SHIELDING, SAFETY ZONING, WORK ROTATION, AND PERSONAL MONITORING FOR RF-EMF EXPOSURE REDUCTION IN RADAR-RELATED OCCUPATIONS: AN BAYESIAN NETWORK META-ANALYSIS BASED ON RECONSTRUCTED STUDY-LEVEL COMPARATIVE DATA

Abstract

Military and radar workplaces can expose personnel to radiofrequency electromagnetic fields (RF-EMF), especially during maintenance, proximity to transmitting elements, or poorly controlled duty cycles. Published occupational surveys suggest that routine radar exposures are usually below occupational limits, but hotspot scenarios and accidental overexposure remain plausible. To compare four workplace RF-EMF mitigation strategies relevant to radar-related occupationsshielding, safety zoning, work rotation, and personal monitoringusing an  Bayesian network meta-analysis (BNMA). A structured scoping review of authoritative guidance and occupational exposure literature was used to define intervention nodes, realistic operating ranges, and the primary outcome: time-weighted RF-EMF exposure expressed as a fraction of the applicable occupational limit. A reconstructed dataset of 14 two-arm comparative studies was analyzed using a Bayesian random-effects network model with weakly informative priors and 50,000 posterior draws obtained through numerical integration over between-study heterogeneity. As results, shielding produced the largest reduction versus usual practice (mean difference -16.0 percentage points of the occupational limit, 95% credible interval [CrI] -19.5 to -12.4), followed by safety zoning (-11.0, 95% CrI -14.6 to -7.4), work rotation (-7.7, 95% CrI -11.3 to -4.2), and personal monitoring (-7.3, 95% CrI -11.0 to -3.7). Shielding had a 99.2% probability of being the best intervention and a SUCRA of 0.998. Posterior between-study heterogeneity was low (mean tau 0.011; 95% CrI 0.001 to 0.033). In this  framework, engineering control through shielding outperformed administrative and surveillance-based strategies, while safety zoning was the strongest non-engineering option. Because the evidence network is reconstructed rather than derived from a mature corpus of comparative field trials, the findings should be used for prioritization, teaching, and local study designnot for direct regulatory adoption without site-specific measurements.