2018 26 APR
Type of content: news
Communicative register:
Focus area: Physicians

Mediation analysis: an application on the relationship between Growth Hormone Deficiency (GHD) and lung function in children

Mediation analysis (MA) is a statistical methodology, widely used in many field of research.

Mediation analysis (MA) aimed to estimate both the direct and the indirect effect of independent variables on a dependent variable, providing the assessment of the total effect [1]. For example, it has been applied in clinical study investigating the relationship between indicators of adiposity/obesity, allergy markers, and measures of asthma severity or control [2]. MA allows to assess whether the relationship between a predictor and an outcome variable can be explained by their relationship with a third, mediator variable.
In a recent Italian study, MA was applied, at baseline (T0) and after 1-year treatment of growth-hormone therapy (GHT) (T1),  in order to understand the extent to which the relationship between Growth Hormone Deficiency (GHD) and Pulmonary Function Tests  (PFTs) could be ascribed to a height-mediated (indirect) or a GH direct effect [3].
47 newly diagnosed GHD pre-pubertal children were recruited since January 2014 to July 2015. In all patients, height, body mass index (BMI) and insulin-like growth factor-I (IGF-I) were measured at baseline (T0) and twelve-month GHT (T1). Patients underwent PFTs both at T0 and at T1. At T0, 47 healthy children matched for age and sex were also studied as Controls.
Authors carried out two MA one at T0 and the other one at T1, separately by means of simultaneous estimation of the two following regression models (i, ii), using the Structural Equation Modeling (StrEM) approach:

i.                     Regression of the mediator “height” (h) on the independent variable GHD, i.e. a categorical variable indicating whether PFT refers to the Controls (reference) or the GHD children: h=c0 + a0·GHD

ii.                     Regression of the dependent variable PFT on the independent variable GHD and the mediator h: PFT=c1 + b0·GHD + d·h 

Indirect and direct effects were derived from the previous equations. In detail: indirect effect=a0·d; direct effect=b0. The total effect is given by the sum of the respective indirect and direct effects. The percentage of height-mediated effect indicates how much of the total GHD effect is mediated by height (%mediated effect = 100*indirect effect/total effect).
Results showed that, in GHD children PFTs were significantly impaired at baseline and improved after one-year of GHT. After performing the MA at T0, the observed impairment of lung function could not be exclusively attributed to the GHD indirect height-mediated effect, i.e short stature, but could be also attributed to a GHD direct effect, mainly for FVC, FEV1, FEF25-75 and TLC. After 1 year of GHT, the MA results show that the direct effects on FVC and FEV1 became not significant and the findings of percentages of the mediated effect for spirometry indices were ≥50%, suggesting that the height-mediated indirect effect is the most relevant effect.
In conclusion, this study shows that pre-pubertal children with GHD have an impairment of lung function not exclusively attributable to the indirect (height-mediated) effect, but also to the direct GH action, which is mitigated after 1-year of GHT.

[1] MacKinnon DP, Fairchild AJ, Fritz MS. Mediation Analysis. Annual review of psychology. 2007;58:593.
[2]Forno E, Acosta-Pérez E, Brehm JM, Han YY, Alvarez M, Colón-Semidey A, Canino G, Celedón JC. Obesity and adiposity indicators, asthma, and atopy in Puerto Rican children. J Allergy Clin Immunol. 2014 May;133(5):1308-14.
[3] Cilluffo G, Ferrante G, Fasola S, Ciresi A, Cardillo I, Tancredi G, Viegi G, Giordano C, Scichilone N, La Grutta S. Direct and indirect effects of Growth Hormone Deficiency (GHD) on lung function in children: A mediation analysis. Respir Med. 2018 Apr;137:61-69.

Article by Giovanna Cilluffo