The annual number of emergency consultations continuously increased over time (Table 1). From 1981 to 2017 there was a 3.85 time increase in patient referrals, with a 131% increase between 1981 and 1992, a 9% increase between 1992 and 2002, and a 37% increase between 2002 and 2017. The differences between the mean ages of subjects over time were statistically significant (χ² = 21.867, df = 2, p < 0.001). On average, youths admitted in 1981 were 1 year 7 months older than those admitted in 2017. Gender ratio was stable over time, around 0.55 for male.

The comparison of the source of referral for an emergency visit over time is presented in Table 2. The proportion of consultations requested by the family and/or relatives almost doubled between 1981 and 2017 (respectively 29.4 and 57.8% of the annual number of visits). The consultations at the youth's request represented 6% of the total number of referrals in 1981 but only 0.3% in 2017.

Agitation and/or aggressive behavior were the main reason for referral to the emergency consultation service for all the periods considered (around a quarter of the visits). The consultations for suicidal attempt and/or non-suicidal self-injury (NSSI) decreased from 12.7% in 1992 to 9.2% in 2017. The referral for anxiety and depressive mood sharply increased over time with, respectively a four times and a three times increase between 1992 and 2017. The referral for school problems and/or learning difficulties decreased from 6.1 to 1.7% during the same period. The main reason for referral differed across gender (Figures 1, 2).

The annual number of youths diagnosed with an anxiety disorder as the primary psychiatric disorder sharply increased over time (Table 3). This diagnosis was seven times more likely in 2017 compared to 1992. A substance use disorder was four times more frequent in the youths admitted in the emergency consultation service in 2017 compared to 1992. The proportion of youths with a discharge diagnosis of schizophrenia or other psychotic disorder in 2017 was half of that in 2002.

The admission to an inpatient unit directly following the emergency visit represented three-quarters of the total annual number of consultation in 1981 but only 15.6% in 2017 (Table 4). In absolute terms, this decrease represents 64 annual admissions in 1981 vs. 48 in 2017. In parallel, the orientation toward outpatient services became the most frequent outcome. In 2017 59% of the annual visits were directed to a community mental health service and 13.3% were addressed to a post-emergency consultation.

This study has a number of strengths, such as the assessment of emergency activities over more than three decades. But the study also has limitations. First, the naturalistic design of our study does not permit a standardized measure of clinical symptoms and diagnostic procedure that could result in assessment bias. Comparison of the discharge psychiatric diagnoses between the four time periods therefore requires caution. Second, as our study is monocentric selection bias exists. Therefore, findings should be interpreted in the context of the CAPD-PSH. For example, the development of specific care pathways for youths with severe developmental disorder in the CAPD-PSH explains that this group is underrepresented in our sample. Third, as a medical record review, the data were not originally collected for research purposes and therefore we may have under or overestimated certain patient characteristics based on data unavailability. Fourth, as no community help-seeking sample or no severity assessment was available here, we cannot properly tested whether the changes observed here are truly specific to emergency activity. Moreover, measures of duration of untreated illness, symptom severity and onset of illness (acute/insidious) would have given some insights to better understand the reasons for seeking consultation in emergency. Finally, the time intervals between each annual data can be seen as arbitrary. It cannot be excluded that significant changes have not been identified using a 10-year time scale. In particular, the 5-year period leading up 2017 has been of much interest given the evidence for marked increases in youth presenting for affective difficulties and suicidality during this time period in North America.

We used naturalist clinician-reported data collected in 1981, 1992, 2002, and 2017 to examine the evolution of the types of youth mental health problems observed in an emergency unit in a child and adolescent psychiatric department over the last three decades.

In line with our first hypothesis, we found that since 1981 the number of patients presenting to the emergency service rose steadily, with an increase of 3.85-fold between 1981 and 2017. This is in line with the observations made in pediatric EDs in the US (5, 6, 14, 15), in Canada (4, 12), and in Spain (13). As mentioned in the introduction, in the context of the French national mental health system this finding supports the view that financial reasons are not a determining factor in the increase in the use of emergency services in children and adolescents. This is the first study using a period of observation extended over more than 30 years, compared with studies of 4 years (12, 14), 6 years (15), 9 years (6), and 10 years (4, 5). We found a constant increased use of emergency consultations over the study period, although not at the same pace, with the largest growth during the 1980s. To ensure that these numbers were not explained by demographic changes, we verified that the size of service catchment area of the CAPD-PSH remained stable over the study period using official demographic data (Table S1).

In line with our second hypothesis, we found that the proportion of consultations for a more frequently severe problem (e.g., delusion) did not increase to the same extent compared to common/more frequently benign forms of mental health problems (e.g., anxiety) over the study period. In fact, the number of consultations for delusion decreased between 1992 (8.7%, n = 17) and 2017 (4.4%, n = 13). Unfortunately, no specific measure of the severity of clinical cases (e.g., with triage or acuity assessments) was available here, while the use of the discharge diagnoses as a proxy measure for clinical severity has previously been used (4). We are uncertain whether this increased number of ED psychiatric diagnoses of mood and anxiety disorders reflects increased prevalence of these diseases, increased severity of these diseases, change in clinical practice with increased diagnosis in emergency consultation, heightened awareness of these diagnoses by the family the youth himself, or finally a decreased access to outpatient care for youths with mood and anxiety disorders. As our study did not assess currently available outpatient resources, clinical practices, or epidemiological data from non-help-seeking population the question remained unsolved. Although, in our study, preliminary evidence seem to show that the proportion of consultations for common/more frequently benign reasons have progressively increased during the last decades. This finding indirectly supports the assumption that the increased use of emergency structures in children and adolescents is partly explained by the shortages of mental health providers. Moreover, it is possible that the overutilization of emergency resources for the common/more frequently benign forms of mental health problems affect access to mental health emergency services for more severe form of mental illness.

We found in our sample that the consultations for anxiety and depressive mood have risen steadily over the three decades. This is in line with previous studies conducted in EDs (4, 12). This is also consistent with an increase in self-reported symptoms of depression during adolescence reported in general populations (22), and longitudinal data collected in both primary and secondary health care services (23). Interestingly, in our study such increase was not associated with a corresponding rise in the number of diagnosed depressive disorders. It may be due to changes in clinical practices, in particular the better recognition that mood symptoms can occur as part of emotional regulation difficulties in youths with externalizing problems (24). Unfortunately, this hypothesis cannot be properly tested as psychiatric comorbidities were not taken into consideration in this research. Moreover, anxiety disorders became the most frequent discharge psychiatric disorder in youths admitted in the emergency unit, rising from 5% in 1981 to 34% in 2017. In a context where consultations at the request of the family became the main source of referral, one might ask whether an increased attention to pediatric emotional disorders in scientific debate and in the general media could impact the way the parents expressed their concerns about their child, and ultimately increase the number of emergency visits for anxiety and depressive mood. However, it is worth remembering that the decrease in referral for suicidal behavior in emergency consultations over the last decades observed in our clinical sample does not tell us the evolution of this phenomenon in general population. In particular, given that the emergency consultation service studied is open only weekdays during daily hours and Saturday morning, the proportion of youth presenting for suicidal behavior is expected to be lower than in pediatric ED structure open overnight.

As illustrated by the Figures 1, 2, gender discrepancies in the main reasons for referral to the emergency consultation have been found: the visits for aggressive behaviors were more frequent for boys, while girls were more likely to consult for affective symptoms. Such findings are consistent with preexisting literature, (e.g., 5, 6, 15). Regarding our two principal hypotheses we did not find any significant gender effect, i.e., the growth in emergency consultations across time and the increase proportion of common/more frequently benign forms of mental health problems were observed both in boys and girls.

Between 1981 and 2017 the admission rate to an inpatient structure decreased both in absolute and relative terms, which is consistent with the political context of constant inpatient unit beds reduction for children and adolescents (9). For example the closure of child psychiatric inpatient facilities that took place in the 1970s led to a 70% decrease of the number of beds between 1986 and 2000 (9). The current admission rate to an inpatient structure in our study was lower than those reported in previous American studies [36% (4) and 52% (16)] but was comparable to those found in a Spanish sample [19% for (13)]. Interpretation of these findings should be cautious as all of these studies were conducted in 24 h/7 day pediatric EDs and not in a consultation service of a child and adolescent psychiatry department.

Our finding supports a shift to a more comprehensive use of alternatives to direct hospitalization. In 2017, the discharge to a community mental health service was the most frequent outcome after the emergency visit. It should be noted that the delay for an appointment in a community mental health service varies greatly across Paris areas, and there are virtually no outpatient structures in France that can provide same-day evaluations. It appears that one out of 10 patients presenting to the emergency consultations was admitted to a pediatric inpatient unit until a bed in a child and adolescent psychiatric ward became available. The extent of this practice (called “delay hospitalization”) is not without consequences on the quality of the mental health care provided (7). The development of specific training for the medical and paramedical pediatric teams who receive these patients represents a possible area of improvement (2, 25).

We have confirmed the growing use of mental health emergency consultations and the changes in the types of mental health problems of children and adolescents over the last three decades. Indirect evidence support the assumption that this change is caused by growing patient and family demands in a context of shortage of mental health providers rather than demographic, economic or epidemiological changes. Further research should (i) use specific measures of clinical severity and comorbid psychopathology and (ii) contrast the finding from a help-seeking sample with a community-based sample. This would help to more precisely understand the socio-demographic and clinical factors that contribute to the evolution observed. These findings may, in turn, aid the development of intervention in the model of programs being developed and piloted across the country to provide quick evaluations in an outpatient setting (26).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.