Delta Neutrophil Index (DNI) as a novel diagnostic and prognostic marker of infection: a systematic review and meta-analysis

Supplementary Material- Figures and tables

Jae Hyon Park1*, HyeongJuByeon1*, Geum Hwa Lee2, Jong Wook Lee3,4, Andreas Kronbichler5, Michael Eisenhut6, and Jae Il Shin2

1.Yonsei University College of Medicine, Seoul, Korea

2.Department of Pediatrics, Yonsei University College of Medicine, Seoul, Korea

3.JincheonSungmo Hospital, Jincheon, Korea

4.Research Institute of Bacterial Resistance, Yonsei University College of Medicine, Seoul, Korea

5.Medical University Innsbruck, Department of Internal Medicine IV (Nephrology and Hypertension), Innsbruck, Austria

6.Luton & Dunstable University Hospital NHS Foundation Trust, Luton, United Kingdom

Note: All reference numbers refer to the reference list in the main manuscript text.

*These authors contributed equally to this work

Supplementary Figure S1. Deek’s funnel plot asymmetry test for publication bias. Potential bias is considered to exist for P<0.05
For prognostic factor study (left); For predictive factor study (right)
Deek’s Funnel Plot Asymmetry Test: A non-zero slope coefficient is suggestive of significant small study bias (p<0.05)
2-tailed P-value: 0.582 (left); 0.521 (right)
Acronym: DNI: delta neutrophil index
Left: publication bias of the analysis result of DNI as a prognostic marker
Right: publication bias of the analysis result of DNI as a predictive marker
Supplementary Figure S2. Summary receiver operating characteristic curve of DNI as a predictive factor
Acronym: DNI: delta neutrophil index
Supplementary Table S1. Summary of positive likelihood ratio (LR+) and negative likelihood ratio (LR-) of DNI as predictive factor
Author (Year) / LR+ (95% CI) / % Weight / LR- (95% CI) / % Weight
Ahn (2014) / 8.382 (5.442-12.909) / 16.17 / 0.587 (0.419-0.822) / 16.23
Park (2014) / 5.318 (1.471-19.223) / 10.90 / 0.215 (0.100-0.459) / 11.01
Lee (2014) / 2.303 (1.719-3.085) / 16.72 / 0.558 (0.463-0.674) / 17.65
Kim (2014) / 8.296 (3.878-17.748) / 14.31 / 0.124 (0.064-0.241) / 12.21
Pyo (2013) / 4.416 (2.191-8.900) / 14.67 / 0.521 (0.375-0.723) / 16.34
Seok (2011) / 34.909 (14.305-85.192) / 13.46 / 0.279 (0.160-0.486) / 13.51
Park (2011) / 9.055 (3.878-21.143) / 13.75 / 0.210 (0.116-0.381) / 13.05
Pooled LR / 7.152 (3.553-14.398) / 0.334 (0.224-0.498)
Analysis Report / Heterogeneity chi-squared (Cochran-Q): 53.57
(d.f. =6) p=0.000 (<0.001)
Inconsistency (I-square) = 88.8%
Estimate of between study variance (tau-squared): 0.7397 / Heterogeneity chi-squared (Cochran-Q): 39.58
(d.f. =6) p=0.000 (<0.001)
Inconsistency (I-square) = 84.8%
Estimate of between study variance (tau-squared): 0.2245
Acronyms: LR: likelihood ratio; DNI: delta neutrophil index; d.f.: degrees of freedom
Both positive likelihood ratio and negative likelihood ratio were generated under random effects model
Supplementary Figure S3. Diagnostic odds ratio (OR) of DNI assay as a prognostic marker (left) and as a predictive marker (right)
Acronym: DNI: delta neutrophil index
Left: analysis result of DNI as a prognostic marker
Right: analysis result of DNI as a predictive marker
Supplementary Figure S4. Forrest plot of DNI % in predicting infection in patients. The overall pooled odds ratio was 2.091 (0.588-7.440)

Heterogeneity: (Random Model)
Q-value: 5.062 (Degrees of freedom: 2)
I-squared: 60.490 (P-value= 0.080)
Tau-squared: 0.765; Standard error: 1.352; Variance:1.829; Tau: 0.875
Acronym: DNI: delta neutrophil index
Supplementary Figure S5. Funnel plot of standard error, funnel plot of precision by logarithmic odds ratio, and results of Egger’s linear regression test evaluating publication bias. (for Supplementary Figure S4)
Egger’s linear regression test:
Intercept: 1.466; T-value: 2.007 (P-value 2 tailed: 0.294)
Acronym: DNI: delta neutrophil index
Supplementary Figure S6. Sensitivity and specificity of DNI assay as a prognostic factor for death as outcome
Analysis of Diagnostic Threshold:
Spearman correlation coefficient: -0.086 (P value=0.872)
Acronym: DNI: delta neutrophil index
Supplementary Figure S7. Summary receiver operating characteristic curve for DNI as a prognostic factor
Acronym: DNI: delta neutrophil index
Supplementary Table S2. Summary of positive likelihood ratio (LR+) and negative likelihood ratio (LR-) of DNI as prognostic factor
Author (Year) / LR+ (95% CI) / % Weight / LR- (95% CI) / % Weight
Kim (2014) / 2.963 (2.036-4.312) / 57.53 / 0.317 (0.134-0.751) / 20.67
Kim (2014) / 3.800 (2.060-7.010) / 21.60 / 0.330 (0.141-0.772) / 21.32
Lim (2014) / 4.053 (1.919-8.557) / 14.50 / 0.491 (0.287-0.841) / 53.07
Lee (2013) / 5.067 (1.641-15.639) / 6.38 / 0.238 (0.041-1.386) / 4.94
Pooled LR / 3.386 (2.547-4.500) / 0.398 (0.269-0.589)
Analysis Report / Heterogeneity chi-squared (Cochran-Q): 1.48
(d.f. =3) p=0.688
Inconsistency (I-square) = 0.0%
Estimate of between study variance (tau-squared): 0.0 / Heterogeneity chi-squared (Cochran-Q): 1.49
(d.f. =3) p=0.685
Inconsistency (I-square) = 0.0%
Estimate of between study variance (tau-squared): 0.0
Acronyms: LR: likelihood ratio; DNI: delta neutrophil index; d.f.: degrees of freedom
Both positive likelihood ratio and negative likelihood ratio were generated under random effects model
Supplementary FigureS8. Fagan nomogram of DNI test as a prognostic marker in predicting mortality
Acronym: DNI: delta neutrophil index
Supplementary Figure S9. Forrest plot of DNI % predicting mortality in infected patients (ER and GW patients). The overall pooled hazards ratio was 1.085 (1.000-1.178)

Heterogeneity: (Random Model)
Q-value: 8.028 (Degrees of freedom: 2)
I-squared: 75.088 (P-value= 0.018)
Tau-squared: 0.003; Standard error: 0.005; Variance: 0.000; Tau: 0.055
Acronym: DNI: delta neutrophil index
Supplementary Figure S10. Funnel plot of standard error, funnel plot of precision by logarithmic hazard ratio, and results of Egger’s linear regression test evaluating publication bias (for Supplementary Figure S9)
Egger’s linear regression test:
Intercept: 2.249;T-value: 1.586 (P-value 2 tailed: 0.358)
Acronym: DNI: delta neutrophil index
Supplementary Figure S11. Forrest plot of DNI % predicting mortality in infected patients as a prognostic factor (subgroup analysis of studies that analyzed the general ward patients). The overall pooled hazard ratio was 1.959 (0.520-7.378)

Heterogeneity: (Random Model)
Q-value: 7.879 (Degrees of freedom: 1)
I-squared: 87.309 (P-value= 0.005)
Tau-squared: 0.812; Standard error: 1.316; Variance:1.731; Tau: 0.901
OR, 95%CI and P-value of a study that analyzed the emergency room (ER) patients
Hwang (2015): hazards ratio: 1.068; 95%CI: 1.01-1.13; P-value: 0.0209
Acronym: DNI: delta neutrophil index
Analysis of publication bias using Egger’s linear regression test and funnel plots could not be performed because the number (n=2) of studies analyzed is less than three.
Supplementary Figure S12. Forrest plot of DNI % predicting mortality in infected patients. The overall pooled odds ratio was 10.355 (0.055-1965.776)

Heterogeneity: (Random Model)
Q-value: 4.334 (Degrees of freedom: 1)
I-squared: 76.925 (P-value= 0.037)
Tau-squared: 11.639; Standard error: 21.397; Variance: 457.853; Tau: 3.412
Acronym: DNI: delta neutrophil index
Analysis of publication bias using Egger’s linear regression test and funnel plots could not be performed because the number (n=2) of studies analyzed is less than three.