Abstract

The present study attempted to quantitatively summarize the psychological, physiological, and ergogenic effects of anabolic-androgenic steroids (AAS)s in healthy humans. Using the format of a meta-analysis, data from 127 studies and 35 outcomes was converted into 894 effect sizes and summarized across studies. In order to account for any significant variability within effect sizes, each finding was coded for 15 moderator variables. Unfortunately, due to a lack of adequate findings, no psychological outcome was included in the analysis. As indicated by the magnitude of the average effect size, large negative treatment effects of AASs were observed for sperm concentration, testicle size, follicle-stimulating hormone, and luteinizing hormone. Moderate positive treatment effects were observed for lean body mass, thigh circumference, bench press, and squat; while moderate negative effects were found for Apo-AI, high-density lipoprotein, sex hormone binding-globulin, and albumin. Small positive treatment effects were observed for hemoglobin, hematocrit, body weight, biceps circumference, low-density lipoprotein, serum estrogen, and serum testosterone. Significant heterogeneity among effect sizes for high-density lipoprotein (HDL), low-density lipoprotein (LDL), estrogen (E2), follicle-stimulating hormone (FSH), luteinizing hormone (LH), and testosterone (T) indicated that these findings were not reliable and that further analysis of moderator variables was required. In general, type of drug and route of administration were identified as being the two most important moderators of variability in the six heterogeneous outcomes. Specifically, testosterone and its esters produced higher serum levels of HDL, E2, and T and lower levels of LDL and LH than other synthetic steroids. Also, when compared with oral compounds, the use of parenterally-administered steroids resulted in higher serum levels of T and lower levels of FSH and LH. Models incorporating indicators of study quality and subjects' experience with AASs along with treatment variables (i.e., drug, form, dose, and cycle length) sufficiently accounted for the heterogeneity in HDL, LDL, and FSH. However, excess heterogeneity in effect sizes prevents conclusions of any certainty being made about the relationship between AASs and E2, LH, or T