Effect of Novel Breastfeeding Smartphone Applications on Breastfeeding Rates

Abstract

Background: Low-income women are less likely to breastfeed than high-income women. Technology-based interventions demonstrate promise in decreasing health disparities. We assessed whether increased use of breastfeeding smartphone applications (apps) impacts breastfeeding rates for low-income women.

Materials and Methods: This is a secondary analysis of a randomized control trial (RCT), including nulliparous, low-income women. Women were randomized to one of two novel apps: control app containing digital breastfeeding handouts and BreastFeeding Friend (BFF), an interactive app containing on-demand breastfeeding educational and video content. App usage was securely tracked. The highest quartile of BFF and control app users were combined and compared to the lowest quartile of app users. The primary outcome was breastfeeding initiation. Secondary outcomes included breastfeeding outcomes and resource preferences through 6 months.

Results: In the RCT, BFF and control app median uses were 15 (interquartile range [IQR] 4–24) and 9 (IQR 5–19) (p = 0.1), respectively. Breastfeeding initiation did not differ with app usage (84.1% in highest quartile versus 78.2% for lowest quartile; p = 0.5). Rates of sustained and exclusive breastfeeding through 6 months were similar between groups. Among both groups, smartphone apps were the most preferred breastfeeding resource at 6 weeks. Low quartile users also preferred alternative online breastfeeding resources: >50% of all users preferred technology-based breastfeeding resources.

Conclusions: Increased usage of breastfeeding apps did not improve breastfeeding rates among low-income women. However, technology-based resources were the most preferred breastfeeding resource after hospital discharge, indicating ongoing development of technology-based interventions has potential to increase breastfeeding in this high-needs population. clinicaltrials.gov (NCT03167073).

Introduction

Breastfeeding is beneficial to both maternal and neonatal well-being. Specifically, for mothers, breastfeeding reduces rates of postpartum depression and chronic disease such as type 2 diabetes, hypertension, and cardiovascular disease.^1–3 For infants, breastfeeding has been associated with lower rates of childhood obesity, type 1 diabetes, asthma, and susceptibility to certain infections.^1–5 Given these benefits, the World Health Organization, American College of Obstetricians and Gynecologists, and American Academy of Pediatrics recommend exclusive breastfeeding through 6 months of age for all women without contraindications to breastfeeding.^6–8

Breastfeeding is a unique lifestyle intervention, in that, it is not only available to most women postpartum but also has been shown to increase family and national resources.⁹ Despite these health and economic benefits, low-income and African American mothers have significantly lower rates of breastfeeding compared to Caucasian and upper class women.^10,11 It is known that prenatal lactation interventions are effective in increasing exclusive breastfeeding rates^12,13; however, the majority of interventions studied were labor intensive, requiring in-person appointments. Research on social determinant of health has demonstrated that low-income women face barriers limiting their participation in in-person reproductive health care, including reliance on public transportation and the need to bring additional family members to appointments.¹⁴

Web-page (online) education and smartphone applications (apps), also known as Mobile health (mHealth)-based interventions, have been shown to be both feasible and efficacious for providing on-demand resources to patients, leading to improvement in long-term management of chronic diseases such as diabetes, hypertension, and chronic lung disease.¹⁵ Importantly, in the United States, as many as 96% of Americans 18–29 years of age, including ∼2/3 of women earning less than $30,000 annually own a smartphone regardless of race or ethnicity,¹⁶ making mHealth interventions accessible to most marginalized populations, including African Americans and low-income women.

It is hypothesized that smartphone apps have the potential to improve breastfeeding outcomes^17–20; however, these studies have demonstrated mixed results. Specifically, Wheaton et al. and Farr et al. demonstrated that women who were exposed to an app-based breastfeeding education curriculum had increased rates of breastfeeding compared to women who were not exposed^17,19; however, these studies were limited. Specifically, Farr et al. required participants to complete formal breastfeeding education before study enrollment.¹⁹ In addition, neither was the study controlled nor were the studies designed to confirm or calculate app usage in participants.^17,19

Lewkowitz et al. performed a randomized control trial (RCT) of a novel breastfeeding app entitled BreastFeeding Friend (BFF) compared to a control app containing digital breastfeeding handouts.²⁰ In this RCT, app usage for both the BFF and control apps was monitored using an embedded usage tracker. This RCT did not demonstrate a significant difference in breastfeeding rates with app usage. One possible explanation for the variance between this and prior studies was Lewkowitz et al. encountered overall low rates of app usage.²⁰

In this study, we looked to re-examined data from Lewkowitz et al.'s trial²⁰ to examine whether increased use of breastfeeding smartphone apps would impact breastfeeding rates for low-income women. We hypothesized that breastfeeding initiation and exclusive and sustained breastfeeding rates at 6 months postpartum would be increased in the highest quartiles of breastfeeding app users compared to lowest quartiles of breastfeeding app users.

Materials and Methods

This was a secondary analysis of a previously described RCT at a single academic medical center between July 2017 and December 2018.²⁰ Briefly, eligible participants were English-speaking, nulliparous women with singleton pregnancies to prevent confounding by experiences with prior infant feeding practices.² Women with multiple gestations, major fetal anomalies, lack of desire to initiate breastfeeding, or contraindications to breastfeeding (such as active recreational drug use or diagnosis of Human Immunodeficiency Virus) were excluded from the parent study. Recruitment occurred at ∼36 weeks' gestation during routine obstetric appointments at a prenatal clinic serving women with Medicaid or no health insurance. Recruitment was limited to this single site to maximize enrollment of our target population of low-income women. The project was approved by the Washington University in St. Louis Human Research Protection Office (Institutional Review Board #201704147), and the parent study protocol was registered on clinicaltrials.gov (NCT03167073).

Following consent for enrollment, participants completed an in-person survey of baseline demographic data and infant feeding intent. Participants were randomized by computer-generated randomization sequence²¹ to either the BFF or the control app.

The BFF app was a novel smartphone app designed based on a cross-sectional breastfeeding survey and feedback from focus groups of postpartum, low-income African American women.²² Specifically, a multidisciplinary team of lactation consultants, perinatologists, and neonatologists designed the BFF app. All educational content was at the fifth grade reading level and was tested and approved by focus groups comprising low-income, predominantly Black pregnant and postpartum women. The educational content included interactive advice on overcoming breastfeeding challenges, educational content on breastfeeding benefits, normal infant behavior and maternal postpartum physiology, strategies to optimize breastfeeding and pumping at work or school, hyperlinks to on-demand videos of tips and troubleshooting for successful latching, common breastfeeding positions, and using and cleaning a breast pump, as well as links to breastfeeding nonprofits, in-person and online support groups, Facebook breastfeeding pages, and Instagram handles.

In contrast, the control app contained digital versions of breastfeeding handouts provided at routine third-trimester prenatal care visits. Importantly, both the BFF and control app included built-in usage trackers that recorded the study identification number and phone number each time the app was opened. Participants were unaware of the usage tracker to minimize potential error from the Hawthorn effect, when research participants modify behaviors based on their knowledge they are being observed.²²

All participants were given a complimentary Android smartphone with assigned breastfeeding app preloaded by a telecom provider. To ensure all women had access to the internet, prepaid internet service was provided for 1 month before and 3 months after delivery. At the completion of the study, the participants maintained ownership of the phone and access to the study app for future use. Both participants and researchers were blinded from study intervention. Each patient was individually familiarized with their assigned app by a research assistant who was not involved with collection or analysis of the study data. After enrollment, all women received standard prenatal, intrapartum, and postpartum care, including access to in-hospital lactation support, as prescribed by their primary obstetrical care team.

After delivery, outcome data were collected through self-reported surveys modified from the validated Centers for Disease Control and Prevention's Infant Feeding Practices Study II survey.^20,23 These surveys were completed in person at 2 days and 6 weeks postpartum. At 3 and 6 months, follow-up surveys were completed through telephone. Exclusive breastfeeding was defined as the use of only breast milk for infant nutrition, whereas sustained breastfeeding included infant feeding practices that included both exclusive breastfeeding and use of breast milk with formula and/or solid food supplementation.

In the parent study and a subsequent secondary analysis that assessed for breastfeeding rates in those who intended to exclusively breastfeed, there were no differences in breastfeeding outcome between women randomized to the BFF and control apps, despite women stating the app was their best breastfeeding resource postpartum at 6 weeks postpartum.^20,24 In this secondary analysis, women were stratified into usage quartiles. Specifically, the highest quartile of BFF and control app users was combined into one group. The lowest quartile of each app's users, including participants who did not utilize their assigned study app after randomization, was combined into the comparison group.

Breastfeeding outcomes were then compared between the highest and lowest quartiles of breastfeeding app user to assess if higher use of on-demand access to an electronic breastfeeding app, regardless of content, increased breastfeeding rates. Given that BFF is an interactive platform and the control app provided text alone, we also conducted a sensitivity analysis limiting the study population to those randomized to BFF users to compare outcomes between high-quartile BFF users and low-quartile BFF users.

Baseline characteristics were compared between highest and lowest quartiles of app users using the chi-squared test or Fisher's exact test for categorical variables and the Student's t test or Mann-Whitney U-test for continuous variables. Chi-square or Fisher's exact test was used to assess rates of exclusive and sustained breastfeeding and other dichotomous secondary outcomes as appropriate. Continuous or ordinal outcomes were compared using the Student's t-test, Mann-Whitney U-test, analysis of variance, or Kruskal-Wallis' test as appropriate. Relative risks with 95% confidence intervals (CIs) were calculated. All tests were two sided and the significance level was set at <0.05. Analyses were performed using STATA (special edition 16; StataCorp LP; College Station, TX).

Results

In the parent study, of the 253 first-time mothers who were assessed for eligibility, 170 women were consented and randomized, and 169 women were included in the parent study (1 woman was excluded due to third-trimester stillbirth)²⁰ (Fig. 1). App usage of both apps was compared. Women randomized to BFF were more than twice as likely to not open their study app than women randomized to the control app, although this did not achieve statistical significance (n = 18 [21.4%] versus n = 9 [10.6%], p = 0.05).

FIG. 1.

Study flowchart.

Among women who accessed their study apps, usage was similar for both the BFF and control apps. For BFF, the median number of uses was 15 uses (interquartile range [IQR] 4–24 uses) with the top 10% of app users opening the app at least 50 times. Conversely, the median of app use for the control app was 9 uses (IQR 5–19), with the top 10% of app users opening the app at least 30 times (p = 0.1). Of the 44 highest quartile users (21 BFF users and 23 control app users), the median app usage was 37 (IQR 25–52), and 10% of users opened their app at least 92 times. The lowest quartile of users comprised 41 women (21 BFF and 20 control app users), and median usage of 0 uses (IQR 0–2): the majority of women in this quartile (27 women) did not use their assigned app after randomization, and none opened her app more than four times.

No significant differences were identified regarding demographics or intended infant feeding method between the highest and lowest quartile users (Table 1). Specifically, most participants were Black, and more than half had an annual household income of less than $25,000. Over half of participants in each group planned to exclusively breastfeed and a similar proportion of women in each group planned to exclusively formula feed. In addition, there was no difference in maternal reporting of anticipated age of formula or food supplementation, with the majority planning introduction of supplementation after 3 months of age.

Table 1.

Baseline Sociodemographic Characteristics and Planned Infant Feeding Methods Among Women in Highest Versus Lowest Quartile of Breastfeeding Smartphone App Use

	Highest use (n = 44)	Lowest use (n = 41)	p-Value
Sociodemographic characteristics
Maternal age at due date [mean years (±SD)]	22.7 (5.1)	21.8 (4.3)	0.5
Reported prepregnancy body mass index
<25	20 (45.5)	19 (46.3)	0.6
25.0–29.99	6 (13.6)	9 (22.0)
30.0–34.99	6 (13.6)	4 (9.8)
35.0–39.99	3 (6.8)	2 (4.9)
≥40	3 (6.8)	7 (17.1)
Decline to answer	6 (13.6)	0 (0.0)
Race/Ethnicity [n (%)]
White	3 (6.8)	3 (7.3)	0.3
Black	37 (84.1)	35 (85.4)
Hispanic	3 (6.8)	0 (0.0)
Asian	0 (0.0)	0 (0.0)
Other	1 (2.3)	3 (7.3)
Education [n (%)]
Less than high school	6 (13.6)	7 (17.1)	0.3
High school degree	29 (65.9)	19 (46.3)
Some college (no degree)	5 (11.4)	12 (29.3)
College degree	1 (2.3)	2 (4.9)
Professional or graduate degree	2 (4.6)	1 (2.4)
Declined	1 (2.3)	0 (0.0)
Relationship
Married or living with partner	11 (25.0)	12 (29.3)	0.7
Have partner, not living together	13 (29.6)	10 (24.4)
Single/not significantly involved	19 (43.2)	18 (43.9)
Other	1 (2.3)	0 (0.0)
Decline	0 (0.0)	1 (2.4)
Annual household income [n (%)]
Under $25,000	30 (68.2)	22 (53.7)	0.3
$25,001 to $50,000	4 (9.1)	9 (22.0)
Over $50,001	1 (2.3)	0 (0.0)
Declined	9 (20.5)	10 (24.4)
Reported possession of smartphone [n (%)]	41 (93.2)	39 (95.1)	0.9
Employment
Full-time employment	12 (27.3)	11 (26.8)	0.5
Part-time employment	9 (20.5)	5 (12.2)
Self-employed	2 (4.5)	1 (2.4)
Temporarily unemployed	1 (2.4)	2 (4.9)
Student or disabled	0 (0.0)	2 (4.9)
Other	0 (0.0)	1 (2.4)
Decline to answer	20 (45.5)	19 (46.3)
Planned infant feeding methods
Planned infant nutrition method
Breastfeeding only	24 (54.6)	20 (48.8)	0.2
Breastfeeding and formula feeding	0 (0.0)	3 (7.3)
Formula feeding only	17 (38.6)	17 (41.5)
Unsure	3 (6.8)	1 (2.4)
Best way to feed infant is
Breastfeeding	38 (86.4)	28 (68.3)	0.1
Breastfeeding and formula feeding	1 (2.3)	5 (12.2)
Formula feeding only	0 (0.0)	1 (2.4)
Breastfeeding and formula feeding are equally good	5 (11.4)	7 (17.1)
Expected infant age at introducing formula or other food
<1 Months	3 (6.8)	8 (19.5)	0.2
1–3 Months	4 (9.1)	5 (12.2)
3–6 Months	22 (50.0)	12 (29.3)
>6 Months	15 (34.1)	16 (39.0)

SD, standard deviation.

Participants were surveyed 2 days postpartum regarding their infant feeding practices to date. There was no significant difference in breastfeeding initiation between the highest and lowest quartiles of app users (Table 2). At 2 days, 6 weeks, 3 months, and 6 months postpartum, rates of exclusive or sustained breastfeeding rates were similar between the two groups. When asked about barriers to breastfeeding, there were no differences in reasons mother cited for not initiating breastfeeding (Table 3). However, women in the highest quartile of app users were more likely to report issues with latching at 2 days postpartum compared to women in the lowest quartile of users (1.81 [1.05–3.11], p = 0.02), but, interestingly, were not significantly more likely to report pain with breastfeeding as a reason for formula supplementation at 6 weeks.

Table 2.

Infant Feeding Practices Among Women in the Highest and Lowest Quartiles of Breastfeeding Smartphone Application Usage

	Highest quartile (n = 44a)	Lowest quartile (n = 41a)	Relative risk (95% CI)	p-Value
Primary outcome
Breastfeeding initiation	37 (84.1)	32 (78.2)	1.08 (0.88–1.33)	0.5
Secondary infant nutrition outcomes
2 Days postpartum	n = 43	n = 40
Exclusive breastfeeding	13 (30.2)	15 (36.6)	0.83 (0.45–1.51)	0.5
Sustained breastfeeding	34 (82.9)	33 (76.7)	1.09 (0.93–1.15)	0.5
6 Weeks postpartum	n = 43	n = 41
Exclusive breastfeeding	7 (16.3)	7 (17.1)	0.95 (0.37–2.48)	0.9
Sustained breastfeeding	23 (53.5)	17 (41.5)	1.29 (0.82–2.04)	0.3
3 Months postpartum	n = 42	n = 38
Exclusive breastfeeding	5 (11.9)	7 (18.4)	0.65 (0.22–1.87)	0.4
Sustained breastfeeding	11 (26.2)	13 (34.2)	0.77 (0.39–1.50)	0.4
6 Months postpartum	n = 30	n = 34
Exclusive breastfeeding	3 (10.0)	3 (8.8)	1.13 (0.25–5.20)	0.9
Sustained breastfeeding	6 (20.0)	6 (17.7)	1.13 (0.41 3.14)	0.8

^aUnless otherwise noted.

CI, confidence interval.

Table 3.

Patient-Reported Factors Impacting Infant Nutrition Decision-Making Among Women in the Highest and Lowest Quartiles of Breastfeeding Smartphone Application Usage

	Highest quartile (n = 44a)	Lowest quartile (n = 41a)	Relative risk (95% CI)	p-Value
Postpartum day 2
Reasons to not initiate breastfeedingb	n = 7	n = 9
Baby was sick and could not breastfeed	1 (1, 4)	3 (1, 4)	—	0.6
Concern for lack of milk	1 (1, 4)	2 (1, 3)	—	0.3
Believe formula is better than breast milk	2 (1, 3)	2 (1, 3)	—	0.8
Needed someone else to feed baby	1 (1, 3)	2 (1, 3)	—	0.6
Had to return to work or school	3 (1, 4)	4 (4, 4)	—	0.1
Issues with breastfeeding [n (%)]	n = 36	n = 32
No issues	9 (25.0)	10 (31.3)	0.80 (0.37–1.72)	0.6
Issues with latching	23 (62.2)	11 (34.4)	1.81 (1.05–3.11)	0.02
Not enough milk	10 (27.8)	9 (28.1)	0.99 (0.46–2.12)	1.0
Milk too long to come in	7 (19.4)	7 (21.9)	0.89 (0.35–2.26)	0.8
Postpartum week 6
Reasons for formula supplementation at home [n (%)]	n = 43	n = 41
Told by health care professional to do so	4 (9.3)	4 (9.8)	0.95 (0.26–3.56)	0.9
Believe formula is better than breast milk	2 (4.7)	1 (2.4)	1.91 (0.18–20.23)	0.6
Friends or relatives recommended formula	4 (9.3)	3 (7.3)	1.27 (0.30–5.34)	0.7
Concern baby was hungry	9 (20.9)	7 (17.1)	1.23 (0.50–2.99)	0.7
Had to return to work and had no pump	5 (11.6)	4 (9.8)	1.19 (0.34–4.13)	0.8
Given formula by WIC	5 (11.6)	7 (17.1)	0.68 (0.23–1.98)	0.5
Breastfeeding was too painful	3 (7.0)	6 (14.6)	0.48 (0.13–1.78)	0.3
Postpartum month 3
Age formula was introduced	n = 44	n = 38
Same day as baby was born	14 (33.3)	13 (34.2)	0.97 (0.53–1.80)	0.9
Day after baby was born to 13 days after baby was born	6 (14.3)	6 (15.8)	0.90 (0.32–3.57)	0.9
2 to 6 weeks after baby was born	11 (26.2)	3 (7.9)	3.32 (1.00–11.00)	0.03
More than 6 weeks after baby was born	8 (19.1)	9 (23.7)	0.80 (0.35–1.87)	0.6
Never fed formula	5 (11.9)	7 (18.4)	0.65 (0.22–1.87)	0.4

Bold text indicates statistically significant findings.

^aUnless otherwise noted.

^bAs measured from an importance scale of 1–4 with 1 being “not at all important” and 4 being “very important.” Data represent median (interquartile range) as calculated with Kruskal-Wallis Test, with chi-square as ties when appropriate.

CI, confidence interval; WIC, women, infants, and children.

Although ∼50% of women in both groups reported a perceived issue with their milk supply, including low supply of milk and delay in milk coming in, there was no significantly difference in issues with milk supply (Table 3). While there was no difference in reason for formula supplementation between the two groups, the most common reason for formula supplementation in both groups was maternal concern that the infant was hungry. In addition, women in the highest quartile of users were more likely to initiate formula supplementation between 2 and 6 weeks postpartum than women in the lowest quartile of users (3.32 [1.00–11.00], p = 0.03).

During postpartum hospitalization, there were no differences in women's opinions of the best breastfeeding resource between the two groups, with women in both groups reporting in-person health care professionals as being their best breastfeeding resource immediately postpartum (Table 4). By 6 weeks postpartum, the highest rated resource for both the highest and lowest quartile users was the breastfeeding app, with a significantly higher proportion of women in highest quartile of users reporting the app as their best breastfeeding resource (n = 20 [60.6%] versus n = 11 [32.4%]; relative risk 1.87 [95% CI 1.07–3.27]; p = 0.02). Interestingly, 20% of lowest quartile users reported that their best breastfeeding resource was an online resource that was not the app, while no users in the highest quartile identified a similarly rated online resource (0 [0.0%], p = 0.006).

Table 4.

Patient-Reported Best Breastfeeding and Pumping Resources Among Women in the Highest and Lowest Quartiles of Breastfeeding Smartphone Application Usage

	Postpartum day 2				Postpartum week 6				Postpartum month 6
	Highest quartile (n = 44a)	Lowest quartile (n = 41a)	Relative risk (95% CI)	p-Value	Highest quartile (n = 44a)	Lowest quartile (n = 41a)	Relative risk (95% CI)	p-Value	Highest quartile (n = 44a)	Lowest quartile (n = 41a)	Relative risk (95% CI)	p-Value
Best breastfeeding resourcea [n (%)]	n = 37	n = 31			n = 33	n = 34			n = 20	n = 27
Health care professional	24 (64.9)	22 (71.0)	0.91 (0.66–1.27)	0.6	3 (9.1)	9 (26.5)	0.34 (0.10–1.16)	0.06	3 (15.0)	3. (11.1)	1.35 (0.30–6.00)	0.7
In-person breastfeeding support group	3 (8.1)	3 (9.7)	0.84 (0.18–3.86)	0.8	2 (6.1)	2 (5.9)	1.03 (0.15–6.89)	1.0	1 (5.0)	1 (3.7)	1.35 (0.090–20.30)	0.8
Online resource (not app)	1 (2.7)	2 (6.5)	0.42 (0.040–4.40)	0.5	0 (0.0)	7 (20.6)	—	0.006	0 (0.0)	4 (14.8)	—	0.07
Breastfeeding app	7 (18.9)	2 (6.5)	2.93 (0.66–13.11)	0.13	20 (60.6)	11 (32.4)	1.87 (1.07–3.27)	0.02	9 (45.0)	9 (33.0)	1.35 (0.66–2.78)	0.4
Other	2 (5.4)	2 (6.5)	0.84 (0.13–5.61)	0.9	8 (24.2)	5 (14.7)	1.65 (0.60–4.52)	0.3	7 (35.0)	10 (37.0)	0.95 (0.44–2.05)	0.9
Best pumping resource since hospital discharge [n (%)]b					n = 34	n = 33			n = 23	n = 28
Health care professional					14 (41.2)	14 (42.4)	0.97 (0.55–1.71)	0.9	14 (60.9)	12 (42.9)	1.42 (0.83–2.43)	0.2
In-person breastfeeding support group					2 (5.9)	2 (6.1)	0.97 (0.15–6.49)	1.0	1 (4.4)	3 (10.7)	0.41 (0.045–3.64)	0.4
Online resource (not app)					3 (8.8)	7 (21.2)	0.42 (0.12–1.47)	0.2	0 (0.0)	3 (10.7)	—	0.1
Breastfeeding app					9 (26.5)	5 (15.2)	1.75 (0.65–4.67)	0.3	4 (17.4)	1 (3.6)	4.87 (0.58–40.59)	0.09
Other					6 (17.7)	5 (15.2)	1.16 (0.39–3.45)	0.8	4 (17.4)	9 (32.1)	0.54 (0.19–1.53)	0.2

Bold text signifies statistically significant findings.

^aFor postpartum day 2, asked about best breastfeeding resource in the hospital. For postpartum week 6 and month 6, asked about best breastfeeding resource after hospital discharge.

^bBest pumping resource was assessed at 6 weeks and 6 months postpartum. The proportion of who reported pumping was similar in each group (For BFF, n = 64 [80.0%]; for control app, n = 62 [74.7%]; relative risk 1.07 [95% CI 0.91–1.26]; p = 0.4).

BFF, BreastFeeding Friend; CI, confidence interval.

By 6 months postpartum, the breastfeeding apps remained the preferred breastfeeding resource for nearly half of the women in each group, but there was no longer a significant difference in preference for the app between the two groups. Also, there remained a statistically significant preference in the lowest quartile group for an alternative nonstudy app online resource compared to higher quartile users (4 [14.8%] versus 0 [0.0%], p = 0.007). While there were no significant differences in participant-reported preferences for pumping resources after hospital discharge, at 6 months postpartum, there were nonsignificant trends toward the highest quartile users preferring the app and lowest quartile users preferring an alternative online resource.

Of note, in the sensitivity analyses limiting the analytic population to high- and low-quartile users of the BFF app, there were fewer than 20 respondents for each infant nutrition outcome. Overall, the results were unchanged from the primary analysis; however, data are not presented here, given the small analytic population produced analyses that were significantly underpowered to observe accurate trends, much less statistical significance.

Discussion

This study was a secondary analysis of an RCT that evaluated breastfeeding outcomes and participant opinions of acceptable breastfeeding resources between the highest and lowest quartile users of two novel smartphone apps. Our results demonstrated that there was no difference in breastfeeding rates among high- and low-quartile app users at any time point assessed. Importantly, although, at 6 weeks postpartum, the apps were rated as the most preferred breastfeeding resource in both high- and low-quartile groups. Interestingly, in addition to the smartphone apps, 20% of low-quartile users also preferred the use of an alternative online breastfeeding resource. When combining apps and alternative online resources, >50% of women in both groups favored mHealth resources for breastfeeding support. This suggests that technology-based breastfeeding education is not just an acceptable means of breastfeeding support for low-income women after hospital discharge, but is a preferred method. Additional research with smartphone apps is needed to further examine their effect on breastfeeding support and infant nutrition outcomes.

Our results highlight some important considerations for future breastfeeding mHealth platforms. First, while >60% of women in the highest quartile reported issues with latching, only 7% of high-quartile users indicated that pain with breastfeeding led them to supplement with formula. This may suggest that while the apps did not result in changes in breastfeeding rates, the novel smartphone apps in this study may have been sufficient to enable women to successfully improve latching mechanics such that pain with feeding did not become a major contributor to failure of exclusive breastfeeding.

Future studies could expand the embedded tracker to calculate time spent on each topic to determine what areas of lactation education women are utilizing the most. In addition, end-user feedback could be solicited to determine if the current resources were indeed sufficient to improve infant latch and decreased pain with breastfeeding. Issues with milk supply were another area of high concern for women in both groups, with ∼50% of women in each group reporting a perceived issue with their milk supply such as low supply or delay in milk coming in. Interestingly, for women in both groups, the most common reason for initiating formula supplementation was concern their baby was hungry. This indicates that educational interventions surrounding infant feeding behaviors, such as cluster feeding, and the timeline for lactogenesis have the potential to impact exclusive breastfeeding rates and should be the focus of ongoing research.

Second, our results demonstrate that women preferred in-person breastfeeding support while in the hospital and mHealth resources after discharge. It is known that in-person breastfeeding interventions improve breastfeeding outcomes.^12,13 Unfortunately, transportation is a known barrier for accessing lactation services for low-income women.²⁵ This may be a significant factor to why participants preferred mHealth resources by 6 weeks postpartum. Taken together, development of digital one-on-one personalized breastfeeding support (telelactation) within an educational smartphone app may be an excellent means to minimizing barriers such as lack transportation and childcare and loss of insurance postpartum for low-income mothers, while maximizing the benefits of face-to-face breastfeeding support.

Although research on telelactation services is limited, current challenges to the delivery of lactation services during the COVID-19 pandemic, including limitations on in-person visits, have lead to a renewed interest in telelactation.^26,27 Studies assessing Pacify, a telelactation mobile app utilized at some Women, Infants, and Children nutrition program centers, have shown that telelactation is not only feasible but also acceptable to patients. Importantly, the use of Pacify was associated with increased duration, intensity, and maternal confidence of breastfeeding.^28,29

Telelactation, similar to on-demand smartphone apps, may increase access to lactation services for low-income women by minimizing barriers such as cost, transportation, and time constraints as has been previously shown in rural populations.^30,31 Similar to on-demand smartphone apps, there are limitations of telelactation due to inability of a provider to physically assist patients with issues such as latching and infant positioning. Initial studies and anecdotal experiences by experienced lactation consultants using innovative means, such as puppets for patient latching education, indicate that telelactation can provide sufficient education to improve breastfeeding experiences for individual women.^27,32

mHealth resources can only improve outcomes if they (i) contain useful education content and (ii) are utilized by patients. In this study, women were provided with both a phone and a prepaid internet service to ensure financial concerns did not limit patient access²⁰; however, recent studies have shown that, in addition to access, the second major barrier to adoption mobile health apps is concerns about privacy.³³ Our study did not explore patient's rationale for not using the app, but concern for privacy may explain why a substantial number of participants never utilized their smart phone app and that, while 50% of lowest quartile users preferred mHealth resources, half preferred an alternative online resource, not a smartphone app. It will be important in future studies to elucidate specific barriers to using smartphone apps in low-income women.

This study has notable strengths. The parent study is the first app-based breastfeeding intervention RCT and is the first to track app usage by participants. In contrast to our parent trial and subsequent secondary analyses that did not demonstrate an increase in breastfeeding rates with app usage,^20,24 other studies demonstrate an improvement in breastfeeding rates with technology-based breastfeeding education.^17–19 However, these studies were unable to ensure their app interventions were utilized and played a role in the increased breastfeeding rates. In addition, Farr et al.'s study required patients complete a formal breastfeeding education course before enrollment, further confounding the effects of the technology-based intervention.¹⁹ Further controlled studies are needed to determine if apps can improve breastfeeding outcomes, including differences in rates with and without combined telelactation services.

Our study is not without limitations. First, while participants were unaware of the usage trackers embedded in the app, there remains a risk of the Hawthorne effect as participants knew their responses to surveys were being evaluated. In addition, since data were collected by self-reported surveys at varying time points, our results are subject to recall bias. While our embedded tracker recorded app usage, it did not mark date, time, length of use, or record content accessed. Given observations as discussed above regarding concerns about infant latch, pain with breastfeeding, and milk supply, future studies looking at the specific content and timing of app usage would enable the development of improved mHealth resources.

Since we combined BFF and control app users, it is unclear whether specific educational medias (i.e., video versus text) were the most helpful to women. Although we compared high- and low-quartile BFF users, this analysis was underpowered to observe accurate trends or statistically significant outcomes. Larger future studies using the BFF app need to be performed to assess which educational media prove most useful to women. Finally, like all secondary analyses, our study population may not have been adequately powered to examine differences in outcomes. This is possible, although less likely, given our analyses uncovered multiple significant differences between study groups.

Conclusion

Among low-income first-time mothers, high-frequency use of two novel breastfeeding apps did not increase breastfeeding rates compared to low-frequency use. However, more than half of women in the high usage group reported the app to be their best resource after hospital discharge. Furthermore, among low-frequency users, 50% indicated that an electronic resource provided the best breastfeeding support after hospital discharge, suggesting that low-income women are highly receptive to technology-based education. This indicates that further research into technology-based interventions would benefit from feedback not only into educational content but also regarding barriers to app usage.

Authors' Contributions

L.B.G. was responsible for study design and data interpretation, as well as composing and editing the article. J.D.L. was responsible for study design and data interpretation, as well as editing the article. M.L.R. was responsible for study design and data interpretation, as well as editing the article. G.A.M. was responsible for study design and data interpretation, as well as editing the article. A.G.C. was responsible for study design and data interpretation, as well as editing the article. A.K.L. was responsible for study design, statistical analysis, data abstraction, data interpretation, and editing the article.

Disclaimer

Washington University School of Medicine in St. Louis owns the apps' intellectual property. Neither app is commercially available or patented. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official view of the National Institutes of Health.

Disclosure Statement

No competing financial interests exist.

Funding Information

Both novel smartphone applications created for this research project were funded by a combination of A.K.L.'s personal research discretionary funds and grants. This project was supported by Washington University in St. Louis's Institute of Clinical and Translational Sciences (Grant 3125-17429) and Center for Diabetes Training and Research (Grant 3125-89725A). A.K.L. was supported, in part, by a National Institutes of Health training grant T32-HD-55172-9.