Validation of the WHO-5 as a first-step screening instrument for depression in adults with diabetes: Results from Diabetes MILES – Australia

https://doi.org/10.1016/j.diabres.2017.07.005Get rights and content

Highlights

  • This is the first study of its kind among English-speaking adults with diabetes.

  • The WHO-5 has satisfactory reliability (α = 0.90) and convergent and factorial validity.

  • A WHO-5 cut-off of <13 is optimal to screen for likely depression.

  • The WHO-5 is a suitable first-step screening instrument for likely depression.

Abstract

Aims

Screening for depression is recommended internationally. The World Health Organization’s 5-item Well-being Index (WHO-5) is used clinically to screen for depression but its empirical suitability for this purpose is not well documented. We investigated the psychometric properties of the WHO-5 and its suitability for identifying likely depression in Australian adults with diabetes.

Methods

The Diabetes MILES – Australia study dataset provided a sample of N = 3249 who completed the WHO-5 (positively-worded 5-item measure of emotional well-being) and the PHQ-9 (9-item measure of depressive symptoms). Analyses were conducted for the full sample, and separately by diabetes type and treatment (type 1, non-insulin-treated type 2, and insulin-treated type 2 diabetes). Construct (convergent and factorial) validity and reliability of the WHO-5 were examined. ROC analyses were used to examine the sensitivity and specificity of the WHO-5 as a depression screening instrument, comparing two commonly used WHO-5 cut-off values (≤7 and <13) with the PHQ-9.

Results

For the whole sample, the WHO-5 demonstrated satisfactory internal consistency reliability (α = 0.90) and convergent validity with the PHQ-9 (r = −0.73, p < 0.001). Confirmatory factor analysis partially supported factorial validity: Χ2(5) = 834.94, p < 0.001; RMSEA = 0.23, 90% CI 0.21–0.24; CFI = 0.98, TLI = 0.96; factor loadings = 0.78–0.92. The AUC was 0.87 (95% CI: 0.86–0.89, p < 0.001). The sensitivity/specificity of the WHO-5 for detecting likely depression was 0.44/0.96 for the ≤7 cut-off, and 0.79/0.79 for the <13 cut-off, with similar findings by diabetes type and treatment.

Conclusions

These findings support use of a WHO-5 cut-point of <13 to identify likely depression in Australian adults with diabetes, regardless of type/treatment.

Introduction

Successful diabetes management requires attention not only to biomedical markers, co-morbidities and diabetes-related complications, but also to the impact of living with diabetes on psychological outcomes. The prevalence of depression is approximately two-fold higher in people with type 1 diabetes and type 2 diabetes than the general population [1]. Living with co-morbid diabetes and depression is associated with less engagement in diabetes self-care and treatment, sub-optimal glycaemic outcomes, increased rates of diabetes-related complications, decreased quality of life, and greater mortality [2], [3]. Yet, it is common for depression to go undetected and untreated in clinical practice [4].

Though a clinical diagnostic interview is required to formally diagnose depression, short screening instruments are a useful first-step, to quickly and systematically identify people who may be experiencing clinically relevant depressive symptoms (referred from here-on as people with likely depression). Depression screening is a recommended care practice [5], [6] and previous studies have shown that the use of a brief instrument to screen for likely depression is an acceptable practice within diabetes clinical settings [7], [8], [9]. Despite the recommendations, there is limited evidence regarding which screening instruments are optimal for use in diabetes [10]; this leaves diabetes health practitioners with a difficult choice regarding which screening instrument to use in clinical practice [11]. Numerous brief self-report screening instruments for depression exist, including (but not limited to) the Patient Health Questionnaire Nine (PHQ-9; nine items) [12] and the Centre for Epidemiologic Studies Depression Scale (CES-D; 20 items) [13]. Such instruments are psychometrically sound and relatively short, and useful in clinical practice for identifying people with diabetes who may experience depressive symptoms [11].

It has been our observation, when developing questionnaires or observing participants’ completion of questionnaires in clinical studies, that people with diabetes find the content of many depression screening tools to be too negative, and people without (or unaware of) depressive symptoms do not find depression screening relevant to their diabetes care. The acceptability of such questionnaires in non-psychiatric settings has been questioned by others [14]. It is possible that positively-worded questionnaires (e.g. asking to what extent the person has felt calm, relaxed and cheerful), may be more acceptable to people with diabetes, and the absence of positive mood may indicate depressive symptoms [15]. The World Health Organisation Well-being Index Five (WHO-5) [16] is one such example. It was developed as a brief measure of general emotional well-being in a sample of people with diabetes [16]. Its original development and validation study is published in a conference abstract [17], thus a comprehensive description of the original study methods and psychometric results are unavailable. Given its use in large, influential initiatives such as DAWN, DAWN-2 and DAWN-MIND [18], [19], there appears to be increasing demand for clinical use of the WHO-5 as a screening instrument for likely depression in people with diabetes.

The WHO-5 has been used in over thirty diabetes studies, some of which have made efforts to examine its psychometric properties [20]. For example, some exploration of the sensitivity and specificity of the WHO-5 as a screening instrument for depression has been undertaken with non-English speaking people with diabetes, with promising results [21], [22], [23], [24], [25]. However, there has been no comprehensive validation of the English-language WHO-5 among people with diagnosed diabetes. Additionally, the cross-cultural generalizability of the WHO-5 as a depression screener in people with diabetes is not well established. One Dutch study has called for further investigation of the psychometric performance of the WHO-5 in other cultural groups [24]. Furthermore, the lived experiences and risk of impaired well-being (including depression) are distinctly different for people with type 1, non-insulin-treated type 2, and insulin-treated type 2 diabetes [3]. Yet, there has been no investigation to date of the extent to which the WHO-5’s depression screening ability is generalizable across diabetes types and treatment regimens. As argued by Henkel and colleagues, “it is neither safe nor justified to assume that diagnostic validity would be homogenous…across subgroups” [27 p.191].

Therefore, the aim of the present study was to investigate the psychometric performance of the WHO-5 among Australian (English-speaking) adults with type 1, non-insulin-treated type 2, and insulin-treated type 2 diabetes. Our objectives were to examine the WHO-5 in terms of: (1) construct (convergent and factorial) validity and internal consistency reliability in these three groups; and (2) its sensitivity and specificity for detecting likely depression across groups, using the PHQ-9 [12] as the reference standard, as it is an accurate and widely used measure in diabetes research and clinical practice [27].

Section snippets

Materials and methods

Data were extracted from the Diabetes MILES (Management and Impact for Long-term Empowerment and Success) – Australia dataset. This national survey of adults with type 1 diabetes and type 2 diabetes focused on the psychosocial and behavioural aspects of living with diabetes. Full details have been published elsewhere [28]. In summary, the survey was posted to 15,000 adults with type 1 or type 2 diabetes aged 18–70 years, selected randomly from the National Diabetes Services Scheme database; the

Results

A total of 3338 Australians aged 18–70 years with type 1 diabetes or type 2 diabetes participated in the Diabetes MILES – Australia survey. Of these, 89 participants were excluded from the current study due to missing WHO-5 or PHQ-9 data.

Discussion

This is the first study to demonstrate that the WHO-5 has satisfactory psychometric properties and utility as a suitable screening instrument for likely depression among English-speaking adults with diabetes (type 1, non-insulin-treated type 2, and insulin-treated type 2 diabetes). The findings provide support for the internal consistency, reliability and convergent validity of the WHO-5, with a high Cronbach’s alpha and a strong negative correlation of expected magnitude and direction with the

Funding

The Diabetes MILES – Australia 2011 Survey was funded by a National Diabetes Services Scheme (NDSS) Strategic Development Grant. The NDSS is an initiative of the Australia Government administered with the assistance of Diabetes Australia. The Diabetes MILES Study was also supported by an unrestricted educational grant from Sanofi-Aventis. The funding sources had no role in the design, data collection, analysis or interpretation of the study, or in the decision to submit the manuscript for

Author contributions

JS and CH devised the concept for this manuscript. JH conducted the data analysis, and prepared a first draft and subsequent revisions; except for the CFA, which was conducted and text drafted by LB. CH, LB, JB, GN, FP and JS critically reviewed the manuscript for important intellectual content and all authors approve the final version as published. JS conceived the Diabetes MILES – Australia Study and together with FP developed the Diabetes MILES Study International Collaborative. JS and JB

Conflict of interest

None to declare.

Acknowledgements

The authors thank the Diabetes MILES – Australia 2011 survey participants, study team volunteers, and reference group.

References (49)

  • Royal Australian College of General Practitioners (RACGP): General practice management of type 2 diabetes 2016–18. East...
  • D. Young-Hyman et al.

    Psychosocial care for people with diabetes: a position statement of the American Diabetes Association

    Diabetes Care

    (2016)
  • K.M. van Steenbergen-Weijenburg et al.

    Validation of the PHQ-9 as a screening instrument for depression in diabetes patients in specialized outpatient clinics

    BMC Health Serv Res

    (2010)
  • Hendrieckx C, Bowden J, Halliday J, Colman P, Cohen N, Jenkins A, et al. An audit of psychological well-being in adults...
  • F. Pouwer

    Should we screen for emotional distress in type 2 diabetes mellitus?

    Nat Rev Endocrinol

    (2009)
  • F. Pouwer et al.

    Measuring and assessing depression in people with diabetes: implications for clinical practice

  • K. Kroenke et al.

    The PHQ-9 – validity of a brief depression severity measure

    J Gen Intern Med

    (2001)
  • L.S. Radloff

    The CES-D scale a self-report depression scale for research in the general population

    Appl Psychol Meas

    (1977)
  • E. Tarolla et al.

    Measuring depression with questions about well-being: a study on psychiatric outpatients

    Rivista di Psichiatria

    (2012)
  • C.E. Lloyd et al.

    Measuring psychological well-being in South Asians with diabetes; a qualitative investigation of the PHQ-9 and the WHO-5 as potential screening tools for measuring symptoms of depression

    Diabet Med

    (2012)
  • P. Bech

    Measuring the dimension of psychological general well-being by the WHO-5

    Quality of Life Newsletter

    (2004)
  • S. Shea et al.

    Routine assessment of psychological well-being in people with diabetes in primary care-validation of the WHO-5 Well-being Index in six countries (Abstract)

    Diabetologia

    (2003)
  • Novo Nordisk. The DAWN programmes, http://www.dawnstudy.com/dawn-programmes/about-dawn.html; 2015 [accessed 24 November...
  • F.J. Snoek et al.

    Monitoring of individual needs in diabetes (MIND): baseline data from the cross-national diabetes attitudes, wishes, and needs (DAWN) MIND study

    Diabetes Care

    (2011)
  • Cited by (54)

    View all citing articles on Scopus
    View full text