SRQ-20 Questionnaire Holds Up as Mental Health Screener for Older Northeastern Brazilians, Bifactor Study Finds

Brazil is aging faster than most people realize, and it is aging unevenly. The country’s Northeast region, historically marked by persistent socioeconomic disadvantage, now faces a rapid demographic transition toward older populations at the exact moment when its primary care system must decide how to detect psychological distress among millions of older adults. A new psychometric study published in Trends in Psychology by Josevânia da Silva, Karla Santos Mateus, and Emerson Do Bú tackles a deceptively simple question with considerable technical rigor: does the World Health Organization’s Self-Reporting Questionnaire, the SRQ-20, actually work as designed when administered to older adults in Northeastern Brazil? The answer, delivered across three integrated studies involving more than 1,300 participants, is a carefully qualified yes, and it carries immediate implications for mental health screening in Brazil’s Unified Health System, known as the SUS.

The SRQ-20 is a twenty-item self-report screener developed by the WHO in the 1980s for detecting common mental disorders, chiefly depression and anxiety, in primary care settings in low- and middle-income countries. Its appeal is obvious: it is short, free, translatable, and scorable by non-specialists. In Brazil, it has been used for decades, and a cutoff of eight or more positive responses has become the conventional threshold for flagging probable cases among adults. But conventions travel poorly by default. Older adults report distress differently than younger ones, often emphasizing somatic complaints, fatigue, and cognitive slowing rather than explicitly psychological language, and some researchers have proposed lower cutoffs for older populations on the assumption that the instrument behaves differently with age. Whether that assumption holds in a specific, disadvantaged, rapidly aging region had rarely been tested directly.

The research program addressed the question in three stages, each attacking a different layer of the measurement problem. Study 1, with 401 older adults, examined the latent structure of the instrument using polychoric exploratory factor analysis, the appropriate factor-analytic approach for ordered categorical items, combined with two-parameter logistic item response theory modeling. The analysis recovered a four-factor solution: Depressive Affect, Somatic Anxious Complaints, Loss of Vital Energy and Cognitive Slowing, and Hopelessness with Self-harm content. Item response theory calibration showed that measurement precision was concentrated in the low-to-moderate range of the distress continuum, which is precisely where screening decisions are made, since individuals with extreme scores are typically already visible to services.

Study 2, with 580 participants, moved from exploration to model comparison. The team contrasted four competing structural models: a unidimensional model treating all twenty items as a single distress scale, a correlated four-factor model, a second-order model in which the four specific factors load onto a higher-order general dimension, and a bifactor model in which every item loads simultaneously on a general distress factor and on one of the four specific factors. The bifactor model fit best. This result matters because bifactor evidence, when the general dimension is strong and the specific factors add little unique reliable variance, licenses the use of a simple total score; when specific factors retain substantial independent strength, total scores can be misleading. Here the findings supported a strong general distress dimension coexisting with meaningful specific factors, a structure that justifies scoring the SRQ-20 by its total while acknowledging its multidimensional content.

Study 3, with 355 older adults, anchored the interpretation of SRQ-20 scores to an external reference. The researchers used the Kessler Psychological Distress Scale, the K10, with a threshold of fifteen or more, as a concurrent self-report criterion and applied receiver operating characteristic analysis to SRQ-20 totals. The area under the curve reached .84, indicating good discrimination between older adults above and below the reference threshold. The optimal balance of sensitivity and specificity fell at a cutoff of eight, converging with the long-standing adult convention rather than with the lower cutoffs sometimes proposed for older adults. In other words, the portable convention appears to survive transplantation to this population, at least for screening purposes.

The authors are careful, and rightly so, about what that cutoff means. Because the reference standard was itself a screener rather than a structured clinical interview, the threshold of eight is a screening rule, not a diagnostic boundary. This distinction is a familiar trap in cutoff research: when optimal thresholds are selected against imperfect criteria, sensitivity and specificity estimates can be biased, a mechanism documented in the methodological literature. A positive SRQ-20 at this threshold flags someone who warrants further assessment, not someone who has a disorder. Within the logic of primary care triage, however, that is exactly what a screener should do, and the finding gives SUS practitioners a defensible, evidence-based rule for the population they serve.

The technical machinery behind these conclusions deserves emphasis because it represents current best practice in psychological measurement. Polychoric correlations respect the ordinal nature of yes/no symptom items rather than treating them as continuous variables. Parallel analysis guided dimensionality assessment against chance factors. Item response theory located each item’s difficulty and discrimination along the distress continuum, revealing where on the scale the instrument is informative. Bifactor modeling, evaluated with the statistical indices developed by Reise, Rodriguez, and Haviland, provided the most stringent test of whether a total score is psychometrically defensible. And ROC analysis, with the DeLong method for correlated curves, quantified classification accuracy without relying on a single arbitrary operating point. Datasets and supplementary materials are publicly available on the Open Science Framework, supporting transparency and reproducibility.

Context amplifies the significance of the work. Older adults in Brazil’s Northeast live at the intersection of rapid population aging and structural disadvantage, conditions that shape how distress is experienced, expressed, and recognized. Prevalence studies in the region have reported substantial rates of suspected common mental disorders among older residents, and social determinants such as food insecurity and informal work are repeatedly linked to psychological distress in low- and middle-income settings. Ageism, documented by the WHO as a global threat to older people’s health, further risks normalizing distress as a natural feature of aging, allowing treatable conditions to go unrecognized. A validated, scalable screener with a locally examined cutoff is a practical countermeasure: it gives community health agents a tool whose behavior in this population has been tested rather than assumed.

The study also makes a broader conceptual contribution that extends beyond one questionnaire and one region. Screening conventions are often treated as portable defaults, applied across populations because they were validated somewhere else, sometimes decades earlier. This research demonstrates that portability is an empirical property that can be examined rather than presumed. By integrating latent structure, item functioning, and score interpretation in a single program, the authors model how any imported instrument can be interrogated for a specific population, an approach relevant to geriatric mental health wherever demographic transition meets measurement uncertainty.

For Brazil’s SUS, the practical takeaway is straightforward: the SRQ-20 total score, with its conventional cutoff of eight, is supported for scalable screening of common mental disorders among older adults in Northeastern primary care, with positive screens followed by fuller clinical assessment. For the international field, the message is more subtle and arguably more viral: the tools we trust are only as trustworthy as the evidence behind them in the hands that use them, and a twenty-item WHO questionnaire, tested with bifactor models and ROC curves in one of the world’s most disadvantaged aging regions, can still surprise us by confirming that sometimes the old conventions were right all along.

Subject of Research: Screening Common Mental Disorders in Older Northeastern Brazilians: Bifactor Evidence, Item Functioning, and Score Interpretation of the SRQ-20

Article Title: Screening Common Mental Disorders in Older Northeastern Brazilians: Bifactor Evidence, Item Functioning, and Score Interpretation of the SRQ-20

Article References: da Silva, J., Mateus, K. S., & Do Bú, E. (2026). Screening Common Mental Disorders in Older Northeastern Brazilians: Bifactor Evidence, Item Functioning, and Score Interpretation of the SRQ-20. Trends in Psychology. https://doi.org/10.1007/s43076-026-00542-7

Image Credits: AI Generated

DOI: 10.1007/s43076-026-00542-7

Keywords: Screening, Common, Mental, Disorders, Older, Northeastern, Brazilians, Bifactor, Evidence, Item, Functioning, Score