The flow diagram of publications in Figure 1, is illustrated according to the Quality of Reporting of Meta-analyses statements. Overall, 13 study were included at present systematic review and meta-analysis. Two articles reported mortality in dementia patients who had PEG feeding tube (16, 17), 10 studies reported survival rate (18-27) and one reported both of mortality rate and survival in this papulation (28).
At the primary databases searching, 11377 related data were identified (762 in PubMed and 10610 in Scopus). In continue, fallowing manual search of related articles, 5 additional studies were enrolled in the present systematic review and meta-analysis. In next step, the duplicated studies were determined and removed using Endnote software (n=742).
The title and abstract of remained publications (10635) were reviewed to determine included studies. The fallowing studies were excluded from this work (totally 10574):
Unrelated studies (n=7891), the disorder other than dementia (872), the Data from patients with dementia were combined with other diseases (n=166), animal or in vitro studies (n=1053), case reports (261) and review articles (n=249). Sixty-one articles selected and their eligibility were evaluated exactly trough review their full-text. Finally, 13 articles included to meta-analysis after studies with the following characteristics were excluded (totally n=48): Full text not found (n=5), non-english full-texts (n=12), without expected outcomes (n=15), PEG was administered in combination with naso-gastric tube (NGT) or other alternative nutrition (n=8) and the data from patients with dementia were combined with other diseases (n=8).
The included studies characteristics are presented in Table 2. Based on search strategy in this mete-analysis, the relevant data were enrolled up to Aug 2019. A total of 1020 dementia patients who had PEG feeding tube, and 1296 patients in control group (408 demented patients with oral nutrition (ON) or NGT nutrition support, 678 patients with secondary dementia as a result of stroke with PEG and 210 patients with other disease and PEG nutrition support) participated in current meta-analysis. The mean age of participants were 78.5 years old. Among included studies, 3 studies were performed in United States (16, 19, 21), 3 studies in Japan (24, 26, 27), 2 studies in Israel (18, 22), 1 study in United Kingdom (17), 1 study in France (20), 1 study in Sweden (23), 1 study in Colombia (25) and 1 study in Italy (28). Across the eligible articles, eight studies reported survival or mortality rate in dementia patients with PEG feeding as illustrated in Kaplan-Meier graph or table, one reported only 180-day mortality and in 7 articles median survival days was extractable from the study text or Kaplan-Meier graph (16-28). All of studies were performed on both of male and female except one (21). The feeding method, underling disorder, complications and predictors of poor survival of participants are presented in Table 2. In the 4 studies the complications were not mentioned (19, 23, 26, 28). In 8 articles age has been evaluated as a survival predictor (17, 18, 20, 22, 23, 26-28), albumin serum levels in 3 articles (18, 20, 27) and 1 articles dementia stage (21), in other included studies, various factors have been mentioned and in two studies it has not mentioned (16, 24). Six studies had prospective and 7 studies had retrospective design. A significant increase in mortality rate in dementia patients with PEG tube feeding was reported in 3 articles (16, 17, 28), higher survival was observed in 3 studies (18, 23, 24), while Rimon et al (22), Atencio et al (25) and Ticinesi et al (28) reported shorter survival in these patients. Five included articles found no significant differences in median survival in dementia patients who receive PEG in compare to control group (19-21, 26, 27) (Table 2).
The 30-day mortality in demented patients with PEG
As presented in figure 2, the risk ratio preformed on 8 studies, showed PEG intervention had no statistically significant effect on 30-day mortality (RR: 1.16; 95% CI: 0.59 to 2.28; P= 0.66). In addition, a significant heterogeneity was observed among studies (I2 =81%, P<0.001).
To identify the between study heterogeneity sources, subgroup analysis was conducted on control group intervention (oral, NGT or PEG), diseases, and age (Table 3). The subgroup analysis found that in 30-day mortality, the PEG method intervention in control group, as well as disease (dementia or other disorder), and 80> age was detected as the potential sources of heterogeneity. However, among these subgroups, no significant reduction in 30-day mortality was found after subgroup analysis based on feeding method, disease and age (Table 3).
The 90-day mortality in demented patients with PEG
The overall RR from 8 studies showed that PEG intervention exerts no significant reduction in 90-year mortality (RR: 1.13; 95% CI: 0.60 to 2.16; P = 0.70), with a considerable between studies heterogeneity (I2 =93%, P<0.001) (Figure 3). Subgroup analysis showed that PEG intervention in control group, disease (dementia or other diseases) and age (80> or 80≤) are considered as heterogeneity sources. Fallowing subgroup analysis based on feeding rout in control participants, a significant reduction in 90-day mortality was found in NGT group (RR: 0.51; 95% CI: 0.31 to 0.82; P=0.005) and increasing in 90-day mortality in oral feeding in (RR: 1.70; 95% CI: 1.06 to 2.74; P=0.03) comparison with PEG receiving patients with dementia.
In addition, no significant differences in 90-day mortality was observed after subgroup analysis based on other diseases or age (Table 3).
The 180-day mortality in demented patients with PEG
According to figure 4 illustrated RR of PEG feeding on 180-day mortality rate in patients with dementia preformed on 9 studies (505 cases and 857 controls), PEG intervention exerts no statistically significant reduction on 180-day mortality rate (RR: 1.07; 95% CI: 0.75 to 1.53; P=0.70). There was a significant heterogeneity among included studies (I2 =88%, P<0.001). Fallowing subgroup analysis based on feeding rout, disease and age as identified heterogeneity sources, a significant reduction in 180-day mortality was observed in NGT receiving group (RR: 0.59; 95% CI: 0.43 to 0.81; P=0.001) comparison with PEG intervention in patients with dementia. also, no significant differences in 180-day mortality was found after subgroup analysis based on other diseases or age (Table 3).
The 1-year mortality in demented patients with PEG
At the present meta-analysis, based on overall effect sizes of 8 included data, the PEG intervention had no significant effect on 1-year mortality reduction (RR: 1.01; 95% CI: 0.77 to 1.33; P = 0.94), with a considerable heterogeneity across enrolled studies (I2 =92%, P<0.001) (Figure 5). After subgroup analysis, a significant reduction in 1-year mortality rate was detected in NGT receiving group (RR: 0.67; 95% CI: 0.46 to 0.97; P=0.04) comparison with PEG intervention in dementia patients (Table 3).
The 2-year mortality in demented patients with PEG
The overall effect sizes that preformed on 5 studies illustrated that PEG intervention couldn’t exert any statistically significant differences on 2-day mortality rate in patients with dementia (RR: 1.02; 95% CI: 0.77 to 1.34; P= 0.91) (Figure 6). As presented in figure 6, a significant between-studies heterogeneity was detected (I2 =96%, P<0.001). After subgroup analysis, a significant reduction in 2-year mortality rate was detected in NGT receiving group (RR: 0.63; 95% CI: 0.50 to 0.79; P=<0.001) comparison with PEG intervention in dementia patients (Table 3). In addition, in patients 80≤ years, peg intervention significantly increases 2-years mortality (RR: 1.46; 95% CI: 1.21 to 1.75; P<0.001).
The median survival days in demented patients with PEG
The figure 7 illustrated the WMD of median survival days in demented patients with PEG method. The overall effect from the random-effect model that preformed on 7 studies showed PEG intervention had no statistically significant effect on patient’s median survival (WMD: 9.77; 95% CI: -22.43 to 41.98; P= 0.55) (Figure 7). In addition, no significant between studies heterogeneity was identified among included studies (I2 =0%, P= 0.48) (Table 4).