Time trends in adolescent school absences and associated bullying involvement between 2000 and 2019: a nationwide study

Education is a central determinant of adolescent health. School absences and bullying involvement jeopardize educational attainment. We analyzed time trends in school absenteeism over two decades and examined the association of absenteeism with bullying involvement. We analyzed data from the nationwide School Health Promotion study, with self-reported data from Finnish middle school students in grades 8 and 9 (ages 14–17, N = 1 000 970). Questionnaires assessed frequency of illness absences (IA), truancy, frequency of bullying victimization (BV), bullying perpetration (BP), and involvement in both bullying perpetration and victimization (BPV). Frequent school absences were de�ned as occurring at more than 3 days during the prior month (2000–2015), or least weekly (2017–2019). Frequent IA increased from 12% in 2000 to 22% in 2015. In 2017–2019 frequent IA was reported by 3.5%. Frequent truancy declined from 9% in 2000 to 4% in 2015, and 2017–2019. BV was reported at least weekly by 6.9%, BP by 5.4% and BVP by 1.9% of participants. In a logistic regression model, all kinds of bullying involvement increased odds for both IA and truancy. Since bullying involvement was associated with both IA and truancy, particular concern should be raised for adolescents involved in bullying, and for their social and educational functioning. The concurrent increase in IA and decrease in truancy may re�ect destigmatization of mental health problems or other changing patterns in reporting absenteeism.


Introduction
School attendance is one of the key determinants of healthy development in children and adolescents (Heyne et al., 2019;Kearney & Graczyk, 2020). Absenteeism is associated with poor academic performance, social isolation, mental health problems, increased risk for substance use, and future unemployment (Heyne et al., 2019). The reasons for absenteeism are heterogenous and often complex (Heyne et al., 2019).
School absences are often divided into legitimate/excused and illegitimate/unexcused absences (also referred to as truancy). Most absences are legitimate (Kearney, 2016). A majority of them are due to common illnesses (illness absences, IA) veri ed by either a caregiver or medical professional, while some reasons for school absence are approved by the caregivers and school (e.g., participation in family or sports events). Truancy refers to absences that neither the caregiver nor the school have approved (Heyne et al., 2019).
Absence rates vary as a function of school system (Keppens & Spruyt, 2018.), culture (Kearney & Graczyk, 2020; OECD, 2020), operationalization and informant (Keppens et al., 2019). In a cross-sectional study, 32-35% of 14-to-15-year-old Dutch students reported IA during the prior 30 days (Eaton et al., 2008). In Sweden, recurring IAs were reported by 9.5% of students in middle school (Brolin Låftman et al., 2020). In the Programme for International Student Assessment (PISA) report from 2019, 21% of 15-year-old students internationally, and 13% of students in Finland reported truancy at least once during the previous two weeks, boys more often than girls (OECD, 2020). Severe truancy, i.e. more than three times during the previous two weeks, was reported by 2.5% of high school students in Finland (Keppens & Spruyt, 2018).
Large population studies on temporal changes in the frequency of legitimate and illegitimate school absences are infrequent. A recent government report using register-data in Scotland found no temporal change in IA in secondary schools, rates being 4.3% in 2010, and 4.4% in 2019, reported as % of total annual school time (Scottish Government, 2022), but truancy showed a slight increase from 2.1% in 2010 to 2.9% in 2019. In the 2019 PISA report, truancy increased by 1% from 2015 to 2018 internationally (OECD, 2020), yet in Finland a 23% decline in self-reported truancy occurred concurrently (OECD, 2020). In the United States, self-reported truancy rates remained stable at approximately 11% among adolescents aged 12 to 17 between 2002 and 2014, with higher rates among females, older adolescents, and among Hispanic adolescents (Maynard et al., 2017).
The present study aims to examine temporal changes in the prevalence of IA and truancy among adolescents in a nationally representative sample gathered over a period of 20 years. The second aim was to study the associations of IA and truancy with BP, BV and BVP.

Procedure
The study is based on a large, biannual survey, The School Health Promotion study (SHP), conducted by the Finnish Institute of Health and Welfare (THL). All students in grades 8 and 9 in Finland are invited to participate during a school day in the spring term. Participation is voluntary and anonymous. Prior to 2013, the study was implemented in Southern, Eastern and Northern Finland in even-numbered years and in Western and Central Finland in odd-numbered years. After 2013, data has been collected from the entire country biannually and students have had the possibility to respond electronically. Previously, and also as a complement after 2013, a paper version of the survey was used. Annual response rates varied between 63-84%. In 2015, technical problems led to a lower response rate. The typical age for 8th and 9th graders is 13-16 years, therefore responses with age below 13 or over 17 were excluded, as nonserious replies. The study was approved by the THL Working Group on Research Ethics.

Demographic and socioeconomic background
The survey includes questions on several demographic and socio-economic background variables. Demographic variables in the present study are gender (male, female), grade (grade 8 or 9), and maternal educational level(1 = comprehensive school or equivalent, 2 = upper secondary school, high school or vocational education institution, 3 = occupational studies in addition to upper secondary school, high school or vocational education institution and, 4 = university, university of applied sciences or other higher education institution).

Illness absences and truancy
Self-reported IA and truancy were measured each year. From 2000 to 2015, the following item was used: "During the past 30 days, how many days have you been absent from school due to truancy/illness/other reason?". Responses were categorical: none, one day, 2-3 days, more than 3 days. In 2017 and 2019, the question read: "During this school year, how often have you experienced the following: being late, being absent without permission: skipping school, or being absent due to illness?". The response options were: not at all, a few times in the year, every month, every week, daily or almost daily.
To facilitate analyses across the whole study period, frequency of IA was coded into three categories: infrequent (none or one day per month, or not at all or a few times in the year); moderate (2-3 days per month, and every month); or frequent (more than 3 days per month, or every week, daily or almost daily during the last school year). Truancy was also coded into three categories: infrequent (none and not at all) moderate (maximum 2-3 days; and a few times per year or every month) and frequent (more than 3 days and every week, daily or almost daily).

Involvement in bullying
Bullying involvement items were derived from a World Health Organization study on youth health (King, 1996), based on Olweus' de nition of bullying (Olweus, 2013). First, a de nition of bullying was provided: "We say a student is being bullied when another student (or group of students), say or do nasty things to him or her. It is also bullying when a student is being teased repeatedly in a way she or he does not like.
But it is not bullying when two students of about the same strength quarrel or ght." BV was then measured with one item: "How often have you been bullied at school during this semester?". BP was measured with one item: "How often have you participated in bullying other students during this semester?". Response options for both items were: several times a week, about once a week, less frequently, not at all.
Students who reported both BV and BP at least once a week were coded as BVP for the analyses of associations; however, this category was not included in the time trend analyses.

Data analyses
All analyses were conducted using SAS 9.3 software. Descriptive data are presented as frequencies, and cross-tabulations with chi-square statistics. In the Spearman correlational analyses, to analyse if the strength of the association between bullying and absence had changed over time, gender, grade, and maternal education level were included as covariates. We used cumulative logistic regression (LR), which yields estimated adjusted odds ratios (OR) with 95% con dence intervals (CI), to model the effects of gender, age, maternal education, study year, bullying involvement on IA or T.
Over the entire study period, frequent IA was reported by 12% of students, moderate by 20%, and 32% of students reported infrequent/no IA (Table 2). Frequent and moderate IA were somewhat more common among girls over the entire study period (χ² = 807,15, p < 0,001). 18% reported some truancy, and 4% of students reported frequent truancy. Truancy in general was more common among girls, whereas frequent truancy was more common among boys (χ² = 496,01, p < 0,001).
During the entire study period, seven percent of students reported BV once a week or more often (Table 2). BV was more frequent among boys than girls (8% vs. 5%. χ² = 5540.09. p < 0.001; Table 2). Boys also reported BP more often than girls (9% vs. 3% weekly level χ² = 47893.86. p < 0.001). Boys reported BP more often than BV, whereas girls reported more BV than BP. Two percent of the study population reported BVP, with 3% of boys and less than 1% of girls reporting at least weekly occurrence (χ² = 5343.71. p < 0.001). Note. IA: infrequent (none and 1 day. and not at all; and a few times per year). moderate (2-3 days and monthly) or frequent (more than 3 days or weekly; and daily or almost daily). Truancy: infrequent (none and not at all) moderate (1 day or 2-3 days; and a few times per year or monthly) and frequent (more than 3 days and weekly; and daily or nearly daily). BV = bully victimization. BP = bullying perpetration. BVP = bully victim-perpetrator. All gender differences ( 2 were signi cant at level p < 0.001. Time trends in illness absences and truancy χ Figure 1 depicts the absence rates for each study year. Overall, IA increased between 2000 and 2015, with a pronounced increase in frequent IA for both boys (from 12-21%) and girls (from 12-23%; Fig. 1, Panel A). In 2017-2019rates of frequent IA were 3.4 − 4.3%. Truancy rates declined between 2000 and 2015.
Frequent truancy declined for all students from approximately 8-4%, and moderate truancy for boys (grade 9 from 23-13%) and girls (26% to15%; Table 4, Fig. 1, Panel B). Correlations between illness absences, truancy, and bullying The correlations between BV, BP and BVP and school absences were small (Table 3). Correlations were slightly higher between truancy and BP than between the other variables. The correlation between truancy and BV tripled over the study period (from 0.04 to 0.12), but was still small in size, when controlling for age, gender and maternal education.

Results from multivariate analyses
In the logistic regression, odds for IA were higher for older students (grade 9 compared to grade 8), females, years 2006-2015, and any type of bullying experience (Table 4).
Older age, female gender, and any type of bullying experience were associated with increased odds for truancy.
Both BV and BP were associated with both IA and truancy. The odds for absence increased by 45% if BV was reported several times a week. When BP was reported several times a week, odds were higher for truancy (OR = 8.75), and also signi cantly higher (OR = 1.85) for IA.

Discussion
School absenteeism is an increasing concern worldwide. However, the suggested increase has not been , adolescents may be more prone to disclose their emotional di culties to their parents. Parental attitudes towards illness absences due to these reasons may also have changed. Indeed, "mental health days", meaning taking a day off for rest from school, are often referred to on social media, or in everyday language of adolescents. As parents/caregivers give permission to these days, both parents and adolescents may consider them as IA. Thus, what once was considered truancy might now be viewed as IA due to a shift in mindset. In 2017-2019, 3-4% of students reported weekly IA. Monthly IA was reported by 18-24% of students. What appears to be a decline is likely explained by the change in wording of the item. Also,weekly absences are understandably less common compared with more than 3 days in a month, due to the fact that the latter may also capture periods of u for instance, with consecutive absence days. Another possible explanation is, that a markedly increased number of adolescents were not at school during the time of data collection, which would explain the big drop in occurrence. Such a drastic change seems unlikely.
The truancy rate found in the current study is in line with that reported in other studies: 13% in Finland reported by the OECD (OECD, 2020), with a 23% decline in self-reported truancy in Finland between 2015 and 2018 (OECD, 2020). The present data also shows a declining trend in truancy levels during the study period. This decline may be explained by the implementation of anti-bullying programs, which have been widely implemented from 2007 onwards (Salmivalli et al., 2011). Other preventive programs aiming to increase wellbeing and a new curriculum based on positive pedagogy has also been implemented during the study period (Finnish National Agency for Education, 2022). The use of electronic student management systems to track absences has increased nationally in Finland during the study period, with both parents and teachers reporting absences in the same system.
Associations between absenteeism and BP, BV, and BVP were clearly established. Odds increased most for truancy, when BP occurred. This nding suggests a need for broad support measures for students with disruptive behavior and school absences. The results showed an increase, albeit small, over years in the correlation between BV and truancy, when controlling for covariates such as gender, grade and mother's educational level. This increase should be taken seriously, as the consequences of both BV and absenteeism might be severe. The increase may be interpreted as a tendency for bullied students to increasingly also skip school, or for students who skip school to be bullied, or a third factor, such as mental health problems, to increasingly contribute to both. Also, the strong connection between truancy and BP should be considered, since adolescents with such externalizing behaviors could bene t from a holistic assessment and support The strengths of the current study were the large, nationally representative cohorts and recurring items in the survey. The School Health Promotion study reaches a large proportion of Finnish adolescents. Adolescents with excessive absences may, however, have been absent from school on the survey day. In conclusion, we found signi cant temporal changes in both IA and truancy in Finland between 2000 and 2019. Rates of IA increased signi cantly during most of the study period, whereas truancy rates declined during the time period. However, a relatively stable 3-4% of students reported frequent truancy.
Both IA and truancy were associated with bullying involvement. A follow up on absenteeism during and after the COVID-19 pandemic would be important, as it could show even higher IA, due to increased mental health problems, as well as a change in the thinking about when IA is needed.
Declarations Figure 1 Time trends in absences. Panel A Illness abence separately per gender and grade, adn Panel B, truancy separately per gender and grade.