The End Of An Error

A three-part series that looks at how understanding the root causes of mistakes involving infant feeding management can help prevent them in the NICU. Originally published in Neonatal Intensive Care, The Journal of Perinatology-Neonatology: Part 1 (page 38), Part 2 (46), and Part 3 (40).

Part 1: Introduction

We all know that the NICU is an intense and complex clinical environment where you are dealing with the most precious of patients. This complexity also presents many opportunities for error, specifically involving infant feeding management.

At this time, there are no universally accepted national standards to regulate safety management for preparation and administration of infant feedings in hospitals. As an added complication, literature available in the feeding management space is limited when compared to other topics in the healthcare industry. Furthermore, when you review what is available much of the citations speak only to the dangers of these errors; infection risks, HIPAA violations, Joint Commission sentinel events, etc. While that information is incredibly important and well documented, I would prefer to look at the root causes of these errors.

To put this in perspective, the few published studies that have completed an FMEA for infant feeding management found that there are between 32-282 potential failure modes in feeding preparation and administration that require a human to detect and prevent (see Steele and Oza-Frank).

A recent large scale Six Sigma study put the overall risk for breast milk error to be 1 in 10,000 feeds, which is the monthly feed volume for most average-sized NICUs (see Luton). Articles focused on these topics regularly reference this study even though that rate is actually a predictive model based on blood transfusion administration error rates.

This was the best information we had, but also let’s think about how controlled the blood administration process is. Unfortunately, this is certainly not the case with infant feeding management.

Specific to our topic, when identified, the most common types of errors are feeding an infant the wrong milk, feeding an infant expired milk, and errors in milk preparation. All with potentially significant consequences.

The topic of errors is multifactorial and this three-part article series will cover the many reasons we have errors and what can be done to mitigate them. Here in part 1, we’ll cover cognitive behaviors contributing to errors, human factors, and lastly the current state of electronic management and opportunities to improve safety in these processes.

Cognitive Demands

To work through the many pieces of this puzzle, we need to look at all the cognitive reasons errors occur. We can look at this from five angles: distractions, prospective memory, cognitive fatigue, inattentional blindness, and mental validations.

Distractions in healthcare delivery and their implications for patient safety are well established. One well-documented example of these phenomena is errors during medication preparation and administration.

In 2012 the Institute of Safe Medication Practices (ISMP) published an article speaking to distractions in healthcare related to medication preparation and administration.

Nurses, pharmacists, and technicians are distracted and interrupted as often as once every two minutes (see ISMP, Relihan, and Sliver) Medication error risk increases by 12.7% with each interruption (Westbrook).

As a general rule, I hesitate to apply or extrapolate data for topics on which the studies were not originally conducted. However, to date, I have not found any publications addressing interruptions during infant feeding preparation and administration. Experience has shown me how frequent and similar disruptions during these two points of patient care can be.

When interruptions occur our prospective memory, or the ability to remember to do something that must be deferred, is impaired (see ISMP, and Relihan). When an individual makes a plan to complete a task, a signal is set to remind them to actually complete that task. In the case of a distraction, the individual has pulled away from the task they are set to perform. If that signal is encountered in the future, that reminder is supposed to be triggered (see ISMP, and Grundgeiger). What if that signal does not happen? What if that reminder is not triggered?

Example: A RN is in the middle of mixing a feed. He or she then has to leave the feeding prep process and area (regardless of where that is) to attend to a crisis alarm. What happens next? What is the cue to remember to go back to what they were doing (or delegate it out)? Do they see the clock and realize the feed is late? Does someone ask them if they need their patient fed? Do they see another staff member feeding a baby and remember? Do they remember where in the process they were? Do they remember at all?

In the event an individual does remember to go back to the initial task, they risk omitting or duplicating steps. In certain situations, the entire workflow may need to be repeated which can be extremely problematic depending on what they were doing. Adding insult to injury, in an attempt to complete the new task the individual has an increased likelihood of committing an error with either of the tasks because “the stress of the distraction or interruption causes cognitive fatigue,
which leads to omissions, lapses, and mistakes.
” (see ISMP)

Inattentional blindness, a phenomenon in which the brain fails to see an object or detail when attention or focus is not completely centered on the object. Also known as perceptual blindness, the brain will interpret what it expects to see or conversely, looks at something and does not really see it. This is something we probably all have done. We expect to see a certain patient name or a medication and see what is not actually there.

Mental Validation

Now let us build a little more on these phenomena. On top of the interruptions and associated cognitive failures, you have to take into account all of the mental validations needed to manage the processes for infant feedings.

Pulling from an FMEA at a large urban hospital they found as many as 15-20 mental validations are required to prepare feedings for one infant. Validations such as; patient verification, order verification, complex recipe management, combing for volume and fortification, expiration updates, parsing out feedings, planning for real-time and future feeding times, thawing milk, and freezing milk.

NICU nurses generally have 2-4 infants in their care, eating 3-4 times in a 12 hr shift. A nurse/tech could realistically prepare 12 feeds in one day. This would be >180 mental validations in one shift for one clinician specifically related to feeding preparation.

These tasks can overlap multiple times throughout a typical shift, leaving every validation point open to potential error. When you consider the need for validation amidst the likely impact of distractions and interruptions the opportunity for error is high.


The science behind infant-specific recipe management continues to gain momentum and we learn more about tailoring these feeding plans to the individual infant.

As we have started to touch on, infant feeding management is extremely complex and involves multiple processes and players. With the added complexity of more intricate recipes and feeding regimens, we have to be ready to handle the workload safely.

Amongst all the distractions, we are relying on humans to not only detect but prevent dozens of potential failure points. Considering what we have discussed regarding cognitive failures and mental validations how can we expect to be successful?

What impacts errors and our ability to detect and prevents them continues to be a complex conversation. Many more factors come into play and need consideration in addition to what we have reviewed here.

We have only scratched the surface of these issues. Part 2 of The End of an Error will focus on what human factors influence these processes and contribute to or prevent errors.

Part 2: Introduction

In Part 1 of this article series, we covered some cognitive behaviors that allow for errors to occur or fail to be prevented. Here, in part 2 of “The End of an Error,” we will continue to explore the why behind these errors related to the human factors that support them. We will then switch gears — to discuss the specific operational factors that influence errors in feeding management.

Human Factors

Normalization of deviance, or what I like to call operational deviance, was first coined in 1996 by psychologist Diane Vaughn after the Challenger shuttle disaster.

“The gradual process through which unacceptable practice or standards become acceptable. As the deviant behavior is repeated without catastrophic results, it becomes the social norm for the organization.” (see Banja)

Commonly referred to as a “workaround”, the term describes the way we operationalize deviations or departures from established rules.

Inherently, workarounds are created when staff does not have the right tools or processes to do their jobs. A workaround is created because they feel they “have to” and this becomes the accepted practice. More often than not, where there are workarounds there are errors which defeats the entire purpose of implementing policies, practices, or technologies.

As nurses, it is in our DNA to root out flaws or deficiencies with any new process or technology. I feel confident in saying the collective “we” become frustrated when mandated to follow a process or use a particular technology if; it hinders our workflow without positively impacting patient safety, or unnecessarily takes precious time away from our patients and families.

Speaking generally, I believe it is natural for nurses to create workarounds once we have established workflow inefficiencies. In 2010 an article was published titled, “The Normalization of Deviance in Healthcare Delivery” (see Banja). This publication discussed the following classifications of errors:

Active Error, which refers to the failure of a healthcare professional to follow established procedures, practices, or standards.

Latent Errors, defined as laws of weaknesses in a system’s defenses that allow the former to break said defenses, reach patients, and cause harm.

While we can all quickly find examples of these weaknesses, below are some of the proposed reasons latent errors occur: (see Banja)

  • Rules are perceived as stupid or inefficient.
  • The work itself, along with new technology, can disrupt work behaviors and rule compliance.
  • The belief that I am breaking the rules for the good of my patient.

The latent error is where we truly uncover the systemic issue; when an event which started as an Active Error is deemed acceptable and not quashed, others will follow suit perpetuating the normalization and acceptance of the workaround. Additionally, the article discusses how the occurrence of these deviations becomes part of the operational norm. A weakened safety net allows for the acceptance of future active errors, continuing the cycle of deviance.

There is a delicate balance between operational deviance and simply trying to work in a safer, more efficient way. One could argue these points from either side. What is established as the appropriate way to do things may not, in reality, be the best practice.

Ultimately, you have to ask: if the processes and systems were better, leaving no need to circumvent the established procedure, would the deviance still occur?

Error Identification

As previously stated, the root causes of errors are multifactorial. So far, we have covered some of the contributing cognitive and
human factors. What other factors impact the occurrence of errors in these processes?

I am sure that many of you are familiar with the FMEA, but let’s review. You begin with mapping out the steps in a process, then work through failure modes or at what points could something go wrong. Failure causes — why would this happen and then the failure affects — what would happen if this event occurred. Each item is given a score or priority, and then teams decide to evaluate change based on their findings.

Why are we talking about this? There are a handful of studies that have documented the utilization of this methodology for infant feeding management. Each found that there are between 32-282 potential failure modes that require a human to detect and prevent while the authors believed that 55% are unlikely to be discovered (see Steele and Oza-Frank).

The reason to touch on this here is that the different institutions found a varying number of and specific potential failure points. It is not that one or the other is incorrect, it simply highlights that in every institution the workflows and resources are vastly different. These differences have a significant impact on people and their processes, creating the potential for errors.

We will not go in-depth to review each of those studies, (see Steele and Oza-Frank) here, we are focusing on the why. In these examples, thorough process evaluation and then targeted interventions occurred. Yet, even after the intervention, errors still happened. Was something missed? Probably not. Still, this shows us that with the vast complexity of the infant feeding management process simply correcting for process errors is not sufficient. The cognitive challenges and human factors need to be a part of the conversations, evaluations, and interventions.

Contributing Factors

Let’s review some day to day workflows and operational challenges experienced by NICU staff. This is where we will see some specific error points come to life. Below you’ll find examples of workflow challenges and some of the causes.

The bottle lifecycle is incredibly complex; Figure X depicts one permutation of the path a bottle can travel. See how many hands, state and/or location changes can occur?

Figure X

Bottle Pathway in Figure Y – a further development of the life cycle. In its simplest iteration, it is still a highly complex process. Though the set up may be different and various staff could own separate parts of this process, every institution experiences this. With this level of complexity, and considering everything that has been discussed, it would be foolish to not expect errors.

Figure Y
Figure Y

Storage and Transport – Practically every institution faces challenges regarding a lack of ample storage for maternal and donor milk. Some institutions are forced to limit the amount a family can store at the hospital.

Many freezers have bins with open lids (either because they are too full or the lids have long since disappeared). Patient bins may be poorly labeled or, worse, not labeled at all. Milk bottles are placed directly on the shelves or have fallen out of the container. How can a safe process be maintained under these conditions?

Bottles are then transferred from home to the hospital. Milk is received, stored, prepared, and delivered. Refer back to Figure X and Figure Y for the physical movement of a bottle.

Handwritten labels – In most situations, mothers hand write date and time pumped on the label. One study showed that 18% of errors occurred from handwritten labels, wrong information written, illegible handwriting, etc. Staff then have to receive those bottles into the hospital and manage the entire process from this point. If they start with incorrect information, how can the process be executed accurately and safely?

Manual system entry – If an institution is using any form of scanning system where data needs to be manually entered transcription of this information, incorrect or incomplete data, are all opportunities for error.

Patient identification challenges – Unique naming is complicated in the neonatal population, especially when you have multiples. The new Joint Commission requirement for infant naming convention has improved this issue, but challenges persist.


We have covered identified failure points for all facets of feeding management, the majority of which go undetected, unappreciated, unreported, and unresolved. Given this, the true breadth of the issues is not realized.

Part 3: Introduction

Part 1 of “The End of an Error” covered the cognitive failures and mental validations that impact the potential for error.

In Part 2, we focused on how human factors influence processes and contribute to errors as well as detailed some examples of failure points.

In Part 3 we will discuss the current state of electronic management of infant feeding and opportunities to improve safety in these processes.

Baby-Bottle Scan Insufficiency

Many institutions (or are being mandated to) employ EMRs that treat breast milk as a medication in an effort to check the safety box and move on. While we can all, philosophically and clinically, agree this to be true, it is extremely lacking when considered in terms of safety management.

Why are EMRS not stepping up? Well they are, however, most EMRs utilize their existing medication management system as a proxy for infant feeding safety scanning. Is this better than nothing? Absolutely, but building on the medication module does (or does not do) a number of things.

To utilize the scanning functionality often infant feeding orders have to be built as medication orders. Medication orders are typically fairly standard. These builds do not allow for the complex recipes and feeding regimens that we all use. With a standard tech platform, heavy customizations of functionality
need to occur and I think we can all agree that competing for institutional IT resources is a herculean task in the best-case scenario.

These systems simply provide the final confirmation of a baby-to-bottle match during feeding, which although a critical step, misses the multitude of other preparatory steps – each as important as the one before it. You do not have the ability to track issues along the prep path, evaluate practice and make improvements. If nothing else it is a false sense of security for families and staff.

  • Some of the processes that are missed
  • Patient verification
  • EBM/DHM state change rules
  • Recipe calculations and verification
  • Expiration verifications
  • Feeding order validation at administration

The cognitive failures and mental validations are back to haunt us.

As we have continued to explore, infant feeding management is extraordinarily complex and involves multiple processes and players. Isn’t it unusual to have a complex process without some mandatory and/or regulated process around it?

Bar Code Medication Administration (BCMA) is the most established process that we can compare infant feeding scanning to. Remember that BCMA is a CMS core measure with an incentive program.

Let’s take a quick look at the similarities and differences. To meet the requirement institutions have to be able to electronically document the “5 rights” of medication administration. I am not trying to downplay the criticality or the significance of each of these steps. But it is far fewer than what we need for feeding
administration and this list grows when you are talking about the integration of scanning technology and its capabilities or needs thereof. Also, medication administration is a standardized process. Follow your 5 rights for safe medication administration often now including a scanning step. Infant feeding management is so complex and every institution does it differently.

Leadership Challenges

What steps can we take to change the culture around feeding management safety? How do we as leaders help others understand these issues need attention?

Establishing and leading a culture of safety, more specifically safe communication. We need to realize that oversight and monitoring for rule compliance are never-ending and in some circumstances incredibly difficult to do, but that does not make them unnecessary. We need to be attentive to signals from our staff and never assume that policies are procedures are always being followed to the letter.

Validation of need — We each need to look at our institutional practices complete our own FMEA, process mapping, error rates evaluation etc. and publish. As we have discussed there is very little research on the true state of these issues or validate that there are any.

Consensus — We need to come together, data in hand, and truly talk about standards of care and establish the best practices.

Research — Do it and publish it. Get the conversations started and the eyes and ears open to your cause.


What impacts errors and our ability to detect and prevent them continues to be a complex conversation. Throughout this series, we have worked through some of the pieces to this puzzle. We have reviewed cognitive reasons errors occur—distractions, prospective memory, cognitive fatigue, inattentional blindness, and mental validations.

We then looked at the normalization of deviance, active and latent errors and the day to day workflows and operational challenges experienced by staff. Amongst all the distractions and mental validations, we are relying on humans to not only detect but prevent dozens of potential failure points, the majority of which go undetected, unappreciated, unreported, and unresolved.

So now we are back to our problem statement. To date, there exist no universally accepted, national guidelines or standards that regulate the safety management of preparation and administration of infant feedings in hospitals. There are huge organizational governing bodies for Medication Safety Management, Donor Milk treatment and storage policies, enteral nutrition safety, formula preparation and mixing standards, infection control, and the list goes on.

Why does no one “own this?” Who would own this? Who should own this? If there is no governing body doing research, making recommendations, lobbying for change how will anything happen? It is up to us as clinical and administrative leaders who are invested in improving infant feeding safety to lead the charge.