Journal of Affective Disorders 253 (2019) 460–466

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Is teachers’ mental health and wellbeing associated with students’ mental T

health and wellbeing?☆
Sarah Hardinga, , Richard Morrisa, David Gunnella, Tamsin Fordb, William Hollingwortha,
⁎

Kate Tillinga, Rhiannon Evansc, Sarah Bella, Jillian Greyc, Rowan Brockmana, Rona Campbella,
Ricardo Arayad, Simon Murphyc, Judi Kidgera
a
Population Health Sciences, Bristol Medical School, University of Bristol, Canynge Hall, 39 Whatley Road, Bristol BS8 2PS, UK
b
University of Exeter Medical School, South Cloisters, St Luke's Campus, Exeter EX1 2LU, UK
c
DECIPHer, School of Social Sciences, Cardiff University, 1-3 Museum Place, Cardiff CF10 3BD, UK
d
London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK

ABSTRACT

Background: Factors within the school environment may impact young people's mental health and wellbeing. The aim of this study was to understand the association
between teacher and student mental health and wellbeing. Further, it seeked to identify possible explanations by examining whether the strength of any association is
weakened once quality of teacher-student relationships, teacher presenteeism and absence are considered.
Methods: Cross-sectional data were collected from 3215 Year 8 (aged 12–13 years) students and from 1182 teachers in 25 secondary schools in England and Wales.
The association between teacher wellbeing (measured by Warwick Edinburgh Mental Wellbeing Scale (WEMWBS)) with student wellbeing (WEMWBS) and with
student psychological distress (Total Difficulties Score (TDS)) was assessed using Random Effects Mixed Models. Analyses were repeated using teacher depression
(measured by Patient Health Questionnaire) as the explanatory variable.
Results: Better teacher wellbeing was associated with (i) better student wellbeing (standardised effect = 0.07, 95% CI = 0.02–0.12) and (ii) lower student psy-
chological distress (standardised effect = −0.12, 95% CI = 0.18 to −0.05). Teacher presenteeism and the quality of the teacher-student relationship appeared to be
on the pathway of these relationships. Higher levels of teacher depressive symptoms were associated with poorer student wellbeing and psychological distress
(standardised effect = −0.06, 95% CI = −0.11 to −0.01 and 0.10, 95% CI = −0.03, 0.16). This association did not withstand adjustment for teacher presenteeism.
Limitations: Cross sectional in design so unable to establish temporal associations.
Conclusions: Associations were found between teacher wellbeing and student wellbeing and psychological distress. There was also an association between teacher
depression and student wellbeing. Both may be partially explained by teacher presenteeism and quality of teacher-student relationships.

1. Introduction Factors within the school environment have been found to have an
impact on young people's mental health (Kidger et al., 2012). For ex-
The mental health of children and young people appears to be de- ample, supportive teacher-student relationships are associated with
teriorating (Collishaw, 2015). A recent survey in England found 37% of lower student depression (Kidger et al., 2012; Plenty et al., 2014). The
girls and 15% of boys in ear 10 (14–15 years) were psychologically importance of a positive teacher-student relationship is supported by
distressed (Department for Education, 2016). Positive mental health the findings of a systematic review and meta-ethnography (Jamal et al.,
and wellbeing during adolescence is associated with positive social 2013) which found positive relationships with teachers and a feeling of
relations, the development of a healthy lifestyle, reduced risk of adverse safety are important for student wellbeing within schools. This is in line
socioeconomic outcomes, psychiatric disorders, self-harm, and suicide with theories regarding health promotion in schools. For instance, Baric
in later life (Hawton and Harriss, 2007; Fergusson and Woodward, (1992, 1993) reasoned that in order to promote healthy organisations
2002; Fava et al., 2010). Almost 75% of adults with depression report (such as schools) there should be a focus on interactions within the
that their mental health problems started in adolescence (Kim-Cohen organisations. Additionally, the importance of a positive teacher stu-
et al., 2003; Joinson et al., 2017). Therefore, it is important to identify dent relationship resonates with the explanatory framework for un-
risk factors for poor mental health among this age group. derstanding how schools may intervene to promote students health put

DOI of original article: https://doi.org/10.1016/j.jad.2019.03.045; https://doi.org/10.1016/j.jad.2018.08.080

☆
Declarations of interest: none.
⁎
Corresponding author.
E-mail address: [email protected] (S. Harding).

https://doi.org/10.1016/j.jad.2019.03.046

Available online 02 April 2019

0165-0327/ © 2019 Published by Elsevier B.V.
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

forward by Markham and Aveyard (2003). This outlines that the pri- considered. Although these are possible explanations for any associa-
mary focus of health promotion in schools should be the realisation for tions that exist, it is also possible that an association would be due to
practical reasoning and affiliation with other humans. Students having shared features of the school environment impacting on the wellbeing
a positive relationship with their teachers may contribute to school of both. Thus, school-level factors are included as potential con-
connectedness which is defined as an environment in which students founders.
believe that adults in the school care about their learning and about
them as individuals (Blum et al., 2004). This has also been linked to 2. Methods
student wellbeing (Aldridge and McChesney, 2018). In addition to
fostering good quality relationships, teachers may also contribute to This study is cross-sectional in design and multi-level as participants
student mental health and wellbeing through identification of and in- were clustered within schools.
tervention with students at risk of mental health problems (Rothì et al.,
2008; Kidger et al., 2010). 2.1. Sample
Teachers themselves are consistently reported to be at increased risk
of common mental health disorders compared to those in other occu- The student, teacher and school data were taken from 25 schools
pations (Stansfeld et al., 2011; Johnson et al., 2005; Kidger et al., which participated in the WISE project (Kidger et al., 2016b). In brief,
2016a). Poor teacher wellbeing may be problematic not only for tea- WISE is a cluster randomised controlled trial with secondary schools as
chers’ longer term mental health (Melchior et al., 2007) but also for that the unit of randomisation. A group of teachers in intervention schools
of their students. Teacher wellbeing and student wellbeing could be were given Mental Health First Aid training for students and a further
linked though complex and interrelated factors. Indeed, poor wellbeing group were given Mental Health First Aid training for colleagues. More
and depressive symptoms are associated with teachers’ self-rated pre- information on the project can be found in the protocol paper (Kidger
senteeism (Kidger et al., 2016a) which is defined as an employee under- et al., 2016b) and on the study website (https://www.bristol.ac.uk/
performing at work as a result of a health problem (Henderson et al., population-health-sciences/projects/wise/).
2011) (for example a teacher having symptoms of poor physical or The schools were recruited from 4 four local authorities in the
mental health but still being present at work). Presenteeism may have South-West of England and from 10 local authorities in South-East and
an impact on student mental health through teachers not being able to South-Central Wales. They varied by size, socioeconomic catchment
develop a positive and supportive school environment and finding it area and academic performance. Details of the recruitment procedure
more difficult to manage classrooms effectively (Jennings and can be found in the WISE study protocol (Kidger et al., 2016c). All
Greenberg, 2009). Additionally, teachers experiencing poor mental students in school year 8 (aged 12–13 years) and all teachers currently
health and wellbeing may find it difficult to develop and model good working at the school were invited to take part.
quality relationships with students (Kidger et al., 2010; Jennings and Results presented here are from the baseline data collection – ad-
Greenberg, 2009). They may also be linked through higher rates of ministered prior to randomisation to the intervention or control group
teacher absence at schools which may prevent students and staff from and prior to intervention delivery. Student and teacher data were col-
fostering supportive relationships (Jamal et al., 2013). Furthermore, lected via self-report surveys, administered during lesson time (stu-
where teachers experience poor wellbeing, this reduces their belief that dents), meeting times (teachers) or via an online survey (teachers). Data
they can help students with emotional problems (Sisask et al., 2014). collection took place in June/July 2016. School-level data were ob-
Wellbeing covers two perspectives; firstly, the subjective experience tained from publicly available routine data sources.
of happiness and life satisfaction (the hedonic perspective) and sec-
ondly, positive psychological functioning, good relationships with 3. Measures
others and self-realisation (the eudaimonic perspective) (Ryan and
Deci, 2001; Stewart-Brown). Key concepts include positive affect psy- 3.1. Outcome measures
chological functioning (autonomy, competence, self-acceptance, per-
sonal growth) and interpersonal relationships (Tennant et al., 2007). 3.1.1. Student wellbeing
Depression is an internalising mental disorder (American Psychiatric The Warwick Edinburgh Mental Wellbeing Scale (WEMWBS)
Association 2013) characterized by persistent sadness and a loss of in- (Tennant et al., 2007) was used to measure wellbeing. This has been
terest in activities that one normally enjoys, accompanied by an in- validated for use in adolescents and has been shown to be reliable (test
ability to carry out daily activities, for at least two weeks (World Health re-test score = 0.83) (Clarke et al., 2011) . This scale consists of 14
Organisation). A large number of studies have shown that depression items (statements) and participants were asked to tick the box which
and wellbeing are two different constructs/dimensions of mental health best describes their experience of each statement over the past 2 weeks
(Greenspoon and Saklofske, 2001; Keyes et al., 2008; Lamers et al., using a five-point Likert scale (Tennant et al., 2007). A total score is
2015; Antaramian et al., 2010; Lyons et al., 2012) suggesting a dual- derived from these 14 items; higher scores signify greater wellbeing
factor model. Indeed, studies have shown that the two constructs have (possible total score ranging from 14 to 70).
different causal determinants and mediating mechanisms (Kinderman
et al., 2015) and respond to different interventions or treatments 3.1.2. Student psychological distress
(Trompetter et al., 2017). Owing to this dual-factor model, both well- Strengths and Difficulties Questionnaire (SDQ) (Goodman, 2001)
being and depression are distinct from one another and need to be was used to measure psychological distress. The SDQ is a brief 25-item
considered separately. The same is also true for student wellbeing and scale covering four main domains of difficulties that can trouble ado-
student psychological distress. lescents (i.e., emotional symptoms, conduct problems, hyperactivity-
Despite the likelihood that teacher and student wellbeing and inattention, and peer problems). A Total Difficulties Score (TDS) was
mental health are linked, evidence for this is currently lacking in the derived by adding the scores from 4 sub-scales: emotional symptoms
literature. Drawing on self-report survey data collected from teachers (anxiety and depressive symptoms), conduct problems, hyperactivity/
and year 8 students (12–13 year olds), this paper investigates whether inattention, and peer relationship problems. The score ranges from 0
mean school-level scores for teacher wellbeing and depression are as- (low difficulties) to 40 (high difficulties). This has been shown to be a
sociated with individual student wellbeing and psychological difficul- valid and reliable measure for use in adolescents (Goodman, 2001;
ties. Further, it seeks to identify possible explanations by examining Muris et al., 2004). For example, Goodman (2001) presented a Cron-
whether the strength of any association is weakened once the quality of bach's alpha score of 0.82 and the results suggest that the questionnaire
teacher-student relationships, teacher presenteeism and absence are can discriminate between a clinical (mental health clinic) and non-

461
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

clinical sample. their own finances. Wales does not have academy schools, so this only
applies to the English schools. Welsh schools were marked as not being
3.2. Explanatory variables an academy.
School attainment: GCSE results (examination results for students at
3.2.1. Teacher wellbeing age 16) the year of data collection was used as an indicator for school
WEMWBS was used as a measure of wellbeing (described above). attainment. Summary data were recorded differently for England and
This has been validated for use in adults (Tennant et al., 2007) Wales. For the English schools, this was the percentage of pupils
(goodness of fit >0.9) and has been shown to be reliable (test re-test achieving a GCSE at A*-C in English and Maths at the year of data
score = 0.83). collection. For the Welsh schools, this was the percentage of pupils
achieving A*-C (or equivalent qualification) in English/Welsh, Maths
3.3. Teacher depressive symptoms and Science. Data were obtained from government websites (Welsh
The 8 item Patient Health Questionnaire (PHQ-8) was used to Goverment; GOV.UK). A binary variable was created which indicated
measure depressive symptoms in teachers (Kroenke et al., 2009). This whether the school attainment was above or below average for each
has been shown to be valid measure when compared to a standard di- country.
agnostic algorithm (Kroenke et al., 2009). This questionnaire asks
participants to rate on a 4-point scale how much they have experienced 3.5. Ethics
8 depressive symptoms in the 14 days prior to evaluation.
The study was approved by the University of Bristol's Faculty of
3.4. Confounding factors Medicine and Dentistry Ethics Committee (reference 2852). An opt-out
consent procedure was used for students whilst parents were given the
Student socio-economic deprivation: Self-reported receipt of free opportunity to opt their child out of completing questionnaires (Kidger
school meals was used as a measure of individual level deprivation. et al., 2016b).
Students are eligible for free school meals if their parents/guardians
receive any benefits such as income support. 3.6. Missing data
Ethnicity: Measured by self-report questionnaire, participants were
asked “what is your ethnic group?” and the possible responses were: Published guidance (NHS Health Scotland; Youth in Mind) was
White, Mixed, Asian or Asian British, Black or Black British, Chinese or followed where individual items were missing on the SDQ and
other ethnic group. WEMWBS. For SDQ, the mean score for each sub-category was entered
Quality of teacher-student relationships: This was measured via a for the missing items within that sub-category, but only if ≥3 of the 5
question created by the study team. Students were asked to rate the items were completed. For WEMWBS, missing items were completed
following statement, “teachers and students generally have good re- using the mean score of all other items but only if ≤3 items were
lationships at this school.” The score ranged from 0 (strongly disagree) missing. For the PHQ-8 scale, the mean value of the other items was
to 3 (strongly agree). used to complete the missing item if ≤1 item was missing. Once these
Teacher absence: This was measured by asking teachers “during the procedures were followed, there were minimal missing data. Therefore,
last four working weeks, how many days did you miss from school there was no need for statistical imputation. After following these
because of health problems?”. The four-week period was an adaptation procedures, only students with no missing data for the variables used in
of the WPAI Work Productivity and Activity Impairment Questionnaire the models were included in the analysis of this paper – i.e., complete
(WPAI) questionnaire which asked about the last month. As data were case analysis.
collected shortly after the school Easter holidays it was decided to word
it as four working weeks rather than one month. 3.7. Statistical analysis
Teacher presenteeism: This was measured using an adapted version
of the presenteeism measure from the WPAI (Reilly et al., 1993). The Data on all outcome variables were normally distributed so para-
relevant question asks participants to rate to what extent health pro- metric statistics were used. Owing to the possibility of clustering within
blems have affected their productivity at work from 0 (no effect on my schools (n = 25), random effects mixed models were used. For each of
work) to 10 (completely prevented me from working) over the previous the teacher variables (wellbeing, depressive symptoms, presenteeism
four working weeks. This score was only calculated for teachers with no and absence), individual scores within each school were combined to
absent days, as it is not applicable if the teacher had absent days in the provide a school mean as students are taught by a range of teachers. For
previous four weeks. the teacher-student relationship variable, we did not combine the score
School size: Number of students for each school was used as a for each school because data was collected on an individual student
measure of school size. Teacher-student ratio was also included in the level.
models (number of students/number of teachers). Analyses were carried out with student wellbeing as the outcome
School-level deprivation: The percentage of students eligible for free variable and then repeated with student psychological distress as the
school meals was used as a measure of school-level deprivation. These outcome variable. Initial univariable models assessed the association
data were obtained from government websites (Welsh Goverment; between teacher wellbeing and student wellbeing/psychological dis-
GOV.UK). tress (Model 1). Individual student variables (gender, free school meal
School performance rating: Schools’ most recent report from the eligibility, ethnicity) and school variables (teacher-student ratio,
independent inspectorate for schools (Ofsted in England and Estyn in number of students, percentage of students eligible for free school
Wales). The potential ratings are outstanding/excellent, good, requires meals, Ofsted/Estyn rating, academy status, school attainment and re-
improvement/adequate and inadequate/unsatisfactory for England and gion) were then added (Model 2). As teacher presenteeism, teacher
Wales respectively. These were scored as 0, 1, 2 and 3 respectively, with absence, and teacher-student relationship may be on the causal
outstanding/excellent (0) being using as the reference. pathway between teacher wellbeing and student outcomes, these were
School region: The schools were either based in England (n = 13) or added sequentially to the models, as follows:
Wales (n = 12).
School academy status: Whether the school is an academy or not. Model 3 – Multivariable model including all individual student
Academies are publicly funded schools that are self-governing (as op- measures, all school-level factors, teacher wellbeing and teacher-
posed to under local education authority control) and have control over student relationship

462
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

Model 4 – Multivariable model including all individual student Table 2

measures, all school-level factors, teacher wellbeing and teacher Mean and standard deviation of the continuous variables included in models
presenteeism (student, teacher and school variables).
Model 5 – Multivariable model including all individual student Variable Mean (SD)
measures, all school-level factors, teacher wellbeing and teacher
absence Student wellbeing (WEMBWS), range = 14–70 47.37 (9.24)
Student psychological difficulties (TDS), range = 0–34 12.28 (6.07)
Model 6 – Multivariable model including all individual student
Teacher-student relationship, range = 0−3 1.78 (0.65)
measures, all school-level factors and all teacher measures Teacher wellbeing (Warwick Edinburgh Mental Wellbeing 46.81 (8.41)
Scale), range 14–70
The above models were repeated using teacher depression as the Teacher depression (PHQ-8), range = 0–24 6.37 (4.92)
Teacher presenteeism, range = 0–10 2.04 (2.36)
key explanatory variable (in place of teacher wellbeing).
Teacher absence (number of absent days in the previous 4 0.44 (1.75)
School performance rating was ordinal, region and attainment were weeks)
binary; the remaining variables were continuous. Standardised effects Number of students at the school 869.28 (264.56)
estimates were created by creating standardised values across the Percentage of pupils eligible for free school meals at the school 18.80 (10.39)
sample for each variable (mean = 0, standard deviation = 1) and the School teacher student ratio 15.64 (3.49)

models were repeated using standardised values. Unless specified, all

*SD = standard deviation, WEMBWS = Warwick Edinburgh Mental Wellbeing
results here are based on the unstandardised data. Data were checked
Scale, TDS = Total Difficulties Score, PHQ = patient health questionnaire,
for homoscedasticity. All data were analysed using Stata Version 14. teacher-student ratio = number of students/number of teachers), teacher pre-
senteeism: teachers rated the extent to which health problems have affected
4. Results their productivity at work from 0 (no effect on my work) to 10 (completely
prevented me from working). Results are presented to 2 decimal places or to the
4.1. Participants nearest significant figure.

Of the 3596 eligible Year 8 students, 3408 completed the ques-

tionnaire and of these, 3215 had no missing data for the variables of of 0.06. This association also remained with the addition of the teacher-
interest and were thus included in the analysis (89.4% response rate). student relationship to the model (B = 0.31, 95% CI = 0.04–0.59) and
Of the 1357 teachers who were eligible, 1182 (87.1% response rate) when teacher absence was included in the model (B = 0.37, 95%
from 25 secondary schools completed the questionnaire. Table 1 de- CI = 0.08–0.66). However, the association disappeared when teacher
scribes the sample. Of the 25 schools, 13 (52%) were based in England presenteeism was added to the model (B = 0.10, 95% CI = −0.23 to
and 12 (48%) based in Wales, 9 (36%) of the schools were an academy, 0.43) and in the fully adjusted model (B = 0.08, 95% CI = −0.22 to
12% of schools rated as outstanding/excellent, 32% rated as good, 44% 0.38), suggesting that teacher presenteeism may be on the pathway
as requires improvement/adequate and 12% rated as inadequate/un- between teacher and student wellbeing. Results shown in Table 3.
satisfactory. 52% of the schools were below the national attainment In the fully adjusted model, better-quality teacher-student re-
average. The mean and standard deviation of the other variables are lationships was associated with better student wellbeing (B = 3.85,
shown in Table 2. 95% CI = 3.39–4.32) and higher teacher presenteeism and teacher
absence were associated with poorer student wellbeing (B=−2.05,
4.2. Association between teacher wellbeing and student wellbeing 95% CI = 3.34 to −0.76, B=−1.92, 95% CI = −3.14 to −0.70 re-
spectively). Results shown in Table 1 supplementary material.
In the univariable model, better teacher wellbeing was associated
with better student wellbeing (B = 0.35, 95% CI = 0.08–0.63)
equivalent to a standardised effect of 0.07. This association remained 4.3. Association between teacher wellbeing and student psychological
after individual student factors and school-level factors were adjusted distress
for (B = 0.34, 95% CI = 0.03–0.64); equivalent to a standardised effect
There was a crude inverse association between teacher wellbeing
Table 1 and student psychological distress (B=−0.41, 95% CI = −0.63 to
Sociodemographic characteristics of student and teacher participants. −0.19) equivalent to a standardised effect of −0.12. This association
Category n (%) remained but was attenuated after individual student factors and
school-level factors were included in the model (B = −0.27, 95%
Students Gender Male 1523 (47.4) CI = −0.49 to −0.05); this is equivalent to a standardised effect of
Female 1692 (52.6)
−0.08. The association remained but was weakened with the inclusion
Ethnicity White 2729 (84.9)
Other 486 (15.1)
of teacher absence (B = −0.28, 95% CI = −0.50 to −0.07). and
Eligible for Free School Meals Yes 556 (17.3) quality of teacher-student relationships in the model (B = −0.25, 95%
No 2659 (82.7) CI = −0.44 to −0.06). Teacher presenteeism appeared to be on the
Teachers Gender Male 429 (36.3) pathway between teacher wellbeing and student psychological distress
Female 751 (63.5)
as the association did not remain when teacher presenteeism was in-
Ethnicity White 1139 (95.6)
Other 34 (2.9) cluded in the model (B = −0.09, 95% CI = −0.33 to 0.15 and
Age ≤25 77 (6.5) B = −0.08, 95% CI = −0.28 to 0.12, with teacher presenteeism in-
26–35 412 (35.0) cluded in the model (model 4) and the fully adjusted model, respec-
36–45 387 (32.9)
tively). Results shown in Table 3.
46–55 236 (20.0)
56–65 65 (5.5)
In the fully adjusted model (Table 1 supplementary material), a
≥ 65 1 (0.1) better teacher-student relationship was associated with lower student
psychological distress (B = −2.36, 95% CI = −2.67 to −2.06). Higher
*Two teachers had missing data for gender, 9 teachers had missing data for teacher presenteeism and higher teacher absence were associated with
ethnicity, 4 teachers had missing data for their age. Results are presented to one higher student psychological distress (B = 1.46, 95% CI = 0.62–2.31,
decimal place or to the nearest significant figure. B = 1.25, 95% CI = 0.45–2.05).

463
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

Table 3
Associations between teacher wellbeing and (i) student wellbeing (ii) student psychological distress.
Student wellbeing unstandardized Psychological distress unstandardized
coefficients (B (95% CI)) coefficients (B (95% CI))

Crude association (model 1) 0.35 (0.08–0.63) −0.41 (−0.63 to −0.19)

Adjusted for: Individual student factors +school-level factors (model 2) 0.34 (0.03–0.64) −0.27 (−0.49 to −0.05)
Individual student factors +school-level factors + teacher-student 0.31 (0.04–0.59) −0.25 (−0.44 to-0.06)
relationship (model 3)
Individual student factors +school-level factors + teacher 0.10 (−0.23 to 0.43) −0.09 (−0.33 to 0.15)
presenteeism (model 4)
Individual student factors +school-level factors + teacher absence 0.37 (0.08–0.66) −0.28 (−0.50 to −0.07)
(model 5)
Fully adjusted model (model 6) 0.08 (−0.22 to 0.38) −0.08 (−0.28 to 0.12)

Results are presented to 2 decimal places or to the nearest significant figure. CI = confidence interval, *Individual student factors include: gender, eligibility for FSM
(free school meals) and ethnicity, school-level factors include: number of student at school, teacher-student ratio, school performance rating, percentage of pupils
eligible for FSM, academy status, school region and school attainment. N = 3215.

4.4. Association between teacher depressive symptoms and student student relationship was included (B = 0.34, 95% CI = 0.01–0.67) and
wellbeing when teacher absence was included in the model (B = 0.32, 95%
CI = −0.06 to 0.70). The association did not withstand adjustment for
There was a crude inverse association between teacher depressive teacher presenteeism (B = −0.07, 95% CI = −0.47 to 0.34) nor did it
symptoms and student wellbeing (B = −0.60, 95% CI = −1.15 to remain in the fully adjusted model (B = −0.12, 95% CI = −0.47 to
−0.05), which is equivalent to a standardised effect of −0.06. This 0.23).
association remained when individual student factors and school-level In the fully adjusted model, the teacher-student relationship, tea-
factors were adjusted for (B = −0.51, 95% CI = −1.02 to 0.01), when cher presenteeism and teacher absence were associated with student
the quality of the teacher-student relationship was included psychological distress (B=−2.35, 95% CI = −2.66 to −2.05,
(B = −0.56, 95% CI = −1.01 to −0.10), and when teacher absence B = 1.84, 95% CI = 0.94 to 2.74, B = 1.38, 95% CI = 0.55–2.21 re-
was included (B = −0.51, 95% CI = −1.01 to −0.01); these are spectively) – see Table 2 supplementary material.
equivalent to standardised effect of −0.05. The association between
teacher depression and student wellbeing did not withstand when
5. Discussion
teacher presenteeism was included in the model (B = −0.06, 95%
CI = −0.61 to 0.50). Results are shown in Table 4.
The results of this paper suggest that better teacher wellbeing is
In the fully adjusted model, higher teacher presenteeism was asso-
associated with better student wellbeing and with lower student psy-
ciated with poorer student wellbeing (B = −2.18, 95% CI = −3.55 to
chological difficulties as well as lower teacher depressive symptoms
−0.82) and a better-quality teacher-student relationship was associated
being associated with better student wellbeing. The findings also sug-
better student wellbeing (B = 3.85, 95% CI = 3.39–4.32) – see Table 2
gest that teacher presenteeism and the teacher-student relationship may
supplementary material.
be mediating factors in these relationships. Additionally, the results
show an association between the quality of the teacher-student re-
4.5. Association between teacher depressive symptoms and student lationship, teacher presenteeism and teacher absence with student
psychological distress wellbeing and psychological distress.
The associations between teacher wellbeing and depressive symp-
There was a crude association between teacher depressive symp- toms, and student wellbeing and distress were weakened when teacher
toms and student psychological distress (B = 0.70, 95% presenteeism was included in the models. Kidger et al. (2016a) found
CI = 0.25–1.15), which is equivalent to a standardised effect of 0.10 that poor teacher wellbeing was associated with high teacher pre-
(Table 4). However, this association was attenuated once individual senteeism, and the current study found an association between teacher
student factors and school-level factors were included in the model presenteeism and student wellbeing and psychological difficulties.
(B = 0.32, 95% CI = −0.07 to 0.70), when the quality of the teacher- Therefore, it may be that teacher presenteeism is on the causal pathway

Table 4
Association between teacher depression and (i) student wellbeing (ii) psychological distress.
Student wellbeing unstandardized Psychological distress Unstandardized
coefficients (B (95% CI)) coefficients (B (95% CI))

Crude association (model 1) −0.60 (−1.15 to −0.05) 0.70 (0.25–1.15)

Adjusted for: Individual student factors +school-level factors (model 2) −0.51 (−1.02 to 0.01) 0.32 (−0.07 to 0.70)
Individual student factors +school-level factors + teacher-student −0.56 (−1.01 to-0.10) 0.34 (0.01–0.67)
relationship (model 3)
Individual student factors +school-level factors + teacher −0.06 (−0.61 to 0.50) −0.07 (−0.47 to 0.34)
presenteeism (model 4)
Individual student factors +school-level factors + teacher absence −0.51 (−1.01 to −0.01) 0.32 (−0.06 to 0.70)
(model 5)
Fully adjusted model (model 6) −0.03 (−0.56 to 0.50) −0.12 (−0.47 to 0.23)

Results are presented to 2 decimal places or to the nearest significant figure. CI = confidence interval, *individual student factors include: gender, eligibility for FSM
(free school meals) and ethnicity, school-level factors include: number of student at school, teacher-student ratio, school performance rating, percentage of pupils
eligible for FSM, academy status, school region and school attainment. N = 3215.

464
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

between teacher and student mental health. Poor wellbeing and higher 5.1. Strengths and limitations
levels of depressive symptoms may lead to teachers under performing at
work (Beck et al., 2011; Jain et al., 2013), which may affect student This is the first study to examine the association between teacher
wellbeing and psychological distress. For example, teachers may be less wellbeing and depressive symptoms, with student wellbeing and psy-
able to engage in positive classroom and behaviour management chological distress. The outcome measures used have been shown to
(Jennings and Greenberg, 2009) or be more likely to display negative have good reliability and validity (Tennant et al., 2007). The study
emotions or behaviours (de Moraes et al., 2015). includes a large and representative sample recruited from 25 different
This study found that a better teacher-student relationship is asso- schools in England and Wales, which were stratified at the sampling
ciated with better student wellbeing and with lower student psycholo- stage to ensure a representative range of socioeconomic catchment
gical distress. This resonates with previous studies. For example, areas. Both teacher and student response rates were high. There were
Cornelius-White (2007) showed in a meta-analysis that positive tea- minimal missing data, so the likelihood of sample bias is low.
cher-student relationships are associated with positive student out- The study is limited by the cross-sectional design. Thus, it is im-
comes (affective, behavioural and cognitive), and studies have also possible to establish the temporal direction of any associations, which
shown that supportive teacher-student relationships predict lower stu- could conceivably operate in both directions. Longitudinal studies or
dent depression (Kidger et al., 2012; Hughes and Kwok, 2007). It also randomised controlled trials are needed to further understand the asso-
resonates with the results of Sisask et al. (2014), which suggested that ciation between teacher and student wellbeing, and the extent to which it
poor wellbeing reduces teachers' belief that they can help students with is explained by the quality of teacher-student relationships and teacher
emotional or behavioural problems. A potential explanation for this presenteeism. School-level mean scores were used for the teacher-related
association is that students who have a better relationship with their variables. This was for the pragmatic reason that students at secondary
teachers may have higher levels of connectedness and belongingness school are taught by several teachers and we cannot identify which
with their school, which has previously been associated with higher students are taught by which teachers. However, it meant that these data
levels of student wellbeing (Aldridge and McChesney, 2018). The re- were limited to 25 data points, and therefore had less statistical power to
sults reported here indicate that the quality of teacher-student re- identify associations. Additionally, the measure of teacher-student re-
lationships may partially explain the association between teacher lationship was developed for the WISE project and has not been tested for
wellbeing and student psychological difficulties: teachers with poor validity or reliability. Finally, student and teacher measures were self-
wellbeing may be less able to develop supportive relationships reported. The responses for teacher absence and teacher presenteeism
(Jennings and Greenberg, 2009; Jennings et al., 2013). may have been influenced by recall bias. Further, students with poor
A positive teacher-student relationship is also likely to be important wellbeing may be more likely to rate other aspects of school negatively,
for teacher wellbeing (Spilt et al., 2011; Milatz et al., 2015; Hargreaves, which could partially explain some of the results, such as the association
2000). Hargreaves (2000) qualitative study of teachers in Canada found between student wellbeing and the teacher-student relationship.
that teachers’ relationships with their students was an important source
of enjoyment, motivation and positive emotions. Spilt et al. (2011) 6. Conclusions
outlined the importance of the teacher-student relationship for teacher
wellbeing, suggesting it may in part be explained by teachers’ need for This paper demonstrates cross-sectional associations between tea-
relatedness (the need to feel related/connected) with their students. cher wellbeing and depression, and student mental health and well-
Additionally, Milatz et al. (2015) found an association between emo- being outcomes. These associations appear at least in part to be due to
tional exhaustion in teachers and quality of relationships with students. the quality of teacher-student relationships and teacher presenteeism.
Thus, the findings of this and previous studies suggest that a focus on Therefore, interventions to improve these aspects of school life, possibly
improving the teacher-student relationship may have a positive impact by addressing teacher wellbeing and symptoms of depression, may
on both student and teacher wellbeing. While teacher absence was as- improve outcomes for students. However, longitudinal studies are
sociated with students’ wellbeing and psychological distress, it does not needed to understand these associations more fully.
appear to be on the explanatory pathway between teacher and student
outcomes. Acknowledgements
The relationships between teacher wellbeing, the quality of teacher-
student relationships, teacher presenteeism and student mental health The study team thanks the teachers and students who gave up their
outcomes are clearly complex and likely to be inter-related. The results time so willingly to complete the questionnaires. We thank all staff
of this study suggest that improving teacher wellbeing may lead to members who assisted with data collection alongside Harriett Fisher
better student wellbeing via more supportive relationships or reduced and Camilla Sapsworth for management of the participant lists. We are
teacher presenteeism. It is important to note that this is a cross-sectional also thankful for the support of Bristol Randomised Trials
study and longitudinal or randomised controlled trials are needed to Collaboration, a UKCRC-registered unit in receipt of NIHR support. This
understand these relationships further. However, if additional research project was funded by the National Institute for Health Research Public
supports these results this could have implications for practice such as Health Research (NIHR PHR) Programme grant number: 13/164/06.
informing the development of a school-level intervention, which aims The views and opinions expressed therein are those of the authors and
to improve the mental health and wellbeing of teachers and students do not necessarily reflect those of the NIHR PHR Programme or the
within secondary schools. Department of Health.
It is worth noting that the effect sizes in this paper are small. For
example, the crude association between teacher depression and student Supplementary materials
wellbeing (−0.60 (−1.15 to −0.05)) is interpreted as every 1 unit
change in teacher depression is associated with a 0.60 change in stu- Supplementary material associated with this article can be found, in
dent wellbeing on a scale where the score can range from 14 to 70. This the online version, at doi:10.1016/j.jad.2019.03.046.
may appear small when on an individual level. However, it may be
meaningful or practically important at a population level (Rose, 2001). References
If a public health intervention were to shift a whole distribution in a
favourable direction this could have favourable effects on the on a Aldridge, J.M., McChesney, K., 2018. The relationships between school climate and
population despite only a small change on an individual level (Rose, adolescent mental health and wellbeing: a systematic literature review. Int. J. Educ.
Res. 88, 121–145.
2001).

465
S. Harding, et al. Journal of Affective Disorders 253 (2019) 460–466

American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental cross sectional study in English secondary schools. J. Affect. Disord. 192, 76–82.
Disorders (DSM-5®). American Psychiatric Pub. Kidger, J., Evans, R., Tilling, K., Hollingworth, W., Campbell, R., Ford, T., et al., 2016c.
Antaramian, S.P., Scott Huebner, E., Hills, K.J., Valois, R.F., 2010. A dual-factor model of Protocol for a cluster randomised controlled trial of an intervention to improve the
mental health: toward a more comprehensive understanding of youth functioning. mental health support and training available to secondary school teachers - the WISE
Am. J. Orthopsychiatry 80 (4), 462–472. (Wellbeing in Secondary Education) study. BMC Public Health 16 (1), 1089.
Baric, L., 1992. Promoting health new approaches and developments. J. Inst. Health Educ. Kidger, J., Evans, R., Tilling, K., Hollingworth, W., Campbell, R., Ford, T., et al., 2016b.
30 (1), 6–16. Protocol for a cluster randomised controlled trial of an intervention to improve the
Baric, L., 1993. The settings approach—implications for policy and strategy. J. Inst. mental health support and training available to secondary school teachers—the WISE
Health Educ. 31 (1), 17–24. (Wellbeing in Secondary Education) study. BMC Public Health 16 (1), 1089.
Beck, A., Crain, A.L., Solberg, L.I., Unützer, J., Glasgow, R.E., Maciosek, M.V., et al., 2011. Kidger, J., Gunnell, D., Biddle, L., Campbell, R., Donovan, J., 2010. Part and parcel of
Severity of depression and magnitude of productivity loss. Ann. Family Med. 9 (4), teaching? Secondary school staff's views on supporting student emotional health and
305–311. well-being. Br. Educ. Res. J. 36 (6), 919–935.
Blum, R.W., Libbey, H.P., Bishop, J.H., Bishop, M., 2004. School Kim-Cohen, J., Caspi, A., Moffitt, T.E., Harrington, H., Milne, B.J., Poulton, R., 2003.
connectedness—Strengthening health and education outcomes for teenagers. J. Prior juvenile diagnoses in adults with mental disorder: developmental follow-back of
School Health 74 (7), 231–235. a prospective-longitudinal cohort. Arch. Gen. Psychiatry 60 (7), 709–717.
Clarke, A., Friede, T., Putz, R., Ashdown, J., Martin, S., Blake, A., et al., 2011. Warwick- Kinderman, P., Tai, S., Pontin, E., Schwannauer, M., Jarman, I., Lisboa, P., 2015. Causal
Edinburgh Mental Well-being Scale (WEMWBS): validated for teenage school stu- and mediating factors for anxiety, depression and well-being. Br J Psychiatry 206 (6),
dents in England and Scotland. A mixed methods assessment. BMC Public Health. 11 456–460.
(1), 487. Kroenke, K., Strine, T.W., Spitzer, R.L., Williams, J.B., Berry, J.T., Mokdad, A.H., 2009.
Collishaw, S., 2015. Annual research review: secular trends in child and adolescent The PHQ-8 as a measure of current depression in the general population. J. Affect.
mental health. J. Child Psychol. Psychiatry 56 (3), 370–393. Disord. 114 (1), 163–173.
Cornelius-White, J., 2007. Learner-centered teacher-student relationships are effective: a Lamers, S.M., Westerhof, G.J., Glas, C.A., Bohlmeijer, E.T., 2015. The bidirectional re-
meta-analysis. Rev. Educ. Res. 77 (1), 113–143. lation between positive mental health and psychopathology in a longitudinal re-
de Moraes, A.C.F., Carvalho, H.B., Siani, A., Barba, G., Veidebaum, T., Tornaritis, M., presentative panel study. J. Positive Psychol. 10 (6), 553–560.
et al., 2015. Incidence of high blood pressure in children—effects of physical activity Lyons, M.D., Huebner, E.S., Hills, K.J., Shinkareva, S.V., 2012. The dual-factor model of
and sedentary behaviors: the IDEFICS study: high blood pressure, lifestyle and chil- mental health: further study of the determinants of group differences. Canadian J.
dren. Int. J. Cardiol. 180 (1), 165–170. School Psychol. 27 (2), 183–196.
Department for Education. Longitudinal Study of Young People in England cohort 2: Markham, W.A., Aveyard, P., 2003. A new theory of health promoting schools based on
health and wellbeing at wave 2 2016. human functioning, school organisation and pedagogic practice. Soc. Sci. Med. 56
Fava, M., Hwang, I., Rush, A.J., Sampson, N., Walters, E.E., Kessler, R.C., 2010. The (6), 1209–1220.
importance of irritability as a symptom of major depressive disorder: results from the Melchior, M., Caspi, A., Milne, B.J., Danese, A., Poulton, R., Moffitt, T.E., 2007. Work
National Comorbidity Survey Replication. Mol. Psychiatry 15 (8), 856. stress precipitates depression and anxiety in young, working women and men.
Fergusson, D.M., Woodward, L.J., 2002. Mental health, educational, and social role Psychol. Med. 37 (8), 1119–1129.
outcomes of adolescents with depression. Arch. Gen. Psychiatry 59 (3), 225–231. Milatz, A., Lüftenegger, M., Schober, B., 2015. Teachers’ relationship closeness with
Goodman, R., 2001. Psychometric properties of the strengths and difficulties ques- students as a resource for teacher wellbeing: a response surface analytical approach.
tionnaire. J. Am. Acad. Child Adolesc. Psychiatry 40 (11), 1337–1345. Front. Psychol. 1949 (6).
GOV.UK. Compare school performance service 2017 [Available from: https://www. Muris, P., Meesters, C., Eijkelenboom, A., Vincken, M., 2004. The self-report version of
compare-school-performance.service.gov.uk/. the Strengths and Difficulties Questionnaire: its psychometric properties in 8- to 13-
Greenspoon, P.J., Saklofske, D.H., 2001. Toward an integration of subjective well-being year-old non-clinical children. Br. J. Clin. Psychol. 43 (4), 437–448.
and psychopathology. Soc. Indicators Res. 54 (1), 81–108. NHS Health Scotland. Warwick-Edinburgh Mental Well-being scale (WEMWBS) User
Hargreaves, A., 2000. Mixed emotions: teachers’ perceptions of their interactions with guide – Version 2 2015 [Available from: https://www2.warwick.ac.uk/fac/med/
students. Teaching Teacher Educ. 16 (8), 811–826. research/platform/wemwbs/researchers/userguide/wemwbs_user_guide_jp_02.02.
Hawton, K., Harriss, L., 2007. Deliberate self-harm in young people: characteristics and 16.pdf.
subsequent mortality in a 20-year cohort of patients presenting to hospital. J. Clin. Plenty, S., Östberg, V., Almquist, Y.B., Augustine, L., Modin, B., 2014. Psychosocial
Psychiatry. working conditions: an analysis of emotional symptoms and conduct problems
Henderson, M., Harvey, S.B., Øverland, S., Mykletun, A., Hotopf, M., 2011. Work and amongst adolescent students. J. Adolesc. 37 (4), 407–417.
common psychiatric disorders. J. R. Soc. Med. 104 (5), 198–207. Reilly, M.C., Zbrozek, A.S., Dukes, E.M., 1993. The validity and reproducibility of a work
Hughes, J., Kwok, O.-m., 2007. Influence of student-teacher and parent-teacher re- productivity and activity impairment instrument. Pharmacoeconomics 4 (5),
lationships on lower achieving readers' engagement and achievement in the primary 353–365.
grades. J. Educ. Psychol. 99 (1), 39. Rose, G., 2001. Sick individuals and sick populations. Int. J. Epidemiol. 30 (3), 427–432.
Jain, G., Roy, A., Harikrishnan, V., Yu, S., Dabbous, O., Lawrence, C., 2013. Patient- Rothì, D.M., Leavey, G., Best, R., 2008. On the front-line: teachers as active observers of
reported depression severity measured by the PHQ-9 and impact on work pro- pupils’ mental health. Teaching Teacher Educ. 24 (5), 1217–1231.
ductivity: results from a survey of full-time employees in the United States. J. Ryan, R.M., Deci, E.L., 2001. On happiness and human potentials: a review of research on
Occupational Environ. Med. 55 (3), 252–258. hedonic and eudaimonic well-being. Ann. Rev. Psychol. 52 (1), 141–166.
Jamal, F., Fletcher, A., Harden, A., Wells, H., Thomas, J., Bonell, C., 2013. The school Sisask, M., Värnik, P., Värnik, A., Apter, A., Balazs, J., Balint, M., et al., 2014. Teacher
environment and student health: a systematic review and meta-ethnography of satisfaction with school and psychological well-being affects their readiness to help
qualitative research. BMC Public Health 13 (1), 798. children with mental health problems. Health Educ. J. 73 (4), 382–393.
Jennings, P.A., Frank, J.L., Snowberg, K.E., Coccia, M.A., Greenberg, M.T., 2013. Spilt, J.L., Koomen, H.M., Thijs, J.T., 2011. Teacher wellbeing: the importance of tea-
Improving classroom learning environments by Cultivating Awareness and Resilience cher–student relationships. Educ. Psychol. Rev. 23 (4), 457–477.
in Education (CARE): results of a randomized controlled trial. School Psychol. Q. 28 Stansfeld, S.A., Rasul, F., Head, J., Singleton, N., 2011. Occupation and mental health in a
(4), 374. national UK survey. Soc. Psychiatry Psychiatr. Epidemiol. 46 (2), 101–110.
Jennings, P.A., Greenberg, M.T., 2009. The prosocial classroom: teacher social and Stewart-Brown, S., Kulsum Janmohamed. Warwick-Edinburgh Mental Well-being Scale
emotional competence in relation to student and classroom outcomes. Rev. Educ. Res. (WEMWBS) User guide Version 1 2008 [Available from: http://www.
79 (1), 491–525. mentalhealthpromotion.net/resources/user-guide.pdf.
Johnson, S., Cooper, C., Cartwright, S., Donald, I., Taylor, P., Millet, C., 2005. The ex- Tennant, R., Hiller, L., Fishwick, R., Platt, S., Joseph, S., Weich, S., et al., 2007. The
perience of work-related stress across occupations. J. Manag. Psychol. 20 (2), Warwick-Edinburgh mental well-being scale (WEMWBS): development and UK va-
178–187. lidation. Health Qual. life Outcomes 5 (1), 63.
Joinson, C., Kounali, D., Lewis, G., 2017. Family socioeconomic position in early life and Trompetter, H., Lamers, S., Westerhof, G.J., Fledderus, M., Bohlmeijer, E.T., 2017. Both
onset of depressive symptoms and depression: a prospective cohort study. Soc. positive mental health and psychopathology should be monitored in psychotherapy:
Psychiatry Psychiatr. Epidemiol. 52 (1), 95–103. confirmation for the dual-factor model in acceptance and commitment therapy.
Keyes, C.L., Wissing, M., Potgieter, J.P., Temane, M., Kruger, A., Van Rooy, S., 2008. Behav. Res. Therapy 91, 58–63.
Evaluation of the mental health continuum–short form (MHC–SF) in setswana- Welsh Goverment. My Local School 2017 [Available from: http://mylocalschool.wales.
speaking South Africans. Clin. Psychol. Psychotherapy 15 (3), 181–192. gov.uk/.
Kidger, J., Araya, R., Donovan, J., Gunnell, D., 2012. The effect of the school environment World Health Organisation. Mental Health, Depression: let's talk 2018 [Available from:
on the emotional health of adolescents: a systematic review. Pediatrics 129 (5), http://www.who.int/mental_health/management/depression/en/.
2011–2248. Youth in Mind. Scoring the SDQ 2016 [cited 2017. Available from: http://www.sdqinfo.
Kidger, J., Brockman, R., Tilling, K., Campbell, R., Ford, T., Araya, R., et al., 2016a. com/py/sdqinfo/c0.py.
Teachers' wellbeing and depressive symptoms, and associated risk factors: a large

466