A CBC-derived score associated with dengue NS1 antigen positivity among suspected dengue patients in Dhaka: a retrospective model-development study

Background: Early diagnostic triage for suspected dengue is important in high-volume settings, yet confirmatory testing may be delayed or intermittently available. This study aimed to develop and internally validate a simple complete blood count (CBC)-derived score to estimate the likelihood of same-day dengue non-structural protein 1 (NS1) antigen positivity among suspected dengue patients in Dhaka, Bangladesh. Methods: We conducted a retrospective model-development study at Uttara Adhunik Medical College Hospital using data from patients with clinical suspicion of dengue between 19 August and 12 October 2023. The final analysis included 596 unique patients who underwent CBC and NS1 testing on the same day. Least absolute shrinkage and selection operator (LASSO) penalized logistic regression was used for variable selection among routinely available demographic and hematological predictors. Selected predictors were refit using unpenalized logistic regression to derive a CBC score. Internal validation and calibration were assessed using bootstrap optimism correction with 400 resamples. Results: Among 596 patients, 306 (51.3%) were NS1 positive. The final parsimonious model retained platelet count rescaled per 10 x 109/L, white blood cell count, and monocyte count. The optimism-corrected AUC was 0.9429. At the Youden-derived cutoff of 11.825, the final integer score had sensitivity 0.833 (95% CI, 0.787-0.873), specificity 0.966 (95% CI, 0.938-0.983), PPV 0.962 (95% CI, 0.932-0.982), NPV 0.846 (95% CI, 0.802-0.883), and accuracy 0.898 (95% CI, 0.870-0.921). A rounded cutoff of 12 produced similar performance. In benchmark analysis, the final model showed a modest improvement over platelet count alone (AUC 0.944 vs 0.931; Delta AUC 0.012; DeLong p=0.00364). Conclusion: A three-parameter CBC score can estimate the likelihood of NS1 antigen positivity among suspected dengue patients in this single-center dataset. The score should be interpreted as a CBC-only triage adjunct, not as a replacement for confirmatory dengue testing. External validation in independent cohorts is needed before routine clinical use.