According to the keywords (“premature OR immature delivery OR low birth weight (LBW) OR very low birth weight (VLBW)”) AND (“colostrum oral feeding OR colostrum oral intervention OR colostrum oral drip OR colostrum oral daub OR Oropharyngeal Administration”) AND (“gastrointestinal function OR digestive function”). With the language limited to English and Chinese, we retrieved the related articles from Cochrane Library, PubMed, EBSCO, CINAHL, Embase, Medline, Sino Med, CNKI, Wan Fang and VIP from their database inception to April 2018.

Retrieve the steps: 1) System review or meta-analysis of the oral colostrum intervention in the Cochrane Library was searched. 2) Relevant original documents included in Chinese and English databases such as PubMed and EBSCO were searched, and the title, abstract and key words of the obtained articles were analyzed to further determine the keywords. 3) Retrieve the above database using mesh words OR keywords, and use the “AND”, “OR”, “NOT” operators if necessary. 4) Manually search for references in the published literature.

Study selection was required to meet the following inclusion criteria: 1) The study must be Randomized Controlled trials (RCT). 2) The object of study were premature infants, A live birth of a baby less than 37 week of gestational age, weight is less than 2500 g. 1500 g - 2500 g are called low weight children, <1500 g are extremely low weight children [3] [4] . 3) The intervention method must have two conditions, one is the use of colostrums, and the other is oral intervention. The frequency, dose and duration of intervention should be described. 4) outcome indicator: a) Gestational age at the time of complete non-retention; b) Incidence of feeding intolerance, Patients with feeding intolerance without a clear definition are characterized by gastric retention, vomiting and abdominal distension [8] ; c) Incidence of nosocomial infections, septicemia [9] ; d) The incidence of necrotizing enterocolitis; e) LOS (length of stay); f) Secretary immunoglobulin A (slgA); g) lactoferrin.

Exclusion criteria: 1) The studied that do not include outcome measures; 2) Repeated publication; 3) Unable to extract data or obtain full text; 4) The babies with congenital diseases or malformations or whose guardians cannot provide colostrums.

All literature quality evaluations were completed by 2 researchers trained in an evidence-based nursing course. Firstly, the included literature was evaluated independently according to the evaluation criteria of Cochrane evaluation manual 5.1.0 [10] . When disagreements arise, discuss inclusion or exclusion. If the above criteria are fully met, the probability of occurrence of various biases is small, grade A. if partial; the probability of occurrence of bias is moderate, grade B. The probability of bias is higher, which is grade C. A, B can be used, C can be excluded. Recommended level: A > B > C.

Two researchers developed standardized tables based on literature content and independently extracted data. Including the year, author, sample size, intervention measures, frequency of intervention, dose used for intervention, intervention method, and duration and outcome index of the included literature.

The Revman 5.3 software was used for Meta analysis, Weighted Mean Differences (WMD) was used to analyze the continuous data of the same measurement tools, and otherwise Standardized Mean Differences (SMD) was used. Calculation of relative risk (RR) for binary classification data, all analyses calculated 95% CI, for studies without methodological and clinical heterogeneity; statistical heterogeneity was determined by the calculation of I² by Cochrane Q test. If the heterogeneity between studies was small (P > 0.1, I² < 50%), fixed model was used. If the heterogeneity is large (P ≤ 0.1, I² ≥ 50%), the random model is used. If the sources of heterogeneity cannot be determined, or heterogeneity is significant, only descriptive analysis is conducted.

By April 2018, a total of 2682 literalness had been retrieved, Repeated publication, literature review, and irrelevant literature (n = 2324) were deleted, with 358 remaining articles, Read the title, abstract, and exclude the descriptive study and the study without control group (n = 256), read the full article to exclude data incompleteness, remaining 5 articles, 2 articles were obtained by manually searching references, and 7 articles were finally included [9] [11] - [16] . There are 5 Chinese [12] [13] [14] [15] [16] articles and 2 English articles [9] [11] . The screening process is shown in Figure 1. In one article [15] , the intervention group was divided into high frequency group and low frequency group, In order to make the data more clear and split into two RCT, the basic features of the included articles are shown in Table 1.

The included RCT was evaluated and graded according to the evaluation criteria of Cochrane evaluation manual 5.1.0. The methodological quality of the included articles is mostly medium, and the specific evaluation indicators and results are shown in Table 2.

Note: IG: Intervention group. CG: Control group. A: Gestational age at the time of complete non-retention. B: Incidence of feeding intolerance. C: Incidence of nosocomial infection. D: The incidence of necrotizing enters colitis. E: LOS (length of stay). G: Secretary immunoglobulin A (slgA). H: lactoferrin.

Note: L, Low risk of bias. U, Unclear risk of bias. H, High risk of bias.

A total of 392 premature infants were included in 7 literatures. 3 [12] [13] [14] reported gestational age at the time of complete non-retention, 5 RCT [12] [14] [15] [16] reports feeding intolerance, 4 articles [9] [11] [14] [16] reported hospital infection rate, 4 articles [9] [11] [15] [16] reported hospitalization duration, 2 articles [14] [16] reported the incidence of necrotic enter colitis, and 2 articles [9] [11] reported lactoferrin and slgA. The basic characteristics of the included articles are shown in Table 1.

Sensitivity analysis was performed for all the combined results. When sensitivity analysis was performed on the complete non-retention gestational age, the largest sample size Chen Caiyun was removed and the results changed, it is suggested that the result is not stable. As fewer than 10 articles were included, the funnel plot analysis was not done to evaluate the publication bias.

A total of 7 [9] [11] - [16] articles were included in this study, among which 5 [9] [11] [13] [14] [15] used random methods, 4 (57.1%) [9] [11] [14] [15] explicitly stated the use of random numbers Table 1 [9] (14.2%) reported the method of allocation concealment, and 1 [12] (14.2%) only mentioned allocation concealment but did not specify the specific method. Seven (100%) results use the blind method. Six [9] [11] [13] [14] [15] [16] (85.7%) reported the loss of data, and 1 [9] (14.2%) reported increased sample loss rate. All 6 [9] [11] [12] [13] [14] [16] studies (85.7%) compared baseline levels of premature infants, and baseline consistency (P > 0.05).

Colostrums oral intervention includes two forms, one is the colostrum oral drip, namely after removing the oropharyngeal secretions of premature infants, 0.2 ml of normal temperature colostrum was extracted using 1ml syringe, along the side of the mouth, pointing to children with throat, uniform slowly push 0.1 ml, then moved to contralateral injection of 0.1 ml. The second is the colostrum smear, that is, after removing the oral secretions, use cotton swab to apply the colostrum restored to room temperature to the buccal mucosa of premature infants. 5 [9] [11] [12] [13] [14] used colostrum oral instillation, and 2 [15] [16] used colostrum oral smear. 4 [9] [12] [13] [14] performed on premature infants every 4 hours. 3 [15] [16] [18] performed every 3 hours, and one article [15] performed every 6 hours. 5 articles [9] [11] [12] [13] [14] indicated that each use of 0.2 ml of colostrum, 4 articles [9] [12] [13] [14] lasted for 7 days and 1 article [16] lasted for two weeks.

Premature infants often suffer from stomach retention, abdominal distension and vomiting due to immature digestive system [19] . Gastric retention, also known as gastric emptying delay, the results showed that the residual gastric fluid extracted from the stomach before feeding was more than 50% of the previous feeding. Complete non-retention is an indicator of complete recovery of gastrointestinal function and the amount of gastric residue in the normal range. The combined results of this study showed that the colostrum oral intervention can reduce the gestational age at the time of complete non-retention enhance the digestive function and promote the process of entreat nutrition. However, the stability of the combined results was poor, which may be attributed to the lack of a clear definition of feeding intolerance in some studies, and the inclusion of abdominal distension, vomiting and other factors into the tolerance range, which had an impact on the results.

Premature infants need to face the external environment alone after birth, but the each system is not mature, and the antibodies in breast milk cannot be obtained through mouth to resist external pathogenic bacteria, so the risk of infection is high. Studies have shown that the mother’s colostrum of premature infants contains more cytokines and enzymes than that of full-term infant [18] [20] [21] [22] . These substances can be directly absorbed in the buccal mucosa to help premature infants resist external bacteria and development of premature infants. Therefore, the oral intervention of the colostomy of premature infants is more important than that of full-term infants. The combined results of the study also showed that the colostrum oral intervention can effectively reduce the incidence of nonsocial infection and promote the normal growth of premature infants [5] [11] [23] .

Lactoferrin in colostrum has strong immunomodulatory, bactericidal, anti-infection and anti-virus effects [24] , slgA and lactoferrin can be distributed on the oral mucosa surface and intervene oral colonization in the mouth, have the effect of local immunity, Lactoferrin and slgA are rarely used as evaluation indicator in domestic studies. But in foreign studies that are used as evaluation indexes, their measurement methods are too heterogeneous to be combined with data, According to specific research data, colostrum oral intervention can promote the increase of lactoferrin, but it has no significant advantage for slgA growth [5] [25] . The main reason is that slgA content is the highest one day after giving birth and gradually decreases after that. However, the primary oral cavity intervention is mostly in the 4 - 5 days after giving birth, which is in the decline stage of slgA. Therefore, the primary oral cavity intervention has no significant impact on slgA of premature infants.

1) Only searched published Chinese and English articles, and did not search gray literatures. 2) Less than 10 studies were included without funnel plot analysis. 3) The domestic research in enteral feeding process for ending index, study abroad in the urine, saliva slgA and lactoferrin as the ending index, there is obvious heterogeneity, descriptive analysis only.