. 2019 Aug 12;73(10):977–984. doi: 10.1136/jech-2019-212154

Table 1.

Characteristics of included studies

Study	Year(s) of data collection	Sampling/selection	N and response rate	Study design	Finance measures	Mental health measures	Other outcome measures	Analytical strategy	Main findings
Andrews and Wilding, 2004²	2000–2002	One pre-1992 University in London, England (Royal Holloway). Undergraduates only.	676 students at time point 1 (T1), 351 at time point 2 (T2); response rate 76%	Prospective cohort—two time points. One month before first year and halfway through second year	Financial difficulties: major financial crisis, unable to afford essentials (item taken from adverse life experiences list)	Anxiety and depression (HADS), adverse life experiences (death, separation, etc)	Examination results (at the end of second year) obtained from University Registrar	Logistic regression to determine relative contribution of variables to the prediction of anxiety and depression (separately) midcourse. Adjusted for mental health at baseline, caseness of anxiety or depression midcourse, gender, age and ethnicity.	Found an association between financial difficulties and depression, but not anxiety. Depression mediated the relationship between financial difficulties and subsequent exam performance.
Cooke et a l, 2004¹¹	2000–2003	One pre-1992 University in England (University of Leeds). Undergraduates only.	38% at T1 (n=2146), 23% (n=1360) at T2 and 26% (n=1391) at T3	Prospective cohort—three time points at the end of semester one each year	Anticipated amount of debt (T3 only); financial concern (one question) and debt worry (one question, T3 only)	Mental health functioning, problems and general mental health (CORE-GP)		Effect size differences using z scores (0.4 SD units as threshold).	Found an association between debt worry/financial concern and mental health, but no association between amount of debt and mental health.
Jessop et al, 2005¹⁶	Unknown	Middlesex University, London; overseas students excluded (also conducted in Finland but separate analyses reported here only). Graduate status unclear.	89 students; response rate 95.2% (across London and Finland)	Cross-sectional survey	Amount of debt; financial concern: 6 5-point Likert-scale items, for example, ‘I worry about my financial situation’	General health (SF-12)—includes questions on mental health and its interference with role		Correlations; multiple regressions with amount of debt, age, gender, financial concern, no. of hours worked and units of alcohol as predictors and each of the SF-12 dimensions as outcomes.	Found higher debt and more financial concern were correlated with worse mental health, but financial concern did not predict mental health functioning. Financial concern was positively correlated with amount of debt.
Richardson et al, 2015⁸	2012–2014	All UK student unions contacted (46 of 114 agreed). Undergraduates only.	390 students; response rate unclear (very low)	Prospective cohort—four time points ~2 months apart over first 2 years at university	Fee amount (amount of debt): 0-£2.9k, 3–4k or 8–9k	Anxiety (GAD-7), depression (CES-D), general mental health (CORE-GP), stress (PSS)		Correlation between tuition fee group and student well-being, with no apparent adjustment for confounders.	Found no association between debt and well-being.
Richardson et al, 2015²³	Unknown	All UK student unions contacted (46 of 114 agreed). Undergraduates only.	444 students (completed baseline and at least one other time point); response rate unclear (very low)	Prospective cohort—four time points 3–4 months apart over first 2 years at university	Financial difficulties over past 6 months (IFS)	Attitudes towards food and eating (EAT-26)		Hierarchical linear multiple regressions to test whether IFS at baseline predicted later EAT-26 scores (and vice versa), after adjusting for socioeconomic status, baseline EAT-26, gender, ethnicity and age. Significant findings repeated separately for each gender. Missing data replaced with the mode.	Found that financial difficulties predict more severe long-term eating attitudes, for women only. Mixed findings for a bidirectional relationship.
Richardson et al, 2017¹⁴	2012–2014	All UK student unions contacted (46 of 114 agreed). Undergraduates only.	454 students; response rate unclear (very low)	Prospective cohort—four time points ~2 months apart over first 2 years at university	Financial difficulties: over past 6 months (IFS), considering abandoning course for financial reasons; debt stress and view of loan	Anxiety (GAD-7), depression (CES-D), general mental health (CORE-GP), stress (PSS)		Linear hierarchical multiple regression used to see whether financial variables predicted mental health scores at each time point, adjusting for age, gender, disability, mature student status, ethnicity and mental health at baseline. Regression models with baseline IFS, socioeconomic status, demographics and all mental health measures to predict IFS at T2. Missing data filled in with the mode.	Found an association between financial difficulties and mental health cross-sectionally, but mixed longitudinal findings. Evidence points to a potential bidirectional relationship. Greater subjective stress about debt predicted worse mental health cross-sectionally, and at the longest follow-up point.
Richardson et al, 2018²⁰	2012–2014	All UK student unions contacted (46 of 114 agreed). Undergraduates only.	408 students; response rate unclear (very low)	Prospective cohort—four time points ~2 months apart over first 2 years at university	Amount of debt; financial difficulties: over past 6 months (IFS), considering abandoning course for financial reasons; debt stress	Psychosis risk (Prodromal Questionnaire-Brief Version; PQB)		Hierarchical multiple linear regression with financial variables as predictors and PQB score at baseline as outcome (controlling for gender, age and ethnicity). Significant variables entered into a regression with PQB over time as outcome.	Found that debt stress and amount of debt are not associated with psychosis risk. Some indicators of financial difficulies are associated with psychosis risk (i.e. IFS scores), but others are not (i.e.considering abandoning). No evidence of bidirectionality.
Roberts et al, 1998⁶	‘Over the past year’ —1997–1998 assumed.	Unknown British university. Undergraduates (83%) and postgraduates.	103 students; response rate unclear	Cross-sectional survey	Amount of debt; financial difficulties: difficulty paying bills, considering abandoning course for financial reasons	General mental health (GHQ)		Not reported	Found that there is an association between mental health and financial difficulties. Poorer mental health was significantly related to difficulty paying bills. People who had considered abandoning for financial reasons had significantly poorer mental health.
Roberts et al, 1999²¹	Unknown	Students from two universities in London—one pre-1992 and one post-1992. Undergraduates (90%) and postgraduates	360 students; response rate 65%	Cross-sectional survey	Amount of debt; financial difficulties: difficulty paying bills, considering abandoning for financial reasons	General mental health (SF-36 and GHQ-12)		Regression models with GHQ and SF-36 subscale scores as outcome and hours worked, difficulty paying bills and considering dropping out as predictors. Structural equation modelling of pathways linking financial variables and mental health.	Found that there is an association between mental health and financial difficulties (difficulty paying bills and considering dropping out). Conclude that there are two pathways through which amount of debt is associated with mental health.
Roberts et al, 2000²²	Unknown	Students from two universities in London—one pre-1992 and one post-1992. Undergraduates (87%) and postgraduates	482 students; response rate 66%	Cross-sectional survey	Amount of debt; financial difficulties: difficulty paying bills, considering abandoning course for financial reasons	General mental health (SF-36 and GHQ-12)	Help seeking: whether have consulted a GP in past 2 weeks and satisfaction with most recent consultation	Linear regression models adjusted for age and sex to assess association between financial difficulties and GHQ scores. Regression models to examine the relationship between debt and help seeking (adjusted for age and gender). Structural equation modelling of pathways linking financial variables and mental health.	Found that there is an association between financial difficulties (difficulty paying bills and considering dropping out) and mental health. Two pathways confirmed (as above).
Ross et al, 2006⁷	2004	Undergraduate medical students from one university in Scotland (University of Aberdeen).	352 responses out of 900 students (39% response rate)—334 included.	Cross-sectional survey	Amount of debt; questions on worrying about money	General mental health (GHQ-12)	Examination results (students’ rankings relative to the rest of the year group)	Pearson’s partial correlations controlling for effect of year.	Found that indicating that worrying about money affects performance was associated with poorer mental health. Mixed findings on relationship between debt and mental health.

CES-D, Center for Epidemiological Studies - Depression scale; CORE-GP, General Population version of the Clinical Outcomes in Routine Evaluation Questionnaire; EAT-26, Eating Attitudes Test-26; GAD-7, 7-item Generalised Anxiety Disorder Assessment; GHQ, General Health Questionnaire; GP, General Practitioner; HADS, Hospital Anxiety and Depression Scale; IFS, Index of Financial Stress; PSS, Perceived Stress Scale; SF-12, 12-item Short Form health survey; SF-36, 36-item Short Form health survey.