The search strategy yielded 716 potentially relevant papers.Seven additional potential articles were identified from referencelists. From the inclusion and exclusion criteria, ten studies (sevenwere case control studies and three were prospective cohort studies) published in English between 1993 and 2015 encompassing2092 patients (457 in the UE group; 1635 in the no UE group) wereeligible for inclusion in the meta-analysis (Fig. 1). No randomisedstudy was found.Study characteristicsThe characteristics and factors associated with the ten includedstudies are shown in Tables 2 and 3.For the case control studies, the participants of the no UE groupin three of the studies were randomly selected, others were according to case matching. In all the case matching studies, all of them(Chang et al., 2011; Tung et al., 2001; Chang et al., 2008; Atkinset al., 1997) matched cases and controls on basis of gender, ageand dates in hospitals, while three of them matched cases andcontrols on basis of diagnosis (Chang et al., 2011; Tung et al.,2001; Chang et al., 2008), therefore, these studies were taken outof the analysis to assess the impact of these factors on the riskfor UE. And for the cohort studies, all the patients were includedduring the study period. Four of the studies were from Europe,three were from Asia and three were from the USA.Evaluated factors associated with UE were age (3 studies), alertand oriented (2 studies), APACHE II (4 studies), GCS (4 studies),gender (5 studies), physical restraint (6 studies), oro-tracheal intubation (3 studies), use of sedative (5 studies) and disease category(6 studies). Several parameters were evaluated only once: durationof ventilation, smoking history, days of intubation, temperature,pCO2, BUN, medicine, BMI, Pain score, Riker score, length of stayin medical ICU, years of experience as nurse, level of education ofnurse, work shift at the time of UE, vital signs and serumlaboratory.Quality assessmentThe results of the individual studies based on the NOS are presented in Table 3, indicating that a medium to high standard ofdata was collected.Synthesis of resultsIn all the including studies, over 20 factors were examined forthe UE in ICU adult patients.For each parameter investigated in more than one article, ameta-analysis was attempted. Our meta-analysis showed significant statistical risk factors for UE were as followings:Male (Fig. 2): The meta-analysis found that males have a higherrisk to experience UE in comparison to females in ICU (overall OR:1.54, 95% CI 1.12–2.12, p = 0.008). No heterogeneity was observed(I2 = 0%, p = 0.42).