Faecal Calprotectin in Suspected Paediatric Inflammatory Bowel Disease: an Individual Patient Data Meta-analysis




Yüklə 54.97 Kb.
tarix27.02.2016
ölçüsü54.97 Kb.
eText 1: online-only supplement to “Faecal Calprotectin in Suspected Paediatric Inflammatory Bowel Disease: an Individual Patient Data Meta-analysis”
I. Search strategies for the identification of studies - Electronic searches


  1. Search History  22-04-2011 - Ovid MEDLINE(R) 1948 to April Week 2 2011







Searches

Results

1

exp Inflammatory Bowel Diseases/

50157

2

inflammatory bowel disease*.af.

22593

3

ibd.af.

7455

4

(crohn* adj5 disease*).af.

31430

5

(colit* adj5 ulcer*).af.

28941

6

(colit* adj5 indetermin*).af.

467

7

exp Leukocyte L1 Antigen Complex/

1554

8

calprotectin*.af.

535

9

1 or 2 or 3 or 4 or 5 or 6

60948

10

7 or 8

1659

11

9 and 10

231

12

limit 11 to "all child (0 to 18 years)"

79



  1. Search History  23-04-2011 - EMBASE 1989 to 2011 April 22







Searches

Results

1

exp enteritis/

107363

2

inflammatory bowel disease*.af.

25969

3

ibd.af.

10971

4

(crohn* adj5 disease*).af.

36578

5

(colit* adj5 ulcer*).af.

28043

6

(colit* adj5 indetermin*).af.

654

7

exp calgranulin/

1114

8

calprotectin*.af.

805

9

1 or 2 or 3 or 4 or 5 or 6

113327

10

7 or 8

1519

11

9 and 10

500

12

limit 11 to child

76


  1. Database of Abstracts of Reviews of Effects (DARE) of the Centre for Reviews and Dissemination (http://www.crd.york.ac.uk/crdweb/). The search term calprotectin resulted in three hits.




  1. MEDION database of the University of Maastricht (www.mediondatabase.nl). Using the following input (Database: Site: De medion site; DB: Medion database; Filter: Systematic Reviews of Diagnostic Studies; ICPCcode: Digestive; Signssymp: Laboratory tests, Histology and cytology, Signs and symptoms, Endoscopy and laparoscopy; Abstract: Calprotectin) we got three hits.

III. Logistic regression analysis of variables associated with IBD in a merged patient data set of 545 paediatric patients (output of Stata). Study numbers correspond to those reported in Table 1.

Model including faecal calprotectin, age and study centre



Variable

Coeff

SE

z

P>|z|

OR (95% CI)

Calprotectin (μg/g)

0.0046045

0. 0004718

9.76

0.000

1.004615 (1.003687 to 1.005545)

Age (yrs)

0. 1641661

0. 037006

4.44

0.000

1.17841 (1.095965 to1.267056)

Study 2

-0.4071389

0.5099095

-0.80

0.425

0.6655517 (.2449887 to 1.80808)

Study 3

(omitted)













Study 4

-1.191298

0.5349721

-2.23

0.0296

0.3038268 (0.1064773 to 0.8669521)

Study 5

-2.503004

0.4487185

-5.58

0.000

0.0818388 (0.0339633 to 0.1972011)

Study 6

-1.271858

0.5171681

-2.46

0.014

0.2803104 (0.1017243 to 0 .77242)

Study 8

-1.936595

0.5171681

-3.61

0.000

0.1441941 (0.050353 to 0.412924)

Study 9

-2.974031

0.5367972

-4.94

0.000

0.0510969 (0.0156976 to 0.1663243)

Constant

-1.696893

0.5400406

-3.14

0.002




Log likelihood = -179.36081

ROC area = 0.94 (0.91 to 0.96)


Final model including faecal calprotectin and age



Variable

Coeff

SE

Z

P>|z|

OR (95% CI)

Calprotectin (μg/g)

0.0041326

0.0004439

9.31

0.000

1.004141 (1.003268 to 1.005015)

Age (yrs)

0.1749246

0.0323588

5.41

0.000

1.191156 (1.117957 to 1.269149)

Constant

-3.293874

0.4199941

-7.84

0.000




Log likelihood = -209.20818

ROC area = 0.92 (0.89 to 0.94)



IV. Calculations
The fitted multiple logistic regression model is given by:


The estimated variance for the logit was calculated as:


where the estimated covariance was calculated as follows:



and where R(bi,bj) is the correlation between the regression coefficients bi and bj.

The probability p (and its confidence boundaries) is expressed as:




When calculating the probability in a population with another disease prevalence, the constant β0 should be readjusted as follows:




Verilənlər bazası müəlliflik hüququ ilə müdafiə olunur ©azrefs.org 2016
rəhbərliyinə müraciət

    Ana səhifə