How does study quality affect the results of a diagnostic meta-analysis?

Westwood, Marie E; Whiting, Penny F; Kleijnen, Jos

doi:10.1186/1471-2288-5-20

BMC Medical Research Methodology

Table 3 Results of the regression analysis for dipstick tests for the diagnosis of UTI

From: How does study quality affect the results of a diagnostic meta-analysis?

	Univariate analysis		Multivariate model
Variable*	RDOR (95% CI)	p-value	RDOR (95% CI)	p-value
Nitrite dipstick – weighted (n = 23 studies)
Clinical review bias avoided: yes vs unclear	4.7 (1.7, 12.7)	0.004	4.7 (1.7, 12.7)	0.004
Age <5 years vs <2 years	5.8 (0.3, 101.8)	0.213	Dropped^$
Age <12 years vs <2 years	2.2 (0.1, 35.8)	0.548	Dropped
Age <18 years vs <2 years	3.6 (0.9, 14.9)	0.076	Dropped
Europe vs North America	0.3 (0.1,0.9)	0.041	Dropped
Other areas vs North America	0.3 (0.1, 1.0)	0.050	Dropped
Nitrite dipstick – unweighted (n = 23 studies)
Clinical review bias avoided: yes vs unclear	3.4 (0.9, 13.7)	0.078	Dropped
Age <5 years vs <2 years	7.3 (0.9, 61.8)	0.067	Dropped
Age <12 years vs <2 years	2.9 (0.5, 18.3)	0.243	Dropped
Age <18 years vs <2 years	3.8 (1.1, 13.4)	0.039	Dropped
Europe vs North America	0.3 (0.1, 1.0)	0.044	0.3 (0.1, 1.0)	0.044
Other areas vs North America	0.3 (0.1, 1.2)	0.089	0.3 (0.1, 1.2)	0.089
LE dipstick – weighted (n = 14 studies)
No association at p < 0.10
LE dipstick – unweighted (n = 14 studies)
Test details reported: yes vs no	19.0 (1.9, 192.2)	0.017	Dropped
Age <5 years vs <2 years	5.4 (0.5, 64.2)	0.158	5.4 (0.5, 64.2)	0.158
Age <12 years vs <2 years	28.1 (2.3, 343.3)	0.015	28.1 (2.3, 343.3)	0.015
Age <18 years vs <2 years	1.3 (0.3, 4.7)	0.703	1.3 (0.3, 4.7)	0.703
Nitrite or leukocyte esterase dipstick – weighted (n = 15 studies)
Reference standard details reported	4.5 (0.9, 22.5)	0.064	Dropped
North America vs Europe	5.0 (0.8, 10.5)	0.076	Dropped
North America vs Other areas	1.1 (0.28, 5.0)	0.854	Dropped
Nitrite or leukocyte esterase dipstick – unweighted (n = 15 studies)
No association at p < 0.10

*Only items that showed moderate evidence (p < 0.10) for an association with the DOR in the univariate analysis are included.
^$Items were dropped where there were too few studies in one category to allow a coefficient to be calculated.

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com