Predictive value of qSOFA score for mortality in older patients with dengue fever: a retrospective case-control study

Abstract

Background

Dengue fever (DF) in older patients is associated with higher mortality, but optimal mortality prediction approaches remain unclear. We assessed the utility of the quick Sequential Organ Failure Assessment (qSOFA) as a predictive tool in older DF patients.

Methods

We retrospectively included DF patients aged ≥ 65 years who visited our study hospital between September 1, 2015, and December 31, 2015. Demographics, vital signs, comorbidities, laboratory data, and mortality were collected. Accuracy of qSOFA score ≥ 2 in predicting 30-day mortality was evaluated.

Results

626 older DF patients were recruited (mean age: 74.1 years, nearly equal sex distribution). Mean ± standard deviation of qSOFA score was higher in mortality group vs. survival group (0.6 ± 0.8 vs. 0.1 ± 0, p < 0.01). Logistic regression showed qSOFA score ≥ 2 associated with higher mortality risk (odds ratio: 18.74; 95% confidence interval: 4.37 − 50.64). Adjusted area under receiver operating characteristic was 0.75, and Hosmer-Lemeshow goodness of fit test was 0.58. Sensitivity, specificity, positive predictive value, and negative predictive value of qSOFA score ≥ 2 for predicting mortality were 18.5%, 98.7%, 38.5%, and 96.4%, respectively.

Conclusion

qSOFA score ≥ 2 is a simple and specific tool for predicting 30-day mortality in older patients with DF, with high specificity and negative predictive value. It can be used to rule out patients at high risk of mortality.

Introduction

Dengue fever (DF) is a viral infection transmitted by infected mosquitoes and is a significant public health concern in tropical areas [1]. In recent decades, an estimated 100–400 million people are infected with DF each year, and its incidence has been steadily increasing worldwide [1]. While the majority of cases are mild or asymptomatic, DF outbreaks can strain health systems and consume substantial medical resources [1].

Older population is rapidly growing in the world [2]. Between 2015 and 2050, the proportion of the global population aged 60 years or older is projected to nearly double, increasing from 12 to 22% [2]. Older individuals are more vulnerable to DF and have a higher risk of developing severe forms of the disease, such as dengue hemorrhagic fever, dengue shock syndrome, and mortality [3, 4]. Furthermore, older individuals may present with atypical clinical manifestations of DF, posing challenges for clinicians in terms of diagnosis and outcome prediction. As such, early recognition and appropriate therapy are crucial for older patients with DF [5].

New tools predict mortality in older adults with DF. Huang et al. (2017) identified coma, bedridden status, severe hepatitis, and renal failure as predictors of mortality [6]. In their study, patients without these predictors had a 98.51% survival rate, while those with two or more predictors had a 57.14% mortality rate [6]. However, relying on lab data during DF outbreaks is inconvenient. Another study found that a shock index of ≥ 1 had 97.0% specificity and 96.2% negative predictive value (NPV) for identifying high-risk older DF patients [7]. However, the shock index, originally designed for trauma or hemorrhage, may not be ideal for infection prediction. Further research is needed for better tools during DF outbreaks in older adults.

The quick Sequential Organ Failure Assessment (qSOFA) score is determined by three clinical criteria: (1) systolic blood pressure ≤ 100 mmHg; (2) respiratory rate ≥ 22 breaths/min; (3) Glasgow Coma Scale ≤ 15 [8]. qSOFA has been recognized as a useful predictor in patients with suspected or diagnosed infection, including hospital mortality [9] and early recognition of critical illness with appropriate therapy and disposition in the emergency department [10]. However, there is a data gap on using qSOFA as a mortality predictor in older patients with DF. Therefore, our study aims to assess the utility of qSOFA in predicting mortality in this population.

Methods

Study design, setting, and participants

This retrospective case-control study was conducted at Chi Mei Medical Center, a tertiary medical center in Taiwan with 1,288 beds, providing emergency care for approximately 145,000 patients annually [11]. We collected the medical records of all older patients (age ≥ 65 years) diagnosed with DF at Chi Mei Medical Center from September 1, 2015, to December 31, 2015, retrospectively. The diagnostic criteria for DF included: (1) Laboratory confirmation (i.e., NS1, PCR, IgM, and IgG); (2) residents or those with exposure in dengue-epidemic areas; (3) Fever and two of the following symptoms: leukopenia, nausea or vomiting, rash, aches and pains, positive tourniquet test, and any warning sign [1].

Definitions of the variables and primary outcome

The qSOFA score ≥ 2 was categorized based on previous study [12]. Age subgroups were defined as follows: young elderly (65–74 years old), moderately elderly (75–84 years old), and old elderly (≥ 85 years old). The primary outcome was defined as 30-day mortality [3, 6, 7, 13].

Data collection

The medical records of the participants were retrospectively reviewed by three trained registered nurses. Patients were excluded if their records had (1) incomplete basic demographic data (age, sex, and vital signs), (2) missing information on outcome or treatment in Chi Mei Medical Center, (3) confirmed bacterial infections at emergency department (Fig. 1). All the data, including the three variables in qSOFA (systolic blood pressure, respiratory rate, and Glasgow Coma Scale), were collected during the patient’s initial presentation to our emergency department. This captures the earliest clinical assessment and reflects the patient’s status upon admission, ensuring consistency in how qSOFA parameters were measured. In cases where variables were not documented in the patient’s medical record, they were assumed to be negative.

Fig. 1

Flowchart of this study. DF, dengue fever; qSOFA, quick Sequential Organ Failure Assessment

Ethical statement

This study was conducted in strict accordance with the Declaration of Helsinki and was approved by the Institutional Review Board at Chi Mei Medical Center. All data collected were anonymized to protect patient privacy. As this study was designed retrospectively, informed consent from the participants was waived.

Statistical analysis

We conducted all statistical analyses using SPSS 23.0 for Mac (Chicago, IL, USA). Continuous data are presented as means ± standard deviation. Univariate analyses for continuous variables were performed using either independent-sample t-test or Mann–Whitney–Wilcoxon test. Categorical variables were analyzed using Pearson’s chi-square test or Fisher’s exact test. Logistic regression analysis was performed to investigate the association between qSOFA ≥ 2 and 30-day mortality. The discriminatory performance of qSOFA (quick Sequential Organ Failure Assessment) with a score of ≥ 2 was assessed by calculating the adjusted area under the receiver operating characteristic curve (AUC), after accounting for comorbidities that exhibited significant differences between the mortality and survival groups. The reliability of qSOFA ≥ 2 was assessed using the Hosmer-Lemeshow goodness of fit test. Sensitivity, specificity, positive predictive value (PPV), and NPV of qSOFA ≥ 2 was analyzed. The significance level was set at 0.05 (two-tailed).

Results

A total of 626 older patients with DF were enrolled in this study (Table 1). Among the included patients, 528 (84.3%) were diagnosed with DF based on a positive NS1 antigen test, 25 (4.0%) based on PCR, 21 (3.4%) based on positive IgM, and the remaining 52 (8.3%) based on exposure to dengue-epidemic areas and clinical signs and symptoms of DF. The mean age was 74.1 years, and the overall 30-day mortality rate was 4.3% (27/626). There were no significant differences in sex ratio between the survival and mortality groups. However, the mortality group had higher prevalence rates of diabetes (55.6% vs. 34.2%, p = 0.02), coronary artery disease (29.6% vs. 13.2%, p = 0.02), chronic kidney disease (40.7% vs. 7.8%, p < 0.01), higher levels of glutamic oxaloacetic transaminase (990.8 ± 1685.1 U/L vs. 95.2 ± 130.7 U/L, p = 0.01), higher serum creatinine (3.8 ± 3.8 mg/dL vs. 1.3 ± 1.2 mg/dL, p = 0.05), lower albumin (2.9 ± 0.6 g/dL vs. 3.3 ± 0.5 g/dL, p = 0.01), and higher qSOFA score (0.6 ± 0.8 vs. 0.1 ± 0, p < 0.01). Overall, the qSOFA score of ≥ 2 was observed in 2% of the total patients. The 30-day mortality rate demonstrated an upward trend with increasing qSOFA scores, ranging from 0 (2.8%), to ≥ 1 (12.6%), and ≥ 2 (38.5%) (Fig. 2).

Table 1 Clinical characteristics of older patients with dengue fever in the emergency department

Fig. 2

30-day mortality in qSOFA = 0, qSOFA ≥ 1, and qSOFA ≥ 2. qSOFA, quick Sequential Organ Failure Assessment

The logistic regression analysis revealed that patients with qSOFA ≥ 2 had a higher odds ratio for mortality compared to those with qSOFA < 2 (odds ratio: 18.74; 95% confidence interval [CI]: 4.37 − 50.64) (Table 2). The Hosmer and Lemeshow goodness of fit test yielded a value of 0.58. After adjusting for diabetes, coronary artery disease, and chronic kidney disease, the adjusted AUC of qSOFA ≥ 2 was 0.75 (95% CI: 0.63 − 0.85) (Table 3; Fig. 3). The sensitivity, specificity, PPV, and NPV of qSOFA ≥ 2 for predicting 30-day mortality rate in older DF patients were 18.5%, 98.7%, 38.5%, and 96.4%, respectively (Table 4).

Table 2 Comparison of mortality prediction using qSOFA score ≥ 2 through logistic regression analysis

Table 3 Discrimination of qSOFA score ≥ 2 in older patients with dengue fever in the emergency department

Fig. 3

ROC curve for qSOFA ≥ 2. ROC, receiver operating characteristic; qSOFA, quick Sequential Organ Failure Assessment

Table 4 Performance of qSOFA score ≥ 2 in predicting mortality in older patients with dengue fever in the emergency department

Discussion

This study revealed that older patients with DF and a qSOFA score ≥ 2 had a significantly higher risk of 30-day mortality, nearly 19-fold compared to those with qSOFA < 2. Moreover, qSOFA ≥ 2 demonstrated good performance with adjusted AUC and Hosmer and Lemeshow goodness of fit measures. It demonstrated to be a valuable tool for excluding high-risk mortality patients due to its high specificity and NPV.

The qSOFA, comprising three vital signs directly associated with severe dengue, may account for its high specificity and NPV. Severe dengue, a critical manifestation of DF, is characterized by one or more of the following conditions: (1) significant plasma leakage leading to shock (known as dengue shock) and/or respiratory distress due to fluid accumulation; (2) severe bleeding; and (3) pronounced organ impairment [14]. The persistence of vascular permeability and plasma leakage may progress to dengue shock syndrome, ranging from initial asymptomatic capillary leakage to compensated shock, hypotensive shock, and, ultimately, cardiac arrest [15].

The qSOFA serves as a simple, rapid, and practical tool for predicting mortality in older DF patients, particularly in outbreaks with limited medical resources. However, its low sensitivity and PPV limit its use as a comprehensive diagnostic tool. Clinically, this means that while qSOFA can reliably exclude patients at high risk of mortality due to its high NPV, it is less effective at identifying all potential high-risk cases early on due to its low sensitivity. This emphasizes its role as an initial screening tool for ruling out mortality risk rather than a definitive predictive measure.

In comparison to the shock index proposed in our previous study [7], qSOFA ≥ 2 exhibited a similar AUC (0.75 vs. 0.76), superior Hosmer and Lemeshow goodness of fit (0.58 vs. 0.48), and higher specificity and NPV (98.7% vs. 97.0% and 96.4% vs. 96.2%, respectively). Moreover, the odds ratio of qSOFA ≥ 2 was more pronounced than that of the shock index when compared to the reference group (18.74 vs. 8.49), indicating its greater impact on clinical decision-making. Therefore, the qSOFA holds promise as a valuable tool in predicting mortality in older DF patients.

The qSOFA score is useful for ruling out adverse outcomes in suspected infectious diseases, as supported by a meta-analysis. The pooled sensitivity of qSOFA for mortality was found to be 60.8% (95% CI: 51.4−69.4%), with a pooled specificity of 72.0% (95% CI: 63.4−79.2%) [16]. In the intensive care unit (ICU) population, qSOFA demonstrated a higher sensitivity of 87.2% (95% CI: 75.8−93.7%), compared to 51.2% (95% CI: 43.6−58.7%) in the non-ICU population. Conversely, the pooled specificity of qSOFA was higher in the non-ICU population at 79.6% (95% CI: 73.3−84.7%), as opposed to 33.3% (95% CI: 23.8−44.4%) in the ICU population [16]. Notably, a large study involving 130,595 adult emergency department visits demonstrated that qSOFA ≥ 2 had a superior specificity of 98.9% (95% CI: 98.8 − 99.0%) compared to NEWS ≥ 4 (85.0%, 95% CI: 84.8–85.3%) and Systemic Inflammatory Response Syndrome (79.1%, 95% CI: 78.9 − 79.3%) [9]. Thus, qSOFA is a promising tool for excluding high-risk mortality in patients with less severity, aligning with the population of interest in this study.

The qSOFA score has shown good predictive accuracy in the older population based on a study that included patients with suspected infection outside the ICU. The study found that qSOFA had greater predictive ability for hospital mortality, ICU admission, and bacteremia in patients aged 65 − 80 years and very old patients (> 80 years) compared to younger patients [17]. Another multicenter cohort study evaluated the performance of qSOFA in older patients (> 75 years) with infectious disease in the emergency department, and found that a qSOFA score of ≥ 2 demonstrated a specificity of 94%, a sensitivity of 28%, and an AUC of 0.69, making it the most specific score for predicting 30-day mortality [18].

Using vital signs without laboratory data to predict adverse outcomes is a valuable approach in clinical practice, exemplified by the National Early Warning Score (NEWS), which includes six physiological parameters: (1) respiratory rate, (2) oxygen saturation, (3) systolic blood pressure, (4) heart rate, (5) level of consciousness or new-onset confusion, and (6) temperature [19]. Each parameter is assigned a score from 0 to 3 based on deviations from the normal range, with higher scores indicating a greater risk of adverse outcomes [19]. NEWS has been updated to NEWS-2 and is widely adopted as a standard quality indicator in acute care provision [19]. The aim of NEWS was to improve patient outcomes with acute illness or deterioration, and the impact on outcomes is now becoming apparent. However, paradoxically, an effective response can eliminate the link between the score and the ultimate outcome [19]. In recent years, dynamic or real-time NEWS has emerged as a promising way to monitor a patient’s condition. A study conducted in the United Kingdom in 2022 showed that the AUC (95% CI) for predicting death or ICU was 0.857 (0.852–0.862) for NEWS-2 and 0.906 (0.899–0.914) for dynamic NEWS in patients with respiratory disease [20]. For predicting clinically significant deterioration requiring urgent intervention, the AUC was 0.829 (0.817–0.842) for NEWS-2 and 0.877 (0.862–0.892) for dynamic NEWS [20]. NEWS-2 ≥ 5 had a sensitivity of 88.2% and specificity of 54.2% for predicting clinically significant deterioration requiring urgent intervention, while dynamic NEWS ≥ 0.021 had a higher sensitivity of 93.6% and approximately the same specificity of 54.3% for the same outcome [20]. As assessment tools for the quick and reliable identification of high-risk patients, other tools such as the WHO severe Dengue criteria and NEWS-2 criteria can also be utilized during dengue outbreaks. We hope that future comparative studies will evaluate and compare the diagnostic test accuracy metrics of these tools.

This study’s major strength is its validation of qSOFA in predicting mortality among older DF patients. However, there are several limitations that should be noted. Firstly, the wide 95% confidence interval (4.37 − 50.64) for the odds ratio, likely due to the relatively small number of patients, highlights a limitation in the precision of this estimate. Secondly, the external validity of the study may be limited as it was conducted in a single hospital, which may not be generalizable to other hospitals or nations. Thirdly, the higher mortality rate observed in this study (4.3% vs. the global average of 2.5%) may be attributed to the fact that the hospital is a tertiary care center that primarily admits critically ill patients referred from other healthcare facilities. Fourth, not all included patients were diagnosed using NS1 or PCR, which are considered more definitive diagnostic criteria. Finally, the study only assessed qSOFA once during the initial assessment, while the value may change throughout the course of the disease. In our analysis, we acknowledge that variations in the day of illness may affect the qSOFA score’s predictive accuracy. However, due to the retrospective nature of our study and the available data, we were unable to precisely adjust for the day of illness. Specifically, we recognize that some patients may have presented during different phases of the disease, potentially affecting clinical parameters. Future studies should aim to recruit larger sample sizes from multiple healthcare settings and assess serial qSOFA measurements throughout the clinical course to address these limitations.

Conclusions

This study is the first to show that older DF patients with a qSOFA score of ≥ 2 had a nearly 19-fold increase in 30-day mortality compared to those with a qSOFA score of < 2. The high specificity and NPV of qSOFA ≥ 2 suggest that it can be used to exclude patients at high risk of mortality. It is a simple and easily applicable tool that may be useful in this population, especially during large outbreaks with limited medical resources. However, further studies with larger sample sizes and validation are needed to confirm the potential utility of qSOFA in the future.