsychotics—much lower than for significant weight gain and associated dysmetabolic effects. Both are common complications of antipsychotic treatment in young people that may persist into adulthood (Libowitz & Nurmi, 2021), making them a more logical counterweight to the proportionately significant clinical need for antipsychotics in managing behavioral emergencies or serious neuropsychiatric symptoms over the longer term (Gerson et al., 2019; Zuddas et al., 2011). In such cases, and in the absence of an effective alternative, treatment with antipsychotics may be a matter of clinical necessity.

Still, several factors specific to children and adolescents raise the possibility of increased risk of NMS. Relative to adults, children, and adolescents have a higher susceptibility to the neuromotor side-effects of both first-generation and atypical antipsychotic drugs (Correll et al., 2006). In addition, there have been well-documented increases over the past 3 decades in the prescribing of antipsychotic drugs to children and adolescents (Halfdanarson et al., 2017; Olfson et al., 2006, 2012; Radojcic et al., 2023; Steinhausen & Bisgaard, 2014). While recent trends suggest that antipsychotic prescribing among youth is declining (Bushnell et al., 2021; Zito et al., 2018), it may still be relatively high within certain patient subgroups including those with neurodevelopmental disorders (Dosreis et al., 2011; Park et al., 2016; Saldana et al., 2014). At present, it is unclear to what degree the frequent prescribing of antipsychotics and other susceptibility factors have influenced the risk or outcomes of NMS in children and youth.

These points highlight the need for high-quality, population-level studies of NMS occurrence, risk factors, and course among antipsychotic-treated children and adolescents. This issue of the Journal of Child and Adolescent Psychopharmacology features one such report by Ray and colleagues (Ray et al., 2024a In Press), who analyzed data from a retrospective cohort study of antipsychotic drugs and mortality in children and youth, aged 5–24 years (Ray et al., 2024b). Antipsychotic exposures (restricted to new users), patients with NMS, and risk factors were identified from electronic prescription records, claims for injection medication administration, and diagnosis codes. Altogether, the cohort consisted of 1,032,084 patients (82% between the ages of 5 and 17 years) who were treated with antipsychotics over a 10-year period. A total of 131 cases of NMS were observed, translating to 8.9 new cases/100,000 person-years of follow-up. Potential risk factors for NMS included higher antipsychotic dose (>200 mg in chlorpromazine equivalents), use of a first-generation antipsychotic, having a neurodevelopmental or a primary psychotic disorder diagnosis, having multiple psychiatric diagnoses, and being older (aged 18–24 years, as opposed to 5–17 years). The incidence rates of NMS rose as the number of risk factors increased.

Several methodological and clinical factors support the approach taken by Ray and colleagues. First, the cohort study from which the present analysis was conducted used electronic data from Medicaid Analytic Extract (MAX) files representing beneficiaries from 30 states, capturing approximately 90% of antipsychotic prescriptions within that catchment, with over 1.4 million person-years of follow-up. The use of MAX files thus enabled the construction of a very large cohort of antipsychotic-treated youth with detailed longitudinal follow-up, a methodologic imperative given how rarely NMS occurs, the scale of which would not be feasible using prospective study designs. Second, absent the ability to confirm medication adherence, electronic prescription records have still been shown to have good concordance with patient self-report of medication use or medical record review as reference standards (reviewed in Bobo et al., 2014), and may therefore provide reasonable “upper-bound estimate(s)” of medication use or adherence (Crystal et al., 2007). Third, because NMS is a medical emergency, such cases are highly likely to come to clinical attention, especially in emergency room or hospital settings—characteristics of NMS that make it an attractive outcome for studies using claims data. And fourth, MAX and other claims data have limited to no susceptibility to recall and non-response biases or to difficulties in ascertaining clinical information from younger children or patients with severe mental health or neurodevelopmental disorders (Crystal et al., 2007).

These and other factors likely converged in producing plausible results despite a lack of direct contact with patients or access to health records, including a low estimated incidence of NMS, a rational set of independent NMS risk factors, and the identification of NMS complications (including rhabdomyolysis) that comport with known characteristics of adult NMS and with those suggested by reviews of NMS in children and adolescents (Neuhut et al., 2009; Pileggi & Cook, 2016; Silva et al., 1999; Strawn et al., 2007). As would be expected, haloperidol was associated with the highest incidence of NMS, and more cases of NMS were observed among patients with an autism spectrum disorder or intellectual disability diagnosis than any other diagnosis group, including primary psychotic illnesses. Such patients are simultaneously prone to the psychiatric comorbidities and behavioral disturbances that increase the odds of receiving an antipsychotic drug and to the neuromuscular complications of the antipsychotic drugs, themselves (Shafiq & Pringsheim, 2018; Spencer et al., 2013). Additionally, NMS may be precipitated by the administration of antipsychotic drugs to patients with catatonia (White & Robins, 1991), a condition affecting up to 20% of patients with neurodevelopmental disorders (Moore et al., 2022).

Somewhat unexpectedly, the incidence of NMS was significantly elevated with exposure to ziprasidone but not with other individual atypical antipsychotics or with recent antipsychotic dose increases. Speculatively, the findings with ziprasidone may reflect, to some extent, the use of injectable ziprasidone given its availability for clinical use during the entire study period, 2004–2013, and the known associations of intramuscular antipsychotics with NMS in adults (Strawn et al., 2007; Viejo et al., 2003). The predominance of atypical antipsychotics in this study may have buffered against a significant dose effect given the generally lower overall risk of NMS with atypical agents, as compared with first-generation drugs (Strawn et al., 2007). Additionally, by not restricting the cohort to inpatients, more dose adjustments in the study cohort likely occurred in outpatient settings than in the hospital, where the use of first-generation antipsychotics and rapid antipsychotic dose escalations are likely to be more common.

The work of Ray and colleagues also illuminates common challenges in studying NMS using automated population level data. Despite the large size of the cohort, relatively few cases of NMS were identified, thus limiting the authors’ ability to ascertain novel, age group-specific risk factors for NMS or to validate others. While the authors highlight an acceptably high PPV for the major diagnosis code used to identify NMS in their study, validation of that code was conducted in a separate cohort study of adults (Lao et al., 2020). The same level of performance between the two cohorts cannot be guaranteed.

Perhaps more importantly, a high PPV does not address the possibility of false negatives. The validity of the outcome definition in this study requires both a specific NMS diagnosis and appropriate coding. In clinical practice, NMS may often be overlooked (especially in patients treated with atypical antipsychotics as opposed to first-generation agents) or left undiagnosed despite having a strong clinical suspicion (Modi et al., 2016; Oruch et al., 2017). Case-fatality rates also could not be determined in the current study owing to the low number of identified NMS cases. As such, the authors were unable to take full advantage of linkages between MAX data and data from the National Death Index, although such a strategy will undoubtedly be used in future cohort studies of NMS risk and associated mortality. Finally, the authors could not directly address how well their findings in Medicaid beneficiaries translate to broader populations of antipsychotic-treated youth.

In summary, Ray and colleagues present important new work on the incidence and risk factors for NMS in a large cohort of child and adolescent Medicaid beneficiaries. Although antipsychotic use in this population may be in relative decline, the overall prevalence of antipsychotic use in children and adolescents remains high. This fact, and the potential for severe general medical complications and fatal outcomes associated with NMS, highlight the public health significance of this report. As to its broader impact, it is unclear whether achieving a better understanding of the incidence of and risk factors for NMS in young people magnifies its importance to the ongoing controversy regarding the use of antipsychotics in children and adolescents. On the other hand, Ray and colleagues provide a useful template for future studies of NMS risk in children and adolescents—the significance of which will grow as increasingly larger volumes of data become available for conducting large-scale, real-world clinical studies (Wettermark, 2013).