Search Results
Literature selection process was shown in the PRISMA flow diagram[21] (Figure 1), including reasons for exclusion. Initially, a total of 296 publications were retrieved from the following 12 electronic databases: Cochrane Library(n = 7), EMBASE(n = 13), Medline(n = 14), Ovid (n = 122)、PubMed(n = 12), Web of Science (n = 31), CINAHL(n = 22), Scopus(n = 11), SinoMed(n = 13), CNKI(n = 18), WanFang(n = 17) and VIP(n = 16). The search of the reference lists of included studies did not yield any additional studies. After eliminating duplicates, 214 references were included. Then, the remaining 214 studies were screened by titles and abstracts yielding an additional 179 being excluded. The rest 35 articles were screened for full text, of which 22 were excluded as they were not appropriate study designs (n = 18) or interventions (n = 4). Moreover, one article was excluded because the data was difficult to extract and we failed to contact with the author. Finally, 12 articles were fulfilled the eligibility criteria and included in the meta-analysis.
Characteristics of the included studies
A total of 12 unique studies (n=1,936 groups of postpartum women and their newborns, there were 970 groups in the experimental groups and 966 groups in the control groups.) from 2 countries (China and Thailand) were included. There were 11 RCTs and 1 quasi-randomized controlled trial. The included studies were reported in English (one trial) and Chinese (eleven trials). No significant difference was observed between the two groups in terms of the studied variables. All the included studies were published from 2017 up to 2019. The intervention time of the studies ranged from 0~3 days to 0~8 weeks. The outcomes included nipple trauma (n = 7), nipple pain (n =8), the correct position of latching (n = 3) and position comfort (n = 4). We set three subgroups according to intervention, deliver mode and maternal category. Studies were divided into two groups according to whether or not they took the skin-to-skin care (SSC) [37] , in another words, the mother and baby had direct skin contact. The group of laid-back breastfeeding was represented by LBBF and LBBF+SSC was used to represent the group of taking skin-to-skin care on the basis of laid-back breastfeeding. In the maternal delivery mode subgroup, according to the category of the included subjects, studies were divided into "vaginal delivery " group and "vaginal delivery & caesarean " group. In the subgroup of the maternal category, studies were divided into "primipara" group and "primipara & multipara" group according to whether the delivery times is restricted. Further details about the included studies were shown in Table 1, Table 4 and Table 5.
Unfortunately, all included studies did not report the age of the infants. We only found that all the researches began with after the babies born. Thus, the baby's age is estimated to be 0 day to 8 weeks based on the periods of the intervention.
Table 1: Characteristics of the studies included in the meta-analysis.
Author
(year)
|
Sample size(n) (E/C)
|
Study design
|
Vaginal delivery/ Caesarean (n)
|
Primipara /Multipara (n)
|
Gestational weeks (Mean±SD) (E/C)
|
Age of mothers(years) (Mean±SD) (E/C)
|
Interventions (E/C)
|
Intervention periods
|
Outcomes
|
E
|
C
|
E
|
C
|
Zhuang
(2019)[38]
|
75/75
|
RCT
|
38/37
|
35/40
|
—
|
—
|
—
|
26.2±4.7/
25.3±4.5
|
Laid-back Breastfeeding (BN) /Traditional breastfeeding position
|
1~8w
|
Nipple pain
|
Shi et al
.(2017)[39]
|
84/84
|
RCT
|
50/34
|
52/32
|
56/28
|
55/29
|
36~39/
36~40
|
27.98±4.25/
28.51±4.69
|
Laid-back Breastfeeding (BN) /Traditional breastfeeding position
|
1~3d
|
Nipple pain, Nipple trauma
|
Li et al.
(2017)[40]
|
100/100
|
RCT
|
64/36
|
68/32
|
60/40
|
64/36
|
38~40/
37~40
|
27.86±4.25/
27.43±4.17
|
Laid-back Breastfeeding (BN) /Traditional breastfeeding position
|
1~3d
|
Nipple pain, Nipple trauma, Position Comfort, The correct posture of latching
|
Yu et al.
(2019)[41]
|
100/100
|
RCT
|
0/100
|
0/100
|
100/0
|
100/0
|
39.06±1.08/ 39.33±1.06
|
27.43±4.14/
27.40±3.81
|
Laid-back Breastfeeding (BN) /Cradle breastfeeding (Traditonal position)
|
1~3d
|
Nipple trauma, Position Comfort
|
Zeng et al.
(2019)[42]
|
60/60
|
RCT
|
42/18
|
44/16
|
—
|
—
|
39.35±1.23/ 39.42±1.13
|
26.75±4.23/
27.28±4.42
|
Laid-back Breastfeeding (BN) /Cradle breastfeeding (Traditonal position)
|
1d~4w
|
Nipple pain
|
Puapornpong et al. (2017)[43]
|
76/76
|
RCT
|
0/76
|
0/76
|
—
|
—
|
38.5±0.9/ 38.6±1.0
|
27.5±5.9/
27.1±6.1
|
Laid-back Breastfeeding (BN) /Side-Lying Breastfeeding (Traditonal position)
|
1d~6w
|
Position Comfort
|
Liu et al.
(2019)[44]
|
49/49
|
RCT
|
—
|
—
|
—
|
—
|
—
|
—
|
Laid-back Breastfeeding (BN) /Traditional breastfeeding position
|
1~4w
|
Nipple pain, Nipple trauma
|
Zhang
(2019)[45]
|
74/70
|
RCT
|
58/16
|
56/14
|
74/0
|
70/0
|
38.89±0.96/ 39.09±0.97
|
25.82±3.43/
26.16±3.62
|
Laid-back Breastfeeding (BN+SSC) /Cradle breastfeeding (Traditonal position)
|
1~42d
|
Nipple pain, The correct posture of latching
|
Zhao
(2019)[46]
|
48/48
|
RCT
|
48/0
|
48/0
|
—
|
—
|
39.57±1.06/ 39.40±1.11
|
26.31±2.82/
27.27±3.83
|
Laid-back Breastfeeding (BN+SSC) /Traditional breastfeeding position
|
1~3d
|
Nipple pain
|
Liang et al.
(2017)[47]
|
200/200
|
RCT
|
200/0
|
200/0
|
—
|
—
|
—
|
—
|
Laid-back Breastfeeding (BN+SSC) /Traditional breastfeeding position
|
1~3d (4h/d)
|
Nipple trauma
|
Wang
(2019)[48]
|
50/50
|
RCT
|
27/23
|
30/20
|
34/16
|
32/18
|
—
|
27.9±2.3/
27.7±2.1
|
Laid-back Breastfeeding (BN+SSC) /Standard care (Traditonal position)
|
1m
|
Nipple pain, Nipple trauma
|
Lu et al.
(2019)[49]
|
54/54
|
Q-RCT
|
—
|
—
|
26/28
|
25/29
|
38.95±0.45/ 39.50±0.50
|
26.88±3.32/
26.76±3.25
|
Laid-back Breastfeeding (BN+SSC) /Standard care (Traditonal position)
|
1w
|
Nipple trauma, Position Comfort, The correct posture of latching
|
E: experimental group
C: control group
h: hour; d:day; w: week; m: month
SD: standard deviation
Q-RCT: quasi-randomized clinical trial
Risk of Bias Assessment
The quality of all included RCT studies was shown in Table 2 and Fig. 2, further details (all the domains) about the risk assessments were shown in Supplementary file 3. For the judgement of "bias arising from the randomization process", five[39-41, 44, 48] of the 11 included RCT didn’t show the detailed information on random components in the sequence generation process, eight studies[38-42, 44, 46, 48] were judged to raise some concerns, while three studies[43, 45, 47] were judged to be at low risk of bias. They concealed the allocation sequence by envelopes which were sequentially numbered, sealed with a tamper-proof seal and opaque. With regard to the judgement of "Bias due to deviations from intended interventions", only two studies[43, 45] reported the compliance to the intervention, and they were judged to be at low risk of bias. All included studies were judged to be at low risk of bias in the judgement of "bias due to missing outcome data". For the judgement of "Bias in measurement of the outcome", three studies[39, 47, 48] judged to raise some concerns and others[38, 40-46] were judged to be at low risk of bias. For the judgement of "bias in selection of the reported result", there was one study[47] judged to raise some concerns because the trial didn’t analyze in accordance with a pre-specified plan, ten studies[38-46, 48] were judged to be at low risk of bias. Overall, three RCTs (27 %) are at high RoB, six (55 %) showed some concerns, and two (18 %) are at low RoB for their outcomes. The quality of the only one quasi-randomized controlled trial included was presented in Table 3.
Table 2: Results of critical appraisal for Randomized Controlled Trials (Cochrane, 2019)
Study (year)
|
Bias arising from the randomization process
|
Bias due to deviations from intended interventions
|
Bias due to missing outcome data
|
Bias in measurement of the outcome
|
Bias in selection of the reported result
|
Overall risk-of-bias
|
Zhuang (2019)[38]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Shi et al. (2017)[39]
|
Some concerns
|
Some concerns
|
Low risk
|
Some concerns
|
Low risk
|
High risk
|
Li et al. (2017)[40]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Yu et al. (2019)[41]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Zeng et al. (2019)[42]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Puapornpong et al. (2017)[43]
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Liu et al. (2019)[44]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Zhang (2019)[45]
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Low risk
|
Zhao (2019)[46]
|
Some concerns
|
Some concerns
|
Low risk
|
Low risk
|
Low risk
|
Some concerns
|
Liang et al. (2017)[47]
|
Low risk
|
Some concerns
|
Low risk
|
Some concerns
|
Some concerns
|
High risk
|
Wang (2019)[48]
|
Some concerns
|
Some concerns
|
Low risk
|
Some concerns
|
Low risk
|
High risk
|
Table 3: Results of critical appraisal for quasi-experimental studies (JBI, 2016)
Questions (potential bias)
|
Lu et al. (2019)[49]
|
1. Is it clear in the study what is the ‘cause’ and what is the ‘effect’? (causation/reverse causation)
|
Yes
|
2. Were the participants included in any comparisons similar? (selection bias)
|
Yes
|
3. Were the participants included in any comparisons receiving similar treatment/care, other than the intervention of interest? (history threat/ systematic difference/contamination bias)
|
Yes
|
4. Was there a control group? (measurement bias)
|
Yes
|
5. Was there multiple measurements of the outcome both pre and post the intervention? (maturation threat, regression to the mean)
|
Yes
|
6. Was follow-up complete, and if not, was follow-up adequately reported and strategies to deal with loss to follow-up employed? (attrition bias)
|
Yes
|
7. Were the outcomes of participants included in any comparisons measured in the same way? (instrumentation / testing effects threats)
|
Yes
|
8. Were outcomes measured in a reliable way? (detection / instrument / measurement bias)
|
Yes
|
9. Was appropriate statistical analysis used? (performance / detection bias)
|
Yes
|
Synthesis of results
Nipple pain
Eight studies[38-40, 42, 44-46, 48], with a total of 1,076 groups of postpartum women and their newborns, compared traditional breastfeeding positions vs. laid-back breastfeeding in nipple pain. There was substantial heterogeneity among these studies (χ2 = 125.27, p < 0.00001, I2 = 94%) by using the heterogeneity test (Fig. 4). It seems that one outlier[38] has a great influence on the results of the overall meta-analysis by using the sensitivity analyses, which is a major source of heterogeneity. After removing it, the result of the heterogeneity test with the remaining 7 studies decreased the I2 to 73% (χ2 = 22.46, p = 0.001, I2 = 73%). Therefore, a random-effects model was utilized for meta-analysis, the result showed that the experimental group had a lower incidence of nipple pain than the control group (RR = 0.24; 95% CI 0.14,0.40; p < 0.00001). It is worth mentioning that the result of the heterogeneity test decreased the I2 to 0% (χ2 = 2.02, p = 0.85, I2 = 0%) after removing the two studies[38, 45], probably because of the different measurement tools of included studies.
Nipple trauma
Seven studies[39-41, 44, 47-49], which included 1,274 groups of postpartum women and their newborns, reported nipple trauma. The result what these trials of the heterogeneity test showed that there was substantial heterogeneity among these studies (χ2 = 40.95,p < 0.00001; I2 = 85%) (Fig. 3), probably because of the different interventions. However, the sensitivity analyses did not show any study to be substantially influencing the heterogeneity. Thus, a random-effects model was utilized for meta-analysis, the result showed that the experimental group had a lower incidence of nipple trauma than the control group (RR = 0.47; 95% CI 0.29,0.75; p = 0.002).
The correct position of latching
There are three studies[40, 45, 49], which included 452 couples of postpartum women and their newborns, that reported the correct position of latching. The incidence of the correct position of latching did not significantly differ between the two groups (χ2 = 0.09, p = 0.95, I2 = 0%), so we used a fixed-effect model to pool the summary outcome, the results showed that the experimental group had a higher incidence of correct position of latching than the control group (RR = 1.22; 95% CI 1.11,1.33; p < 0.00001) (Fig. 5).
Position comfort
Four studies[40, 41, 43, 49] reported the data of position comfort, and these studies including 660 couples of postpartum women and their newborns. There were two dichotomous data[40, 49] and two continuous data[41, 43] among them. The results of the heterogeneity test showed that there was substantial heterogeneity among these studies (χ2 = 25.58, p < 0.0001, I2 = 88%) using the heterogeneity test, probably because of the different deliver modes. Nevertheless, the sensitivity analyses did not show any study to be substantially influencing the heterogeneity. Therefore, a random-effects model was utilized for meta-analysis, the result showed that there was no statistical significance in the two groups (ES = 0.09; 95% CI -0.63,0.81; p = 0.000) (Fig. 6).
Subgroup Analysis
Three subgroups analysis were undertaken according to the intervention, deliver mode and maternal category. We evaluated the reliability of outcomes and the results are presented in Table 4 and Table 5. For the outcome of nipple trauma, the results of subgroup analysis are, on the whole, the same as the overall results, and the direction had no change. There was no evidence of a different effect related to the intervention (p for interaction = 0.24), deliver mode (p for interaction = 0.37), and Maternal category (p for interaction = 0.37). The subgroup analysis of different interventions showed that the experimental group had a lower incidence (RR = 0.68; 95% CI 0.47,0.99) of nipple trauma than the control group according to the group of Laid-back Breastfeeding (LBBF), but there was no statistical significance (RR = 0.24; 95% CI 0.04,1.31) in the group of taking skin-to-skin care on the basis of Laid-back Breastfeeding (LBBF+SSC). The subgroup analysis of different deliver modes and Maternal categories showed that the group of vaginal delivery & caesarean (RR = 0.50, 95% CI 0.30,0.82) and primipara & multipara (RR = 0.50, 95% CI 0.30,0.82) both had a slightly higher incidence of nipple trauma than the group of caesarean (RR 0.32; 95% CI 0.13,0.76) and primipara (RR = 0.32; 95% CI 0.13,0.76). Similarly, the results of subgroup analyses of nipple pain indicate that there was no evidence of a different effect related to the intervention (p for interaction = 0.51), deliver mode (p for interaction = 0.97), and Maternal category (p for interaction = 0.14). Overall, the results of subgroup analysis showed that all the experimental group had a lower incidence of nipple trauma than the control group, and there was no change about the results of study.
Table 4: Subgroup Analysis of the Effect of Intervention Elements on Nipple trauma.
Subgroups
|
Number of studies
|
Participants(n)
|
The test for heterogeneity
|
Effects model
|
Meta-analysis
|
Interaction
p-value
|
(Experiments /Controls)
|
p-value
|
I2
(%)
|
RR(95%CI)
|
Intervention
|
|
|
|
|
|
|
0.24
|
LBBF
|
4[39-41, 44]
|
333/333
|
0.023
|
68.5%
|
Random
|
0.68 (0.47,0.99)
|
|
LBBF +SSC
|
3[47-49]
|
304/304
|
0.000
|
90.2%
|
Random
|
0.24 (0.04,1.31)
|
|
Deliver mode
|
|
|
|
|
|
|
0.37
|
Caesarean
|
1[41]
|
100/100
|
—
|
—
|
Random
|
0.32 (0.13,0.76)
|
|
Vaginal delivery & Caesarean
|
6[39, 40, 44, 47-49]
|
537/537
|
0.000
|
85.8%
|
Random
|
0.50 (0.30,0.82)
|
|
Maternal category
|
|
|
|
|
|
|
0.37
|
Primipara
|
1[41]
|
100/100
|
—
|
—
|
Random
|
0.32 (0.13,0.76)
|
|
Primipara & Multipara
|
6[39, 40, 44, 47-49]
|
537/537
|
0.000
|
85.8%
|
Random
|
0.50 (0.30,0.82)
|
|
Table 5: Subgroup Analysis of the Effect of Intervention Elements on Nipple pain.
Subgroups
|
Number of studies
|
Participants(n)
|
The test for heterogeneity
|
Effects model
|
Meta-analysis
|
Interaction
p-value
|
(Experiments /Controls)
|
p-value
|
I2
(%)
|
RR(95%CI)
|
Intervention
|
|
|
|
|
|
|
0.51
|
LBBF
|
5[38-40, 42, 44]
|
368/368
|
0.000
|
96.9%
|
Random
|
0.23 (0.06,0.94)
|
|
LBBF +SSC
|
3[45, 46, 48]
|
172/168
|
0.096
|
57.3%
|
Random
|
0.39 (0.21,0.71)
|
|
Deliver mode
|
|
|
|
|
|
|
0.97
|
Vaginal delivery
|
1[46]
|
48/48
|
—
|
—
|
Random
|
0.27 (0.15,0.51)
|
|
Vaginal delivery & Caesarean
|
7[38-40, 42, 44, 45, 48]
|
492/488
|
0.000
|
94.9%
|
Random
|
0.28 (0.12,0.66)
|
|
Maternal category
|
|
|
|
|
|
|
0.14
|
Primipara
|
1[45]
|
74/70
|
—
|
—
|
Random
|
0.57 (0.39,0.84)
|
|
primipara & multipara
|
7[38-40, 42, 44, 46, 48]
|
466/466
|
0.000
|
95.6%
|
Random
|
0.24 (0.08,0.70)
|
|
Publication bias
We used Egger linear regression tests to detect the publication bias arising from various influencing factors quantitatively for there is a concern remains that visual interpretation of funnel plots is inherently subjective. The p-values of the correct position of latching (p = 0.152) and position comfort (p = 0.138) were greater than 0.05, which indicates that there is no significant publication bias. Although the p-values of nipple pain (p = 0.008) and nipple trauma (p = 0.013) were less than 0.05, there is no missing ‘counterparts’ to filled after analyzed by trim and fill method. It indicates that the results of the two outcomes are stable and the effect of publication biases is negligible. To sum up, the results showed no significant risk of publication bias among the studies that were included.