Comparison of the diagnostic accuracy of monocyte distribution width and procalcitonin in sepsis cases in the emergency department: a prospective cohort study

This was a single-center prospective cohort study. We enrolled patients from the emergency department of a 3000-bed medical center from July to October 2019. The average ED visits were 15,000 patients per month. The study was approved by the Institution Review Board of Chang Gung Memorial Hospital (No. 201900442B0) and registered to the ClinicalTrial.gov (NCT04322942).

A research coordinator worked full-time in the ED to screen all non-trauma patients from 8:30 to 17:30, Monday to Friday. The inclusion criteria were: (1) Adult ≥ 20 years of age, (2) Subjects presenting to the ED with the chief complaints of either fever, altered mental status, hypotension, or dyspnea, and (3) Complete blood cell count with differential testing (CBC/DC) was ordered, at presentation, as part of their standard medical care. The exclusion criteria were: (1) Subjects previously enrolled in this study (subjects could not be enrolled more than once in this study), (2) Referred patients who had received antibiotic treatment, (3) Pregnant women, (4) MDW result not reported due to inefficient monocyte count, and (5) Subjects not able to understand or sign informed consent.

No intervention was given to the patients or the caregivers in this study. The decision of blood test, antibiotic treatment, and patient disposition was taken entirely by the physician based on clinical examination. The patients were not enrolled without blood tests. Physicians in-charge were blinded to the MDW result.

After obtaining the informed consent from the patients, their data, including age, gender, height, body weight, vital signs, comorbidities (diabetes mellitus, hypertension, coronary artery disease, chronic obstructive lung disease, asthma, malignancies, congestive heart failure, and chronic kidney disease.), were recorded. DCBUN, creatinine, blood sugar, Na, K, total bilirubin, prothrombin time, lactic acid, and procalcitonin were all measured at the hospital’s central laboratory. We use Roche Elecsys BRAHMS PCT^® for the PCT measurement. MDW along with CBC and differential count were measured using Beckman Coulter DxH 900^® by the study assistant, according to the manufacturer’s suggestion. The sample was processed and measured within 2 h after collection.

Two separate emergency physicians reviewed the final diagnosis or admission after discharge from the hospital and assigned patients into four groups. The “non-infection” group: patients with no diagnosis related to an infection and did not meet the systemic inflammatory response syndrome (SIRS) criteria; the “infection” group: patients with recognized infectious disease but not meeting SIRS criteria; the “infection + SIRS” group: patients meeting the SIRS criteria with any known infection; the “Sepsis-3” group: patients who met the sepsis-3 criteria of 2017 by which qSOFA score and SOFA score were used accordingly. If the initial assignment did not match, the consensus was made after discussion. Both physicians were blinded to the MDW and procalcitonin data during the process.

The primary outcome was the diagnostic accuracy of both MDW and PCT presented with the AUC, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) in predicting infection with/without SIRS and sepsis-3. The secondary outcome was the diagnostic accuracy of WBC count along with MDW in predicting infection and sepsis-3.

Continuous variables were expressed as mean ± SD (standard deviation), and categorical variables were indicated as frequency (%). Statistical significance was determined with Chisq-square test and Kruskal–Wallis test for categorical and continuous variables, respectively. We used boxplots to draw continuous variables. Logistic regression analysis was performed to identify the effect of six variables on the four groups. Diagnostic ability was evaluated in terms of the AUC, sensitivity, specificity, PPV, and NPV along with their 95% CI values. The initial cut-off values were obtained using the Youden index approach, which was optimized iteratively to maximize the sensitivity and specificity. Both WBC count and MDW were dichotomized to 0 and 1 based on their values falling into the normal or abnormal category. WBC count was normal if the recovered value was between 4000 and 11,000/μL. MDW was normal when its recovered value was < 20. Multivariate logistic regression was performed to see if MDW at certain cutoff is an independent predictor of sepsis.

From June to September 2020, we screened all the patients admitted to the ED during convenient hours with symptoms suggesting potential sepsis. Among all 124,702 ED patients, 709 met the inclusion criteria, however, 307 patients were excluded because of exclusion criteria. The algorithm is shown in Fig. 1. We enrolled 402 patients for data analysis (Table 1). Two hundred and one patients (50%) were male, and the mean patient age was 63.7 ± 18.9 years. Vital signs at the triage presented with mean were: body temperature 37.7 ± 1.3 °C, respiratory rate (RR) 20.0 ± 4.0/min, and heart rate (HR) 105.0 ± 21.1/min. Systolic and diastolic blood pressure at arrival were 130.3 ± 29.5 and 75.0 ± 17.1 mmHg and 26.4% and 38.1% of patients had diabetes and hypertension, respectively. In addition, 29.6% of the patients had known malignancy. Pneumonia (lower respiratory tract infection) was the most common infection focus (26.7%), followed by urinary tract infection (16.2%), intra-abdominal infection (11.5%), and soft tissue infection (4.2%). WBC count and monocyte distribution width were 11.5 ± 17.9/µL and 22.6 ± 5.3, respectively. The average serum lactate and procalcitonin were 20.3 ± 16.3 mg/dL and 2.5 ± 11.1 ng/mL, respectively. Blood culture was positive in 59 patients (14.7%), and overall mortality was 9.4%.

Based on the chart review by two separate emergency physicians blinded to both MDW and PCT results, we assigned the patients to four distinct groups as follows (Table 2), the “non-infection” group (n = 64, 15.9%), the “infection” group (n = 82, 20.4%), the “infection + SIRS” group (n = 202, 50.2%), and the “sepsis-3” group (n = 54, 13.4%).

Among the four groups, age and gender distribution were similar. Patients in the infection + SIRS group had the highest initial body temperature (38.2 ± 1.2 ℃, p < 0.001). Patients in the sepsis-3 group had a higher respiratory rate (24.6 ± 4.6/min) and lower systolic blood pressure (114.8 ± 30.66 mmHg, p < 0.001). There was no difference regarding the underlying condition of the patients. Laboratory examinations, including BUN, sodium, potassium, total bilirubin, alanine aminotransferase, were not different between groups. Infection + SIRS and Sepsis-3 groups had higher serum creatinine and INR (p < 0.05). WBC count in each group was 8.8 ± 5.0, 7.9 ± 3.3, 12.1 ± 6.0, and 18.3 ± 46.4 1000/uL, respectively (Fig. 2A, p < 0.05). MDW in each group was measured to be 18.5 ± 3.3, 20.8 ± 4.0, 23.5 ± 4.6, and 26.8 ± 7.0, respectively (Fig. 2B, p < 0.05). The estimated PCT in each group was 0.2 ± 0.3, 0.4 ± 1.5, 1.9 ± 7.9, and 10.3 ± 24.6 ng/mL, respectively (Fig. 2C, p < 0.05). Additional file 2: Table S1 reported continuous variables in median.

Next, we evaluated the diagnostic accuracy of MDW, PCT and MDW + WBC with AUC analysis. For predicting infection + SIRS, the AUC of MDW and PCT was 0.753 (0.701–0.804) and 0.704 (0.65–0.759), respectively (Table 3 and Fig. 3A). The best MDW cut-off values were estimated to be 19.3, which is similar to the suggested cut-off in previous studies. Based on the earlier reports and our data, we defined normal WBC count in the range of 4000 to 11,000/µL [28]. When normal WBC count and normal MDW were used as the cut-off, the AUC predicting infection + SIRS and sepsis-3 was 0.784 (0.738–0.829), and 0.699 (0.631–0.768), respectively. The sensitivity, specificity, PPV, and NPV of MDW using 20 as the cutoff in predicting infection + SIRS were 86.4%, 54.2%, 76.4%, and 70%, compared to 32.9%, 88%, 82.5%, and 43.4% using 0.5 ng/mL as the PCT cutoff value. On combing MDW and WBC count, the sensitivity and NPV further increased to 93.4% and 80.3%, respectively. In predicting sepsis-3, the AUC of MDW and PCT was 0.722 (0.652–0.792), and 0.733 (0.663–0.802), with the best MDW cut-off value of 23.4 and 0.31 ng/mL (Fig. 3B). When the WBC count was added to the model, the AUC was 0.700 (0.631–0.768). MDW, using 20 as cutoff, exhibited sensitivity, specificity, PPV, and NPV of 90.6%, 37.1%, 18.7%, and 96.1%, respectively, compared to 49.1%, 78.6%, 26.8%, and 90.6% when 0.5 ng/mL PCT was used as cutoff. When WBC count was combined with MDW, the sensitivity and NPV were 86.8% and 95.3%, respectively. Multivariate logistic regression showed that MDW > 20 is an independent predictor of both infection + SIRS and sepsis-3. The odds ratio is 3.991 (2.152–7.402) and 6.472 (2.195–19.09) (Additional file 2: Table S2), respectively.