Measurement invariance testing of the PHQ-9 in a multi-ethnic population in Europe: the HELIUS study

Henrike Galenkamp,Eske M Derks,Karien Stronks,Marieke B Snijder

doi:10.1186/s12888-017-1506-9

Abstract

BackgroundIn Western European countries, the prevalence of depressive symptoms is higher among ethnic minority groups, compared to the host population. We explored whether these inequalities reflect variance in the way depressive symptoms are measured, by investigating whether items of the PHQ-9 measure the same underlying construct in six ethnic groups in the Netherlands.MethodsA total of 23,182 men and women aged 18–70 of Dutch, South-Asian Surinamese, African Surinamese, Ghanaian, Turkish or Moroccan origin were included in the HELIUS study and had answered to at least one of the PHQ-9 items. We conducted multiple group confirmatory factor analyses (MGCFA), with increasingly stringent model constraints (i.e. assessing Configural, Metric, Strong and Strict measurement invariance (MI)), and regression analysis, to confirm comparability of PHQ-9 items across ethnic groups.ResultsA one-factor model, where all nine items reflect a single underlying construct, showed acceptable model fit and was used for MI testing. In each subsequent step, change in goodness-of-fit measures did not exceed 0.015 (RMSEA) or 0.01 (CFI). Moreover, strict invariance models showed good or acceptable model fit (Men: RMSEA = 0.050; CFI = 0.985; Women: RMSEA = 0.058; CFI = 0.979), indicating between-group equality of item clusters, factor loadings, item thresholds and residual variances. Finally, regression analysis did not indicate potential ethnicity-related differential item functioning (DIF) of the PHQ-9.ConclusionsThis study provides evidence of measurement invariance of the PHQ-9 regarding ethnicity, implying that the observed inequalities in depressive symptoms cannot be attributed to DIF.

Highlights

In Western European countries, the prevalence of depressive symptoms is higher among ethnic minority groups, compared to the host population
Measurement invariance analyses Three different factor models were compared, to obtain an adequate baseline model for further analysis (Table 3): a one factor model, a two-factor model based on the literature, and a two-factor model based on EFA
The one-factor model had good model fit according to Comparative Fit Index (CFI), and adequate model fit according to Root Mean Square Error of Approximation (RMSEA) after residual covariances were added to the model

Summary

Introduction

In Western European countries, the prevalence of depressive symptoms is higher among ethnic minority groups, compared to the host population. Meta-analyses showed that individuals with low socioeconomic status (SES) have a higher risk of Increased depression rates among ethnic minority populations are of particular concern, since increases in migration were observed for most western European countries over the past decades [8]. In the Netherlands, considerable differences in the prevalence of Galenkamp et al BMC Psychiatry (2017) 17:349 depression and depressive symptoms were reported for Moroccan and Turkish immigrants, compared to the Dutch population [5, 9]. The 1-month prevalence of depressive disorders was 4% in adults of Dutch ethnic origin, whereas it was 7% in adults of Moroccan origin and even 15% among adults of Turkish origin [5]

Objectives

Methods

Results

Conclusion