Dengue fever (DF) is still a major challenge for public health, especially during massive outbreaks. We developed a novel prediction score to help decision making, which has not been performed till date.

We conducted a retrospective case-control study to recruit all the DF patients who visited a medical center during the 2015 DF outbreak. Demographic data, vital signs, symptoms/signs, chronic comorbidities, laboratory data, and 30-day mortality rates were included in the study. Univariate analysis and multivariate logistic regression analysis were used to identify the independent mortality predictors, which further formed the components of a DF mortality (DFM) score. Bootstrapping method was used to validate the DFM score.

In total, a sample of 2358 DF patients was included in this study, which also consisted of 34 deaths (1.44%). Five independent mortality predictors were identified: elderly age (≥65 years), hypotension (systolic blood pressure <90 mmHg), hemoptysis, diabetes mellitus, and chronic bedridden. After assigning each predictor a score of “1”, we developed a DFM score (range: 0–5), which showed that the mortality risk ratios for scores 0, 1, 2, and ≥3 were 0.2%, 2.3%, 6.0%, and 45.5%, respectively. The area under the curve was 0.849 (95% confidence interval [CI]: 0.785–0.914), and Hosmer–Lemeshow goodness-of-fit was 0.642. Compared with score 0, the odds ratios for mortality were 12.73 (95% CI: 3.58–45.30) for score 1, 34.21 (95% CI: 9.75–119.99) for score 2, and 443.89 (95% CI: 86.06–2289.60) for score ≥3, with significant differences (all p values <0.001). The score ≥1 had a sensitivity of 91.2% for mortality and score ≥3 had a specificity of 99.7% for mortality.

DFM score was a simple and easy method to help decision making, especially in the massive outbreak. Further studies in other hospitals or nations are warranted to validate this score.