Effects of Covid-19 confinement on the mental health of children and adolescents in Spain

In Spain, in order to control COVID-19 transmission, one of the strictest confinement measures in the world for children and teenagers has been implemented. From 14 March to 26 April 2020 underage Spaniards were not allowed to leave their homes, except for reasons of force majeure. This could have consequences on their mental health in both the short and the long term. Thus, the aim of the present study was to explore the consequences of confinement on the mental health of Spanish children and teenagers, at the time when minors had been locked down in their homes between 8 and 10 days. The sample was composed of 590 confined Spanish children and teenagers between 8 and 18 years old. The scales of Depression, Self-esteem, Anxiety, Problems with Emotional Regulation, Rage Control Problems, Integration and Social Competence, Somatic Complaints, Rebellious Behaviour, as well as Awareness of the Problems of the Assessment System for Children and Adolescents (SENA) were used. The results revealed that, during confinement, children and adolescents showed emotional and behavioural alterations. This study, as far as we know, is the first one to explore the psychological consequences of lockdown in minors while it was taking place, with them being the ones directly assessed.

www.nature.com/scientificreports/ quarantine with the scales of the technical manual of the SENA 37 (Assessment System for Children and Adolescents), which is the resource used in the present study to evaluate the state of mental health of minors. The second goal is to investigate if confinement has affected the levels of anxiety, depression, self-esteem, rebellious behavior, somatic complaints integration and social competence, rage control problems, as well as greater emotional regulation problems in children or teenagers. Liu et al. 22 found different impacts of the pandemic on adolescents than on children, with adolescents reporting worse mental health. Something similar is found by Zhou et al. 3 , although they compare college adolescents with elementary school children. However, Ravens-Sieberer et al. 28 find that younger children are more affected by the pandemic than adolescents.
Lastly, the third goal is to differentiate between the mental health of children and adolescents based on gender and level of education, since these were some of the main risk factors for anxiety and depression in Chinese adolescents 3 . Regarding gender differences, the results are controversial. Some studies find that female children and adolescents report worse mental health during confinement 3,28 . But others find no gender differences on the mental health of confined minors 22,23 .
Our main hypotheses were: Hypotheses 1: The sample of confined children and adolescents presents worse mental health than those who were not confined.
Hypotheses 2: Confined adolescents present worse mental health than confined children. Hypotheses 3: Female children and adolescents present worse mental health than male children and adolescents.
Hypotheses 4: Children and adolescents in courses with highter academic pressure present worse mental health than children and adolescents with less pressure.

Methods
Participants. The sample consisted of 788 minors, 440 of whom were between 8 and 12 years old and 348 between 13 and 18 years old. After removing the participants that had not answered 100% of the raised items, 590 minors were left, 325 between 8 and 12 years old and 265 between 13 and 18 years old (teenagers). It is an incidental sample, instead of a random one. The selection of participants was carried out through an anonymous survey conducted on the Internet, social networks and the press.
In the case of the children (between 8 and 12 years old) 158 were girls (51.7%) and the average age was 9.95 years old (SD = 1. 40). Regarding the place of residence, they were from all the Spanish communities; the most represented was Castilla y León, where more than half of the participants lived (71.75%), followed by The Basque Country (7.7%) and Madrid (6.5%). A total of 81.5% lived in urban areas and 43.4% were studying the intermediate level of primary education. 25.8% had no siblings, 54.5% had one sibling and 19.7% had two or more siblings. At the time of confinement, 17 children (5.2%) lived with only one person in their home, 82 with two people (25.2%), 166 with three people (51.1%) and 60 with more than three (18.5%). 20.9% had a garden in their home, 39.1% had a terrace or balcony and 39.7% had neither. The average square meters of their home was 113.2 (SD = 71.1). 17.8% had a dog in their home.
In the sample of the teenagers (between 13 and 18 years old), 168 were female (63.4%), with an average age of 15.42 years old (SD = 1.70). They also came from all the Spanish communities, being Castilla y León the most represented again (55.8%), followed by Andalucía (10.9%) and Madrid (6%). 66.0% lived in urban areas, 57.7% studied compulsory secondary education, 33.6% high school and 8.7% vocational training. 19.6% had no siblings, 58.1% had one sibling and 23.3% had two or more siblings. At the time of confinement, 19 adolescents (7.2%) lived with only one person in their home, 59 with two people (22.3%), 142 with three people (53.6%) and 45 with more than three (17.0%). 23.4% had a garden in their home, 42.3% had a terrace or balcony and 34.3% had neither. The average square meters of their home was 119.0 (SD = 67.0). 22.3% had a dog in the home.
The inclusion criteria were: 1. To be between 8 and 18 years of age 2. Parents or legal guardians consent to their children's participation The exclusion criteria were: 1. Not to be between 8 and 18 years of age 2. Parents or legal guardians do not consent to their children's participation.
Instruments. Assessment System for Children and Adolescents 37,38 (SENA) is a resource focused on the detection of a wide range of emotional and behavioural disorders from 3 to 18 years old. It allows one to detect areas of vulnerability that predispose people to develop psychopathological problems and the presence of psychological resources that can act as protective factors to different problems. In our study, self-reporting questionnaires aimed at Elementary (6-12 years old) and High School (13-18 years old) students were used, and some of them were chosen so that they did not take too much time, keeping in mind the state in which the minors were. www.nature.com/scientificreports/ people who listen to me"). It shows a response format of a 5-point Likert-type scale. Both children and teenager scales were reliable, producing α Cronbach internal consistency rates between 0.754 and 0.914. They present scores in their technical manual on both normal population and clinical samples (anxiety, depression, learning difficulties, ADHD, etc.). SENA has shown suitable psychometric traits referring to evidence of validity, as well as test-retest reliability and internal consistency on both normal and clinical population 38 . In addition, the SENA has been related as expected to a wide variety of assessment instruments such as the BASC-2 (Behavior Assessment System for Children 39  None of the variables show relevant influence of gender, age, socioeconomic status of the family or the educational level of the parents 37 . Despite the fact that the evaluation tool is validated from 3 year olds on up, in the present work the sample for 8 year olds on up has been chosen, so that the children were able to adequately answer the questions asked them on their own, without adult collaboration 38 . In addition, for younger children (7 year olds or younger), the SENA assessment is conducted in an interview format, not self-report 37 .

Procedure.
A cross-sectional study was conducted through surveys between 22 and 25 of March 2020 (after 8-10 days of confinement at home) by means of an anonymous survey carried out on the Internet, social networks and press, using the onlineencuesta platform (https:// www. onlin eencu esta. com/), which took approximately 15 min. To try to avoid sample bias, we sent the evaluation questionnaires to multiple destinations and through very different platforms, in order to impact as many people as possible and with as much variability as possible. At first, we sent it, for example, to national news agencies, national and regional newspapers that published the links on their websites, to more than 20 "educational influencers" (primary or secondary school teachers with websites or Youtube channels about education with a large number of followers) some of whom disseminated it in their media and social networks, to more than 500 students from four different Spanish universities, or to five sports clubs in three different cities. In all of them, we urged people not only to have their children participate, but also to disseminate it to as many people as possible.
The investigation was sanctioned by the Research Ethics Committee of the University of Burgos. All parents or legal guardians gave voluntary informed consent to the participation in the study, which was explained to them as an investigation of the effects of confinement on the physical and emotional condition of minors. The study complied with the ethical criteria on research with human beings (informed consent and the right to information, personal data protection and guarantees of privacy, non-discrimination, gratuity and the possibility of leaving the study at any stage).

Data analysis.
First, we analyzed the percentage of children and adolescents scoring moderately high (between one and two standard deviations above the mean) and high (more than two standard deviations).
Then, we conducted a one-sample Student's t test to compare the confined children and adolescents with the SENA validation sample (Hypothesis 1). Effect sizes were estimated using Cohen's d (low effect size: d = 0.2, medium: d = 0.5, high: d = 0.8) 46 .
Next, a MONAVA was carried out to compare the influence of confinement on the mental health of children vs. adolescents (Hypothesis 2).
Finally, individual ANOVAs were conducted for each of the mental health dimensions assessed to establish differences between children and adolescents, as a function of gender and school year (Hypotheses 3 and 4). For the school year, the post-hoc tests were DMS tests. Effect sizes were estimated using Partial eta squared (low effect size: η p 2 = 0.01, medium: η p 2 = 0.059, high: η p 2 = 0.14) 46 .

Results
Prevalence. The correction criteria of the technical manual of SENA 37 are used in the study of prevalence. Table 1 presents the number and percentage of Spanish children and teenagers in strict confinement who are between 1 and 2 standard deviations above average, more than 2 standard deviations above average, and the ones that scored higher than the clinical sample of SENA scales. In the aspects of self-esteem, integration and social competence and awareness of the problems, the distance in standard deviations is below average, and those participants who score lower than the SENA clinical sample are also shown. Following, the results of the consequences of confinement are described first on children (8-12 years old) and then on the teenager sample (13-18 years old).

Consequences of COVID confinement in children between 8 and 12 years old.
First of all, Student t-tests are carried out as a sample in which the one of our study is compared with the technical standards of each of the SENA scales ( Table 2).
As can be seen, once the confinement week is over, there are already significant differences concerning the scales in every dimension besides self-esteem. In all of them, apart from somatic complaints, the clinical situation of children between 8 and 12 years old is significantly worse than the sample of the scales. They show more rage control problems and rebellious behavior with medium effect size. They also show more anxiety, depression, problems with emotional regulation, somatic complaints and worse integration and social competence with low effect size. www.nature.com/scientificreports/ The effect sizes found here are higher than the ones published in the technical manual of SENA 37 regarding the differences between the scores of normal population and the clinical sample in rage control problems (d z = 0.61 vs d clinical = 0.39) and rebellious behaviour (d z = 0.75 vs d clinical = 0.35).
Consequences of COVID confinement in teenagers from 13 to 18 years old. As can be seen in Table 2, after being confined for a week, teenagers show significant differences regarding the scales in all nine evaluated scales. In all of them, besides awareness of the problems, the clinical condition of teenagers between 13 and 18 years old is significantly worse. All effect sizes are low.
The effect sizes of the present study are greater than the ones published in the technical manual of SENA 37 regarding the differences between the scores of the normal sample and the clinical one in Anxiety (d z = 0.

Differences between children and teenagers in Covid-19 confinement.
In relation to the second goal, in order to compare data on children and teenagers, the scores of all the evaluated dimensions were standardised taking the average and standard deviation of the SENA scales for children (8-12 years old) and teenagers (13-18 years old). Then, a MANOVA was carried out with all the mental health variables set out in the study and shared with both groups, that is, all of them except for awareness of the problems.
The differences turned out to be significant (Pillai's trace = 0.157, F = 13.551, p = 0.0001, η 2 = 0.157) with a high effect size. The inter subject tests can be seen in Table 3. Teenagers show higher levels of anxiety, depression, problems with emotional regulation and somatic complaints, as well as lower self-esteem and integration and social competence levels than children. As for children, their sample reaches higher scores in rebellious behaviour than teenagers' . All effect sizes are low, except in Rebellious Behaviour where the effect size is medium.   www.nature.com/scientificreports/

Differences by gender and level of education in children between 8 and 12 years old. Regard-
ing the third goal, ANOVAs of a factor with SENA dimensions based on gender and educational cycle were carried out (Table 4). There are not significant differences regarding gender in any of the nine evaluated variables in children who are in quarantine. According to the educational level, the youngest of the middle elementary level revealed greater problems of emotional adjustment than those of the upper level (F = 9.061, p = 0.003, η 2 = 0.028) with a low effect size. www.nature.com/scientificreports/

Differences by gender and level of education in teenagers between 13 and 18 years old.
Regarding gender, the results show that there are significant differences between teenagers (Table 5), with females presenting higher levels of anxiety, less self-esteem, more problems with emotional regulation and more somatic complaints than their masculine counterparts. Then again, males show a significantly lower score on integration and social competence than females. All effect sizes are low.
Regarding the level of education, there are significant differences among anxiety, depression, self-esteem, and awareness of the problems with a medium effect size, and also in problems with emotional regulation, and somatic complaints with a low effect size. In all of them, teenagers who study in ESO show a better clinical condition that the ones who study in high school and professional training, except for the scale of awareness of the problems (Table 6).

Conclusions
The results of the present report show that the strict confinement situation of children and teenagers already reveal, from 8 to 10 days, significant consequences on the mental health of both of them, although we still do not know the long-term effect (Hypothesis 1). It appears that the consequences of confinement on children are mostly in the affective area, this also being reflected at the behavioural level. They show problems of rebellious behaviour (d z = 0.75), rage control (d z = 0.61) and emotional regulation (d z = 0.27) to a greater extent. As opposed to adults, children do not clearly identify these altered conditions in themselves, and it is frequent that symptoms like irritability or aggression appear as a warning signal of more chronic disorders for this age group. We have also discovered that during confinement they showed higher levels of anxiety (d z = 0.14), depression (d z = 18), and less integration and social competence (d z = 16), although with lower effect sizes. However, it should be pointed out that such high percentages as 33.2% and 22.8% of the children in confinement score higher than the clinical sample of SENA on anxiety and depression respectively.
The only variable that revealed improvement in children during confinement was somatic complaints (t = − 5.571, gl = 324, p = 0.0001, d z = − 0.31). One possible explanation is that somatic complaints usually appear as an involuntary expression of psychological discomfort that children use to seek attention or affection from their parents. Franco, Pérez and de Dios 47 discover differences between parenting styles and the somatization of their children, proving that parents with less disciplined educational behaviour or situations in which one of the parents takes fewer caring tasks, increase the somatization in children. In our results, somatic complaints in children decreases, perhaps because parents are closer and spend more time with their children.
In the case of teenagers, there are significant differences regarding the scales of every evaluated variable (anxiety, depression, rebellious behaviour, somatic complaint problems with emotional regulation and rage control) being the effect size average except in awareness of the problems and somatic complaints, which is low.
In all the dimensions, the results show a worse clinical situation than in teenagers after being between 8 and 10 days in confinement, except in awareness of the problems (t = 3.50, gl = 264, p = 0.001, d z = 0.21), that is, they were more aware during confinement that some things were not going well, that they were having a difficult time, so they could have needed help (Hypothesis 2).
Regarding gender, the results point in the same direction as the ones found by Zhou et al. 3 in Chinese teenagers: Spanish girls between 13 and 18 years old have been more affected by confinement than boys (Hypothesis 3). They show more anxiety, less self-esteem, more problems with emotional regulation and more somatic complaints. As for male teenagers, they show lower levels of integration and social competence. Because of this, it seems they have greater difficulty being accepted and loved by others than females through the methods which social interaction in confinement allow (video calls, chats, phone calls, etc.). This might well be because teenage girls tend to use on-line means of socialization more, give more importance to them than boys do, and they affect their well-being more 48 . Consequently, they are more used to this socializing agent than boys, who use digital media more often for gaming 48 . Nevertheless, confinement does not seem to have a differential impact based on gender in children between 8 and 12 years old. www.nature.com/scientificreports/ Regarding the school year, Zhou et al. 3 discovered that teenagers during years with more academic pressure in students showed more anxiety and depression, since the COVID-19 outbreak interrupted their normal learning process. In our study the results are similar. Teenagers in high school (prior to examinations for University admission) showed lower levels of self-esteem and more anxiety, depression and somatic complaints than middleschool students. Professional training students also showed worse symptomatology (more anxiety, depression, problems with emotional regulation and less self-esteem) than the ones in middle school. Professional training students noticed how their internship came to a halt due to coronavirus, perhaps the part of their studies which they consider the most important and the one they value the most in this kind of education. Without a doubt, this meant a period of uncertainty to them since they did not know the terms and conditions in which they could carry out their studies. In short, the differences in teenagers' mental health took into account high school and professional training students, as well as those in middle school, a school age that does not imply great academic pressure in Spain (Hypothesis 4).
In children between 8 and 12 years old, there were only differences in problems with emotional regulation, where middle-school students reached a higher score than the ones in high school. This coincides with the study that indicates that emotional control grows with age and that not until they are 10 years old, are children mature enough to understand states such as emotional ambivalence (experimenting contradictory emotions in the same situation) 49,50 . This would doubtlessly make emotional understanding and the ability to adequately face the confinement situation difficult for young children.
The present report represents a first approach to the consequences of confinement on Spanish children and teenagers, which, nevertheless, must be interpreted in the light of some limitations. First of all, the selection of the sample was for the sake of convenience and following snowball sampling. However, in future confinement situations, a probability sampling should be obtained, since the sample of the present study is not representative of Spanish children and adolescents, and therefore, special caution should be exercised in the interpretation of the results. Besides, the evaluation was transversal, which prevents establishing causal relationships. It should also be noted that the results displayed here only comprise the consequences of the first 8-10 days of confinement. This is the research, as far as we know, with the earliest evaluation in the confinement of minors. And from what we have found here, it seems that its effects are felt from the first week. Its long-term evolution still needs to be established. In this sense, the detection and intervention of children and adolescents at risk, when identified early, can help prevent the onset or worsening of psychopathologies 28 .
In short, looking at these results, it seems clear that strict confinement situations affect the mental health of children and teenagers between 8 and 18 years old. And it seems that the impact of confinement is different in children and adolescents, at least in the early stages of confinement. And, as far as we know, all previous research evaluating both age groups in Europe has done so jointly. Future research should study the mental health of minors according to their stage of development, as this could influence a different process of return to normality, different needs and vulnerabilities and/or the design of specific support programs.
Our results suggest a very early impact of strict home confinement on children and adolescents, which has implications for health policies. Countries which are less restrictive in their COVID-19 containment measures, such as Germany, appear to have received less impact on the mental health of minors 28 . In addition, the literature on the topic reflects that the COVID-19 pandemic affects young people with and without any underlying physical or mental health conditions, and that they experience feelings and emotions similar to those experienced by adults 16 . This indicates the need to carefully balance strict confinement measures with measures needed to curb the spread of contagious diseases. Our results-along the same lines as other recent studies in Germany, China or the USA 24,28,51 -make the need for specific mental health interventions for children and adolescents explicit. Future research should study prevention strategies for the deterioration of mental health and support in the promotion of mental health from the beginning of periods of confinement to cushion its impact.