Expert reports indicate that the mental health of children, adolescents and young people has deteriorated in recent years.1,2 This trend can be measured in the general population using incidence rates among individuals who were previously without a mental disorder3 or using indicators showing a change in the use of psychiatric resources among individuals who were previously diagnosed with a mental disorder.4,5 In a Beveridge-type healthcare system6, such as the Spanish system, it may manifest as a referral from primary care to psychiatry or as hospitalization.5,7 In 2020, the most notable development in the use of mental health resources was a decrease in psychiatric consultations and hospitalisations.8 Although limited access to health services during the pandemic reduced the recorded incidence of new cases in 2020, these negative conditions led to a rebound effect in 2021 and 2022. Furthermore, it has been noted that patients with a mental disorder decompensated more frequently, resulting in a higher rate of hospitalisations in subsequent years;9 however, most studies have been based on hospital clusters,7,10 with only a few using population-level data.8,11 Generally, previous research has compared the pandemic period with previous years.7,12 The analysis of decompensation trends should incorporate psychiatric comorbidities, given their role in patient evolution.13 However, there is a lack of studies addressing this issue.1

Health inequalities, which are defined as systematic and unfair differences in health outcomes and their determinants within a population, have worsened in recent years.14 It is known that socioeconomic status and gender not only determine the risk of mental disorder in young people and adolescents, but also inequity in mental health care.15,16 As the impact of the pandemic was worse for structurally marginalised groups,17 it is reasonable to suppose that these groups experienced poorer mental health care as a result.18 This increased need should have resulted in greater demand for mental health referrals from primary care and psychiatric hospitalisations among children, adolescents and young adults. In light of the lack of population studies in Spain, studies measuring changes in resource use before, during and after the pandemic are required in order to ascertain whether mental health services in the post-pandemic era are addressing possible inequalities relating to age, comorbidity, sex and socioeconomic status.19

This study aimed to explore trends in psychiatric hospitalisations and referrals for psychiatric consultations among individuals aged 0–30 years with a pre-existing mental disorder, categorised by mental disorder cluster, socioeconomic status, psychiatric comorbidity, sex and age, between 2019 and 2022.

Table 1 shows the total population aged 0–30 years, as well as the subpopulation with a previous diagnosis of mental disorder, for the years 2019, 2020, 2021 and 2022. In 2019, the total population was 608,984, of whom 97,962 had a mental health diagnosis. Of these individuals, 757 (0.77%) were admitted to a psychiatric ward and 9243 (9.44%) were referred to a psychiatrist from a primary care setting.

Table 2 shows the baseline characteristics of the population with a mental disorder who had a psychiatric admission or were referred to a psychiatrist in 2019. Statistically significant differences were observed in age and socioeconomic status between patients who decompensated (were hospitalised and/or referred to mental health services) and those who did not. The percentage of individuals with hospitalisations and primary care mental health referrals was higher in the low socioeconomic status group. Those hospitalised were, on average, 1.96 years older than those not hospitalised, while those referred were, on average, 2.05 years younger. Comorbidity significantly increased the risk of decompensation, with the group of patients with ≥3 comorbidities containing 2.76% of all patients and accounting for 58.78% of hospitalisations and 8.76% of referrals. The characteristics are broken down by pre-pandemic cluster in 2019 in Tables S2 and S3 in Supplementary Material, for hospitalisations and referrals, respectively.

The likelihood of hospitalisation was assessed using logistic regression analysis in each cluster of mental disorders (see Table S4 in Supplementary Material). In 2020, there was a reduction in the risk of hospitalisation that was statistically significant overall, although not in all clusters. By contrast, there were widespread increases in 2021 and 2022, with different odds ratios (OR) whose 95%CI included 1 in several clusters. Figure 1 shows a forest plot summarising these models, comparing OR and their 95%CI by cluster in 2022 with the 2019 baseline values, adjusted for age, sex, socioeconomic status and comorbidity. Table 3 shows the full models for the group with any mental disorder, as well as for the anxiety and self-harm clusters, in which the association of the pandemic was most pronounced. In 2021, the self-harm cluster had an OR of 1.85 (95CI: 1.51-2.28), rising to 1.97 (95CI: 1.61-2.42) in 2022. The OR was significant in the anxiety cluster in both 2021 (OR: 1.33; 95CI: 1.17-1.50) and 2022 (OR: 1.47; 95CI: 1.30-1.66). In the substance use cluster, the changes in 2021 and 2022 were not significant. Also noteworthy in these tables is the substantial association of comorbidity, with an OR of 42.2 (95CI: 37.9-47.0) across all clusters. Figure 2, a forest plot of the OR and their 95CI, illustrates this finding.

Figure 1.

Forest plot of odds ratios and their confidence intervals for each cluster comparing hospitalisations in 2022 with baseline (2019) adjusted for age, sex, socioeconomic status and comorbidity. ADHD: attention-deficit/hyperactivity disorder.

Figure 2.

Forest plot of odds ratios and their confidence intervals for each cluster comparing hospitalisations in the category of comorbidity ≥3 with the category of comorbidity 1 adjusted for age, sex, socioeconomic status and year. ADHD: attention-deficit/hyperactivity disorder.

Table 4 and Table S5 in Supplementary Material show the results of all the logistic regression models that analysed the likelihood of a mental health referral, disaggregated across all clusters. Due to their ability to organise and summarise the trend in hospitalisations and referrals over the years analysed, the OR derived from these logistic regression models should be understood as descriptive statistics, rather than as predictions or evidence of causality. Figure S1 in Supplementary Material provides a clear visualisation of the results by showing the forest plot comparing 2022 with 2019 (baseline). This comparison is adjusted for age, sex, socioeconomic status and comorbidity. A rebound effect, characterised by the highest probability of referral, was apparent in the clusters relating to ADHD, self-harm and substance use, but not in the other clusters or overall. There was a statistically significant decrease in referrals among patients with a mental disorder in 2020, with an OR of 0.86 (95CI: 0.83-0.88), followed by an increase in 2021, with an OR of 1.09 (95CI: 1.05-1.12). In 2022, however, the differences were not significant. Patients with a mental disorder from a low or middle socioeconomic status background were more likely to be referred, with OR of 1.22 (95CI: 1.19-1.25) and 1.41 (95CI: 1.36-1.45), respectively. This higher referral risk in patients with low or middle socioeconomic status is statistically significant in almost all clusters. Comorbidity is also a statistically significant risk factor for referral from primary care, with the OR decreasing by about half in the no comorbidity category relative to the one comorbidity category, and increasing to OR of 2.5 or more in the category with ≥3 comorbidities. Figure S2 in Supplementary Material illustrates this association, which was significant, albeit smaller than that observed for hospitalisations.

The main finding of this study is that psychiatric comorbidity has an enormous association on the need for mental health hospitalizations in children, adolescents and young adults in the post-pandemic peak era. A gradient according to comorbidity category was observed, describing a profile of young patients with multiple severe disorders and an average age of around 22 years. The group with ≥3 comorbidities represented 2.78% of the total population studied, accounting for 58.78% (445 out of 757 hospitalized cases), which translated into a huge OR of 42.2. Differences in mental health referrals (OR: 3.58; 95CI: 3.42-3.76) suggest that crises in these patients are managed through hospitalisation rather than outpatient care. There are several reasons why the pandemic had a much greater effect on those who were more vulnerable, i.e., those with more psychiatric comorbidities. Firstly, it disrupted the continuity of the outpatient clinical management of psychiatric patients in general.22 Secondly, this group was less resilient to the stress generated by restrictions on social mobility, making them more susceptible to decompensation.23 Thirdly, stigma surrounding mental health issues can hinder sociability and encourage the use of social networks, potentially increasing suicidal and self-harming behaviours among young people.11

The use of mental health resources due to decompensation in adolescents and young patients with a diagnosed mental disorder differed significantly before and after the COVID-19 pandemic as a function of the type of resource used and mental disorder cluster. The risk of hospitalisation increased significantly in 2021 and 2022 compared to 2019, following a decrease in 2020. In contrast, the total number of referrals to psychiatry by primary care physicians remained stable in 2021 and 2022, although they increased in some clusters (self-harm, eating disorders, substance use and ADHD). Two other notable findings are the higher risk of decompensation in low- and middle-socioeconomic status clusters, and in patients with psychiatric comorbidities, especially given the size of this cluster in the general population (2.76%). We hypothesise that the mental health situation continued to deteriorate during the pandemic, amplifying existing trends. Our results suggest that young people with a history of mental disorders experienced more decompensations in response to post-COVID-19 stress, especially those with more severe conditions.

To our knowledge, this is the first population-based study to show that the impact of the pandemic on mental health care was more marked in hospital-based care than in community care. A study measuring the incidence of hospitalisations in the Canadian population aged up to 20 years found similar increases in anxiety and personality disorders, as well as suicide in women, and eating disorders in both sexes, during the period of high SARS-CoV-2 infection prevalence, but an overall decrease in mental health hospitalisations between the pre- and pandemic periods.8 Another two time series studies analysing data from French population also found significant increases in post-pandemic psychiatric resource use, including hospitalizations, among young people with a history of mental health issues.11,24 Both studies highlighted the potential role of social media in increasing the risk of decompensation in adolescents and girls due to their greater usage and subsequent higher exposure to interpersonal stress.25 Similarly, our results also indicated a higher risk of hospitalisation among females than males, and among adolescents than young adults. The finding of declining referrals during the peak of the pandemic is consistent with previous literature;26 however, unlike other studies, we did not observe an increase in referrals in the post-peak period of the pandemic.17,27

Notably, the self-harm cluster showed the greatest and most consistent increase in the probability of decompensation in the post-peak years of the pandemic, in terms of both hospitalisations and referrals. This finding is consistent with the 25% increase in suicide attempts in Catalonia between September 2020 and March 2021,28 and the 50% increase in medical emergencies involving suicidal ideation and self-harm in the young US population between 2019 and 2022.29 Another cluster characterised by increased use of resources was eating disorders.

The clusters of psychosis, depression, anxiety and conduct disorders behaved similarly to the overall sample of patients with a history of psychiatric illness in that they showed an increase in hospitalisations but not referrals. Previous studies have also reported increases in acute hospitalisations for psychosis during the pandemic in the British and Australian populations30–32 and in young people in France for any mental disorder.11,17 These results contradict the trend observed over the past 22 years of a general decline in psychiatric hospital activity.33,34

Inequalities in the incidence and prevalence of mental disorders by socioeconomic status in our population have been well documented.15,16 Analysis by socioeconomic status raises the question of why, in the period 2019–2022, people with low socioeconomic status had the same risk of hospitalisation as those with high socioeconomic status, but were more likely to be referred from primary care. On the other hand, statistically significant differences in lower risk were only found for the middle socioeconomic status group for total hospitalisations. One possibility is that the noted inequality is manifested in the incidence rate, but not in the probability of hospitalisation. The explanation for the differences in the detection of decompensation in primary care is simpler. The gradient in referral risk may be due to 20% of the population having dual health coverage (public and private), and people with higher incomes being those who are most likely to take out private insurance and therefore seek treatment from private providers.35 Along similar lines, the lack of data on private sector resource use could mean that our figures underestimate psychiatric hospital admissions. Nevertheless, although private psychiatric consultations are common, there are few private psychiatric hospitals in the Basque Country, and we do not believe that the use of these resources explains our results.

The increase in hospitalisations following the peak of the COVID-19 pandemic indicates a deterioration in the clinical course of children, adolescents and young adults with a previous mental health diagnosis. While the causes may be directly or indirectly linked to the pandemic, the indicators suggest the need for more intensive monitoring, especially of patients with more psychiatric comorbidity.

The owner of the data is the Basque Health Service, which maintains a policy of restricted data access for ethical reasons. Our agreement with the Basque Health Service establishes that individual-level data cannot be shared publicly because of confidentiality. Applications for access to the dataset must be made with the approval of the Basque Health Service (https://www.osakidetza.euskadi.eus/servicios-para-profesionales/webosk00-sercon/es/).

Cristina Candal Pedreira.

The corresponding author, on behalf of the other authors guarantee the accuracy, transparency and honesty of the data and information contained in the study, that no relevant information has been omitted and that all discrepancies between authors have been adequately resolved and described.

L. Mar-Barrutia and A. González-Pinto conceived and designed the research. J. Mar, S. García de Garayo, and M. Gascón obtained the data and carried out the statistical analysis, interpreted the data and drafted the methods and results section. L. Mar-Barrutia, A. González-Pinto and J. Mar drafted the manuscript and approved the final version. A. García, J. P. Chart-Pascual and G. Cano-Escalera reviewed the literature, provided text to the introduction and discussion sections of the manuscript and approved the final version. All authors revised the manuscript for important intellectual content and approved the final version. Further, they all had full access to all the data used in the study and accept responsibility to submit for publication.