OUP user menu

The relationship between personal debt and specific common mental disorders

Howard Meltzer, Paul Bebbington, Traolach Brugha, Michael Farrell, Rachel Jenkins
DOI: http://dx.doi.org/10.1093/eurpub/cks021 108-113 First published online: 20 March 2012


Background: Personal debt is now recognized as one of the many factors associated with common mental disorders (CMD). We aim to estimate the prevalence of ‘specific’ mental disorders based on ICD-10 research diagnostic criteria by type of debt and quantify the additional influence of addictive behaviours. Method: A random probability sample comprising 7461 respondents were interviewed for the third national survey of psychiatric morbidity of adults in England carried out in 2007. The prevalence of CMD was estimated from the administration of the CIS-R. Respondents were asked about sources of debt and their borrowing choices. Results: In 2007, 8.5% of adults were in arrears. Adults in debt were three times more likely than those not in debt to have CMD. The increased likelihood of CMD among those in arrears was found for all CMD and was irrespective of source of debt—housing, utilities and purchases on credit. The situation was exacerbated among those with addictive behaviours—alcohol or drug dependence or problem gambling. Those with multiple sources of debt and who had to obtain money from pawnbrokers and moneylenders had the highest rate of CMD, ∼50%. Conclusions: Debt is one of the major risk factors for CMD. This has practical implications for both health services and financial services, which both need to be alert to the association and adapt and train their respective services accordingly so that people in debt can access help for mental disorders and people with mental disorders can access help for debt.


Increasing rates of unemployment, cuts in pensions and benefits and higher prices in the shops are all factors that are likely to contribute to the financial stress put on individuals and families. Some will cope by looking for opportunities to reduce the impact of the financial strain; others may fall, or fall further, into debt. Problems with financial indebtedness and the impact that financial stress has on family well-being have loomed large in media coverage of the consumer credit market in recent years, heightened by the impact of the ‘credit crunch’.1

There has been some research on the relationship between debt and mental disorders but quite often different approaches have been used. The two principal methodological challenges have been the measurement of debt and the assessment of mental health problems.

In large epidemiological surveys, it is rarely possible to include questions about specific amounts of money relating to individual items of expenditure (housing costs, heating costs and weekly shopping bills), as they are time consuming to collect and increase respondent burden. In many cases, respondents are merely asked whether they can make ends meet or details of their budgeting strategies. These subjective assessments are then used as proxies for actual indebtedness and for underlying household budgetary problems.

The second limitation of previous studies is restriction to the relationship of debt with depression1,2 or with psychological distress3,4 as measured by the GHQ12,5 as distinct from specific mental disorders based on ICD-10 diagnostic criteria.

Transcending these two methodological limitations is the conceptual problem of reciprocal causality. In one interpretation, people get into debt for a variety of reasons (gambling, substance abuse, compulsive shopping, marital or relationship breakdown and redundancy), and these factors, alone or in combination, then increase the risk of anxiety and depression. The four most frequently stated reasons for debt by clients of Citizens Advice Bureaux in 2008 were overcommitment, job loss, low income and illness.6 Theodossiou (1998) hypothesized direct mental health consequences from economic distress, particularly low pay and unemployment.7 In the converse interpretation, individuals with mental disorders are less likely than others to obtain or maintain employment, and may also find it difficult to apply for benefits and to budget, leading to indebtedness. Their debt may also be exacerbated by a failure to appreciate its degree.

It is likely that both sets of mechanisms occur—people with debts are more likely to have mental health problems and people with mental health problems are more likely to be in debt.8

The third national survey of psychiatric morbidity among adults in England carried out in 20079 provides an opportunity for a more refined analysis of the relationship between debt and common mental disorders (CMD), as it included standard procedures for identifying CMD and detailed enquiry about the nature and occasions of debt, together with questions about payment arrears in relation to a list of 20 or so areas of indebtedness.10

The aim of this study is to look at the association between debt and mental health problems by examining the relationship between characteristics of debt and ‘specific’ mental disorders based on ICD-10 research diagnostic criteria. For current purposes, we are taking CMD as the outcome variable and the debt characteristics as the risk factors. We hypothesized that (i) debt would be associated with an increased prevalence of all CMD, (ii) debt would interact with addictive behaviour (alcohol dependence, drug dependence and problem gambling) to further increase the likelihood of CMD and (iii) arrears in paying for utilities would provoke more anxiety and depression than delays in credit card repayments, but number of debts would be the key correlate. The rationale behind this hypothesis was that individuals with utilities debt will be at a heightened risk of developing a CMD because they will be fearful and anxious about such arrears and this could affect their willingness to put on the heating, turn on the lights or use water—key essentials of living. We also hypothesized that those borrowing money from pawnbrokers (who offer loans to people, with items of personal property used as collateral) and from moneylenders (who offer small personal loans at high rates of interest) as distinct from family and friends would tend to have higher rates of CMD.


Sampling procedures

This analysis is based on a stratified multistage random probability sample, selected for the third national survey of psychiatric morbidity among adults in England carried out in 2007.9

In the first stage of sampling, postcode sectors (on average 2550 households) were stratified by socio-economic status within a regional breakdown. Postcode sectors were then sampled from each stratum with a probability proportional to size (where size was measured by the number of delivery points). In this way, a total of 519 postal sectors were selected.

In the second stage of sampling, 28 delivery points were randomly selected within each of the selected postal sectors. This yielded a total sample of 14 532 delivery points. Interviewers visited these addresses to identify private households containing at least one person aged 16 and over. Within the potentially eligible sample of 12 694 addresses, one person was randomly selected in each household where contact was made to take part in the survey, using the Kish grid method.11 The residents of 57% of all eligible households agreed to take part in an interview: 7461 people.

Interviewers and interviewing procedures

Experienced interviewers from the National Centre for Social Research were selected to work on the survey, many of whom had worked previously on health-related surveys.12 They were fully briefed on the administration of the survey. Topics covered in the 1-day survey-specific training included introducing the survey, the questionnaire content, confidentiality and how to handle respondent distress. The fieldwork took place over the course of 1 year.



Survey respondents were asked: ‘Have there been times in the past year when you were seriously behind in paying within the time allowed for any of these items?’ The list included rent, gas, electricity, water, goods bought on hire purchase, mortgage repayments, council tax, credit card payments, mail order payments, telephone, other loans, TV licence, road tax, social fund loan, child support or maintenance. Being behind in any of these payments was regarded as having a debt.

For purposes of analysis, debts were grouped into three main categories: utilities (gas, electricity and water rates), housing-related debts (mortgage, rent and council tax) and shopping-related debt (credit card, higher purchase and mail order debt). The remaining sources of debt were included in an ‘other’ category.

Personal, family and household characteristics

Information on gender, age, ethnicity and marital status of all household members was collected, as well as the socio-economic circumstances of the household: educational level, employment status and housing tenure.


Diagnoses of CMD were derived from responses to the revised Clinical Interview Schedule.13 Diagnostic algorithms based on ICD-10 criteria14 were applied to the data to identify six categories of CMD—generalized anxiety disorder, depression, obsessive-compulsive disorder (OCD), phobia, panic disorder and mixed anxiety and depressive disorder.

Alcohol dependence

The severity of alcohol dependence questionnaire (SADQ-C) was used to estimate the prevalence of alcohol dependence in the past 6 months.15 The SADQ-C consists of 20 items, covering a range of dependence symptoms. Answers to all questions are scored from 0 to 3, and summed to give a total score ranging from 0 to 60. All dependence categories (scores of four or more) were amalgamated to produce a dichotomous variable.

Drug use and dependence

Questions about drug use covered lifetime experience of 13 types of named drugs together with their use in the past year. For each of eight drug types (cannabis, amphetamines, crack, cocaine, ecstasy, tranquillizers, opiates and volatile substances), reports of past year use were followed up by five questions, based on the Diagnostic Interview Schedule16 and designed to assess drug dependence. These questions asked about the past month and year and covered daily use for 2 weeks or more, a sense of need or dependence, an inability to abstain, increased tolerance and withdrawal symptoms. A positive response to any of the items in the past year was used to indicate drug dependence.

Problem gambling

Initial questioning asked whether respondents had spent money on gambling in the last year. Examples of gambling activities were provided. All those who gambled even occasionally were administered a 10-item gambling screen based on DSM-IV criteria.17 The number of DSM-IV criteria endorsed was summed to generate a score: five or more positive responses were considered to be indicative of problem gambling.

Statistical analysis

SPSS (version 18.0) was used to analyse the survey data as it allows the use of clustered data inherent in complex survey designs. Initially, both univariate and multivariate logistic regression analyses were carried out to examine the association between socio-demographic variables and debt in relation to any CMD. All the socio-demographic correlates were then carried forward as potential confounders in further multivariate logistic regression modelling to investigate the relationship between indebtedness and the six categories of CMD. Three derived variables were then created in order to encapsulate the interaction of debt with three addictive behaviours (alcohol dependence, drug dependence and problem gambling), and these were used in multivariate logistic regression analysis to examine their effect on the prevalence of CMD. Cross-tabulations were used to compare the prevalence of CMD by number and source of debts and by the number and type of people lending money to those in debt.

Data were weighted to take account of non-response so that the results were representative of the household population aged 16 years and over in England. Weighting occurred in three steps. First, sample weights were applied to take account of the different probabilities of selecting respondents in different sized households. Secondly, weights were created to reduce household non-response bias. Finally, weights were applied using the techniques of calibration weighting based on age, sex and region to weight the data up to represent the structure of the national population, to take account of differential non-response between regions, and age-by-sex groups. The population control totals used were the Office for National Statistics (ONS) 2006 mid-year household population estimates.


In England, 8.5% of adults reported being in debt in 2007 (excluding planned mortgage repayments, but including arrears in mortgage repayments). Among those in debt, 38% were assessed as having any CMD of some kind, almost three times the rate of the no debt group (13.9%). Several socio-demographic and socio-economic factors that increased the odds of having a CMD emerged from the univariate logistic regression analysis: being young, female, non-married, not working and living in rented accommodation as well as being in debt (table 1). When all of these factors were entered into a multivariable logistic regression model, debt was retained as the correlate with the greatest odds ratio [OR = 2.83, 95% confidence interval (CI) 2.34, 3.43, P < 0.001].

View this table:
Table 1

Debt and socio-demographic correlates of CMD

NPercentage with CMDUnadjusted OR95% CIP valuesAdjusted ORa95% CIP values
    Not in debt667813.91.001.00
    In debt62338.03.823.20–4.55<0.0012.832.34–3.43<0.001
Age (years)
Marital status
    Married or cohabiting460513.81.001.00
    Widowed, divorced or separated105321.01.661.40–1.96<0.0011.391.15–1.680.001
Employment status
    Economically inactive264718.61.441.27–1.64<0.0011.631.39–1.91<0.001
    Owner occupiers509812.81.001.00
  • a: Adjusted for all other socio-demographic variables shown in table

Indebtedness was associated with increased rates of each ICD category of CMD (table 2). After adjustment for confounders, those in debt were nearly four times as likely to have phobic disorders (social phobia and specific isolated phobias) (OR = 3.83, 95% CI 2.43–6.05, P < 0.001), three times more likely to have panic disorder (OR = 3.14, 95% CI 1.79–5.52, P < 0.001) and more than twice as likely to have OCD (OR = 2.27, 95% CI 1.32–3.90, P = 0.002), depressive disorder (OR = 2.36, 95% CI 1.59–3.50, P < 0.001) and generalized anxiety disorder (OR = 2.51, 95% CI 1.85–3.41, P < 0.001).

View this table:
Table 2

Debt correlates of six common mental disorders

Unadjusted OR95% CIP valuesAdjusted ORa95% CIP values
    Not in debt1.001.00
    In debt7.234.81,10.88<0.0013.832.43–6.05<0.001
    Not in debt1.001.00
    In debt4.792.94–7.70<0.0012.271.32–3.900.002
Depressive episode
    Not in debt1.001.00
    In debt4.082.87–5.81<0.0012.361.59–3.50<0.001
Panic disorder
    Not in debt1.001.00
    In debt3.812.28–6.40<0.0013.141.79–5.52<0.001
Generalized anxiety disorder
    Not in debt1.001.00
    In debt3.492.65–4.60<0.0012.511.85–3.41<0.001
Mixed anxiety and depressive disorder
    Not in debt1.001.00
    In debt2.612.10–4.55<0.0012.101.65–2.66<0.001
  • a: Adjusted for age, sex, marital status, employment status and tenure

Being in debt and having an addictive behaviour—alcohol dependence, drug dependence or problem gambling—increased the likelihood of having a CMD. Those with any of the addictive behaviours and being in debt were about seven or eight times more likely to have a CMD than the no addiction, no disorder group (table 3). Controlling for socio-demographic and socio-economic factors made very little difference to the strength of the relationship.

View this table:
Table 3

Debt and addiction correlates of any CMD

Unadjusted OR95% CIP valuesAdjusted ORa95% CIP values
Debt and alcohol dependence profile
    No debt, not dependent1.001.00
    In debt, not dependent3.863.20–4.65<0.0012.942.41–3.63<0.001
    No debt, dependent3.012.38–3.81<0.0013.402.65–4.37<0.001
    In debt, dependent7.844.73–12.98<0.0017.094.18. 12.05<0.001
Debt and drug dependence profile
    No debt, not dependent1.001.00
    In debt, not dependent3.432.84–4.14<0.0012.602.12–3.18<0.001
    No debt, dependent2.131.50–3.03<0.0012.101.46–3.03<0.001
    In debt, dependent7.414.87–18.58<0.0018.445.02–14.19<0.001
Debt and problem gambling profile
    No debt, no problem1.001.00
    In debt, no problem3.783.16–4.52<0.0012.812.31–3.42<0.001
    No debt, with problem4.152.11–8.16<0.0014.342.16–8.70<0.001
    In debt, with problem7.412.92–18.79<0.0016.952.67–18.13<0.001
  • a: Adjusted for age, sex, marital status, employment status and tenure

Although the source of debt had little effect on the prevalence of CMD, the number of debts did seem to be a relevant factor (table 4). Taking the absence of indebtedness as the reference category, about 14% of this group had a CMD. The rate doubled to around 30% among those with one or two debts and nearly doubled again to around 55% of the group with three or more debts. The chi-squared linear-by-linear statistic was 21.89 (df = 1, P < 0.001) indicating a significant trend.

View this table:
Table 4

Prevalence of CMD by source of debt, number of debts, source of loan and number of lenders

Source of debtNo debtHousingaUtilitiesbShoppingcOther debtsd
Any CMD (%)13.941.844.144.840.8
Number of debts01234+
Any CMD (%)13.932.327.054.354.3
Source of loanFamilyFriendsPawnbrokerMoneylender
Any CMD (%)34.244.345.357.5
Number of lenders0123
Any CMD (%)13.934.842.752.2
  • a: Housing debts comprise arrears in rent, mortgage and council tax

  • b: Utilities debt comprise arrears in gas, electricity and water

  • c: Shopping debts comprise arrears in hire purchase, credit card and mail order

  • d: Other debts include arrears in telephone, TV licence, road tax and child maintenance

Focusing on the sample who reported being in arrears and who then borrowed money to assist repayment, the CMD distribution varied somewhat by type of lender, but the analysis is to some extent complicated by the fact that 26% of those who borrowed money did so from more than one source. Nevertheless, those who borrowed from moneylenders had the highest rate of CMD, 58%, and those who borrowed from multiple sources also tended to have higher rates of CMD than single source borrowers.


This article extends understanding of the relationship between debt and mental disorder by quantifying the relationship between debt and specific ICD diagnoses and by quantifying the additional influence of addictive behaviours. It also enables comparison of the rates of indebtedness between 2000 and 2007. In England, 8.5% of adults reported being in debt in 2007. The previous psychiatric morbidity survey in England in 2000 reported a rate of 11.6%18 but that figure was based on a sample of 16–74 year olds, whereas the 2007 survey had no upper age limit. Comparing the 2007 and 2000 data using the same age range shows that the prevalence of debt among 16–74 year olds in 2007 decreased from 11.6 to 9.2%. The slight fall in debt from 2000 to 2007 is somewhat surprising given the common assumption that Britain has become a more indebted society in recent years, although only a future survey will indicate if there have been increases in debt since 2007.

The increased likelihood of debt among particular socio-demographic and socio-economic groups found in this study (16–34 year olds, women, non-married adults, unemployed people and those in rented accommodation) has also been reported in health economic studies.1

Many other factors besides debt have been reported as independent correlates of CMD. In addition to CMD, debt has been found to be independently associated with poor housing quality, job stress, lower levels of social support, recent stressful life events, domestic violence and caring responsibilities.19

The increased prevalence of each category of CMD among the indebted population may to some extent reflect internal comorbidity within the overall category of CMD. Nevertheless, the size of the relationship varies with the type of ICD category, with phobia being the most pronounced, possibly exacerbated by fear or worry about being incapable of paying the money back on time or of something going wrong and acting as a stumbling block to fulfilling repayment obligations.

The relationship between indebtedness and OCD may be influenced by compulsive shopping as this has been reported in case studies of individuals with OCD.20

Debt and addictive behaviour

The association between debt and addictive behaviours and CMD adds quantifiable epidemiological evidence to the plethora of anecdotal material about the financial problems associated with alcohol or drug dependence and with depression. Debt counselling services and addiction treatment centres allude to alcohol and drug problems causing, exacerbating or resulting from debt.

However, our analysis indicates that there is not a multiplicative effect of debt and additive behaviours. It would only be a multiplicative effect if the presence of both factors increases the likelihood of CMD by more than the multiple of the individual OR. This is not the case. Therefore, it looks as though addictive behaviours have an effect on CMD separate from debt. One might expect the interaction between debt and gambling to be more immediate and therefore larger than that between the debt and the other addictive behaviours, but there was no related evidence.

Little has been published in the scientific literature on the relationship between substance abuse, disposable income, debt and depression derived from large, national, epidemiological studies. Data have been published on the socio-economic characteristics of those with alcohol and drug dependence,21 but the emphasis has been on low income rather than debt.22 In the numerous studies on the causes of debt (redundancy, relationship breakdown, gambling and depression), some reference has also been made to alcohol and drug problems.4 The few studies focussing on the relationship between alcohol and drug problems and debt and mental health problems have involved particular groups in which substance abuse, debt and CMD are known to be major problems—such as students,3 homeless people23 and prisoners.24

Sources of debt and borrowing behaviour

Contrary to our hypothesis, source of debt had little effect on the prevalence of CMD. However, this may reflect the equivalence of the perceived negative consequences of being disconnected from electricity or gas and the stress or pressure of receiving hourly phone calls, multiple letters and text messages from credit card companies wanting repayments.

The increasing rate of CMD with increasing numbers of debt may reflect the magnitude of debt. Although data on the actual amounts of money in arrears were not collected, one can hypothesize that being in arrears in multiple areas especially for the necessities of life (housing and heating) must increase the stress of coping with these financial difficulties. Similarly, borrowing from moneylenders must be psychologically more difficult to deal with than borrowing from friends or family.

Advantages and limitations of the study

The main advantages of our study include the sampling procedures that provided a large nationally representative community sample across the age spectrum, the ability to control for important confounding factors and the use of well-validated instruments and epidemiological methods to measure mental disorders and debt. Although the response rate of 57% might potentially influence the results, weighting procedures were performed to reduce any potential non-response biases. The main drawback of the study is the cross-sectional design, thereby limiting the ability to form aetiological inferences or examine maintaining factors. It would be worthwhile carrying out further longitudinal research with well-validated instruments of mental disorders as well as detailed assessments of debt, looking at onset, amount, source and duration. Services treating adults presenting with these disorders should also routinely enquire about debt and consider the value of debt counselling.


Debt and particular sources of debt are key correlates of all CMD, both as a whole and analysed by separate ICD category. The direction of causation is likely to run in both directions, namely that people in debt are more likely to develop a mental disorder and people with a mental disorder are more likely to get into debt.

The relationship between debt and CMD should be taken into account in national policy dialogue about the role of debt in society and in the training of staff of companies and agencies involved with giving loans and managing and recovering debts. Some training material already includes such guidance.25,26 A high proportion of people presenting for online, telephone or face-to-face assistance with debt are likely to have a mental disorder, and so in addition to knowledgeable finance staff, clinical support from primary and secondary care services should be available and signposted.

This need was highlighted in the recent Foresight project27,28 and since then considerable efforts have been made to address these issues.29 More remains to be done, and will need to be integrated into the new health service and financial service reforms.


The data on adult psychiatric morbidity in England, 2007 were derived from a national survey commissioned by the NHS Information Centre for Health and Social Care.

Conflicts of interest: None declared.

Key points

  • People in debt have three times the rate of CMD than those not in debt.

  • CMD rates increase with the number of debts incurred.

  • People in debt are nearly four times as likely to have phobic disorders (social phobia and specific isolated phobias) than the adults not in debt.

  • The relationship between debt and CMD should be included in the training manuals of companies and agencies involved with giving loans and managing and recovering debts.


View Abstract