How Data Analytics Can Help Breast Milk Banks Be More Efficient

June 06, 2022 - In the wake of the recent baby formula shortage, parents and providers are becoming increasingly interested in other methods for providing adequate nutrition to infants who may need it. Breast milk donation banks have long played a role in helping to feed one of our most vulnerable populations. Now more than ever, they are looking for ways to make their services more efficient.

Many healthcare organizations have turned to data analytics to improve patient care, and breast milk banks are no exception. Montana Wagner-Gillespie, manager at Mothers' Milk Bank at WakeMed Health & Hospitals, and Natalia Summerville, PhD, data scientist at SAS and lecturer at MIT and Duke, shared with HealthITAnalytics how data analytics has significant potential to help milk banks improve their services.

THE IMPORTANCE OF DONOR MILK AND MILK BANKS

Outside of a formula shortage, the purpose of milk banks is primarily to serve vulnerable babies, particularly babies in the neonatal intensive care unit. Like blood banks, milk banks screen donors, accept donations, process and test the donations to ensure they’re safe, and dispense donations to hospitals.

Breast milk is critical for NICU babies because they have immature immune systems and cannot fight off illnesses as well as healthy babies can. Breast milk contains bioactive proteins that can help prevent and treat certain illnesses, which can be lifesaving for these infants, according to Wagner-Gillespie. She further noted that an exclusive human milk diet is linked to healthier babies and a shorter length of NICU stay.

Some milk banks may have more milk than is needed by their local hospitals, in which case they can dispense milk to outpatients as well. WakeMed’s milk bank has partnered with two local pharmacies to distribute its excess milk to outpatients.

DATA COLLECTION AND INSIGHT GENERATION

Data collection varies somewhat among milk banks, but the Human Milk Banking Association of North America, the accrediting body for milk banks, requires information such as donor health history, donor demographics, and to whom milk is dispensed to be collected.  

At WakeMed’s milk bank, a software program also tracks information about how many batches have been processed in a certain time period, the characteristics of those batches, what donors were in that batch, volumes of dispensation over time, and waste. Waste is a critical metric, as it generates insights about what causes milk to be wasted and what can be done to minimize waste in the future.

HOW DATA ANALYTICS ENHANCES MILK BANK EFFICIENCY

Summerville and her students are currently researching bacteria in breast milk, which is the primary reason for milk waste. They are utilizing over 2,400 records from WakeMed’s milk bank to determine if there are any donor patterns or characteristics that may make them more probable to test positive for bacillus, a type of bacteria that often leads to milk being thrown away. With this data, the milk bank could potentially test milk from high-risk donors and remove any bacillus-positive milk before it is pooled with other donations for processing and testing.

Analyzing where the milk is coming from also provides a wealth of data that milk banks can use.

WakeMed’s milk bank utilizes remote drop-off sites for milk donation, which makes donating significantly more convenient for interested donors and can potentially increase the number of donations received by the bank.

Data gathered about these drop-off sites not only generates insights about the amount of milk donated but can also provide information about how successful these sites are and what quality improvement initiatives might help make them more successful going forward.

Analytics around milk dispensation are used to help inform decision-making in times of shortage or reduced donor milk supply. WakeMed’s milk bank operates on a priority dispensing model in which NICU babies are prioritized. Using data analytics, the bank can compare milk supply between the NICU and outpatient usage to decide when to dial back on the number of outpatients, if any, who can be served.

CHALLENGES TO USING DATA ANALYTICS IN MILK BANKS

A major challenge to the application of data analytics in breast milk banks, like in many other healthcare settings, is the amount of unstructured data involved and the migration to EMRs.

“I think we're probably the last department in the hospital to move from paper charts to Epic. Epic is wonderful and it does a lot of cool things, and you can generate all kinds of neat reports. I know that they pool all kinds of meaningful data for the hospital system through Epic but getting something built specifically for milking purposes is really difficult because it's an outpatient Epic setup… the questionnaires that we're required to do are different than any other health questionnaire that will be done by a patient,” Wagner-Gillespie said.

As part of the required donor screening process, WakeMed’s milk bank uses an initial online questionnaire and a follow-up questionnaire. As part of these questionnaires, donors must consent to information sharing between their OB/GYN and their baby’s pediatrician with the milk bank, which allows the bank to gather data about the baby’s health and growth while consuming the milk.

The data is collected alongside donor health history, which includes factors that would prevent someone from donating, like a positive drug screening during pregnancy. Blood work and other test data are also included.

Integrating all of these data points in a meaningful way remains challenging.

“Trying to get all of those things to talk to each other in Epic took a really long time. It took us a year to try to develop this questionnaire and get it up and running. We're in the process of converting all of our donors over to Epic,” Wagner-Gillespie stated. “At any given time, we have about 350 active donors. So, I have 350 people who have already gone through the process with paper charts. And then I have probably another 150 that are in process at any time, potentially before we transition to Epic or after we transition to Epic. By this time next year, I think it'll be a lot cleaner, but trying to deal with paper charts and then also Epic has been a challenge.”

This challenge is also an issue for Summerville and her students trying to optimize bacillus prediction for the milk bank. Since WakeMed’s milk bank is transitioning to the EMR, some of the records used for Summerville’s research are digital, but some are still on paper.

“Right now, I'm working on getting volunteers that will help us to pass these 2,400 PDFs into structured data, such that then we can do analysis on it. And I'm looking both ways. I'm checking with some coworker experts on, is there a computer vision technique that can help us to pass it through automatically? As well as getting high school student volunteers to do it manually,” Summerville said.

Despite these challenges, the transition from paper records to digital is largely helping drive research like Summerville’s, and other milk banks are also becoming interested in sharing their data. In doing so, Summerville’s prediction models can become more robust and generalizable, making them useful to banks other than WakeMed’s.

These partnerships between milk banks and analytics professionals are key to improving breast milk banking.

“It's easy to be passionate about what my role is here: feeding sick babies. That's easy. But I've been in this role for seven and a half years, and the passion does wane after a while," Wagner-Gillespie said. "It's a job at [the] end of the day. But to me, having students and having access to students in this way and being able to partner with [Summerville] has really kept my wheels turning and thinking about strategic planning and opportunities for improvement, the kind of consistent quality improvement initiatives that we can use this data for. I think it's really meaningful.”