Data sources CENTRAL, MEDLINE, Embase and CINAHL.Study selection Controlled studies (randomised or non-randomised) that evaluated the effect of full-body PPE on healthcare workers (HCW) exposed to highly infectious diseases, assessed which method of donning and doffing PPE was associated with reduced risk of contamination or infection for HCW, and which training methods increased compliance with PPE protocols.Data extraction and synthesis Two reviewers independently screened the titles and abstracts for inclusion of studies. Full text articles were subsequently assessed for eligibility and disagreements were resolved through consensus. Using criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions, pairs of review authors independently assessed risk of bias for each randomised study and rated each potential source of bias as high, low, or unclear.ROBINS-I tool was used for the assessment of risk of bias in non-randomised intervention studies. Where appropriate, random effects meta-analyses were conducted.Results A total of 24 studies (randomised controlled trials [RCT] [n = 14]; Quasi-RCT [n = 1] and non-randomised design [n=9]) with 2278 participants were included. Included studies compared types of PPE (n = 8), evaluated modified PPE (n = 6), procedures for donning and doffing PPE (n = 8), and types of training (n = 3). Twenty-two studies were simulation studies, of which 18 simulated exposure of HCW to contaminated body fluids using fluorescent markers or harmless microbes and measured contamination outcomes, and four studies provided modified PPE or procedures and measured compliance with donning and doffing procedures. Types of PPE Powered, air-purifying respirator with coverall may protect against the risk of contamination better than a N95 mask and gown (risk ratio [RR] 0.27, 95% confidence interval [CI] 0.17 to 0.43) but was more difficult to don (non-compliance: RR 7.5, 95% CI 1.81 to 31.1). Gowns compared to aprons may protect better against contamination (MD) -10.28, 95% CI -14.77 to -5.79). Breathable types of PPE are more comfortable and may increase user satisfaction, however with little impact on contamination. Modified PPE versus standard PPE Appropriate modifications to PPE design may lead to less contamination compared to standard PPE. For example, contamination can be reduced using a sealed gown and glove combination so that they can be removed together and cover the wrist area (RR 0.27, 95% CI 0.09 to 0.78), tight fitting gown around the neck, wrist area and hands (RR 0.08, 95% CI 0.01 to 0.55) and added tabs to facilitate doffing of masks (RR 0.33, 95%nCI 0.14 to 0.80) or gloves (RR 0.22, 95% CI 0.15 to 0.31).Guidance on PPE use: following the guidance and recommendations from the Centres for Disease Control and Prevention for doffing PPE compared to no guidance may reduce self-contamination (MD -5.44, 95% CI -7.43 to -3.45). One-step removal of gloves and gown compared to separate removal (RR 0.20, 95% CI 0.05 to 0.77), double gloving compared to single gloving (RR 0.34, 95% CI 0.17 to 0.66) and sanitising gloves before doffing with quaternary ammonium or bleach (but not alcohol-based hand rub) may decrease contamination. Additional verbal instructions may lead to fewer errors in doffing (MD -0.9, 95% CI -1.4 to -0.4). User training To a vast extent, face-to-face training may reduce non-compliance with doffing guidance (odds ratio 0.45, 95% CI 0.21 to 0.98) compared to solely providing folders or videos. In addition, computer simulation may lead to fewer errors in doffing (MD -1.2, 95% CI -1.6 to -0.7) and video lecture on donning PPE may lead to better skills scores (MD 30.70, 95% CI 20.14 to 41.26) compared to traditional lectures.Conclusions The more body parts are covered with PPE the better protection it offers. However, this is also associated and increased difficulty in donning and doffing PPE, and the PPE is less comfortable. Coveralls are the most difficult PPE to remove but may offer the best protection, followed by long gowns, gowns and aprons. The included studies had a high or unclear risk of bias, indirectness of evidence in simulation studies and small participant numbers. This increases the uncertainty about the estimates of effects, and it is likely that the true effects may be substantially different from the ones reported in this review.