Skip to main content

Table 4 Out of range (OOR) of different simulation methods verified with test data set

From: Comparison of four algorithms on establishing continuous reference intervals for pediatric analytes with age-dependent trend

Model

Total of OOR (%)

Lower OOR (%)

Upper OOR (%)

Serum creatinine of boys

 Improved non-parametrica

6.50b

3.34

3.16

 RS

4.87

2.39

2.47

 LMS

5.18

2.62

2.57

 GAMLSSc

5.03

2.38

2.65

 Discrete reference interval

4.32

2.08

2.25

 CALIPER

7.58

0.57

7.01

Serum creatinine of girls

 Improved Non-parametric

6.62

3.34

3.28

 RS

5.21

2.60

2.62

 LMS

4.94

2.48

2.46

 GAMLSS

5.45

2.85

2.60

 Discrete reference interval

4.17

2.00

2.18

 CALIPER

6.13

0.84

5.28

Alkaline phosphatase of boys

 Improved non-parametric

7.01

3.69

3.31

 RS

5.26

2.75

2.50

 LMS

4.74

2.46

2.29

 GAMLSS

5.04

2.51

2.53

 Discrete reference interval

5.66

3.03

2.63

 CALIPER

6.81

4.41

2.39

Alkaline phosphatase of girls

 Improved non-parametric

6.72

3.59

3.13

 RS

4.53

2.30

2.23

 LMS

4.94

2.48

2.46

 GAMLSS

5.02

2.43

2.59

 Discrete reference interval

4.90

2.32

2.59

 CALIPER

10.22

7.10

3.12

  1. OOR Out of range, GAMLSS General Additive Model for Location Scale and Shape method, CALIPER Canadian Laboratory Initiative in Pediatric Reference Intervals, LMS Lambda-Median-Sigma method, RS radial smoothing method
  2. a Smoothing parameters are selected by visual inspection of non-parametric method
  3. b All values are calculated as mean of the results of 100 times simulation
  4. c The models of GAMLSS method are selected from Table 3, which AIC is smallest
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us