Abstract
BACKGROUND: Given persistent and unpredictable delays in antiretroviral therapy (ART) initiation globally, this study develops a decision preference scale to identify patients at risk of such delays and evaluates its psychometric properties. METHODS: Guided by the intertemporal decision-making framework, the scale development followed a triphasic approach. In Phase I, we conceptualized decision-making preferences for ART initiation through theoretical synthesis, systematic literature review, and in-depth qualitative interviews. In Phase II, we generated and optimized the items through group discussions, Delphi expert consultations, and Pilot test. Phase III implemented psychometric evaluations: (1) content validity (Item-Content Validity Index [I-CVI] and Scale-Content Validity Index [S-CVI]); (2) structural validity (Exploratory Factor Analysis [EFA] and Confirmatory Factor Analysis [CFA]); (3) predictive validity (e.g., Odds Ratio [OR], Receiver Operating Characteristic [ROC] curve with Area Under the Curve [AUC]); (4) reliability. RESULTS: The scale comprises 16 items across two theoretically derived dimensions: short-term costs and long-term benefits. The content validity is adequate (I-CVIs range from 0.830 to 1.000; S-CVI = 0.949). EFA extracted two factors with a cumulative variance explanation of 66.667%, and CFA shows an excellent model fit. The adjusted scale performed well in predicting ART initiation (OR = 0.745, p < 0.001 logistic regression; AUC = 0.895, p < 0.001 ROC curve analysis). The optimal classification threshold (≥ 2), derived from the Youden index achieved balanced diagnostic accuracy (sensitivity = 81.5%; specificity = 80.9%; precision = 92.0%). Finally, the reliability of the scale is acceptable, with a Cronbach’s α of 0.786 for the total scale, a split-half reliability of 0.908, and a test-retest reliability of 0.837. CONCLUSIONS: The Decision-Making Preferences Scale for ART Initiation is a validated and reliable tool that helps providers systematically identify high-risk patients enabling timely interventions to reduce delays. Future research is warranted to evaluate its external validity in other geographical settings and among broader populations. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12889-026-27217-3.