Are the PHQ-9 and GAD-7 Suitable for Use in India?: A Psychometric Analysis

Jeroen De Man; Pilvikki Absetz; Thirunavukkarasu Sathish; Allissa Desloge; Tilahun Haregu; Brian Oldenburg; Leslie C M Johnson; Kavumpurathu Raman Thankappan; Emily D Williams

doi:10.3389/fpsyg.2021.676398

Are the PHQ-9 and GAD-7 Suitable for Use in India? A Psychometric Analysis

Jeroen De Man
, Pilvikki Absetz
, Thirunavukkarasu Sathish
, Allissa Desloge
, Tilahun Haregu
, Brian Oldenburg
, Leslie C M Johnson
, Kavumpurathu Raman Thankappan
, Emily D Williams

Health Sciences

Research output: Contribution to journal › Article › Scientific › peer-review

67 Citations (Scopus)

22 Downloads (Pure)

Abstract

Background: Cross-cultural evidence on the factorial structure and invariance of the PHQ-9 and the GAD-7 is lacking for South Asia. Recommendations on the use of unit-weighted scores of these scales (the sum of items' scores) are not well-founded. This study aims to address these contextual and methodological gaps using data from a rural Indian population.

Methods: The study surveyed 1,209 participants of the Kerala Diabetes Prevention Program aged 30-60 years (n at risk of diabetes = 1,007 and n with diabetes = 202). 1,007 participants were surveyed over 2 years using the PHQ-9 and the GAD-7. Bifactor-(S - 1) modeling and multigroup confirmatory factor analysis were used.

Results: Factor analysis supported the existence of a somatic and cognitive/affective subcomponent for both scales, but less explicitly for the GAD-7. Hierarchical omega values were 0.72 for the PHQ-9 and 0.76 for the GAD-7. Both scales showed full scalar invariance and full or partial residual invariance across age, gender, education, status of diabetes and over time. Effect sizes between categories measured by unit-weighted scores versus latent means followed a similar trend but were systematically higher for the latent means. For both disorders, female gender and lower education were associated with higher symptom severity scores, which corresponds with regional and global trends.

Conclusions: For both scales, psychometric properties were comparable to studies in western settings. Distinct clinical profiles (somatic-cognitive) were supported for depression, and to a lesser extent for anxiety. Unit-weighted scores of the full scales should be used with caution, while scoring subscales is not recommended. The stability of these scales supports their use and allows for meaningful comparison across tested subgroups.

Clinical Trial Registration: Australia and New Zealand Clinical Trials Registry: ACTRN12611000262909http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=336603&isReview=true.

Original language	English
Article number	676398
Journal	Frontiers in Psychology
Volume	12
DOIs	https://doi.org/10.3389/fpsyg.2021.676398
Publication status	Published - 2021
Publication type	A1 Journal article-refereed

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Publication forum classification

Publication forum level 1

Access to Document

10.3389/fpsyg.2021.676398Licence: CC BY

fpsyg-12-676398Final published version, 781 KBLicence: CC BY

https://urn.fi/URN:NBN:fi:tuni-202201121284Licence: CC BY

Cite this

@article{e10ce1ee32214279bec7f2d892285bc8,

title = "Are the PHQ-9 and GAD-7 Suitable for Use in India?: A Psychometric Analysis",

abstract = "Background: Cross-cultural evidence on the factorial structure and invariance of the PHQ-9 and the GAD-7 is lacking for South Asia. Recommendations on the use of unit-weighted scores of these scales (the sum of items' scores) are not well-founded. This study aims to address these contextual and methodological gaps using data from a rural Indian population.Methods: The study surveyed 1,209 participants of the Kerala Diabetes Prevention Program aged 30-60 years (n at risk of diabetes = 1,007 and n with diabetes = 202). 1,007 participants were surveyed over 2 years using the PHQ-9 and the GAD-7. Bifactor-(S - 1) modeling and multigroup confirmatory factor analysis were used.Results: Factor analysis supported the existence of a somatic and cognitive/affective subcomponent for both scales, but less explicitly for the GAD-7. Hierarchical omega values were 0.72 for the PHQ-9 and 0.76 for the GAD-7. Both scales showed full scalar invariance and full or partial residual invariance across age, gender, education, status of diabetes and over time. Effect sizes between categories measured by unit-weighted scores versus latent means followed a similar trend but were systematically higher for the latent means. For both disorders, female gender and lower education were associated with higher symptom severity scores, which corresponds with regional and global trends.Conclusions: For both scales, psychometric properties were comparable to studies in western settings. Distinct clinical profiles (somatic-cognitive) were supported for depression, and to a lesser extent for anxiety. Unit-weighted scores of the full scales should be used with caution, while scoring subscales is not recommended. The stability of these scales supports their use and allows for meaningful comparison across tested subgroups.Clinical Trial Registration: Australia and New Zealand Clinical Trials Registry: ACTRN12611000262909http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=336603\&isReview=true.",

author = "\{De Man\}, Jeroen and Pilvikki Absetz and Thirunavukkarasu Sathish and Allissa Desloge and Tilahun Haregu and Brian Oldenburg and Johnson, \{Leslie C M\} and Thankappan, \{Kavumpurathu Raman\} and Williams, \{Emily D\}",

note = "Copyright {\textcopyright} 2021 De Man, Absetz, Sathish, Desloge, Haregu, Oldenburg, Johnson, Thankappan and Williams.",

year = "2021",

doi = "10.3389/fpsyg.2021.676398",

language = "English",

volume = "12",

journal = "Frontiers in Psychology",

issn = "1664-1078",

publisher = "Frontiers Media S.A.",

}

TY - JOUR

T1 - Are the PHQ-9 and GAD-7 Suitable for Use in India?

T2 - A Psychometric Analysis

AU - De Man, Jeroen

AU - Absetz, Pilvikki

AU - Sathish, Thirunavukkarasu

AU - Desloge, Allissa

AU - Haregu, Tilahun

AU - Oldenburg, Brian

AU - Johnson, Leslie C M

AU - Thankappan, Kavumpurathu Raman

AU - Williams, Emily D

PY - 2021

Y1 - 2021

N2 - Background: Cross-cultural evidence on the factorial structure and invariance of the PHQ-9 and the GAD-7 is lacking for South Asia. Recommendations on the use of unit-weighted scores of these scales (the sum of items' scores) are not well-founded. This study aims to address these contextual and methodological gaps using data from a rural Indian population.Methods: The study surveyed 1,209 participants of the Kerala Diabetes Prevention Program aged 30-60 years (n at risk of diabetes = 1,007 and n with diabetes = 202). 1,007 participants were surveyed over 2 years using the PHQ-9 and the GAD-7. Bifactor-(S - 1) modeling and multigroup confirmatory factor analysis were used.Results: Factor analysis supported the existence of a somatic and cognitive/affective subcomponent for both scales, but less explicitly for the GAD-7. Hierarchical omega values were 0.72 for the PHQ-9 and 0.76 for the GAD-7. Both scales showed full scalar invariance and full or partial residual invariance across age, gender, education, status of diabetes and over time. Effect sizes between categories measured by unit-weighted scores versus latent means followed a similar trend but were systematically higher for the latent means. For both disorders, female gender and lower education were associated with higher symptom severity scores, which corresponds with regional and global trends.Conclusions: For both scales, psychometric properties were comparable to studies in western settings. Distinct clinical profiles (somatic-cognitive) were supported for depression, and to a lesser extent for anxiety. Unit-weighted scores of the full scales should be used with caution, while scoring subscales is not recommended. The stability of these scales supports their use and allows for meaningful comparison across tested subgroups.Clinical Trial Registration: Australia and New Zealand Clinical Trials Registry: ACTRN12611000262909http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=336603&isReview=true.

AB - Background: Cross-cultural evidence on the factorial structure and invariance of the PHQ-9 and the GAD-7 is lacking for South Asia. Recommendations on the use of unit-weighted scores of these scales (the sum of items' scores) are not well-founded. This study aims to address these contextual and methodological gaps using data from a rural Indian population.Methods: The study surveyed 1,209 participants of the Kerala Diabetes Prevention Program aged 30-60 years (n at risk of diabetes = 1,007 and n with diabetes = 202). 1,007 participants were surveyed over 2 years using the PHQ-9 and the GAD-7. Bifactor-(S - 1) modeling and multigroup confirmatory factor analysis were used.Results: Factor analysis supported the existence of a somatic and cognitive/affective subcomponent for both scales, but less explicitly for the GAD-7. Hierarchical omega values were 0.72 for the PHQ-9 and 0.76 for the GAD-7. Both scales showed full scalar invariance and full or partial residual invariance across age, gender, education, status of diabetes and over time. Effect sizes between categories measured by unit-weighted scores versus latent means followed a similar trend but were systematically higher for the latent means. For both disorders, female gender and lower education were associated with higher symptom severity scores, which corresponds with regional and global trends.Conclusions: For both scales, psychometric properties were comparable to studies in western settings. Distinct clinical profiles (somatic-cognitive) were supported for depression, and to a lesser extent for anxiety. Unit-weighted scores of the full scales should be used with caution, while scoring subscales is not recommended. The stability of these scales supports their use and allows for meaningful comparison across tested subgroups.Clinical Trial Registration: Australia and New Zealand Clinical Trials Registry: ACTRN12611000262909http://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=336603&isReview=true.

U2 - 10.3389/fpsyg.2021.676398

DO - 10.3389/fpsyg.2021.676398

M3 - Article

C2 - 34054677

SN - 1664-1078

VL - 12

JO - Frontiers in Psychology

JF - Frontiers in Psychology

M1 - 676398

ER -