Increasing referrals to emergency department for psychiatric consultation and treatment among 50,056 adolescents and young adults: Predictors and implications

Abstract Objectives To identify predictors and time trends over 10 years of psychiatric consultation or treatment in adolescents and young adults referring to Emergency Department (ED). Methods Real‐world cohort data from 50,056 adolescents and young adults referring 105,596 times to ED between 2007 and 2016. We tested whether gender, age, triage code (red, yellow, green, white with decreasing severity), and referral modality predicted primary (psychiatric consultation) or secondary outcomes (anxiolytic treatment, sedative treatment, psychiatric admission), and whether these outcomes increased over the last 10 years. Results Mean age was 19.57(SD = 2.52), female percentage was 48.77%. Overall, 6.93% underwent psychiatric consultation, treatment, or admission. Among 2,547 adolescents and young adults undergoing a psychiatric consultation, 58.07% had either yellow or red triage code, and 47.2% were brought by ambulance. Female gender predicted psychiatric consultation and anxiolytic treatment, male gender predicted sedative treatment, suggesting gender differences in help‐seeking behaviors. Older age predicted all outcomes. Severe triage presentation and being brought by ambulance increased the risk of primary and secondary outcomes. Psychiatric consultation (1.77% to 3.64%), anxiolytic (3.04% to 6.15%), or admission (0.40% to 0.98%) roughly doubled, and sedative (0.27% to 1.23%) treatment had a four‐fold increase from 2007 to 2016. Conclusions Among adolescents and young adults aged 15 to 24 years old ED appears to be necessary for young help‐seeking subjects given the severe presentations and the increasing number of adolescents referring to ED. More studies should assess whether ED might be helpful in detecting subjects with sub‐threshold or early psychiatric symptoms, or at risk for severe mental illness.


INTRODUCTION
Patients with severe mental illness (SMI) have a decreased life expectancy compared with general population of around 15 to 20 years (Hjorthøj et al., 2017;Laursen et al., 2012). Natural causes (not suicide) account for more than 70% of lost life-years in patients with SMI (Jayatilleke et al., 2017). Among medical comorbidities, compelling evidence supports an association between severe mental illness and diabetes (Vancampfort et al., 2016), metabolic syndrome (Vancampfort et al., 2015), and cardiovascular disease (Correll et al., 2017). Given the frequent medical comorbidity patients with SMI frequently access to Emergency Department (ED) for medical reasons. In addition to medical emergencies or car crash (Gianfranchi et al., 2017), patients can frequently refer to ED for psychiatric symptoms when relapses occur. ED might be relevant for young populations in particular, given that the median age of onset of any mental disorder is around 18 years old  (18 for eating disorders , 17 for anxiety disorders , 19 for obsessive-compulsive disorder , 25 for schizophrenia and related disorders , 25 for substance-related disorders Yin et al., 2018], 31 for depressive or bipolar disorders [Caspi et al., 2020;Kessler et al., 2005;Solmi et al., 2021;Vaingankar et al., 2013]). Importantly, beyond accessing to ED for physical symptoms, or for symptoms of full blown mental disorders, patients often present psychiatric subthreshold symptoms well before SMI onset, often between age 15 and 24 years old (Kessler et al., 2007). Concerning evidence from the United States has shown increasing rates of adolescents accessing to ED, with suicidal ideation or because of suicide attempt (Burstein et al., 2019). Also, reports from Italy have shown that psychiatric consultations, suicidal behavior, suicidal ideation, and non-suicidal selfinjury have increased from 7. 7% in 2011 to 19% in 2016, in particular in females (Castaldo et al., 2020). Hence, according to this body of evidence, it could be expected that ED may serve as an early detection opportunity, and that frequently adolescents or young adults presenting to ED can have severe clinical pictures.
Evidence showing that adolescents accessing to ED have severe psychiatric clinical presentations is of upmost clinical relevance, since it defines the necessary role of ED in the pathway to care and clinical services within a wider heterogeneous setting of separate yet integrated services collaborating for prevention/ early intervention for psychiatric disorders in adolescents/young adults.
However, service organization and pathway to care is different across countries, and within regions in the same country. For instance, findings from insurance-based health system cannot be generalized to public funded health systems. Similarly, reports from rural areas might differ substantially from center with high urbanicity. Hence, in order to inform local service organization, local evidence must be generated. Specifically, to inform whether ED can have a role in early detection of adolescents and young adults with symptoms of psychiatric disorders, predictors, rates, and time trends of indicators of psychiatric symptoms, as well as severity of clinical presentation must be identified using data from local context.
The main aim of the present work is to describe which ED referral (i.e., severity of presentation) and subject characteristics (i.e., demographics) predict psychiatric consultation, treatment, or admission (as a proxy of psychiatric symptoms), and to test whether ED referrals due to psychiatric symptoms in young subjects have increased over the last 10 years. Also, we aim to show that specialist psychiatric assessment in ED is sub-optimally delivered. Our hypotheses are that ED is a necessary setting for a safe evaluation of help-seeking subjects referring to ED with psychiatric symptoms (in other words, that ED could not be replaced by different setting with lower medical assistance), that adolescents and young adults frequently access, and that referrals to ED for psychiatric symptoms have increased over the last 10 years.

Population, study design, variables of interest, primary and secondary outcomes
These real-world data were extracted from clinical electronic records from a retrospective cohort study, approved by local Ethical Committee. It included all subjects aged between 15 (adolescents access to ED since age 15) and 24 years old (upper limit widely used for young adults) who accessed for any reason to Padua University Hospital ED � Over the last 10 years, the percentage of adolescents accessing to ED and requiring anxiolytic treatment, or admission in psychiatry unit had about a 2-fold increase.
The adolescent requiring a sedative treatment had a 4.5fold increase � Presentation and need for treatment differ between males and females � Adolescents accessing to ED in need of psychiatric care have severe presentations and triage codes � EDs should be involved in the network of services for prevention and early intervention for mental disorders severity to red, highest severity with potential life-threatening course), referral modality (ambulance with or without doctor, autonomously, other), and gender. More in detail, regarding triage code, if vital functions are compromised the patient receives a red color code and has the highest treatment priority (e.g., cardiac arrest, respiratory distress, and coma). When clinical condition of the patient is stable but symptoms indicate a risk of rapid deterioration she/he receives a yellow color code (e.g., syncope, chest pain, and palpitations); otherwise the patient has no risk priority ad receives a green or white color code on the basis of degree of pain. The assignment of each specific triage code implies an operational response: red-coded patients have immediate access to the treatment areas, yellow-coded patient have access to protected waiting areas where they can wait for medical evaluation at most 20 min. The patients without risk priority can wait in a common waiting room without time limits. Primary outcomes was psychiatric consultation. Secondary outcomes were anxiolytic drug administration (diazepam), sedative drugs administration (haloperidol, promazine, or midazolam), and psychiatric admission.
We estimated which characteristics predicted primary and secondary outcomes, and whether primary and secondary outcomes increased over time. We also calculated rates of subjects not undergoing a psychiatric consultation despite being treated with anxiolytic treatment, as a proxy of sub-optimal psychiatric specialist evaluation delivery.

Statistical analyses
We tested multivariate generalized linear mixed effect models (GLMMs). In the analyses to individuate factors associated with psychiatric consultation and treatment, fixed factors of the multivariate models (and preceding primary and secondary outcomes) were gender, age at the time of arrival in ED, triage code (four levels: white, green, yellow, and red), and the referral modality (four levels: ambulance without medical doctor, ambulance with medical doctor, autonomous, and other ways). Considering that some people underwent into ED multiple times, we set the individual tax code as the clustering variable and the intercept as random factor (i.e., random intercept models). We used time series analysis to investigate the trend of psychiatric consultation, treatment, and admission over 10 years. In particular, we calculated the year-over-year change in ratio between outcomes and total ED accesses from 2007 to 2016, and we tested if this change had an increasing trend.
We did not consider time to event and did not run a survival analysis due to the short time between exposure (referral to ED) and outcome (psychiatric consultation in ED). We computed all the statistical analysis by means of the R statistical software (R Foundation for Statistical Computing, 2019). We used the lme4 package (Bates et al., 2015) to test the GLMMs, tfplot (Gilbert, 2015) and pastecs (Grosjean & Ibanez, 2018) packages to calculate the yearover-year change and the trend statistics, respectively. Finally, we estimated the R 2 of the GLMMs by means of the MuMin package (Barton, 2013). More specifically, we calculated both marginal and conditional R (Laursen et al., 2012), referring, respectively, to the proportion of variance explained by only the fixed factors and by both the fixed and random factors (Nakagawa & Schielzeth, 2013).
Analyses scripts are provided in Appendix S1.

Population
We included 50,056 subjects aged between 15 and 24, who accessed to Padua University Hospital ED 105,596 times. Mean age was 19.57(SD = 2.52), 48.77% were females. Overall patients were assigned with a white triage code in 67.07%, green in 9.46%, yellow in 21.75%, and red in 1.73%. Referral modality was autonomously in 73.80%, in other ways 6.51%, brought by ambulance without medical doctor in 17.52%, and by ambulance with medical doctor in 2.16%. Among the whole sample, 2.41% of subjects aged 15 to 24 years old referring to ED underwent psychiatric consultation, 4.66% underwent anxiolytic treatment, 0.77% underwent sedative treatment, and 0.64% underwent psychiatric admission. Overall, 6.93% of young subjects referring to ED needed at least one among psychiatric consultation, anxiolytic treatment, sedative treatment, or psychiatric admission. Also, among 2,547 subjects undergoing a psychiatric consultation, 30.43% subjects had white triage code, 11.5% green code, 52.53% yellow code, 5.54% red code, and 42.21X% were brought by ambulance without doctor, 4.99% by ambulance with doctor, 45.74% autonomously, and 7.07% in other ways.

Sedative treatment
Factors associated with sedative treatment are reported in Table 3.

Psychiatric admission
Psychiatric admission was independent from gender (

SUBOPTIMAL PSYCHIATRIC CONSULTATION DELIVERY IN ED
We confirm that mental status evaluation is sub-optimally delivered in ED since a psychiatric consultation is requested only in a portion of patients (2.41%) who are treated with anxiolytic treatment (4.66%).  Beyond safety considerations supporting the role of ED in early mental assessment, we also describe gender-differences between help-seeking behaviors in ED. Among other factors, stigma could be one of the reasons why males show less help-seeking behavior, given the association between male gender and stigma, and between stigma and less help-seeking behavior (Clement et al., 2015;Zaninotto et al., 2018). However, reports from other settings actually describe that male gender is associated with higher ED access for mental disorders (Young et al., 2005). Beyond what might obstacle help-seeking behavior in males, our results are relevant since they

Real
show that a gap exists between females and males access to specialist assessment in ED and call for gender-specific strategies to warrant early mental health assessment in ED. Also, it is possible that ED staff might differently address psychiatric symptoms in females versus males, offering a more thorough assessment process to females. Females could also more frequently access to ED because of higher rates of anxious symptoms. One further reason behind higher sedative treatment in males is that males that access to ED more frequently have poor medication adherence which might be associated with more disturbed behavior (Juhás & Agyapong, 2016).
In addition, despite such severe clinical presentations, a psychiatric consultation is requested only in a portion of patients (2.41%) who are treated with anxiolytic treatment (4.66%). Several reasons could explain why medical doctors working in ED do not always require a psychiatric consultation when a subject presents psychiatric symptoms. First, a detailed diagnostic process is not a priority in ED; excluding emerging life-threatening risks is the main aim of ED instead.
Such a priority hierarchy might restrict the clinical focus on managing emerging anxiety symptoms, rather than in starting a thorough diagnostic process. Second, frantic, chaotic, and unpredictable work pace in ED may push the symptoms threshold to request specialist (i.e., psychiatric) consultation well above a typical outpatient service setting.   On the other hand, beyond treatment, improving psychiatric assessment in ED in adolescents remains crucial for preventive psychiatry and early intervention (Hartmann et al., 2019;McGorry et al., 2018), and in particular given the ineffectiveness of prevention services to detect subjects who later develop psychosis (van Os & Guloksuz, 2017). ED could serve as an additional setting to screen for early symptoms of full blown psychiatric disorders, or for subthreshold symptoms. Compared with the creation of ad-hoc units (which of course also have the role of treating patients and reducing psychiatric admissions, going beyond an improvement in psychiatric assessment), cost-effective (Ising et al., 2015(Ising et al., , 2017 alternatives are there to screen adolescents and young adults accessing to ED. For instance, quick self-administered screening questionnaire for risk of psychosis, or other severe mental illness might be considered (Addington et al., 2015;Kline et al., 2015). Among those with a quick and self-report structure, which are more likely implementable in the context of ED, Prodromal Questinonaire-16 (PQ-16) (de Jong et al., 2018;Lorenzo et al., 2018;Pelizza et al., 2019) and PRIME-Revised (Kobayashi et al., 2008)

ETHICAL STATEMENT
The present study adhered to Helsinki Declaration. Subjects consented to use data in aggregated anonymous form.

DATA AVAILABILITY STATEMENT
Data are available upon request with a pre-specified research question.