Hospitalizations of Adolescents with Psychiatric Disorders in a Pediatric Unit: A 10-Year Retrospective Study

1. Introduction

In recent years, adolescent mental health has become an increasingly important public health priority. According to the World Health Organization, about one in seven adolescents aged 10–19 years lives with a mental disorder, and UNICEF has highlighted the broad impact of mental health problems on the well-being of children and adolescents [1,2]. In this context, hospitalization may be regarded as a marker of the most severe forms of psychological distress, because it captures patients with the greatest clinical, organizational, and psychosocial complexity.

International studies have documented a progressive increase in mental-health-related hospitalizations and acute-care use among young people, with further worsening during and after the COVID-19 pandemic [3,4,5,6]. Increases have been reported for depression, anxiety, suicidality, self-harm, and eating disorders, with a disproportionate impact on female adolescents and older age groups [4,5,7,8]. In Italy, several reports have described increasing psychological vulnerability in adolescence and mounting pressure on child and adolescent neuropsychiatry services [9,10,11,12,13].

However, most available data derive from dedicated child and adolescent psychiatric units, whereas longitudinal data from general pediatric wards remain limited. This distinction is not merely descriptive: where inpatient neuropsychiatric beds and community resources are scarce, general pediatric wards increasingly absorb the acute psychiatric demand that cannot be accommodated elsewhere. Trends observed in such a setting may therefore provide a useful sentinel measure of the most severe component of adolescent mental distress, while also quantifying a burden that is often invisible in psychiatry-based statistics. Against this background, the aim of the present study was to describe the evolution of psychiatric hospitalizations among adolescents admitted to a general pediatric ward over a 10-year period (2016–2025), focusing on diagnostic, sex, and age distribution and on pre- versus post-2020 changes, while adopting the individual patient as the primary unit of analysis.

2. Methods

2.1. Study Design and Setting

We conducted a retrospective observational study of psychiatric hospitalizations occurring in the Pediatrics Unit of G.B. Morgagni–L. Pierantoni Hospital, Forlì, Italy, between January 2016 and December 2025. The ward admits patients from the neonatal period through adolescence (0–17 years) and represents the main hospital referral center for acute pediatric conditions in a district area of about 200,000 inhabitants. The study was designed and reported in accordance with the STROBE guidelines for observational studies.

2.2. Participants and Data Sources

We included all hospitalizations for psychiatric disorders among patients aged 10–17 years at admission. Although adolescence is conventionally defined by the World Health Organization as 10–19 years [1], patients aged 18 years or older were not included because the local pediatric ward admits patients only up to 17 years of age. Age at admission was calculated as the interval between date of birth and date of admission. Data were extracted from the hospital discharge database, which systematically records information on each admission episode; for each episode we retrieved an anonymized patient identifier, sex, date of admission, age at admission, and the recorded discharge diagnosis. All data were anonymized before analysis. No missing data were present for the variables used in the analyses (sex, age, year of admission, and diagnostic category were complete for all 210 episodes).

2.3. Diagnostic Classification

Each episode was assigned to a single main clinical category based on the prevailing condition responsible for admission and inpatient care. When multiple diagnoses were present, classification was based on the clinical condition judged to be dominant in determining the need for hospitalization. The categories were suicidal ideation/attempt, eating disorders, psychomotor agitation, anxiety/panic disorder, mood disorder, conduct/behavior disorder, non-suicidal self-harm, alcohol/substance use, conversion/functional disorder, and psychosis/delirium. Episodes classified as suicidal ideation/attempt frequently showed psychiatric comorbidities such as mood or anxiety disorders; to preserve mutual exclusivity, each episode was assigned only to the category considered primarily responsible for admission. Classification was performed by the study investigators on the basis of the discharge diagnosis.

2.4. Unit of Analysis and Outcomes

Because a minority of adolescents were hospitalized more than once during the study period, treating every episode as independent would violate the assumption of independence and could inflate statistical significance. Therefore, the primary analyses used the individual patient as the unit of analysis, retaining the first admission of each patient. Episode-level analyses, in which repeated admissions were counted individually, were retained as a pre-specified sensitivity analysis and interpreted as a measure of inpatient burden rather than of patient-level epidemiology. The primary outcome was the annual number and trend of hospitalizations. Secondary outcomes were the diagnostic distribution, overall and diagnosis-specific temporal trends, the association between diagnosis and sex, age differences across diagnostic groups, the temporal trend in age at admission, and the comparison between the two 5-year periods 2016–2020 and 2021–2025.

2.5. Statistical Analysis

Analyses were performed in Python (version 3.11) using the statsmodels, scipy, and pandas libraries. Temporal trends in annual counts, overall and by diagnosis, were modeled using Poisson regression, with results reported as incidence rate ratios (IRRs) per year with 95% confidence intervals (95% CIs); diagnosis-specific trends were adjusted for multiple testing using the false discovery rate (FDR, Benjamini–Hochberg). The association between diagnosis and sex was assessed with the chi-square test and quantified with Cramér's V; because several diagnostic categories were small, cells with low expected counts were present and the corresponding estimates are interpreted with caution. Age differences across diagnoses were assessed with the Kruskal–Wallis test. The relationship between age and year of admission was evaluated with ordinary least squares (OLS) regression and Spearman correlation. For four pre-specified diagnoses (suicidal ideation/attempt, eating disorders, psychomotor agitation, and mood disorder), multivariable logistic regression models were fitted with sex, age, and year (centered on 2016) as predictors; results are reported as odds ratios (ORs) with 95% CIs and McFadden's pseudo-R², with the number of events per model reported to allow assessment of model stability. The two 5-year periods were compared for sex (chi-square) and age (Mann–Whitney), and admission rates per 1000 total pediatric admissions were computed for each period. All tests were two-sided, and p<0.05 was considered statistically significant. Computations were cross-checked for internal consistency; artificial-intelligence tools were used only for language editing and did not perform the statistical analysis.

3. Results

3.1. Cohort and Overall Trend

From a database of 9,090 patients admitted between 2016 and 2025, 210 psychiatric hospitalizations of adolescents were identified, corresponding to 165 individual patients. Twenty-seven patients were hospitalized more than once during the decade, accounting for 45 repeated admissions. In the primary (patient-level) analysis, females represented 77% of the cohort (80.5% of episodes) and the mean (SD) age was 14.47 (1.73) years (median 15, IQR 13–16; range 10–17). The number of admissions increased progressively over the study period, with a marked rise from 2021 onward (Figure 1). Poisson regression confirmed a significant temporal increase, with an IRR per year of 1.20 (95% CI 1.13–1.27; p<0.001) at the patient level and 1.20 (95% CI 1.14–1.27; p<0.001) at the episode level. 124/165 (75.2%) patients were first admitted in the 2021–2025 period.

3.2. Diagnostic Distribution and Diagnosis-Specific Trends

Suicidal ideation/attempt (27.3% of patients) and eating disorders (15.2%) were the most frequent categories. Both showed a significant upward trend that was robust in the primary patient-level analysis and in the episode-level sensitivity analysis (Table 1). A significant increase was also observed for non-suicidal self-harm and alcohol/substance use. The upward trend for mood disorder, significant at the episode level, was attenuated and no longer statistically significant after FDR correction in the patient-level analysis, and is therefore interpreted with caution. Estimates for the smallest categories (e.g., psychosis/delirium and conversion/functional disorder) are based on very few patients and should be regarded as unstable.

3.3. Sex and Age

Diagnosis was significantly associated with sex in both the patient-level (chi-square=23.98; p=0.004; Cramér's V=0.38) and episode-level (chi-square=25.43; p=0.003; Cramér's V=0.35) analyses (Figure 2). Eating disorders and mood disorders occurred almost exclusively in females, whereas psychomotor agitation showed a markedly higher male proportion than the overall sample. In contrast, age at admission did not differ significantly across diagnostic groups in the primary analysis (Kruskal–Wallis p=0.16; Figure 3); a nominally significant difference was present only at the episode level (p=0.032) and is therefore not emphasized. Across the study period, age at admission showed a small but significant increase, corresponding to +0.14 years per calendar year (p=0.010; Spearman ρ=0.17, p=0.027).

3.4. Comparison Between the Two 5-Year Periods

In 2016–2020, 51 episodes were observed (11.78 per 1000 admissions), compared with 159 in 2021–2025 (33.39 per 1000), corresponding to a rate ratio of 2.83 (95% CI 2.07–3.88); the patient-level rate ratio was 2.75 (95% CI 1.93–3.91) (Table 2; Figure 4). Age at admission was higher in the second period, whereas the sex distribution did not differ significantly between periods. The relative diagnostic composition shifted toward suicidal ideation/attempt and eating disorders, although at the patient level this shift did not reach statistical significance.

3.5. Multivariable Analysis

In multivariable logistic regression (Table 3), suicidal ideation/attempt was independently associated with older age, while psychomotor agitation was strongly and inversely associated with female sex and showed a relative decrease over time. No independent significant associations emerged for eating disorders or mood disorder. Model results were consistent between the patient-level and episode-level analyses. Model fit was modest overall (McFadden pseudo-R² 0.03–0.05 for suicidal ideation/attempt, eating disorders, and mood disorder) and higher for psychomotor agitation (0.14 at the patient level and 0.16 at the episode level). Given the limited number of events in some models, these estimates should be interpreted as exploratory.

3.6. Readmissions

Twenty-seven patients (16.4%) were hospitalized more than once, accounting for 45 readmissions (72 episodes in total; Table 4). These readmissions clustered in the most severe diagnostic categories: suicidal ideation/attempt (16), eating disorders (8), mood disorder (5), psychomotor agitation (4), anxiety/panic disorder (3), conduct/behaviour disorder (3), non-suicidal self-harm (3), psychosis/delirium (2), and conversion/functional disorder (1). Because the primary analyses use each patient's first admission only, these repeated admissions do not contribute to the patient-level estimates.

4. Discussion

According to our data, psychiatric hospitalizations of adolescents in a general pediatric ward increased markedly between 2016 and 2025, with a clear acceleration from 2021 onward. The most relevant feature was not only the trend itself but its magnitude: about three quarters of admissions occurred in the 2021–2025 period, with an approximately threefold increase in the admission rate compared with 2016–2020. Importantly, these findings were confirmed when the individual patient, rather than the hospitalization episode, was used as the unit of analysis, directly addressing the concern that repeated admissions might inflate the observed trends.

The added value of this single-center experience lies less in re-demonstrating the post-2020 rise—now well documented [3,4,5,6,7,8]—than in characterizing it within a general pediatric ward. In settings where dedicated neuropsychiatric beds are scarce, the pediatric ward becomes a de facto safety net for acute adolescent psychiatric presentations, and its admissions provide a sentinel signal of the most severe component of adolescent distress. The observation that the 27 patients with repeated admissions clustered in suicidal ideation/attempt and eating disorders reinforces this interpretation: recurrence concentrated in the conditions of greatest clinical severity, a pattern with direct implications for continuity of care after discharge.

From a diagnostic perspective, the increase was driven mainly by suicidal ideation/attempt and eating disorders, both of which increased significantly and robustly. The prominence of suicidal ideation/attempt confirms that pediatric hospitalization increasingly intercepts the most acute component of adolescent psychological distress, in line with studies describing rising emergency and inpatient presentations for suicidality, self-harm, and internalizing disorders [4,5,17,18]. The share of eating disorders more than doubled between the two periods, consistent with Italian studies documenting increased hospitalizations and referrals for eating disorders in developmental age during the pandemic period [15,16].

Placed alongside the international literature, our trends converge in direction with data from the United States [3,5,17], France [4,18], Finland [6], Spain [7], Denmark [19], and Italy [8,9,14,15,16], all describing rising acute and inpatient mental-health care use among young people, with a disproportionate impact on female and older adolescents. Our estimates are not directly comparable in magnitude, given differences in health-care organization and in the hospitalization setting; the value of the comparison lies in the consistency of the direction rather than in the equivalence of the rates.

The association between diagnosis and sex was confirmed at both the patient and episode levels, with a strong female predominance overall—particularly for eating and mood disorders—and a distinctively higher male proportion for psychomotor agitation. By contrast, we deliberately refrain from emphasizing differences in age across diagnostic groups, since this difference was not significant in the primary patient-level analysis. The modest but significant increase in age at admission over time, together with the concentration of the most severe diagnoses in mid-to-late adolescence, suggests that the greatest need for inpatient care emerges mainly in this age range.

Our findings have organizational implications. Where child neuropsychiatric inpatient beds and community resources are limited, general pediatric wards increasingly care for patients with substantial psychiatric and psychosocial complexity, and pediatric hospitalization may serve as a sentinel indicator of severe adolescent suffering. The results support strengthening integrated pathways among pediatrics, child neuropsychiatry, emergency care, and community services, with particular attention to suicidal crises, eating disorders, and post-discharge continuity of care [13].

Limitations

Several limitations should be acknowledged. First, the single-center design limits generalizability. Second, although we prioritized the individual patient as the unit of analysis, the assignment of complex presentations to a single dominant diagnostic category entails subjectivity and possible misclassification, and formal inter-rater reliability was not assessed. Third, our analyses are based on admission counts and on rates per 1000 hospital admissions, not on population-based incidence; changes over time in referral pathways, service availability, admission thresholds, or local organization may have contributed to the observed trends, so conclusions about disease burden should be interpreted with caution. Fourth, several diagnostic categories were small, and the corresponding trend and regression estimates are consequently unstable and exploratory. Finally, the study captured ward admissions only and did not include psychiatric presentations managed in the emergency department without admission, which may underestimate the overall demand for acute care [14,17]. For these reasons, comparisons with the international literature should be read as convergence in direction rather than full equivalence of estimates [4,6,18].

5. Conclusions

This single-center series shows that psychiatric hospitalizations among adolescents admitted to a general pediatric ward increased substantially over the 2016–2025 decade, with a clear gradient from 2021 and a predominant contribution from suicidal ideation/attempt and eating disorders. These findings, robust to analysis at the patient level, support the view that general pediatric wards increasingly intercept a clinically severe share of adolescent mental distress, and underline the need for stronger integrated pathways between pediatrics, child neuropsychiatry, emergency care, and community services, together with earlier preventive strategies directed particularly at female adolescents and the 14–17-year age range.