2.1 Literature retrieval results
At the initial detection, 2708 articles were selected separately by two evaluators and cross-checked to reach a consensus, In the end, 29 articles on RCT were included and a total of 3617 ARDS patients were included, among which 1828 were prone position ventilation, supine position ventilation in 1789 cases, among them (周俊杰2018) since the two groups were used for different ventilation duration, 12 people were in the observation group and 12 people in the control group, and the analysis was calculated according to the grouping. The screening flow chart is shown in Figure 1.
2.2 Literature quality evaluation
All the included studies were RCT tests, and there was no difference in baseline data. The basic features of the included literature were shown in Table 1, and the bias risk assessment results of the included studies were shown in Figure 2.
2.3 Effect Evaluation
2.3.1 Influence of different duration of prone ventilation on 28-day mortality of ARDS patients
A total of 26 studies reported the effect of prone ventilation on 28-day mortality of ARDS patients.A total of 3454 patients were included, including 1755 patients in prone position and 1699 patients in supine position. According to the different ventilation time in prone position, the study was divided into three subgroups: the <8 hour group, the 8-16 hour group and the >=16 hour group. The fatality rate days of some studies were not accurate 28 days, but the average days of hospitalization were all included. (周俊杰2018) with different time prone position ventilation test, so it was conducted in two subgroups for Meta analysis.The random-effect model was used for all the three groups. After testing, the heterogeneity of the three groups was (P =0.03, I2=0%;P =0.003, I2=61%, P =0.0002, I2=0%), overall heterogeneity (P <0.00001, I2=27%).The differences were statistically significant.When two subgroups without heterogeneity were tested by fixed-effect model, the results showed no significant change.When sensitivity analysis was performed using RevMan 5.3 software (Guerin2013), there was heterogeneity in the subgroup analysis, but it was a multicenter study with high quality of evidence and could not be excluded. Beuret2002, 李华 2016 and 王强2013 may have heterogeneity. Among them, Beuret2002 selected patients were ARDS patients with severe coma, prone time of 4 hours, and sample size was relatively small.李华 2016 had patients with severe ARDS combined with chest trauma, lying prone for 6 hours.王强2013 required treatment with fiberoptic bronchoscopy for aspiration lung injury and prone position for 4-6 hours.The heterogeneity may be caused by the different degree of disease, comorbidities and general treatment, but it meets the inclusion criteria and is a baseline comparable randomized controlled trial, so it cannot be ruled out.The P values of the three subgroups in this group were all <0.05, and the difference was statistically significant, indicating that prone ventilation can reduce 28d fatality rate of ARDS patients regardless of the duration. However, it cannot be said that how long prone ventilation is more beneficial to 28d fatality rate. Figure 3 for the forest map and Figure 5 for the funnel map.
2.3.2 Influence of different duration of prone ventilation on the 3M mortality rate of ARDS patients
A total of 15 studies reported the influence of prone ventilation on the 3M mortality rate of ARDS patients.A total of 2535 patients were included, including 1280 patients in prone position and 1255 patients in supine position.Among them (Taccone2009), the outcome of the study was 6-month mortality. Due to its high quality of evidence, the 3M mortality was incorporated into the study as a long-term efficacy observation. All the studies were divided into three subgroups according to the duration of prone position ventilation, namely the <8 hour group, the 8-16 hour group and the ≥16 hour group, and the random-effect model was used for Meta analysis., the heterogeneity of the three groups was (P =0.93, I2=58%;P =0.72, I2 could not be calculated (Guerin2004 was the only study);P <0.0001, I2=0%), overall heterogeneity (P =0.04, I2=43%).There was no significant change in the outcome of the≥16-hour group after the fixed-effect model test, and the difference in the subgroup was statistically significant.Although the overall difference was statistically significant, there was only one phase study in the 8-16 hour group, so the three subgroups could not be combined and counted, and only each subgroup could be analyzed independently. The 8-16-hour group was a multi-center randomized controlled trial, with a large sample size and high quality of evidence. The difference was not statistically significant, indicating that prone ventilation and supine ventilation had no significant difference in 3M mortality. Three studies were included in the <8 hours group, and the difference was not statistically significant, indicating that prone ventilation in this group had no significant benefit on 3M mortality. When all studies in the ≥ 16-hour group were excluded one by one and then replotted for sensitivity analysis, the results of this subgroup were stable, so it was considered that in the fatality rate statistics for 3 months or longer, only the duration of ventilation in prone position ≥16 hours per day was significantly beneficial. The forest map is shown in Figure 4, and the funnel map is shown in Figure 6