Abstract
Maternal prenatal stress places a substantial burden on mother's mental health. Expectant mothers in low- and middle-income countries (LMICs) have thus far received less attention than mothers in high-income settings. This is particularly problematic, as a range of triggers, such as exposure to traumatic events (e.g. natural disasters, previous pregnancy losses) and adverse life circumstances (e.g. poverty, community violence), put mothers at increased risk of experiencing prenatal stress. The ten-item Perceived Stress Scale (PSS-10) is a widely recognised index of subjective experience of stress that is increasingly used in LMICs. However, evidence for its measurement equivalence across settings is lacking. This study aims to assess measurement invariance of the PSS-10 across eight LMICs and across birth parity. This research was carried out as part of the Evidence for Better Lives Study (EBLS, vrc.crim.cam.ac.uk/vrcresearch/EBLS). The PSS-10 was administered to N = 1,208 expectant mothers from Ghana, Jamaica, Pakistan, the Philippines, Romania, South Africa, Sri Lanka and Vietnam during the third trimester of pregnancy. Confirmatory factor analysis suggested a good model fit of a two-factor model across all sites, with items on experiences of stress loading onto a negative factor and items on perceived coping onto a positive factor. Configural and metric, but not full or partial scalar invariance, were established across all sites. Configural, metric and full scalar invariance could be established across birth parity. On average, first-time mothers reported less stress than mothers who already had children. Our findings indicate that the PSS-10 holds utility in assessing stress across a broad range of culturally diverse settings; however, caution should be taken when comparing mean stress levels across sites.