During the COVID-19 pandemic, the announcement of France’s first period of lockdown (on March 14, 2020) raised concerns about a possible increase in the risk of child abuse [1]. Family relationships were inevitably modified by the almost permanent presence of all family members under the same roof (due to the closure of daycare centers, and with parents teleworking from home), the associated climate of fear, social isolation, and (in some cases) the financial consequences of COVID-19.

In France and some other developed countries, the numbers of reports to child welfare services and emergency department visits for child abuse started to fall when lockdown was introduced [2, 3]. Since levels of domestic violence were unlikely to fall during this uncertain period, the observed decreases led to concerns that abused children were not being identified and cared for [1, 4]. One of the most serious forms of abuse in children aged 0 to 24 months is abusive head trauma (AHT), which can result in severe injury or death [5, 6]. Subdural hemorrhage (SDH) is present in 89% of cases of AHT [7, 8]. Although some recent studies have identified increased child abuse during lockdown as a major public health concern, this phenomenon has not previously been quantified [1, 3, 4]. In view of the results of the above-cited studies, we decided to conduct a nationwide study of data in the French national hospital discharge database (Programme de médicalisation des systèmes d'information, PMSI) and focused on cases of subdural hematoma (SDH) in a context of ATH.

The primary objective of the present study was to compare the incidence of hospital admissions in France for child abuse with SDH during 2020 (the first year of the COVID-19 pandemic), relative to the previous two years. To further analyze the potential impact of lockdowns, we evaluated the incidences of SDH and child abuse for each calendar month.


We performed a retrospective observational study of data in the PMSI database. Children aged up to 24 months with a diagnosis of child abuse and/or SDH following hospital admission anywhere in France between January 1, 2018 and December 31, 2020 were included in the study. The PMSI database was searched with specific International Classification of Diseases, 10th Revision (ICD-10) diagnostic codes (Supplementary Data 1). For 2018, 2019, and 2020, we collected data on the total number of children aged up to 24 months admitted to hospital in France, the number of children admitted for child abuse, and the number of hospital admissions for child abuse + SDH. To better characterize the children hospitalized for child abuse + SDH, we examined the month-by-month number of hospital admissions for the three study years. We chose both 2018 and 2019 as comparator years because (in contrast to 2020) they did not include any periods of lockdown or travel restrictions (Supplementary Data 2). We noted the median age at the time of hospital admission, the child’s sex, the length of hospital stay, the number of admissions to an intensive care unit, and the number of deaths.

The year 2020 was considered to be the first year of exposure to COVID-19, and so the 2020 data were compared with control data from 2018 and 2019. The main exclusion criterion was hospital admission for SDH associated with a code for accidental trauma.

Statistical analysis

Data were expressed as the median (range) or the frequency (percentage). Intergroup differences in qualitative variables were probed using the chi-squared test. The threshold for statistical significance was set to p < 0.05. All statistical analyses were performed using GraphPad Prism 8 software (version 8.4.3 (471), GraphPad Software Inc., La Jolla, CA, USA). In order to maintain the patients’ anonymity, exact (individual) data were not available on the length of stay (for all patients) or the age (for patients aged 12–24 months); only median values could be extracted from the national database. Hence, we could not apply chi-squared tests to these two variables.


The numbers of hospital admissions of children under the age of 24 months, admissions for child abuse, and admissions with SDH + child abuse codes in 2018, 2019 and 2020 are given in Table 1. There was a nonsignificant trend toward a lower total number of hospital admissions per year for the 2020 (COVID-19) group, relative to the 2019 group (decrease: 6.4%) and the 2018 group (decrease: 7.6%) groups.

Table 1 Hospital admissions in France for child abuse with or without subdural hemorrhage in children under 24 months of age, 2018–2020

Of the 913,623 children under the age of 24 months hospitalized in 2018, 1137 were admitted for child abuse and 154 were admitted for SDH + child abuse at least once. In 2019, 902,556 children were hospitalized, 1184 were admitted for child abuse, and 150 were admitted for SDH + child abuse. In 2020 (the COVID-19 year), 844,509 children were hospitalized, 1131 were admitted for child abuse, and 138 were admitted for SDH + child abuse.

There were no significant differences in hospital admissions with codes for child abuse or for SDH + child abuse between the 2020 COVID-19 group and the 2019 and 2018 control groups (Table 1). The SDH + child abuse population (Table 2) showed male predominance in all three years. In all years, most of the children were aged below 12 months. There were significantly fewer hospital admissions in May 2020 than in May 2019 or May 2018, whereas there were significantly more hospital admissions in December 2020 than in December 2019 or December 2018 (Fig. 1).

Table 2 Characteristics of children with SDH + child abuse codes recorded during their hospital stay
Fig. 1
figure 1

The monthly distribution of the incidence of SDH + child abuse codes, showing a significant decrease in May 2020 and a significant increase in December 2020 (marked by an asterisk)

The median age of the children hospitalized in the 2020 COVID-19 group was 107 days; this was lower (albeit not significantly) than the value for 2018 (121 days) or 2019 (108 days). The hospital stay was longer (albeit not significantly) in the COVID-19 group (15.6 days) than in the 2018 group (13.9 days) or the 2019 group (14.1 days). The proportion of children with SDH + child abuse codes requiring admission to an intensive care unit was 22.1% in 2018, 24.7% in 2019, and 22.5% in 2020; the inter-year difference was not statistically significant.

The mortality rate for children with SDH related to child abuse was 0.8% lower in 2020 than in 2018 but this difference was not statistically significant (p = 0.9).


The present study focused on severe forms of child abuse — forms that are frequently associated with AHT and sequelae like SDH [7]. Our nationwide study showed that the number of children aged 0 to 24 months hospitalized for SDH in the context of child abuse was no higher in 2020 than in 2018 or 2019. The demographic profile of our “SDH + child abuse” group was in line with the literature data [8]. Recently, Maassel et al. analyzed the number of hospital admissions for AHT during the COVID-19 period recorded in a US nationwide database; their results were similar to ours [9]. We were surprised not to find an increase in 2020, given the presence of all the risk factors for domestic violence and child abuse during this period of uncertainty [10].

In the literature, a number of studies have highlighted the considerable psychological and psychosocial repercussions of pandemics or other major humanitarian crises, including greater child abuse [11, 12]. In contrast to our present results, Sidpra et al.’s single-center study found a 1493% increase in hospital admissions for AHT in 2020, relative to the three previous years [13]. Several other studies have evidenced increases in the number of hospital admissions for child abuse [14, 15]. The COVID-19 crisis has accentuated tensions between family members; due to confinement and other restrictions, exasperated and/or exhausted adults might be more likely to shake and injure a crying infant. Furthermore, a recent study found that children under the age of two were more likely to be abused when they were cared for at home [16].

Our present results are especially surprising when viewed against the sharp increase in calls to the national abuse hotline (an increase of 56.2% between March 18 and May 10, 2020, relative to the same period in 2019) [17]. This increase prompted the French government to quickly issue media alerts about the increased risk of domestic and child abuse (on television, in newspapers, on radio stations, on social networks, and in shopping malls, together with the implementation of additional national help lines).

However, there are several possible explanations for our present results. Firstly, severe injuries (including SDH) are usually easy to observe clinically and will always require hospitalization; in contrast, bruising, burns, and general violence may go unnoticed. The curfew and lockdowns may have had an impact on the number of consultations with healthcare professionals in general. Indeed, we observed a decrease in all-cause hospital admissions of children aged under 24 months in 2020. Hence, under-reporting might have masked an overall increase in child abuse [3, 9, 18]. Furthermore, some children admitted for SDH might not have been recognized as having suffered AHT. Secondly, the French government’s media campaign might have raised awareness of the risk of AHT among the population. Thirdly, the permanent presence of all family members might have prevented some acts; it is known that AHT is most often perpetrated by an adult alone with the child. This might also explain the decrease in ATH in May (when many adults were working from home) and the increase in ATH in December 2020 (corresponding to the end of the second lockdown and a decrease in working from home) [19]. Lastly, the closure of daycare centers and child care facilities might have reduced the number of reports of abuse and contributed to under-reporting. In Rey-Salmon et al.’s study in 2020, 317 cases of abuse (98.5%) occurred in a private home and only one occurred in a daycare center; the risk of abuse is higher when children are kept at home, as was the case during lockdown. The results of Rey-Salmon et al.’s study contributed to the concern about an increase in child abuse during the COVID pandemic [16].

A better understanding and early identification of risk factors and families’ personal and social resources are essential for avoiding the recurrence of abuse. This is shown by the fact that shaking was repeated (from 2 to 30 times, with an average of 10 times) in 55% of situations in which parents admitted to shaking their infant [20]. In 20% of cases in which an infant cried repeatedly, shaking occurred every days for several weeks [20].

The present study had several limitations, including reporting bias. The study required specific, accurate ICD-10-CM coding for physical child abuse, with a combination of a “child abuse code” and an “SDH code.” Even though healthcare professionals dealing with child abuse victims were especially aware of the need to detect cases of abuse and thus code them accurately, we did not observe an increase in the frequency of SDH codes. It was not possible to identify all the paraclinical examinations (e.g., imaging) performed during the stay. Furthermore, the month-on-month differences between the 2020 COVID-19 group and the 2018 and 2019 groups might have been due to chance. Lastly, the diagnosis of AHT is complex; this might explain the contradictory literature data [14, 21, 22].


Reassuringly, we did not observe a greater risk of hospital admissions for SDH in a context of child abuse during the first year of the COVID-19 epidemic in France (i.e., 2020). Our present data and other studies indicate that the early identification of risk factors (age, sex, etc.) might facilitate the early detection of at-risk situations for child abuse and that could result in SDH.