U.S. baby formula crisis: What bacteria science reveals about how we can prevent the next one

For months, the parents of young children in the United States have struggled to find baby formula. The problem started with an outbreak of the bacteria Cronobacter sakazakii and Salmonella in babies-fed formula manufactured at a factory owned by Abbot Laboratories, which controls 48 percent of the US infant formula market. The company recalled some of the formulas manufactured at Abbot Laboratories' Michigan facility and shut the facility down for months for an investigation and corrective actions, sending ripple effects across the country. 

The U.S. Food and Drug Administration said on May 19 that the plant was on track to open within one or two weeks, Reuters reported.

The company said it would begin production of EleCare and other specialty and metabolic formulas, with an initial EleCare product release to consumers starting on or about June 20, in a press release.

How could one of the main facilities that make such an important product become contaminated? IE sat down with microbiologist and food safety expert Abby Snyder, an assistant professor of food science at Cornell University, to find out how these incidents happen and what researchers and manufacturers are doing to prevent them.

This interview has been edited for length and clarity.

IE: How do experts think about food safety regarding products like baby formula?

Abby Snyder: Historically, we thought that low-moisture foods — like infant formula, peanut butter, and a variety of other products that are sometimes in the news — were relatively low-risk because they're dry. They do not have enough water in them to support microbial growth, unlike milk or meat. But while a lot of these low-moisture food products can't support growth, they do support the persistence of pathogens like salmonella for a long period of time.

Due to outbreaks and recalls in the last 10 or 15 years, the perception of low-moisture foods as low risk has changed. Now, relevant food safety hazards are recognized for those products, too. 

IE: We hear news reports about contamination from microbes like salmonella, listeria, and — more recently — cronobacter. Do these different pathogens affect different kinds of food? 

Snyder: You bring up a number of pathogens that I think are relevant. For a lot of high-moisture food products, listeria is the pathogen of concern for cross-contamination into products. In low-moisture foods, salmonella is typically the pathogen of concern, but you have other examples, too. In infant formula, cronobacter is the pathogen of concern. There have been a number of outbreaks in flour like wheat flour with pathogenic E. coli. 

IE: How do producers manage the risk of contamination.

Snyder: In some of these low moisture foods examples, you have what food processors call a kill step. That's some sort of typically thermal heat inactivation step that is very carefully controlled to deliver a certain lethality that will control the pathogens that are in the raw materials.

For peanut butter, it's the roast step. For milk and dairy products, it's the pasteurization step. For something like wheat flour, there's not. That's a raw product that consumers bake. That's all well and good unless you have continued processing or manipulation of that product after the kill step. Any pathogen that comes from the production environment — from surfaces that cross-contaminate into the food — after the kill step is a problem because there's no subsequent kill step to take care of it for a lot of these ready-to-eat foods. That's where it's largely coming from.

IE: What does the kill step look like in baby formula production?

Snyder: There are slight variations in how infant formula is produced, but to generalize, you have fluid milk coming in and going through s a pasteurization step. That's the kill step. Then there's subsequent manipulation, so the milk is concentrated through evaporation and then spray dried. That's how you get milk powder.

To go from that is kind of the bulk ingredient and infant formula, there's a series of dry blending steps. You take dry ingredients (like milk powders) and other functional ingredients (like vitamins), and they get blended together into the finished product. There are some liquid ingredients that can be combined and then pasteurized together with the fluid milk, but some get combined at the dry blending step.

After it's blended, the product gets packaged. So, after the kill step, all of those are subject to environmental cross-contamination. The blending and even just the packaging step can be places where you get environmental cross-contamination. That's the concern with this and other low moisture food products.

IE: How do food processors manage the risk of contamination after the kill step?

Snyder: Food safety, writ large, is extremely complicated, and a lot of these processes are very complex. We have made big advances in, let's say, the last 50 years in terms of food safety and reducing the incidence of outbreaks and recalls, as well as decreasing the size of outbreaks when they do happen.

That said, there are still places where cross-contamination does happen. That's because where food is grown — in the natural environment — there are lots of sources for cross-contamination. And the facilities where food is processed — food manufacturing spaces — are very complex. They're highly industrialized, and they vary a lot from site to site. So, implementing activities like sanitation to reduce cross-contamination is complex.

There are strategies in place. There are principles of hygienic design. The idea is that you're going to build these manufacturing spaces to be easily cleanable. There are principles of environmental monitoring. So, you do you swab surfaces to try and detect pathogens to say, 'it's here, and we need to clean this better to get rid of it.'

There are novel processing and sanitization technologies to kill pathogens on surfaces or in products, respectively. There are still gaps, but those are the places where there's innovation to improve.

It's just such a complex problem, compounded by the problem of scale. We have an enormous food supply — we produce a lot of food over time — and there's a certain probability of contamination. Even if there's a low probability of it occurring, the fact that we are processing so much food all the time means that there's a certain likelihood of outbreaks. We have these interventions to reduce that.

IE: How are academic researchers and actors in the industry improving food safety in these kinds of production facilities?

Snyder: Some things that the industry does relate to ingredient specifications and supplier requirements. They're getting in dry products, and they want to make sure the producer is making it in a way that minimizes the likelihood that the product is contaminated. If you are not applying a kill step, then you rely on the person who's supplying you with the product to have done it correctly. 

Another area is hygienic design. That's having equipment and surfaces that are readily cleanable and that don't promote harborage. So, if you do get contamination, your sanitation program will take care of it. The third part is sanitation — the kind of tools and approaches that you use to reduce contamination on surfaces. Product testing is also used. 

Infant formula is in a unique category of products where the shelf life is long enough that you can test and hold and hold before release. Finished-product testing is not perfect because it's destructive by definition. You're going to take a small sample and identify if there are pathogens in that sample. It could be that you have contamination elsewhere in the bulk of that product that you miss, but it can be a useful additional kind of verification check.

There are other areas of emerging research or innovative technologies that look at like other kill steps like post-pasteurization that are still in research and development phases. Researchers are improving sampling schemes and environmental detection. Understanding better the sources of contamination and how it moves around the environment for different vectors. There's a framework called 'seek and destroy' where you find contamination, and you get rid of it. 

IE: Reports about baby formula storage have pointed to market concentration as a major aggravating factor. Manufacturing is so concentrated that it just took the closure of one plant to spark a crisis. Is there anything inherent in the manufacturing process that makes this kind of concentration more likely?

Snyder: I'm a microbiologist, so the market dynamics is a little bit out of my area of expertise. I'll say that some technical processes require a lot of investment in technology, and infant formula is very regulated from a safety and nutrition perspective. That can make it harder to have a lot of diverse producers.

IE: Given all of the safety measures we've discussed so far, how does an outbreak actually happen? 

Snyder: There are oftentimes several dynamics that you could point to as needing improvement. In a lot of these instances of cross-contamination, you have some harborage point in the food production environment where you have pathogen take up residence. It resists sanitation, doesn't get removed, and then it inoculates the product. Places, where there's poor sanitation, poor hygienic design, or insufficient environmental monitoring, can contribute to that contamination. It's hard in these to have a single smoking gun, so I think there are probably several competing variables.

IE: Is food safety a trade-off for producers, given that outbreaks are going to happen if a system is processing as much food as our system does? Are they aiming for outbreaks to occur below a certain acceptable frequency? 

Snyder: Again, knowing the mindset of the industry is somewhat outside my area of expertise. I'm not sure it's the case that they're unwilling to invest more in these preventative measures. There is not necessarily a silver bullet fix that's readily available. Some of this is still in the development phase. Like I said, there have been a lot of instances of these kinds of problems recently, but that's not necessarily the case in the long term, historically. So, some of these solutions are still kind of in the incipient phase. I'm a little hesitant to give kind of a concrete answer, but I don't think it's a simplistic trade-off.