Skip to main content

Table 3 Heterogeneity analysis by subgroup analysis and meta-regression for prognostic nutritional index as an acute kidney injury prediction marker

From: Prognostic nutritional index as a predictive marker for acute kidney injury in adult critical illness population: a systematic review and diagnostic test accuracy meta-analysis

Variables

Subgroups (number of studies)

Sensitivity

(95% CI)

Specificity

(95% CI)

Positive LR

(95% CI)

Negative

LR (95% CI)

Diagnostic odds ratio (95% CI)

RDOR (95% CI)#; p value#

Population

Surgical (6)

0.55* (0.40–0.70)

0.77 (0.74–0.86)

2.36 (1.44–3.87)

0.58* (0.43–0.79)

4.13 (1.92–8.92)

0.63 (0.28–1.42); 0.27

Medical (10; 11 cohort)

0.72* (0.64–0.79)

0.73 (0.64–0.80)

2.54 (2.06–3.12)

0.40* (0.32–0.50)

6.46 (4.47–9.33)

 

Procedure

PCI (6)

0.72 (0.61–0.80)

0.70 (0.59–0.80)

2.42 (1.84–3.19)

0.42 (0.31–0.56)

5.91 (3.52–9.91)

1.12 (0.49–2.32); 0.79

Other (10; 11 cohort)

0.63 (0.52–0.73)

0.77 (0.68–0.84)

2.55 (1.84–3.55)

0.48 (0.37–0.63)

5.32 (3.12–8.96)

 

CKD

Including CKD (7; 8 cohort)

0.56** (0.43–0.67)

0.75 (0.65–0.84)

2.12 (1.55–2.88)

0.61** (0.49–0.76)

3.50 (2.15–5.62)

0.44 (0.22–0.88); 0.02

Excluding CKD (9)

0.75** (0.67–0.81)

0.74 (0.64–0.82)

2.81 (2.12–3.72)

0.36** (0.29–0.46)

7.92 (5.15–12.18)

 

Serum albumin##

Alb < 3.5 g/L (5)

0.68 (0.52–0.80)

0.79 (0.67–0.87)

2.84 (2.26–3.57)

0.42 (0.28–0.36)

7.19 (4.60–11.23)

1.52 (0.58–3.97); 0.39

Alb ≥ 3.5 g/L (10)

0.64 (0.52–0.74)

0.73 (0.63–0.81)

2.31 (1.67–3.21)

0.50 (0.39–0.64)

4.68 (2.74–8.00)

 

AKI criteria

KDIGO (7; 8 cohort)

0.63 (0.49–0.76)

0.77 (0.69–0.83)

2.49 (2.05–3.03)

0.4 (0.34–0.64)

5.45 (3.45–8.62)

0.98 (0.44–2.22); 0.39

Other (9)

0.69 (0.61–0.77)

0.72 (0.60–0.81)

2.46 (1.73–3.50)

0.45 (0.35–0.59)

5.58 (3.16–9.88)

 
  1. Diagnostic indices, including sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, and diagnostic odds ratio, are determined through subgroup analysis. The overall diagnostic accuracy of the Prognostic Nutritional Index across subgroups, distinguished by the presence or absence of specific covariates, is evaluated using meta-regression and presented as the relative diagnostic odds ratio
  2. #Relative diagnostic odds ratio (RDOR) with 95% CI and p value are obtained from meta-regression which selective covariates are examined as moderator to compare the overall diagnostic performance between subgroups
  3. Subgroup difference for four diagnostic accuracy indices (sensitivity, specificity, positive LR and negative LR) were examined by subgroup analysis. *p value for subgroup difference < 0.05; **p value for subgroup difference < 0.01
  4. ##Two studies did not report baseline serum albumin level were not included in this subgroup analysis (Hatem 2023; Kurtul 2021). One study with two cohorts was classified into different subgroup (Hu, 2021)
  5. AKI: acute kidney injury; Alb: albumin; KDIGO: Kidney Disease Improving Global Outcomes; LR: likelihood ratio; RDOR: relative diagnostic odds ratio