Fig 1.
STROBE-compliant participant flow diagram.
Study participant recruitment and enrollment flow. A total of 313 patients with metastatic breast cancer were screened for eligibility at NSIA-LUTH Cancer Centre, Lagos, Nigeria (September 2020–February 2022). Four patients were excluded due to incomplete primary outcome data (DT or BDI-II scores), resulting in 309 participants included in the final analysis. The high retention rate (98.7%) demonstrates the feasibility and acceptability of study procedures in this population.
Table 1.
Baseline demographic, clinical, psychosocial, and quality of life characteristics of participants (N = 309).
Fig 2.
Receiver operating characteristic (ROC) curve showing the performance of the Distress Thermometer (DT) in predicting clinically significant depressive symptoms (BDI ≥ 20).
ROC curve analysis demonstrating the diagnostic performance of the Distress Thermometer for detecting clinically significant depressive symptoms (BDI-II ≥ 20) in Nigerian women with metastatic breast cancer (n = 309). The AUC was 0.414 (95% CI: 0.326–0.503), indicating performance significantly worse than chance (diagonal reference line, AUC = 0.5). The curve’s position below the reference line confirms the DT’s inadequate discriminatory ability for depression screening in this population.
Fig 3.
Sensitivity and specificity by DT cutoff scores.
Sensitivity and specificity curves for the Distress Thermometer across all possible cutoff values (0.5–9.5) for detecting clinically significant depressive symptoms (BDI-II ≥ 20). The red line represents sensitivity and the blue line represents specificity. The vertical dashed line indicates the Youden-optimal cutoff (DT = 7.5), which achieved 98.5% specificity but only 2.0% sensitivity. The curves demonstrate the classic inverse relationship between sensitivity and specificity, with no threshold providing clinically acceptable performance (≥80% sensitivity and ≥70% specificity simultaneously).
Table 2.
Diagnostic performance of distress thermometer (DT) cutoffs for predicting clinically significant depression (BDI ≥ 20).
Fig 4.
Forest plot showing the area under the receiver operating characteristic curve (AUC) for the Distress Thermometer’s ability to detect clinically significant depressive symptoms (BDI-II ≥ 20) across different patient subgroups.
Error bars represent 95% confidence intervals. The horizontal dashed line at AUC = 0.5 indicates performance no better than chance, while the dotted line at AUC = 0.7 represents the threshold for acceptable diagnostic accuracy. All subgroups demonstrated poor diagnostic performance with AUC values well below the acceptable threshold, indicating consistent failure of the DT across patient characteristics. P-values compare AUC between subgroups within each category using DeLong’s test for correlated ROC curves. Reference lines: Dashed line (AUC = 0.5) = chance performance; Dotted line (AUC = 0.7) = acceptable diagnostic accuracy threshold. Statistical note: No subgroup achieved clinically acceptable diagnostic performance (AUC ≥ 0.7). The consistently poor performance across all demographic and clinical characteristics suggests fundamental limitations of the DT in this population rather than subgroup-specific issues.