1. Objectives

To assess the effects of various methods of human milk expression on outcomes including maternal satisfaction, adverse events, quantity and nutritional quality of expressed milk, and cost

2. How studies were identified

The following databases were searched in March 2016:

• Cochrane Pregnancy and Childbirth Group’s Trials Register
• CENTRAL (The Cochrane Library)
• MEDLINE
• EMBASE
• CINAHL

Relevant journals, conference proceedings and reference lists were also searched and the authors directly contacted researchers and organizations in the field

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials or quasi-randomized trials, including crossover trials

3.2 Study participants

Women expressing or pumping milk by any method and for any reason

(No exclusion criteria were placed on the health or gestational age of the child, or whether or not the woman was also feeding a child at the breast)

3.3 Interventions

Instructions or equipment provided for hand expression or mechanical breast pumping, or a specific protocol or support programme for milk expression, compared with another expression method or no intervention

3.4 Primary outcomes

• Maternal satisfaction with method (acceptability, comfort, ease of use, achievement of expressing/pumping)
• Adverse outcomes (milk contamination, injury to breast/other anatomy, reduction or cessation of pumping or expressing due to difficulties)

Secondary outcomes included transfer to feeding at the breast (if expressing preceded breastfeeding), quantity of milk expressed, time taken to express milk, nutrient quality of expressed milk, maternal physiological effects of expressing, and economic outcomes (cost of equipment, effect on hospital length of stay, healthcare usage)

4. Main results

4.1 Included studies

Forty-one randomized controlled trials, enrolling 2293 women, were included in this review

• Twenty-five trials were parallel in design and 16 were crossover trials
• Most trials evaluated breast pumps, with 11 studies additionally evaluating hand expression, and 14 trials comparing different pumps or vacuum patterns
• Twenty trials evaluated a protocol or adjunctive behaviour, such as sequential versus simultaneous breast pumping (5 trials), frequency of expression, education and support (3 trials), audio-visual relaxation (3 trials), timing of initiation among mothers of very low birth weight infants, breast massage prior to pumping (2 trials), therapeutic touch, warming breasts prior to pumping, breast shield sizes, breast cleaning protocol, and combined hand expression and pumping versus pumping alone
• Of the 41 trials meeting inclusion criteria, 22 studies involving 1339 women contributed data to analyses

4.2 Study settings

• Australia (2 trials), Brazil, Canada (2 trials), Ecuador, Egypt, India (2 trials), Israel, Malaysia, Mexico, Turkey, the United Kingdom of Northern Ireland and Great Britain (4 trials), the United States of America (23 trials), and one multinational study was conducted in Kenya and Nigeria
• Twenty-six trials (two-thirds of all participants) were conducted in women with preterm or ill infants in neonatal units; 14 trials were conducted in mothers of healthy term infants living at home; and one trial was conducted in both neonatal and healthy older infants

4.3 Study settings

How the data were analysed
Thirteen comparisons were made: i) any type of pump versus hand expression; ii) any manual pump versus hand expression; iii) any manual pump versus any other manual pump; iv) any battery or small electric pump versus any other battery or small electric pump; v) any large electric pump versus hand expression; vi) any large electrical pump versus manual pump; vii) any large electric pump versus battery or small electric pump; viii) any method with a specified protocol of simultaneous versus sequential pumping; ix) any method with a specified relaxation technique versus no specified relaxation technique; x) any method plus specific instruction or support provided versus any method with no specific instruction provided; xi) any method plus breast massage versus no breast massage; xii) any method plus warming the breast versus not warming the breast; and xiii) any vacuum protocol versus vacuum protocol. Studies were not deemed sufficiently similar to combine in meta-analyses. Sensitivity analyses of trial quality were also planned but not conducted. Should sufficient data have been available, the following subgroup analyses were planned:

• Gestational age
• Time since birth when intervention occurred
• Make and model of the pump
• Trial design

Results
Any type of pump versus hand expression
No difference between pump and hand expression groups was found for the outcomes contaminated milk samples (risk ratio (RR) 1.13, 95% confidence interval (CI) [0.79 to 1.61]) or transfer to feeding at the breast (RR 1.30, 95% CI [0.63 to 2.67]) in one trial involving 28 participants. In a trial of 45 women comparing hand expression to electric pumps and manual pumps, one case of nipple damage was found in each of the pump groups, and none in the hand expression group.

Any manual pump versus hand expression
Bacterial level measured by Dornic degrees of acidity was not different between treatment groups (MD 0.20, 95% CI [-0.18 to 0.58], 1 trial/142 women). Mean volume of milk expressed over six days was greater in the manual pump group (MD 212.10 mL, 95% CI [9.39 to 414. 81], 1 trial/48 women), but was not significantly greater on days four to five. In one trial of 118 women, the potassium concentration was higher (MD 1.20 mmol/L, 95% CI [0.04 to 2.36]), and sodium and protein concentrations were lower (MD -6.0 mmol/L, 95% CI [-9.79 to -2.21] and MD -1.30 g/L, 95% CI [-2.56 to -0.04], respectively) with manual pumping.

Any manual pump versus any other manual pump
In a crossover trial including four different pumps and involving 32 women, the Isis pump received better scores for ease of use, comfort, and overall opinion, while Harmony and Little Heart pumps received equal scores, and the Evenflo pump received the lowest scores (p<0.001). Scores for perceived amount of suction favoured Isis, Harmony, and Little Heart over Evenflo (p=0.007). A non-significant preference for squeeze-handle mechanisms over cylindrical mechanisms was also reported. The quantity of milk expressed in 24 hours was greater with the Isis pump compared to the Evenflo pump (MD 30.49 mL, 95% CI [3.40 to 57.58]), and was also greater with the Harmony pump compared to the Evenflo pump (MD 28.5 mL, 95% CI [12.11 to 44.89]).

Any battery or small electric pump versus any other battery or small electric pump
In one trial of 62 women, no overall preference for either a standard or novel electric pump was reported and no preference between the pumps was observed when women chose which pump to keep. In a trial including 40 women, the small electric UNO pump took four minutes longer to express milk than the small electric Swing pump (MD 4.00 minutes, 95% CI [1.19 to 6.81]).

Any large electric pump versus hand expression
Using the Breastfeeding Self-Efficacy Scale, in one trial of 68 women a large electric pump compared to hand expression increased the likelihood that women would not want anyone to see them hand expressing/pumping (MD 0.70 points in a scale of 1 to 5, 95% CI [0.15 to 1.25]). Using the Breast Milk Expression Experience Scale, women were also less likely to feel satisfied with the electric pump instructions (MD -0.40 points, 95% CI [-0.75 to -0.05]). Adverse events, including breast pain rated on a scale from one to ten (MD 0.02 points, 95% CI [-0.67 to 0.71], 1 trial/68 women), bacteria level (MD 0.10 Dornic degrees of acidity, 95% CI [-0.29 to 0.49], 1 trial/123 women), and number of colony forming units/mL (1 trial/16 women; data not provided) were not different between electric pump and hand expression groups. Mean volume of milk pumped over six days was greater with the large electric pump compared to hand expression (MD 373.1 mL, 95% CI [161.09 to 585. 11], 1 trial/43 women), as well as on days two to five. The sodium concentration of milk was reduced with the large electric pump (MD -6.90 mmol/L, 95% CI [-10.58 to -3.22], 1 trial/111 women).

Any large electric pump versus manual pump
Significantly higher maternal satisfaction was reported using a manual pump compared with a large electric pump in one trial of 145 women (ease of use: p=0.03, comfort: p=0.003, pleasant to use: p=0.01, overall opinion: p=0.003, amount of suction: p=0.05). In the same study, similar percentages of women in each group developed sore nipples (7% both groups) or breast engorgement (4% manual versus 6% electric), and two percent of those using the electric pump developed mastitis. In a further study of 60 women by the same group, the same manual pump was also preferred over a small electrical pump (comfort: 73% versus 20%, pleasant to use: 58% versus 20%, overall opinion: 69% versus 42%, with no differences for ease of use: 63% versus 65%, and amount of suction: 67% versus 71%). A significantly greater percentage of women also selected to keep the manual pump at the study’s conclusion (64% versus 34%, p=0.049). However, in another small study including 9 women, the large electric pump was preferred over the manual pump (data not provided). Bacterial level was not different between electric pump and manual pump groups (MD -0.10 Dornic degrees of acidity, 95% CI [-0.46 to 0.26], 1 trial/141 women). The time spent per day pumping was reduced by more than 20 minutes with a large electric pump compared to a manual pump (MD 20.27 minutes, 95% CI [28.30 to -12. 24], 1 trial/145 women). Protein concentration was greater in milk of women using the electric pump (MD 1.40 g/L, 95% CI [0.08 to 2.72], 1 trial/121 women).

Any large electric pump versus battery or small electric pump
In a trial of 71 women, mothers preferred the smaller less costly pump to the larger electric pump based on ease of use and the position of the control button (data not provided). The volume of milk expressed in one expression was greater in the Whittlestone pump compared to the UNO pump (MD 20.0 mL, 95% CI [1.28 to 38.72]) and the time taken to pump was less (MD -6.0 minutes, 95% CI [-8.81 to -3.19], 1 trial/40 women).

Any method with a specified protocol of simultaneous versus sequential pumping
In one study including 25 women, a preference for simultaneous pumping compared to sequential pumping was found, while in two other studies of 49 and 52 women, no difference between groups was reported (data not provided). Simultaneous breast pumping reduced the time spent pumping per week by three and a half hours (MD -3.50 hours, 95% CI [-5.61 to -1.39], 1 trial/32 women).

Any method with a specified relaxation technique versus no specified relaxation technique
In one trial of 55 women, the quantity of milk expressed was greater with use of an audio relaxation recording (MD 34.70 mL, 95% CI [6.10 to 63.30]), and mothers viewed the relaxation technique positively (data not provided). In a crossover trial examining therapeutic touch versus mimic therapeutic touch, neither treatment was perceived as being superior to the other (p>0.05, data not provided). In the same study, more mothers experienced milk leaking (a proxy for oxytocin release) with therapeutic touch (28%) than mimic therapeutic touch (6%) or no therapeutic touch (0%). In one trial of 160 women, the volume of milk expressed with music listening was greater on all measured days (days one, five, ten, 14), with over 500 mLs more milk expressed on day 14 (MD 503.3 mL, 95% CI [410.8 to 595.8]), and a mean over the 14 days among the controls of 166 mL versus 317 mL in the music group. In the same trial, the fat content of the milk was significantly greater with music on days one, five, and ten (8.6 to 14.0 g/L higher), but was not significantly different on day 14 (MD 21.30 g/L, 95% CI [-2.46 to 45.06]). In a trial of 29 mothers of preterm infants, a higher mean volume of breast milk was reported 7.12 mL (SD 1.57) compared to 6.68 mL (SD 1.37) in the group not receiving the music therapy intervention.

Any method plus specific instruction or support provided versus any method with no specific instruction provided
In a trial of 60 women, those who received an education intervention were more likely to express earlier (p<0.004, data not provided) and were twice as likely to transfer to feeding at the breast (RR 2.00, 95% CI [1.25 to 3.21]). Most women receiving a support intervention in one trial reported it to be beneficial for the establishment and maintenance of their breast milk production (education session: 11/14 women; telephone follow-up: 10/12 women; telephone helpline: 2/4 women). No differences between groups were found for quantity of milk expressed or for lipid content of milk. While time taken expressing per day was not significantly different between groups in weeks one to three, in weeks four, five, and six women spent approximately 30 minutes longer expressing if they had received instruction or support (all p≤0.02, 1 trial/33 women).

Any method plus breast massage versus no breast massage
Quantity of milk from two expressions (MD 4.82 mL, 95% CI [1.25 to 8.39]) and the fat content of milk (MD 1.92% creamatocrit, 95% CI [1.02 to 2.82]) were increased with breast massage in one trial of 72 women.

Any method plus warming the breast versus not warming the breast
Quantity of milk was measured over six expressions in one trial of 78 women. Breast warming increased the amount expressed by approximately 10 mL per expression, and this was statistically significant in four of the six measurements.

Any vacuum protocol versus vacuum protocol
In one crossover trial including 128 women, no differences were found in the quantity of milk expressed with differing pump suction patterns. In this study, it was reported that women did not like the suction strength of the new experimental breast pump suction pattern and that it was not as comfortable as the initial experimental breast pump suction pattern (data not provided).

Other interventions
In a study of 20 very low birth weight infants, initiation of milk pumping within 60 minutes of birth resulted in higher mean milk volumes in the first week, more than doubling the total mean volume for the first seven days (1374.7 mL versus 608.1 mL; p=0.05). Increased frequency of pumping (≥4 times/day) produced greater milk volumes than less frequent pumping (≤3 times/day) (mean 342 mL, SD [229], versus 221 mL, SD [141], p>0.02; 1 trial/25 women). In a trial involving 20 women, more milk was expressed when using a breast shield size >24 mm.

5. Additional author observations*

The methodological quality of the included trials was mixed, with high risks of detection bias and attrition bias due to lack of assessor blinding and incomplete outcome data, respectively. The 41 identified trials were conducted in 14 different countries and spanned more than three decades, and findings were unable to be pooled due to heterogeneity in interventions, comparison groups, and outcomes. Many of the data from crossover trials could not be used due to lack of reporting between-mother differences or pair analyses. In 16 of the 30 trials evaluating pumps or products, support from manufacturers was reported.

Overall, few trials reported on maternal satisfaction and no consistent preference for any method of milk expression was observed. Milk contamination did not differ between hand expression and manual pumping or between hand expression, a manual pump and an electric pump. Nipple or breast pain did not appear to differ between methods, but few studies reported on these outcomes. Insufficient studies were identified to report results for infant morbidity in relation to method of milk expression. Higher volumes of milk were produced when using relaxation techniques, massage of the breast before or with pumping, and warming the breast before pumping. No consistent effect on prolactin change or oxytocin release by method of expression was reported.

Further high-quality research comparing methods of milk expression is needed. Such research should be independently funded, and should consider outcomes such as cost and maternal goals for milk expression.