Associations between academic achievement and internalizing disorders in Swedish students aged 16 years between 1990 and 2018

Rising rates of internalizing mental health problems or disorders such as anxiety or mood disorders among adolescents are a growing public health concern in Sweden and other high-income countries [1]. In Sweden, the increase in internalizing problems or disorders has coincided with declining academic achievements and rising rates of school failure in students [2], but possible links between these two worrying trends have so far not been investigated.

It is well known that academic achievement and internalizing problems or disorders are closely interlinked in adolescents. Internalizing problems and disorders can lead to a loss of energy and motivation, poorer self-confidence and withdrawal from school, and may impair learning by hampering students’ ability to focus [3‐5]. Accordingly, students with internalizing problems have on average lower achievements in school, including lower grades and test results and higher risks of school dropout [6‐13]. The causal association may also go in the other direction, such that poor academic achievements undermine self-esteem, generate stress as students’ struggle to keep up, and lead to worries and a sense of hopelessness about the future, thereby increasing the risk of internalizing problems [3, 7, 8, 14‐18]. Not least, school failure – understood as early school leaving, failure to obtain passing grades or meeting minimum academic standards, or failure to obtain the qualifications necessary to enter the next stage of the education system [19] – may be especially distressing due to the social stigma attached to failure and the fear concerning negative future consequences that school failure can bring about [20‐23].

Common to the above referenced studies is that they have investigated associations between academic achievement and internalizing problems within single cohorts of children. However, both schooling and diagnostic criteria and treatment for internalizing problems have undergone substantial changes over recent decades, and these changes may have altered how academic achievement and internalizing problems interrelate. A changing relationship over time would point to changes in the life situation of adolescents in need of further investigation, and to new or increased vulnerabilities in adolescence that need to be addressed by policy.

Most high-income countries have experienced continuous educational expansion, resulting in children spending more of their lives in school, having higher educational expectations, and being more dependent on their academic achievements for their prospects in life [24, 25]. There are also indications that schooling in many countries has become more performance-oriented, with intensified competition and more extensive testing or grading [26, 27]. In Sweden, the context of this study, stricter eligibility criteria and reforms of grading systems have contributed to an intensified focus on achievement in school as well as to higher rates of school failure, with Sweden now having the largest share of students in Europe that leave compulsory school without a certificate or diploma [28]. These developments may have contributed in making low academic achievement a more important risk factor for internalizing problems [29, 30], and may also have further undermined achievement among students with such problems who already struggle in school [31].

On the other hand, psychotherapeutic and, especially, pharmacological treatment for internalizing problems have become more available for adolescents, and the acceptance for seeking treatment has become greater [32]. These treatments, in turn, can improve academic achievement among children with internalizing problems [33]. Moreover, a well-being discourse has gained traction in education policy, emphasising that schools must cater for the full socio-emotional development of children [34, 35]. In Sweden, for instance, schools are since 2010 obliged to provide access to a student health team that should work towards health-promotion, and help students who struggle in school to succeed academically [36]. Both these developments may, in turn, have weakened the association between achievement and internalizing problems.

Existing research on temporal changes in the association between academic achievement and internalizing problems is scarce and fragmentary. A meta-analysis of primarily American research [15] found that the association between achievement and subsequent depression was somewhat, but not significantly, stronger in more recent studies. Swedish studies have found that low-achieving students have experienced a disproportionate decline in school-related well-being in recent years [37, 38]. No study to date has, however, systematically investigated associations between academic achievement and internalizing or other mental health problems across a longer time span.

Against this background, the aim of this study was to investigate the development of the cross-sectional associations between students’ academic achievement and anxiety or mood disorders in the last year of compulsory school (aged 16 years) between 1990 and 2018. We addressed the following research questions:

Research question 1: Has the relationship between students’ grade point average (GPA) and anxiety or mood disorders changed between 1990 and 2018?

Research question 2: Has the relationship between upper secondary school eligibility and anxiety or mood disorders changed between 1990 and 2018?

Register data on specialist psychiatric inpatient and outpatient care from the Swedish National Patient Register as well as on prescriptions of psycholeptic and psychotropic drugs from the Swedish Prescribed Drug Register were linked to data on students’ school grades and school failure from the Swedish Pupil Register. The data were accessed through the Umeå SIMSAM Lab [39].

Average academic achievement was measured as student’s GPA in school year 9 of compulsory school [12]. In Sweden, the 16, or in some cases 17, subjects with the highest grades are summed into an overall grade sum, which is then used to regulate access to upper secondary school. This grade sum corresponds to a grade point average, and we will hereafter refer to it as students’ GPA. GPA in school year 9 carry high stakes in Sweden, and low grades can constrain students’ future opportunities in life. Grades were standardized into percentile ranks within graduation cohorts [2]. Thus, each student’s score shows his or her GPA relative to the GPA of all other Swedish students that graduated the same year. This standardization ensures that the measure of achievement is not affected by changes in the grading system or by grade inflation [2]. Moreover, since students primarily compete with peers from the same graduation cohorts in the transition from compulsory to upper secondary school, this standardization best captures the actual value of the grades for the students themselves. Student’s percentile ranks were then categorized into quintiles, with the 1st quintile representing the lowest 20% of grades, and so on. Categorization into quintiles was used to allow for investigation of non-linear associations between achievement and internalizing disorders, and to identify specific groups at risk of internalizing disorders [40]. Additional descriptive statistics on both the raw (untransformed) GPA scores and the GPA percentiles were included in Online Resource 5.

School failure was measured as a binary variable, coded 1 for students who failed to reach the minimum number of passing grades in school year 9 required for eligibility to national programs in upper secondary school. Although there is no generally agreed upon definition of school failure, this operationalization captures one crucial aspect: failure to obtain the minimum qualifications necessary to continue to the next stage of the education system [19, 23]. The operationalization is also in line with previous Swedish research showing that failure to obtain upper secondary eligibility has adverse consequences for student’s socioeconomic opportunities later in life [41, 42]. This variable was only relevant from 1998 and onwards since there were no formal eligibility requirements before that. Between 1998 and 2010, eligibility necessitated passing grades in the three core subjects: Swedish, English and Mathematics. From 2011, eligibility required passing grades in the three core subjects, along with at least five additional subjects for vocational programs or nine additional subjects for academic programs.

Anxiety and mood disorders were measured as specialized psychiatric in-patient care (available 1990–2018) and specialised psychiatric outpatient care provided by a doctor (available 2001–2018). These were identified by International Statistical Classification of Diseases 10 (ICD-10) codes F30-39 for mood disorders and F40-44 for neurotic and stress-related disorders, hereafter referred to as anxiety disorders. Similar measurements have been used in previous research on academic achievement and mental health [9, 11, 43]. ICD-9 was used in Swedish health care until 1997. In the years 1990–1997, codes 296, 298, 300, 311 were used to identify mood disorders and codes 300, 308, 309 to identify anxiety disorders. Only the patient’s main diagnosis was used. There may be underreporting of data from clinics to the National Patient Register, especially in the early years and especially for out-patient care. Additionally, the National Patient Register does not include patients treated in primary care that are not treated by a psychiatric specialist [44]. Patient data were therefore complemented by data on prescriptions of psycholeptic and psychotropic drugs from the Swedish Prescribed Drug Register (available 2005–2018), which also covers patients who receive pharmacological treatment in primary care and has very high reliability [45]. Anatomical Therapeutic Chemical (ATC) classification system codes N05B, N05C and N06A were used to identify anxiety or mood disorders.

Since data on drug prescriptions were only available from 2005, and since the included ACT groups are used to treat both anxiety and mood disorders, two outcome variables were used in the analyses: (1) in-patient care for anxiety and mood disorders analyzed separately for the years 1990–2018; and (2) combined data on in-patient or out-patient care, or pharmacological treatment, for anxiety and mood disorders combined, for the years 2005–2018. Thus, students were coded 1 if they received care or were prescribed pharmaceuticals according to the previously specified definitions under (1) and (2) at least once in the same year that they graduated from school year 9, and 0 otherwise.

Students’ graduation year or cohort, defined as the year that students graduated from school year 9, was used as a covariate. This variable was categorized into bins in order to reduce random fluctuations over time due to the small number of students with anxiety or mood disorders in some years and GPA quintiles. Four-year bins were used in analyses of in-patient care in the 1990–2018 period, and three-year bins in analyses of in-patient or out-patient care, or pharmacological treatment, in the 2005–2018 period. Larger bins were used in analyses of in-patient care due to the smaller number of number of students in in-patient care. Some students (0.29% of the sample) graduated from school year 9 more than once because of grade retention. For these students, only their grades from the first graduation were included in the analysis so as to make the graduation years more comparable.

The use of register data on diagnoses and treatment has benefits but also drawbacks. Without data on symptoms, it is difficult to distinguish real changes in symptom prevalence from changing treatment practices or care-seeking behaviors [51]. In our data the overall increase in prevalence was, for instance, considerably smaller for in-patient care than for the combined category comprising in-patient or outpatient care or pharmacological treatment. In-patient care requires more resources in terms of personnel and hospital beds and is therefore more difficult and expensive to expand. Patients treated in in-patient care also tend to have more severe psychiatric illness and/or greater care needs [46]. Assuming that scarce resources are rationed based on need, the greater overall increase in the combined category may thus have been driven by a disproportionate increase in the treatment of relatively milder cases of internalizing disorders. This increase is likely related to both greater acceptance for treatment-seeking and new diagnostic criteria and treatment practices, catching previously undiagnosed cases, and to an increase in mental health problems among students, as indicated in studies using self-reported measures [1, 52]. When comparing the relationship between internalizing disorders and academic achievement over time it is thus important to keep in mind that there are likely many undiagnosed students at the beginning of the period that would have been diagnosed later, and that the severity of symptoms in those diagnosed may have decreased over time. This is probably a larger potential problem for the combined category comprising outpatient care and pharmacological treatment than for the in-patient group where the overall increase is more limited. The finding that achievement-related inequalities were larger in in-patient care than in the combined category suggests that average or high-achieving students may have been disproportionally overrepresented among these relatively mild cases. If so, the convergence in relative risks observed at the end of the period may, at least with regard to the combined category, be distorted by unobserved changes in the underlying symptoms. Related to this, the results of the study may not generalize to continuous and self-reported measures of internalizing problems or to symptoms of internalizing disorders that do not lead to medical diagnosis or treatment, nor to non-pharmacological treatment in primary care (which is not covered by the registers). In future research, it would here be of great benefit to replicate the approach used in this register-based study but with suitable multi-cohort survey data on students.

Another limitation is that it was not, due to lack of data on school classes, possible to standardize grades within classes and thereby account for teachers’ tendency to grade individual students based on their relative achievement compared to the rest of the class (“grading on the curve”). It is therefore likely that our measure of achievement (GPA percentile rank) reflects achievement relative to both the national average and the school class. Both may be relevant for mental health: achievement relative to the school class may be more relevant for self-esteem since the school class is the primary point of reference for students [53], while achievement relative to the national average better captures the instrumental value of the grades in the admission to upper secondary school, where students compete with all other students in the same larger region (or even nationally). Lastly, as a descriptive study, all discussions regarding possible causes of the growing achievement-related inequalities in internalizing disorders should be viewed as tentative hypotheses to be tested in future research.