Pediatric ICU Admissions After Adolescent Suicide Attempts During the Pandemic

The data presented here are a subset analysis of a larger retrospective observational multicenter study.¹¹  Members of the German Society of Neonatal and Pediatric Intensive Care and heads of pediatric intensive care departments identified via the homepage of German Society for Pediatrics were inquired via E-mail to participate in the initial study. Inquiries for participation in the initial study were sent out twice between September of 2020 and February of 2021. The 37 centers participating in the initial study were asked to participate in the follow-up in June 2021.

Patients <18 years of age admitted to a German pediatric ICU because of accidents or injuries were eligible. Because no standardized hospital or PICU admission criteria exist in Germany, the decision for PICU admission was at the discretion of the attending physician or hospital. The observation period of the first wave was the first German lockdown (March 16 to May 31, 2020). The corresponding calendar periods of the years 2017 to 2019 served as reference period. For the second wave, the study was extended by the same period of 2021. For this study, we analyzed data on admissions for suicides of patients aged 12 to 17.9 years.

The diagnoses defining eligibility were S00–S99 and T00–T78 according to the German modification of the International Classification of Diseases, 10 Revision. S codes apply for trauma diagnoses and T codes apply for other injury types or damage from external sources. A case was considered a suicide attempt if any of the discharge diagnoses was “suicide attempt,” “acute suicidality,” or self-harm with “suicidal intention.”

Eligible patients were identified via the local hospitals’ medical controlling services.

2020: The first German lockdown came into effect on March 16, 2020 with school and day care closures. Recreational facilities including playgrounds were closed and people urged to reduce their contacts and stay at home. Group gatherings were prohibited. Performing individual sports outside or leaving the house was not prohibited or restricted to a specific distance from home or amount of time. Compared to other European countries like France, Italy, Spain, and the United Kingdom, the German lockdown was less strict and relied on voluntary participation of the population. Unlike in other countries, restructuring of the health care system was not applied to pediatric departments and pediatric intensive care units, which remained fully functional during the lockdown. From the beginning of May, the lockdown measures were gradually relaxed until May 31, 2020.

2021: The second German lockdown came into effect on November 2, 2020 and was designed as a “light” lockdown. It was repeatedly prolonged and gradually tightened until January 19, 2021. From then on, the restrictions were similar to those of the first lockdown described above with no predefined end. The lockdown ended at the end of May 2021.

Local investigators reviewed discharge summaries and entered anonymized clinical data online into a questionnaire hosted at Microsoft Office Forms 365 for institutional users. Alternatively, anonymized discharge letters were transferred to the principal study site (Department of Pediatrics I, University Medicine Essen) and entered by LW and KH (Fig 1). After completing collection, the raw data sets (2017 to 2020 and 2021) were downloaded as Microsoft Office Excel files, imported into SAS Enterprise Guide 8.4, data cleaning performed, the data sets merged, and statistical analyses performed.

We accessed the DIVI (German Interdisciplinary Association of Intensive Care and Emergency Medicine) registry of intensive care beds and extracted the total numbers of PICUs and PICU beds in Germany and the number of PICU beds provided by the study centers.

Some children’s hospitals collapse pediatric and neonatal intensive care capacities in their reports to the DIVI registry. These centers were ignored for calculations, because we assumed that this information is missing completely at random between participating and nonparticipating centers.

To compare our data on nonfatal suicide attempts (SAs) with the number of fatal SAs, we retrieved data on officially registered deaths from suicides via a special inquiry to the Federal Statistical Office of Germany (https://www.destatis.de/EN/Home/_node.html). At the time of inquiry, data were available for each month of 2020 and included cases aged 10 to 17.9 years. For 2021, only preliminary monthly data on the entire population including all ages were available.

No major migratory movements occurred during the study period within, into, or out of Germany. We extracted data of the end-of-year populations for each federal state and 1-year age group between 10.0 and 17.9 years of the years 2016 to 2021 from the homepage of the Federal Statistical Office of Germany (https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Bevoelkerung/Bevoelkerungsstand/_inhalt.html) to verify that the adolescent population was stable across Germany and within each federal state. From these data we calculated nation-wide midyear populations for 2017 to 2021.

For continuous variables we present the median with interquartile range and mean with 95% confidence interval (CI). For discrete variables, absolute and relative frequencies are given.

We estimated standardized morbidity ratios (SMR) for the observation periods based on age- and sex-stratified data. The years 2017 to 2019 served as reference period to calculate the expected number of cases for 2020 and 2021, respectively. The observed number of cases in 2020 and 2021, respectively, was divided by the expected number of cases. An SMR >1 indicates an increase and an SMR <1 a decrease of cases. For the calculation of 95% CIs, we employed the Poisson approximation in case of ≥15 events in the observation period and calculated exact CIs for <15 events.¹²  Three patients with diverse sex, 2 from the reference period and 1 from 2021, were excluded from SMR calculations because of the low case number.

For fatal suicide attempts, we calculated yearly incidence rates and monthly incidence rate ratios (IRR) on the basis of the corresponding midyear populations. No sex-specific information on fatal SAs were available, allowing only the calculation of overall rates. CIs for incidence rates were calculated as described above for SMRs and for IRRs as described by Rothman et al.¹³ 

SAS Enterprise Guide 8.4 (SAS Institute Inc, Cary, NC, USA) was used to perform statistical analyses and produce figures. SISA software¹⁴  was used to calculate exact and Poisson CIs for SMRs. Microsoft Office PowerPoint 2019 (Microsoft Corporation, Redmond, WA, USA) was used to create a map of Germany based on “Karte Deutsche Bundesländer (nummeriert).svg” by David Liuzzo (license: CC BY-SA 2.0 de, https://commons.wikimedia.org/w/index.php?curid=2594472).

The original study and its follow-up were approved by the ethics committee of the Medical Faculty of the University of Duisburg-Essen (20-9560-BO and 20-9560_1-BO). Local ethics committees of the participating centers additionally approved the study if required by local regulations. Patient informed consent for retrospective anonymized data were not required according to national legislation. Data entry and storage in Microsoft Office Forms 365 is in line with the General Data Protection Regulation of the European Union (Regulation [EU] 2016/679).

In the first wave, 37 (23%) of 159 German PICUs participated. Twenty-seven (73%) of the initial PICUs participated in the second wave, comprising 17% of German PICUs and 147 (18.4%) of 801 registered German PICU beds (Fig 2). Thirteen (48%) of the participating PICUs were located in University Hospitals. According to the DIVI registry, the proportion of University Hospitals among German PICUs across Germany is 33 of 159 (21%).

A total of 588 adolescents between 12.0 and 17.9 years was admitted across all studied periods (Table 1). Total admissions because of accidents or injuries declined from an average number of 100 per period in 2017 to 2019 to 74 in 2020 and increased to 214 in 2021 (Fig 3A). Alike, admissions after nonaccidental injuries declined in 2020 (SMR 0.52 [95% CI 0.32–0.81]) and then rose in 2021 (2.87 [2.36–3.46]), applying both for SAs (Table 2, Fig 3B) and nonaccidental nonsuicidal injuries (SMR 2020: 0.44 [0.14–1.03]; 2021: 2.73 [1.86 – 3.89]) (Table 2). Sex-specific differences were present during the first lockdown with a decline in SAs among girls but not among boys (Fig 4).

No fatal SAs were reported in the reference period but in 2020 (n = 2) and 2021 (n = 4). According to national suicide statistics, there was no increase in fatal SAs in 2020 compared to the reference period in adolescents aged 10 to 17.9 years (Table 3, Fig 5, Supplemental Fig 6, Supplemental Table 4).

This longitudinal multisite study across Germany found a dramatic increase in PICU admissions after suicide attempts in adolescents during the second lockdown in 2021, after an initial decrease during the first lockdown in 2020. Likewise, total PICU admissions related to trauma or injuries also increased in the 2021 lockdown after a decline in the 2020 lockdown. Suicide statistics provided no evidence that fatal SAs among adolescents increased compared to prepandemic years or during the course of 2020.

Studies on long-term effects of COVID-19 restrictions showed increasing mental health problems among children and adolescents during the course of the pandemic. Heads of child and adolescent psychiatric departments in Europe perceived the impact of the pandemic on mental health and psychopathology of children and adolescents as “medium” in 2020, with an increase to “strong” or “extreme” in 2021.¹⁵  More than 80% reported an increase in suicidal crises.¹⁵  During a 6 month-lockdown in Australia, mental health-associated presentations increased by 47% and presentations because of suicidality by 59% in the last months of the lockdown.¹⁶  In France, the number of suicide attempts was lowest during the first lockdown (March and April 2020) and highest at the beginning of the second lockdown (November and December 2020) with a 299% increase.¹⁷  Among adolescents in the United States aged 12 to 17 years, the mean weekly number of emergency department visits for suspected suicide attempts were 26.2% higher during summer 2020 and 50.6% higher at the beginning of 2021 than during the corresponding periods in 2019.¹⁸ 

Besides such critical peaks that were typically related to regional peaks of pandemic waves or excessively long lockdowns, a population-based registry on suicide attempts from Catalonia showed a 195% increase in girls’ suicide attempts between September 2020 and March 2021 compared to the 6 months before.¹⁹  An increase during the September to March period was also observed in the prepandemic year, but only by 54%.¹⁹  Our results do not allow the conclusion that the second observation period represents a critical peak related to the lockdown rather than a longer-term trend. However, they generally align with the above findings on SAs not admitted to intensive care and provide new evidence that also serious suicide attempts requiring intensive care treatment increased. The degree of increase for the severe cases captured by this study was similar to non-PICU cases with an SMR of 2.84 in our study versus 200 to 300% increases. This multisite cross-national study shows that adolescent suicidality has evolved as an increasingly serious public health problem that affects various pediatrics subspecialties and requires interdisciplinary treatment efforts in the acute and subacute phases.

An increase in suicidality during the pandemic may result from the interaction of preexisting risk factors such as depression, anxiety, substance abuse, other psychiatric disorders, previous psychological trauma, personality traits, and other stressors.²⁰  Yet, there is clear evidence from around the world that the pandemic exerts stress on adolescents. A minority of children and adolescents whose mental well-being improved during the pandemic provides important insight for potential interventions.²¹  These children and adolescents reported improved relationships, reduced loneliness and exclusion, reduced bullying, better management of school tasks, and more sleep and exercise compared to those who reported no change or deterioration.²¹  Increased efforts to cushion the effects of this global crisis on adolescents are urgently needed and should include measures to promote the above-mentioned protective factors.

Limitations of our study are the overrepresentation of university hospitals, which serve as transferal centers for severe cases, and the retrospective design that did not allow to homogenize PICU admission criteria between the centers. SAs were not verified by a standardized protocol, but based on discharge diagnoses. Further, only nonfatal SAs and those with a time lag between the suicidal action and death could be reported by the PICUs. The small 2020 case numbers in boys make the interpretation of the observed effects less certain compared to girls and the overall cohort. For fatal SAs, official statistics are only available for complete months of 2020 and include cases from 10 to 17.9 years compared to 12 to 17.9 years for the remaining analyses. Because of the limited observation period of 2.5 months per year, seasonal variations of PICU admissions and SAs were not captured by this study.

After an initial decline of trauma- and injury-related PICU admissions and suicide attempts during the first pandemic wave, there was a drastic increase in admissions and SAs in the same period of 2021. As the pandemic continues, preventive measures and psychosocial support must be offered to adolescents to prevent ongoing damage to their mental health.

CI

confidence interval

DIVI

Deutsche Interdisziplinäre Vereinigung für Intensiv- und Notfallmedizin

COVID-19

coronavirus disease 2019

IQR

interquartile range

PICU

pediatric intensive care unit

SA

suicide attempt

SMR

standardized morbidity ratio