The impact of the COVID-19 pandemic on admissions to an adolescent inpatient psychiatric unit

Mental illness and substance use are leading causes of disability in young people, accounting for nearly half of the global burden of disease in those aged 10–24 years. ¹ Adolescence is a critical developmental period, during which early-onset psychiatric disorders can shape long-term health trajectories and affect future generations. ²

The COVID-19 pandemic and associated lockdowns created an unprecedented disruption to adolescent social, educational, and family environments. Young people are particularly vulnerable to such disruptions, given their reliance on peer relationships, schooling, and community supports.^3–5 Evidence from previous public health crises suggests heightened risks of abuse, neglect, and mental health problems during periods of social restriction.^4,6 Melbourne, Australia, experienced one of the world’s longest lockdowns, with restrictions lasting 262 days, raising concerns about its potential impact on youth mental health.^7,8

Emerging international evidence has reported mixed effects of the COVID-19 pandemic on adolescent psychiatric presentations (see Appendix 1). Several studies have found increases in self-harm, anxiety, and eating disorder presentations,^9,10 whereas others found stable or reduced rates of hospital attendance, ¹¹ often attributed to service restrictions and help-seeking behaviour.^12,13 However, most of this research has examined community or emergency department settings, and largely originates from North America and Europe, where lockdown measures were typically shorter and less restrictive than those in Victoria, Australia.

Consequently, little is known about how prolonged lockdowns affected the clinical profile, diagnostic complexity, and severity of adolescents requiring hospitalisation. Inpatient data offer a unique perspective on the most acute and complex presentations, complementing community-based research that may underestimate the pandemic’s impact on severe mental illness.

This present study addresses this gap by comparing sociodemographic, diagnostic, and clinical characteristics of adolescent psychiatric inpatient admissions before and during Victoria’s first 9 months of lockdown. By analysing diagnostic distributions, clinician- and self-reported symptom measures, and hospital variables such as restraint use, the study provides insight into how systemic disruptions may alter both the nature and severity of cases requiring hospital care.

Method

Results

Sociodemographic and admission variables

Between 2019 (n = 443) and 2020 (n = 520), there was a statistically significant increase of 17% in the number of admissions (One-Sample Binomial test, p = 0.014). No significant differences were observed in mean age, gender distribution, rurality, or LOS. Admissions of Indigenous adolescents significantly decreased, and use of physical restraints significantly increased, while seclusions remained unchanged (Table 1).

OPEN IN VIEWER

Table 1.

Comparison of sociodemographic and hospital admission variables.

Variable	2019	2020	Test (df)	p
Mean age in years (SD)	16.26 (1.35)	16.33 (1.25)	t = 0.86 (961)	0.95
Gender
Female  Male  Non-binary	310 (70%) 111 (25%) 22 (5%)	393 (76%) 108 (21%) 19 (4%)	χ2 = 3.93 (2)	0.014*
Indigenous status N (%)	18 (4%)	4 (0.8%)	χ2 = 11.59 (1)	<0.001*
Rural region N (%)	67 (15%)	68 (13%)	χ2 = 0.83 (1)	0.36
LOS (days), mean (SD) a	4.19 (4.64) b	3.86 (2.94)	t = 1.29 (722.18)	0.19
Number of restraints a	18	30	t = −2.39 (766.31)	0.017*
Number of seclusions a	12	13	t = −1.32 (784.97)	0.19

^adf adjusted as equality of variance could not be assumed.

^bOne extreme outlier removed (LOS >3x the next highest LOS).

Bold indicates significant *p < 0.05.

Further breakdown of physical restraints use by diagnostic category or number of individuals restrained was not possible, as episode-level data were not consistently linked to diagnostic entries in the statewide database.

With LOS, one outlier was three times longer than the next highest LOS. This extremely long admission, due in part to post-discharge placement problems, was excluded from the LOS analysis. Variance equality could not be assumed, and t-test parameters were adjusted accordingly (Table 1).

Diagnoses

All diagnoses for the 963 AIPU admissions were extracted, and the key priority diagnosis was identified. Diagnostic groups less than 1% (n = 11) or where the priority diagnosis was an ICD procedural or investigation code (n = 37) were excluded. Personality Disorders (PD) and Autism Spectrum Disorders (ASD) ¹ were the only groups to increase significantly between study periods (Table 2).

OPEN IN VIEWER

Table 2.

Change in priority diagnoses (ICD-10).

Diagnosis	2019 n (%)	2020 n (%)	Total n (%)	Binomial test (p)
Personality disorders	118 (28.2)	196 (39.4)	314 (34.3)	<0.001*
Mood disorders	125 (29.9)	142 (28.6)	267 (29.2)	0.327
Anxiety disorders	46 (11.0)	41 (8.2)	87 (9.5)	0.668
Autism spectrum disorders	20 (4.8)	38 (7.6)	58 (6.3)	0.026*
Adjustment disorders	27 (6.5)	21 (4.2)	48 (5.2)	0.47
Eating disorders	19 (4.5)	19 (3.8)	38 (4.2)	1.00
Substance related disorders	17 (4.1)	13 (2.6)	30 (3.0)	0.584
Schizophrenia and psychoses	17 (4.1)	10 (2.0)	27 (2.9)	0.248
Disruptive behaviour disorders	8 (1.9)	5 (1.0)	13 (1.4%)	0.581
Total	418 (100)	497 (100)	915 (100)	​

NB: Excluding diagnostic groups <1% of total priority diagnoses.

Bold indicates significant *p < 0.05.

Clinician-rated symptoms (HoNOSCA)

HoNOSCA records were matched to admissions. There were 902 valid HoNOSCA admission records (93.6% completion rate) with 398 (89.9%) and 504 (96.9%) HoNOSCA in the pre and the pandemic periods. HoNOSCA total scores did not differ significantly between years (2019 mean 18.23, sd 7.31; 2020 mean 18.54, sd 7.09). However, MANOVA (F = 3.699, df = 15, 866, p < 0.001) revealed significant differences between years for four scales: increases in ‘overactivity and concentration problems’, ‘psychotic symptoms’, ‘somatic complaints’, and a significant reduction in ‘substance-related problems’ (Table 3).

OPEN IN VIEWER

Table 3.

Comparison of mean (sd) HoNOSCA scales by year.

HoNOSCA rating scales	2019	2020	Test statistic		p
	Mean (sd)	Mean (sd)	Test (F)	df
1. Disruptive…	1.17 (1.28)	1.28 (1.25)	1.88	1,900	.171
2. Overactivity…	1.42 (1.16)	1.60 (1.19)	4.69	1,900	.005*
3. Self-injury	2.29 (1.16)	2.32 (1.14)	0.09	1,900	.755
4. Substance use…	1.18 (1.40)	0.76 (1.18)	23.12	1,900	<.001*
5. Scholastic/Language…	0.84 (1.19)	0.74 (1.11)	1.63	1,900	.202
6. Physical illness/Disability…	0.39 (0.92)	0.46 (0.95)	1.03	1,900	.309
7. Hallucinations…	0.70 (1.14)	0.95 (1.28)	8.85	1,900	.003*
8. Somatic…	0.20 (0.63)	0.31 (0.76)	5.54	1,900	.021*
9. Emotional…	2.90 (0.72)	2.93 (0.69)	0.51	1,900	.477
10. Peer…	1.190 (1.97)	1.92 (1.19)	0.06	1,900	.840
11. Self-care…	1.35 (1.36)	1.46 (1.21)	1.65	1,900	.199
12. Family…	2.33 (1.15)	2.30 (1.19)	0.18	1,900	.670
13. School attendance…	1.55 (1.47)	1.51 (1.42)	0.11	1,900	.746
14. Knowledge of difficulties…	0.65 (1.07)	0.64 (1.05)	0.002	1,900	.968
15. Information about management…	0.60 (1.01)	0.51 (0.99)	1.56	1,900	.212
Total score	18.23 (7.31)	18.54 (7.09)	0.42	1,900	0.52

Bold indicates significant *p < 0.05.

Discussion

This study is, to our knowledge, the first in Australia to examine the impact of the COVID-19 pandemic and associated lockdowns on AIPU. By comparing admissions during 2020 with the equivalent period in 2019, it provides a unique perspective on the most severe youth presentations requiring hospitalisation. Overall admission numbers increased whereas international reports showed no widespread rise in adolescent psychiatric hospitalisations during early COVID-19.^18,19 While admissions increased, clinician’s, parent’s and young people’s perception of overall symptom and functioning remained stable. Adolescents admitted to hospital typically represent the most severe and complex presentations, suggesting both a possible ceiling effect for admission, and a potentially reduced capacity for the community to manage severe problems during the pandemic.

Clear diagnostic and symptom shifts were observed. The significant rise in PD aligns with literature highlighting the vulnerability of adolescents with borderline traits during COVID-19 (20). Disruptions to therapeutic alliances, peer relationships, and schooling likely contributed to heightened emotional instability and crisis presentations.^20,21 Beyond individual vulnerabilities, broader family dynamics during lockdowns may also have contributed. Extended time spent at home during prolonged lockdowns likely increased family exposure to adolescents’ emotional and behavioural difficulties, prompting earlier recognition and, in some cases, lower tolerance for distress or dysregulation. Heightened parental stress, financial strain, and reduced access to community and respite supports may have further intensified family dynamics. These contextual factors may have influenced both the timing of hospital presentations and clinicians’ formulation of diagnoses, particularly for disorders characterised by interpersonal and emotional dysregulation. Many of these features could also have contributed to the increase in Autism Spectrum Disorders (ASD) admissions, although it is not clear whether this is confounded by the increasing community-based presentations of ASD over the last decade ²² .

No increase in eating disorder admissions was found, diverging from international evidence of escalating adolescent eating disorder presentations^23,24 and from an Australian study. ²⁵ This discrepancy may reflect local service pathways with medically unstable cases admitted to specialised paediatric eating disorder units/wards,^26,27 or diagnostic overlap where disordered eating behaviours were coded under personality disorder features.^28,29

The decline in substance misuse is consistent with studies showing reduced access to alcohol and illicit substances during lockdowns.^30,31 While this reduction may have been protective in the short term, it raises questions about post-pandemic trajectories once access is normalised.

Increased hyperactivity was a notable and consistent finding across clinician-rated and self-reported measures, echoing international reports of worsening ADHD-related difficulties during the pandemic.^32,33 Loss of daily routines, reduced opportunities for physical activity, and prolonged online learning may have amplified attentional and behavioural regulation problems in adolescents, particularly those with pre-existing vulnerabilities. ³⁴ This finding underscores the importance of environmental structure and routine in managing hyperactivity, and highlights the need for targeted supports during periods of systemic disruption.

The rise in the use of physical restraints during the pandemic was interesting. This trend has also been observed in other inpatient settings during the pandemic. ³⁵ Given that there was not higher severity in disruptive behaviours this may reflect the heightened challenges faced by staff working under the restrictive pandemic conditions. Importantly, this occurred during a period when physical distancing requirements limited close contact, potentially constraining the use of standard de-escalation strategies. This finding raises concern about the effectiveness and sustainability of de-escalation practices during times of systemic stress and underscores the need for additional staff training, reflective supervision, and organisational support to maintain trauma-informed and non-restrictive approaches even under crisis conditions.

Increased clinician-rated psychotic symptoms, without a parallel rise in formal psychotic disorder diagnoses, mirrors local reports of increased first-episode psychosis admissions in Melbourne. ³⁶ This suggests dimensional measures such as the HoNOSCA may capture early shifts in severity not reflected in categorical diagnoses.

Finally, the decrease in Indigenous admissions is concerning and warrants cautious interpretation. The dataset did not include information to differentiate whether this reduction reflected barriers to accessing inpatient care during lockdowns, ³⁷ or protective factors arising from strengthened family and community supports. ³⁸ Both possibilities are plausible. Lockdowns and service restrictions may have compounded pre-existing access barriers for Aboriginal and Torres Strait Islander young people, including mistrust of mainstream services, reduced outreach, or geographic isolation. Conversely, some communities have reported strengthened kinship support and culturally grounded coping responses during the pandemic, which may have reduced the need for hospitalisation. Without more granular data, these explanations remain speculative. This finding nonetheless highlights the importance of culturally safe service design and the need to ensure equitable access to care during public health crises. Further investigation comparing outcomes across other cultural and racial groups would assist in determining whether inequities were specific to Indigenous populations or reflected broader systemic disparities in service reach and responsiveness.

Self- and parent-reported data (SDQ) showed no change in functional difficulties and complexity between the pre-pandemic and pandemic periods, consistent with stable clinician-rated severity. While complexity may have changed within community settings, neither parents, adolescents nor clinicians observed an overall change in this tertiary AIPU. A supplementary finding revealed an increase in over activity/inattention that was noted by parents, adolescents, and clinicians. In times of social restriction, the AIPU could be considered an island of face-to-face interactions.

These findings have several implications for service planning. Aboriginal and Torres Strait Islander young people, a high-risk group, had fewer admissions during the pandemic. It may be that periods of social crisis require additional engagement strategies to maintain accessibility for vulnerable groups. Secondly, increased admissions alongside a requirement to achieve social distancing with the adolescents created additional pressures within the AIPU. Achieving increased social distancing required more active staff intervention. The observed rise in restraint use may be partly consequent to this shift in the milieu rather than simply an increase in admissions with disruptive behaviours. Forward planning addressing increased social restrictions without exacerbating aggressive interactions may be prudent.

Finally, the increase in PD and in ASD admissions raises the question of flexibility in service response. For example, young people with ASD may have been particularly vulnerable to the disrupted routines and supports during lockdowns, while the reliability of the interpersonal relationships of those with a PD may have been impacted by the pandemic. How an inpatient unit could alter its’ therapeutic approach to a changing diagnostic profile, let alone account for the pandemic impact, is a complex matter for consideration. While treatment and milieu approaches were not examined, the question of therapeutic flexibility remains important in achieving a patient-focussed unit.

Ongoing monitoring is required to detect shifts in adolescent presentations, and planned treatment modules should be identified that can respond. Longer term research may distinguish temporary pandemic effects from enduring service needs, but that will not diminish the need to respond to respond appropriately to the short-term changes.

Conclusion

Adolescent psychiatric inpatient admissions did increase during the COVID-19 pandemic, but this was not accompanied by a rise in overall symptom severity. The clinical profile showed more Personality and Autism Spectrum Disorders, increased hyperactivity, and restraint use, alongside reduced substance-misuse and fewer Indigenous admissions.

The rise in restraint at a time of increased volume and stable acuity may have been influenced by the demand for physical distancing rather than behavioural challenges amongst adolescents.

These results suggest that the changes during the pandemic were different to what was discussed in community settings. Ongoing monitoring is more important than ever at an AIPU to ensure appropriate interventions and that marginalised groups are not left behind during a pandemic.

Supplemental material