Response shift–adjusted longitudinal assessment of patient-reported outcomes in older people living with HIV: a multilevel structural equation and Rasch analysis

To investigate the presence and clinical impact of response shift (RS) on longitudinal patient-reported outcome (PRO) assessment among older people living with HIV/AIDS (PLWHA), and to evaluate how not accounting for RS may affect the interpretation of changes in PRO at both domain and item levels.

A two-wave longitudinal survey with a 3-month interval was conducted among older PLWHA (baseline n = 1245; follow-up rate 84.90%) using the PROHIV-OLD instrument. Domain-level RS—including reconceptualization, recalibration, and reprioritization—was examined using Oort’s structural equation modeling (SEM) approach. Item-level recalibration was further assessed within the Mental Status domain using the updated ROSALI–Rasch measurement theory (RMT) algorithm, incorporating clinical subgroups defined by diagnostic duration, disease stage, and HIV-1 RNA status.

At the domain level, a medium uniform recalibration was detected for Physical Symptoms (Est. = 0.65, Δχ² = 86.22), and a small uniform recalibration for Mental Status (Est. = − 0.28, Δχ² = 18.47). Family Relationship showed a small reconceptualization (Est.=0.12) and a medium reprioritization (Est. = − 0.58), while the Treatment Dimension exhibited non-uniform recalibration (Δχ² = 21.73; all P < 0.05). At the item level, when RS was not accounted for, positive RS was associated with underestimation of Mental Status among newly diagnosed patients (Est. = 0.34), AIDS-stage patients (Est. = 0.09), and those with undetectable HIV-1 RNA (Est. = 0.26), whereas negative RS was associated with overestimation among patients with longer diagnostic duration or earlier clinical stages.

RS substantially informs longitudinal PRO assessment in older PLWHA. Accounting for RS effects can improve the interpretation of changes in PROs and support more patient-centered clinical evaluation in aging HIV populations.