Academic pediatric clinical research: factors associated with study implementation duration

Table 3 Analysis of factors associated with the implementation duration using multivariable regression: a- all included studies (n = 139), b- restricted to studies involving drug therapy (n = 35)

a) All studies, multivariable mixed linear model with a random intercept for year of grant ( n = 139)
Variable	Coefficient (95 % CI)	p -value
Intercept	22.3 (18.5;26.1)
Funding mechanism (ref. = Government)		<0.001
External	-7.7 (-11.9;-3.5)	<0.001
Internal	-5.3 (-9.8;-0.8)	0.02
Number of centers ^a	-0.1 (-0.2;0.02)	0.07
Type of study (ref = Observational)		0.003
Interventional involving drug therapy	6.0 (0.7;11.3)	0.03
Interventional without drug therapy	3.5 (-0.3;7.3)	0.06
b) Studies involving drug therapy, using multivariable linear regression ( n = 35)
Variable	Coefficient (95 % CI)	p -value
Intercept	14.9 (6.9;22.9)
Funding mechanism (ref. = Government)		0.08
External	-6.7 (-13.2;-0.2)	0.05
Internal	-2.9 (-9.8;4.0)	0.41
Study population (ref. = Mixed population)		0.01
Pediatric population	7.8 (1.1;14.5)
Placebo-controlled study	6.6 (0.9;12.3)	0.01
Age-appropriate formulation (ref. = Appropriate)		0.05
Intermediate at least for one drug of the study	-3.6 (-13.0;5.8)	0.44
Inappropriate at least for one drug of the study	6.9 (-0.2;14.0)	0.06

^aFor an increase of 1 centre
For all multivariable regression models, P < 0.20 was used for the variable entry criteria. Then, a stepwise selection procedure was applied. Because the statistical analyses were essentially exploratory, factors associated with the considered outcome with P < 0.10 were kept in the final model. Variables dropped at earlier stages were re-evaluated for inclusion in the final model
Values shown are coefficient, 95 % confidence interval (95 % CI). Individual p-value for each category of the variables are shown (in italics), as well as the global p-value of the variable (in bold)

ISSN: 1471-2288